CENTER FOR DRUG EVALUATION AND RESEARCH APPLICATION NUMBER: 205677Orig1s000 STATISTICAL REVIEW(S) U.S. Department of Health and Human Services Food and Drug Administration Center for Drug Evaluation and Research Office of Translational Sciences Office of Biostatistics S TAT I S T I C A L R E V I E W AND E VA L U A T I O N C LINICAL S TUDIES NDA/BLA Serial Number: NDA 205-677 SN000 Drug Name: Tasimelteon Indication(s): Non-24 Hour Disorder in Totally Blind Individuals Applicant: Vanda Pharmaceuticals Inc. Date of Submission 5/31/2013 Review Priority: Priority Biometrics Division: Division of Biometrics I Statistical Reviewer: Jingyu (Julia) Luan, Ph.D. (HFD-710) Concurring Reviewers: Kun Jin, Ph.D. (HFD -710) H.M. James Hung, Ph.D. (HFD -710) Medical Division: Division of Neurology Product (HFD -120) Clinical Team: Project Manager: Devanand Jillapalli, M.D. (HFD-120) Ronald Farkas, M.D. (HFD-120) Ms. Cathleen Michaloski (HFD-120) Keywords: ANCOVA, Permutation ANCOVA Reference ID: 3409480 NDA 205-677, Tasimelteon Page 2 Table of Contents LIST OF TABLES 4 LIST OF FIGURES 4 1 6 EXECUTIVE SUMMARY 1.1 1.2 1.3 1.4 2 INTRODUCTION 2.1 2.2 3 SPONSOR’S EFFICACY ANALYSES SUMMARY OF IMPORTANT EVENTS RELATED TO STATISTICAL ANALYSIS IN STUDY 3201 SUMMARY OF STATISTICAL ISSUES REVIEWER’S EFFICACY ANALYSES OVERVIEW DATA SOURCES STATISTICAL EVALUATION 3.1 DATA AND ANALYSIS QUALITY 3.2 EVALUATION OF EFFICACY 3.2.1 Protocol VP-VEC-162-3201 (SET Study) 3.2.1.1 Study Objectives 3.2.1.2 Study Design 3.2.1.3 Efficacy Measures 3.2.1.4 Statistical Analysis Plan 3.2.1.5 Patient Disposition, Demographic and Baseline Characteristics Disposition of Patients Datasets Analyzed Demographics and Baseline Characteristics 3.2.1.6 Sponsor’s Primary Efficacy Results 3.2.1.7 Sponsor’s Secondary Efficacy Results 3.2.2 Protocol VP-VEC-162-3203 (RESET Study) 3.2.2.1 Study Objectives 3.2.2.2 Study Design 3.2.2.3 Efficacy Measures 3.2.2.4 Statistical Analysis Plan 3.2.2.5 Patient Disposition, Demographic and Baseline Characteristics Patients Disposition Datasets Analyzed Demographic and Other Baseline Characteristics 3.2.2.6 Sponsor’s Primary Efficacy Results 3.2.2.7 Sponsor’s Secondary Efficacy Results 3.2.3 3.3 4 Reviewer’s Analysis 6 7 7 9 10 10 11 11 11 11 11 11 13 15 16 17 17 18 19 20 22 23 23 23 24 24 25 25 26 27 30 31 32 3.2.3.1 Summary of Important Events Related to Statistical Analysis in Study 3201 3.2.3.2 Summary of Statistical Issues Study 3201 Study 3203 3.2.3.3 Graphic Presentation of Clinical Endpoints 3.2.3.4 The Twelve Patients Excluded from Efficacy Analysis by Sponsor for Study 3201 3.2.3.5 Results of ANCOVA Analysis for Clinical Endpoints on ITT Population 3.2.3.6 Permutation ANCOVA Test for Clinical Endpoints for ITT Population 32 33 33 34 34 45 47 49 EVALUATION OF SAFETY 51 FINDINGS IN SPECIAL/SUBGROUP POPULATIONS 4.1 GENDER, RACE, AGE AND GEOGRAPHIC REGION 4.1.1 Study 3201 (SET Study) 4.1.2 Study 3203 (RESET Study) Reference ID: 3409480 51 51 51 55 NDA 205-677, Tasimelteon Page 3 4.2 5 OTHER SPECIAL/SUBGROUP POPULATIONS SUMMARY AND CONCLUSIONS 5.1 STATISTICAL ISSUES AND COLLECTIVE EVIDENCE 5.1.1 Sponsor’s Efficacy Analyses 5.1.2 Summary of Important Events Related to Statistical Analysis in Study 3201 5.1.3 Summary of Statistical Issues 5.1.4 Reviewer’s Efficacy Analyses 5.2 CONCLUSIONS AND RECOMMENDATIONS Reference ID: 3409480 56 56 56 56 57 57 59 60 NDA 205-677, Tasimelteon Page 4 List of Tables Table 1: Non-24 Scale of Clinical Response (N24CRS) .............................................................. 15 Table 2: Summary of Analysis Datasets ....................................................................................... 18 Table 3: Demographics and Baseline Characteristics – All Randomized Patients ....................... 19 Table 4: Summary of Primary Efficacy Endpoints (ITT and Analysis Population) ..................... 21 Table 5: Summary of Secondary Efficacy Endpoints (ITT and Analysis Population) ................. 22 Table 6: Analysis Population Summary -- All Enrolled Patients ................................................. 27 Table 7: Demographic and Baseline Characteristics – All Enrolled Patients and ITT population28 Table 8: Summary of Primary Efficacy Endpoint (ITT Population) ............................................ 31 Table 9: Summary of Secondary Efficacy Endpoints (ITT Population)....................................... 32 Table 10: Summary of Important Events Related to Statistical Analysis in Study 3201 ............. 33 Table 11: The 12 Patients Excluded from Efficacy Analysis (Study 3201) ................................. 46 Table 12: Descriptive Statistics for the 12 Patients Excluded from Sponsor Analysis Population ............................................................................................................................................... 47 Table 13: LS Mean in Hours and P-value from ANCOVA with Site (Study 3201, ITT, n=84) .. 48 Table 14: LS Mean in Hours and P-value from ANCOVA without Site (Study 3201, ITT, n=84) ............................................................................................................................................... 48 Table 15: LS Mean in Hours and P-value from ANCOVA without Site (Study 3203, ITT, n=20) ............................................................................................................................................... 48 Table 16: Descriptive Statistics for Clinical Endpoints (Study 3201, ITT, n=84) ....................... 49 Table 17: Descriptive Statistics for Clinical Endpoints (Study 3203, ITT, n=20) ....................... 50 Table 18: Permutation ANCOVA with Site (Study 3201, ITT, n=84) ......................................... 50 Table 19: Permutation ANCOVA without Site (Study 3201, ITT, n=84) .................................... 51 Table 20: Results of Permutation ANCOVA without Site (Study 3203, ITT, n=20)................... 51 Table 21: Subgroup Analysis by Sex (Study 3201, ITT, n=84) ................................................... 52 Table 22: Subgroup Analysis by Race (Study 3201, ITT, n=84) ................................................. 53 Table 23: Subgroup Analysis for Age Group (Study 3201, ITT, n=84) ....................................... 54 Table 24: Subgroup Analysis for Country (Study 3201, ITT, n=84) ........................................... 54 List of Figures Figure 1: Study Design (US) Figure 2: Study Design (Germany) Figure 3: Flow Diagram of Patients Disposition (All Patients) Figure 4: Study Design Figure 5: Flow Diagram of Patient Disposition (All Patients) Figure 6: Histograms and CDF for LQ-nTST (Study 3201, FDA ITT, n=84) Figure 7: Histograms and CDF for UQ-dTSD (Study 3201, FDA ITT, n=84) Figure 8: Histograms and CDF for MoST (Study 3201, FDA ITT, n=84) Figure 9: Histograms and CDF for CGIC (Study 3201, FDA ITT, n=84) Figure 10: Histograms and CDF for nTST (Study 3201, FDA ITT, n=84) Figure 11: Histograms and CDF for dTSD (Study 3201, FDA ITT, n=84) Reference ID: 3409480 14 14 17 24 26 35 36 37 38 38 40 NDA 205-677, Tasimelteon Page 5 Figure 12: Histograms and CDF for LQ-nTST (Study 3203, n=20) Figure 13: Histograms and CDF for UQ-dTSD (Study 3203, n=20) Figure 14: Histograms and CDF for MoST (Study 3203, n=20) Figure 15: Histograms and CDF for nTST (Study 3203, n=20) Figure 16: Histograms and CDF for dTSD (Study 3203, n=20) Reference ID: 3409480 41 42 43 44 45 NDA 205-677, Tasimelteon Page 6 1 EXECUTIVE SUMMARY This original NDA submission includes two pivotal efficacy studies, Study 3201 (SET Study) and Study 3203 (RESET Study). For both studies, no agreement was reached between the Agency and the sponsor regarding the primary efficacy endpoint(s), analysis populations and analysis methods. The Agency has decided that the primary efficacy evaluation for tasimelteon should be based on the clinical endpoints. According to this reviewer’s post-hoc exploratory analyses without multiplicity adjustment, five out of the six clinical endpoints yield nominal pvalues less than 0.05 or approximately 0.05; thus, the two studies appear to suggest that tasimelteon 20 mg may be beneficial for Non-24 Hour Disorder in Totally Blind Individuals based on all the clinical endpoints except for night Total Sleep Time (nTST, the original primary endpoint). However, the results of these two studies should be interpreted with caution, since they are based on post-hoc exploratory analysis without multiplicity adjustment and thus it is unknown whether the overall type I error is properly controlled. Study 3201 was a multicenter, randomized, double-masked, placebo-controlled, parallel study designed to evaluate the efficacy and safety of 20 mg of tasimelteon versus placebo in patients suffering from Non-24. Eighty-four patients (tasimelteon 42; placebo 42) were randomized to receive tasimelteon (20 mg/day) or placebo. Of these 84 patients who were randomized to study drug, 62 (73.8%) patients completed the Randomization Phase and 22 (26.2%) patients discontinued early. This study was conducted at investigative sites in the US and Germany. The study began with a Pre-Randomization Phase (~5-6 weeks) and was followed by either a Randomization Phase (~26 weeks) or an Open-Label Extension Phase (~26 weeks). Study 3203 was a multicenter, randomized withdrawal, double-masked, placebo-controlled, parallel group study designed to evaluate the long-term maintenance effect and safety of 20 mg of tasimelteon versus placebo in patients with Non-24. Every patient who enrolled in Study 3203 had previously been screened in 3201. Patients who met the inclusion criteria and who had previously participated in, or were screened for Study 3201, were eligible to participate. The study had 2 phases: a Pre-Randomization Phase (consisting of an Open-label tasimelteon Run-in Phase [~6 weeks] and a τ Estimation Phase [~ 6 weeks]), and a Randomized Withdrawal Phase (~8 weeks). Twenty patients in the US were randomized into the study and they all completed the Randomized Withdrawal Phase. For both Study 3201 and Study 3203, no agreement was reached between the Agency and the sponsor regarding the primary efficacy endpoint(s), analysis populations and analysis methods. The Agency has decided that the primary efficacy evaluation for tasimelteon should be based on the clinical endpoints, instead of a melatonin-based biomarker proposed by the sponsor. The sections below present sponsor’s efficacy analyses, summary of important events related to statistical analysis in Study 3201, summary of statistical issues, and this reviewer’s efficacy analyses. 1.1 Sponsor’s Efficacy Analyses Study 3201: The primary efficacy endpoints were the following: Reference ID: 3409480 NDA 205-677, Tasimelteon Page 7 • • The entrainment of the circadian melatonin rhythm as measured by urinary aMT6s (Entrainment is a melatonin-based biomarker and is defined as having a post-baseline τ value less than 24.1 and a 95% CI that included 24.0. τ is circadian period.) (Step-down primary endpoint) The Clinical Response rate corresponding to individuals who had both entrainment of the aMT6s rhythm and a score of ≥3 on the Non-24 Clinical Response Scale (N24CRS). N24CRS is a 4-item scale that includes the Lower Quartilenighttime Total Sleep Time (LQ-nTST), Upper Quartile-daytime Total Sleep Duration (UQ-dTSD), Midpoint of Sleep Time (MoST), and Clinician Global Impression-Change (CGI-C) assessments. The proportion of Non-24 patients who were entrained after tasimelteon treatment during the Randomization Phase was statistically significantly greater than the proportion of Non-24 patients who were entrained after placebo treatment (% difference = 17.4; p = 0.0171). The proportion of patients who were entrained (aMT6s) and had a clinical response rate (N24CRS) ≥3 after tasimelteon treatment during the Randomization Phase was statistically significantly greater than the proportion of patients who were entrained and had a clinical response rate ≥3 after placebo treatment (% difference = 23.7; p = 0.0028). Study 3203 The primary efficacy endpoint was the proportion of non-entrainment of the circadian melatonin rhythm as measured by urinary aMT6s. The proportion of Non-24 patients who became nonentrained to a 24-hour day after randomization to tasimelteon was statistically significantly less than the proportion of Non-24 patients who became non-entrained after randomization to placebo treatment (% difference = -70.0; p = 0.0026). 1.2 Summary of Important Events Related to Statistical Analysis in Study 3201 In the original protocol, the primary endpoint proposed by the sponsor was nTST and the proposed sample size was 160 patients, based on the postulated mean treatment difference of 39 minutes and standard deviation of 66 minutes. In Amendment 6 submitted to the Agency, the sample size was changed from 160 to 100 patients, based on the new postulated mean treatment difference of 30 minutes and standard deviation of 45 minutes. In Amendment 9, the primary endpoint was changed to entrainment and the sample size was reduced to 84 patients. At the time of Amendment 9 (May 21, 2012), 95% of the patients were randomized and 56% of the patients completed the study. Amendment 11 was dated on December 11, 2012 and the trial data was unblinded on December 12, 2012. It is unclear to the Agency how much these changes might have impacted the trial results. 1.3 Summary of Statistical Issues Study 3201 There are three main statistical issues for this study: Primary Efficacy endpoint(s) Sponsor’s primary efficacy analyses were based on a melatonin-based biomarker endpoint “entrainment”. However, throughout the development program for tasimelteon, the Agency repeatedly disagrees that this biomarker is adequate to assess the efficacy for this indication and Reference ID: 3409480 NDA 205-677, Tasimelteon Page 8 requires clinically meaningful primary endpoint(s). No agreement was reached between the Agency and the sponsor regarding the primary endpoint(s). The Agency has decided that the primary efficacy evaluation for tasimelteon should be based on the following clinical endpoints: LQ-nTST, UQ-dTSD, MoST, CGIC, nTST and dTSD. Analysis Populations For Study 3201, 84 patients were randomized (n=84). The ITT Population defined by the sponsor included all patients randomized into the study that had τ calculated post-randomization (thereafter referred as Sponsor ITT, n=78). The Analysis Population defined by the sponsor included all patients in the Sponsor ITT population that had at least 70% of 1 circadian cycle of nTST data reported during each screening and post-randomization (thereafter referred as Sponsor Analysis Population, n=72). The sponsor’s ITT and Analysis Population were selected after the randomization and they are non-randomized subsets. Six patients in the Randomized Population without τ calculated post-randomization were excluded from the Sponsor ITT and six patients in the Sponsor ITT were excluded from the Sponsor Analysis Population due to less than 70% of 1 circadian cycle of nTST data reported during screening and/or post-randomization. Thus, 12 patients in total were excluded from the 84 patients randomized. In sponsor’s efficacy analyses, the Sponsor ITT population was utilized for all circadian rhythm related outcomes and the Analysis Population was used for analyses of all other endpoints including the step-down primary and all other efficacy analyses. However, the 84 randomized patients all took study medication and had at least one baseline and postbaseline assessment. Based on the intent-totreat principal, all the 84 patients should be included in the ITT population (n=84). The ITT population is a randomized population. The efficacy analyses conducted by this reviewer are based on ITT population. In Section 3.2.3.4, this reviewer will present information regarding the 12 patients excluded from the efficacy analyses by the sponsor and explain why these 12 patients should be included in the efficacy analyses. Analysis Methods In sponsor’s efficacy analyses for the clinical endpoints, ANCOVA model was used which includes baseline value as a covariate and treatment group and pooled sites as factors. However, due to small sample size, non-normal distribution of the data, and some heterogeneity in variances, this reviewer thinks permutation ANCOVA is more appropriate than ANCOVA for analyzing the clinical endpoints. In addition, in sponsor’s ANCOVA analysis for Study 3201, study site was included as a factor, but the Study Report shows that the randomization was not stratified by study site. Normally, if the randomization isn’t stratified by study site, in order to comply with the trial design, the site is not necessarily included in the analysis model. Therefore, this reviewer thinks permutation ANCOVA without site is more appropriate than ANCOVA with site. Study 3203 For this study, 20 randomized patients all received the study medication and completed the study, thus all 20 patients were included in the ITT population. The issues related to primary endpoint and analysis method are similar to Study 3201, but please note that for Study 3203, the randomization was not stratified by site and site was not a factor in sponsor’s ANCOVA analysis. Reference ID: 3409480 NDA 205-677, Tasimelteon Page 9 1.4 Reviewer’s Efficacy Analyses For both Study 3201 and Study 3203, no agreement was reached between the Agency and the sponsor regarding the primary efficacy endpoint(s), analysis populations and analysis methods. The analyses presented below were conducted by this reviewer and they are post-hoc exploratory analyses without multiplicity adjustment. In order to visually examine the distribution of the clinical endpoints, this reviewer generated histograms for each of the clinical endpoints in both studies. Please refer to Section 3.2.3.3 for details. For Study 3201, sponsor’s efficacy analyses for clinical endpoints were based on Analysis Population. The Analysis Population included all patients in the Sponsor ITT population that had at least 70% of 1 circadian cycle of nTST data reported during each screening and postrandomization. Based on this definition, 12 patients in the Randomized Population were excluded from the efficacy analysis for clinical endpoints. However, among these 12 patients, 10 patients have more than 50% of 1 circadian cycle of nTST at baseline and 6 patients have more than 50% of 1 circadian cycle of nTST at postbaseline. It seems that the data are not extremely sparse for these 12 patients and this reviewer thinks they should be included in the efficacy analysis. It appears that the inclusion of the 12 patients neither consistently strengthen nor consistently weaken the treatment effect for different clinical endpoints. Please refer to Section 3.2.3.4 for details. This reviewer conducted ANCOVA analysis on ITT for clinical endpoints. In sponsor’s ANCOVA analysis for Study 3201, baseline was included as a covariate if applicable and the pooled sites as a factor (variable name: SITEGR3). However, since the randomization wasn’t stratified by site, this reviewer thinks it isn’t necessary to include site as a factor in the analysis model. Based on the results of ANCOVA analysis without site, it seems that for both studies the nominal p-values were statistically significant or marginally significant for LQ-nTST, UQdTSD, MoST, CGIC and dTSD. The nominal p-value for nTST wasn’t statistically significant in either of the studies. Please refer to Section 3.2.3.5 for details. Because of small sample size, non-normal distribution of the clinical endpoints, and some heterogeneity in the variances of the clinical endpoints, this reviewer thinks permutation ANCOVA would be more appropriate than ANCOVA. It appears that the results of permutation ANCOVA without site is fairly close to those of ANCOVA without site. For LQ-nTST, UQdTSD, MoST, CGIC, and dTSD, the nominal p-values are statistically significant or marginally significant for both studies. The p-value for nTST isn’t statistically significant for either of the studies. Please refer to Section 3.2.3.6 for details. This reviewer performed subgroup analysis by sex, race, age group (<50 and >=50) and country (Germany and US) for Study 3201. It seems that the point estimates of treatment effects are all in the right direction for sex, age group and country for all clinical endpoints. For race, the point estimates of treatment effect for LQ-nTST, MoST and CGIC aren’t in the right direction for nonwhite patients. However, this doesn’t raise concerns since there are only 14 non-white patients in this study. The subgroup analysis for study 3203 wasn’t performed due to small sample size (n=20). Please refer to Section 4.1 for details. Reference ID: 3409480 NDA 205-677, Tasimelteon Page 10 2 2.1 INTRODUCTION Overview Non-24 is a severe, chronic, circadian rhythm disorder characterized by the inability to entrain (synchronize) the master body clock to the 24-hour day. Patients with Non-24 have prolonged periods of misalignment of circadian rhythms, including the timing of melatonin and cortisol secretion and the sleep-wake cycle, which are associated with significant impairments in social and occupational functioning, or marked subjective distress. The majority of reported cases of Non-24 occur in blind patients with no conscious perception of light. As a result of light information failing to reach the suprachiasmatic nuclei (SCN) to synchronize the clock and its outputs, the pacemaker may revert to its endogenous non-24-hour period. Consequently, the timing of physiology and behavior that is controlled by the circadian system (e.g. the timing of melatonin and cortisol production, the core body temperature rhythm, metabolic processes, the sleep-wake cycle, and alertness and performance patterns) becomes desynchronized from the 24-hour day, which has serious consequences on the daily functioning of the patient. The estimated prevalence of Non-24 in the totally blind is approximately 100,000 individuals in the United States. Studies suggest that 50-70% of totally blind individuals with no light perception have Non-24. Currently, there is no approved treatment for Non-24. The sponsor thinks tasimelteon is a circadian regulator that resets the master body clock in the suprachiasmatic nucleus (SCN). Tasimelteon is believed to be a Dual Melatonin Receptor Agonist (DMRA) with selective agonist activity at the MT1 and MT2 melatonin receptors. The development of tasimelteon formerly referred to as VEC-162 and BMS-214778 was initiated by Bristol Myers Squibb (IND Submission #1 December 17, 1997). Vanda Pharmaceuticals Inc. (Vanda) licensed the tasimelteon product in 2004 and was granted Orphan drug designation for the treatment of Non-24-Hour Disorder in blind individuals with no light perception on January 9, 2010. This submission includes two pivotal efficacy studies. Study 3201 was a multicenter, randomized, double-masked, placebo-controlled, parallel study designed to evaluate the efficacy and safety of 20 mg of tasimelteon versus placebo in patients suffering from Non-24. Eighty-four patients (tasimelteon 42; placebo 42) were randomized to receive tasimelteon (20 mg/day) or placebo. This study was conducted at investigative sites in the US and Germany. The study began with a Pre-Randomization Phase (~5-6 weeks) and was followed by either a Randomization Phase (~26 weeks) or an Open-Label Extension Phase (~26 weeks). Reference ID: 3409480 NDA 205-677, Tasimelteon Page 11 Study 3203 was a multicenter, randomized withdrawal, double-masked, placebo-controlled, parallel group study designed to evaluate the long-term maintenance effect and safety of 20 mg of tasimelteon versus placebo in patients with Non-24. Every patient who enrolled in Study 3203 had previously been screened in 3201. Patients who met the inclusion criteria and who had previously participated in, or were screened for Study 3201, were eligible to participate. The study had 2 phases: a Pre-Randomization Phase (consisting of an Open-label tasimelteon Run-in Phase [~6 weeks] and a τ Estimation Phase [~ 6 weeks]), and a Randomized Withdrawal Phase (~8 weeks). For Study 3201, the first patient was enrolled on August 25, 2010 and the last patient completed on October 29, 2012. For Study 3203, the first patient was enrolled on September 15, 2011 and the last patient completed on November 28, 2012. According to the sponsor, the date of data unblinding was December 12, 2012 for Study 3201 and January 16, 2013 for Study 3203. The Statistical Analysis Plan for Study 3201 and Study 3203 were singed-off on December 11, 2012 and December 28, 2012, respectively. 2.2 Data Sources The sponsor’s electronic submission was stored in the directory of \\Cdsesub1\evsprod\NDA205677 of the center’s electronic document room. 3 STATISTICAL EVALUATION 3.1 Data and Analysis Quality For both Study 3201 and Study 3203, no agreement was reached between the Agency and the sponsor regarding the primary efficacy endpoint(s), analysis populations and analysis methods. The Agency has decided that the primary efficacy evaluation for tasimelteon should be based on the clinical endpoints. The analyses conducted by this reviewer were produced from raw data. 3.2 Evaluation of Efficacy Reviewer’s Note: Section 3.2.1 and Section 3.2.2 present the study design, statistical analysis plan and efficacy results excerpted from sponsor’s protocols and Clinical Study Reports for Study 3201 and Study 3203, respectively. Please note that there is no agreement between the Agency and the sponsor regarding the primary efficacy endpoint(s), analysis populations and analysis methods. 3.2.1 PROTOCOL VP-VEC-162-3201 (SET STUDY) 3.2.1.1 Study Objectives The primary objective of this study was: Reference ID: 3409480 NDA 205-677, Tasimelteon Page 12 • • To determine the efficacy of tasimelteon in patients with Non-24 as measured by the proportion of entrainment. (a step-down objective): To determine the efficacy of tasimelteon in patients with Non-24 as measured by the proportion of patients with a clinical response. Clinical response was defined as the coincident demonstration of entrainment of the 6-sulfatoxymelatonin (aMT6s) rhythm and a score of ≥3 on the Non-24 Clinical Response Scale (N24CRS). The key secondary objectives were: • To determine the efficacy of tasimelteon in patients with Non-24 as measured by the proportion of responders with a combined sleep/wake response for nighttime sleep duration and daytime sleep. • To determine the efficacy of tasimelteon in patients with Non-24 as measured by the proportion of entrainment as assessed by urinary cortisol. Additional secondary objectives of this study were: • Subtype I: To determine the efficacy of tasimelteon in patients with Non-24, as measured by the proportion of subtype I responders, defined as an individual who was both entrained and had a significant improvement from screening in Lower Quartile – nighttime Total Sleep Time (LQ-nTST); • Subtype II: To determine the efficacy of tasimelteon in patients with Non-24, as measured by the proportion of subtype II responders, defined as an individual who was both entrained and had a significant improvement from screening in Upper Quartile – daytime Total Sleep Duration (UQ-dTSD); • Subtype III: To determine the efficacy of tasimelteon in patients with Non-24 as measured by the proportion of subtype III responders, defined as an individual who was both entrained and had a significant improvement from screening in Midpoint of Sleep Timing (MoST); • Subtype IV: To determine the efficacy of tasimelteon in patients with Non-24 as measured by the proportion of subtype IV responders, defined as an individual who was both entrained and had a significant improvement from screening in Clinical Global Impression-Change (CGI-C); • To determine the association between treatment response, as measured by entrainment of aMT6s circadian rhythms, and the baseline urinary aMT6s excretion rate in tasimelteontreated patients with Non-24; • To determine the efficacy of tasimelteon in improving subjective nighttime total sleep time (nTST) in patients with Non-24, as assessed by the change from screening in the average of LQ-nTST; • To determine the efficacy of tasimelteon in reducing subjective daytime total sleep duration (dTSD) in patients with Non-24, as assessed by the change from screening in the average of UQ-dTSD; • To determine the efficacy of tasimelteon in patients with Non-24 as measured by the change from the screening in MoST; • To determine the efficacy of tasimelteon to treat Non-24, as assessed by the CGI-C; • To determine the efficacy of tasimelteon in patients with Non-24 as measured by the proportion of entrainment as assessed by urinary analytes under circadian control. Reference ID: 3409480 NDA 205-677, Tasimelteon Page 13 3.2.1.2 Study Design This was a multicenter, randomized, double-masked, placebo-controlled, parallel study designed to evaluate the efficacy and safety of 20 mg of tasimelteon versus placebo in patients suffering from Non-24. Eight-four patients (tasimelteon 42; placebo 42) were randomized in a ratio of 1:1 to receive tasimelteon (20 mg/day) or placebo. This study was conducted at investigative sites in the US and Germany. The study began with a Pre-Randomization Phase (~5-6 weeks) and was followed by either a Randomization Phase (~26 weeks) or an Open-Label Extension Phase (~26 weeks). The Pre-Randomization Phase comprised a screening visit, a circadian period (τ) estimation segment, and a variable-length in-phase transition segment. After the Pre-Randomization Phase, patients who met all entry criteria for the study entered the Randomization Phase. Eligible patients at US sites could participate in the Randomization Phase or the Open-Label Extension Phase; eligible patients at sites in Germany could participate in the Randomization Phase followed by the Open-Label Extension Phase after completing the washout segment or could directly enter the Open-Label Extension Phase. Per Protocol Amendment 6 for US sites, the 2-week washout segment was removed to directly enroll patients into Study VPVEC-162-3203 (a randomized withdrawal study) or Study VP-VEC-162-3204 (a safety study), without interruption of treatment. The Randomization Phase comprised the double-masked evaluation segment and a 2-week washout segment (for German sites and US patients who completed the evaluation segment prior to implementation of Protocol Amendment 6). During the Randomization Phase, patients were asked to take either 20 mg tasimelteon or placebo approximately 1 hour prior to their target bedtime for 26 weeks in a double-masked fashion. Patients reported their nighttime sleep parameters for 2.5 circadian cycles (one circadian cycle = the number of calendar days for the non-24-hour circadian rhythms to complete a 24-hour cycle; e.g., 48 days for a τ = 24.5 hours) or 6 months, whichever was less. Patients also completed 48-hour urine collections for aMT6s assessment, cortisol, and other analyte assessments beginning on Days D14, D21, D28, D35, and D154. During the washout segment, patients were treated for 2 weeks with placebo in a singlemasked fashion. Patients who completed the Randomization Phase of the study were given the opportunity to participate in the optional Open-Label Extension Phase (for German sites). Additionally, patients who had a τ >24.0 and met all entry criteria but were ineligible for the Randomization Phase due to their τ were given the opportunity to participate in the Open-Label Extension Phase. During this phase, patients took open-label 20 mg tasimelteon for 26 weeks. The purpose of the Open-Label Extension Phase was to explore the long-term safety of tasimelteon in patients with Non-24 over 26 weeks. Figure 1 and Figure 2 provide schematics of the study design for sites in US and Germany. Reference ID: 3409480 NDA 205-677, Tasimelteon Page 14 Figure 1: Study Design (US) Source: Figure 1 of sponsor’s Clinical Study Report Figure 2: Study Design (Germany) Source: Figure 2 of sponsor’s Clinical Study Report Reference ID: 3409480 NDA 205-677, Tasimelteon Page 15 3.2.1.3 Efficacy Measures Efficacy measurements included assessments of aMT6s, cortisol, other circadian analytes, Post Sleep Questionnaire (PSQ), Pre-Sleep Questionnaire (PreSQ), and Clinical Global Impression of Change (CGI-C). The primary efficacy endpoints were the following: • The entrainment of the circadian melatonin rhythm as measured by urinary aMT6s (Entrainment is a melatonin-based biomarker and is defined as having a post-baseline τ value less than 24.1 and a 95% CI that included 24.0. τ is circadian period.) • (Step-down primary endpoint) The Clinical Response rate corresponding to individuals who had both entrainment of the aMT6s rhythm and a score of ≥3 on the Non-24 Clinical Response Scale (N24CRS). N24CRS is a 4-item scale that includes the LQ-nTST, UQ-dTSD, MoST, and CGI-C assessments (Table 1). Each assessment on the scale is scored as a 1 or 0 depending on whether the pre-specified threshold was achieved or not. The score for each assessment was summarized with a range of 0 to 4. Table 1: Non-24 Scale of Clinical Response (N24CRS) Source: Table 7 of sponsor’s Clinical Study Report The key secondary efficacy endpoints were the following: • The proportion of responders with a combined sleep/wake response for nighttime sleep duration and daytime sleep duration defined as increase of 90 minutes or greater in LQnTST and decrease of 90 minutes or greater in UQ-dTSD. • The entrainment of the circadian rhythm as measured by urinary cortisol. The additional secondary efficacy outcomes were the following: • The subtype I response rate (sleep time subtype) • The subtype II response rate (daytime sleep subtype) • The subtype III response rate (MoST subtype) • The subtype IV response rate (CGI-C subtype) • Treatment response association with aMT6s excretion rate • The average of LQ-nTST • The average of UQ-dTSD • The average of MoST • The CGI-C • The entrainment of circadian analytes. Reference ID: 3409480 NDA 205-677, Tasimelteon Page 16 3.2.1.4 Statistical Analysis Plan Below is a summary of sponsor’s Statistical Analysis Plan. The following analysis populations were defined for this study: • Intent-to-Treat (ITT) Population: The ITT Population included all patients randomized into the study that had τ calculated post-randomization. • Analysis Population: The Analysis Population included all patients in the ITT population that had at least 70% of 1 circadian cycle of nTST data reported during each screening and post-randomization. The number of non-missing days of nTST data during each of screening and post-randomization had to be at least equal to the number of days that made up 70% of 1 circadian cycle. • Safety Population: The Safety Population included all patients randomized into the study who received at least 1 dose of study drug. The ITT population was utilized for all circadian rhythm related outcomes: the primary endpoint of entrainment, the cortisol entrainment endpoint, and the aMT6s baseline secretion rate analyses. The Analysis Population was used for analysis of all other endpoints including the step-down primary and all other efficacy analyses. The primary efficacy endpoint was defined as the proportion of the number of Non-24 patients who were entrained after placebo or tasimelteon treatment during the Randomization Phase of Study 3201 only. Hence, the primary null hypothesis of the primary efficacy endpoint was that no difference exists in the proportion of people with entrained circadian rhythms between patients receiving tasimelteon and patients receiving placebo. Barnard’s Exact Test was used to test the null hypothesis in the ITT Population. The Fisher’s Exact Test was used as the sensitivity analysis method. If the primary null hypothesis was rejected at an alpha level of 0.05, then the step-down primary null hypothesis was to be tested at an alpha level of 0.05 to assess the efficacy of tasimelteon versus placebo as measured by the Clinical Response Rate. The Clinical Response Rate was summarized and analyzed in the same manner of the primary endpoint. Sensitivity analyses were conducted to verify the results of the step-down endpoint (Clinical Response). The first analysis was conducted for this step-down endpoint but with threshold scores of ≥2 for the N24CRS. The second and the third were conducted for individuals (regardless of entrainment status) with a N24CRS score of ≥3 and ≥2, respectively. The last one was conducted for responders who were entrained in the Randomization Phase of Study 3201 only that also had an N24CRS score of ≥3. This differs from the step-down primary endpoint which defined entrainment as occurring in the Randomization Phase of Study 3201 or the Run-in Phase of Study 3203 for individuals who were originally treated in the tasimelteon arm of Study 3201. The clinical measurements in the N24CRS for all analyses were only derived from the Randomization Phase of Study 3201. Reference ID: 3409480 NDA 205-677, Tasimelteon Page 17 For key secondary efficacy endpoints, the cortisol entrainment rate was summarized and analyzed in the same manner of the primary endpoint in the ITT Population. In addition, the sleep/wake response endpoint was summarized and analyzed in the same manner of the primary endpoint in the Analysis Population. 3.2.1.5 Patient Disposition, Demographic and Baseline Characteristics Disposition of Patients Figure 3 shows the patient disposition of all patients. Figure 3: Flow Diagram of Patients Disposition (All Patients) Source: Figure 3 of sponsor’s Clinical Study Report Reference ID: 3409480 NDA 205-677, Tasimelteon Page 18 Of the 391 patients who were screened for this study, 255 patients failed screening and were not enrolled in the Randomization Phase or the Open-Label Extension Phase, 84 patients were enrolled in the Randomization Phase, and 52 patients failed screening and were enrolled in the Open-Label Extension Phase. An equal number of patients were randomized into each treatment group of the Randomization Phase: 42 patients in the placebo group and 42 patients in the tasimelteon group. Of these 84 patients who were randomized to study drug, 62 (73.8%) patients completed the phase and 22 (26.2%) patients discontinued early. The reasons for discontinuation during the Randomization Phase were comparable between placebo and tasimelteon treatment groups. Datasets Analyzed Data sets analyzed for the Randomization Phase and the Open-Label Extension Phase are presented in Table 2. For the Randomization Phase, all 84 patients received at least one dose of study drug and were included in the Safety Population. The ITT Population included 78 randomized patients who had a τ value calculated post-randomization. The Analysis Population included 72 patients from the ITT Population who had at least 70% of one circadian cycle of nTST data reported during screening and post-randomization. For the Open-Label Extension Phase, 54 patients received at least one dose of study drug and were included in the Safety Population. Table 2: Summary of Analysis Datasets Source: Table 12 of sponsor’s Clinical Study Report Reference ID: 3409480 NDA 205-677, Tasimelteon Page 19 Demographics and Baseline Characteristics Table 3 summarizes demographics and baseline characteristics for all randomized patients. There were 49 males and 35 females in the study. The mean age was 50.7 years with a range of 23 to 74 years. The majority of patients were White (83.3%) and not Hispanic or Latino (96.4%). Overall, the mean LQ-nTST, UQ-dTSD, nTST, and dTSD at screening were 3.2 h, 2.4 h, 5.3 h, and 0.9 h, respectively. The mean τ values as measured by urinary aMT6s and cortisol were 24.47 h and 24.45 h, respectively. It seems that demographic and baseline characteristics were comparable between the groups. Table 3: Demographics and Baseline Characteristics – All Randomized Patients Reference ID: 3409480 NDA 205-677, Tasimelteon Page 20 Source: Table 13 of sponsor’s Clinical Study Report 3.2.1.6 Sponsor’s Primary Efficacy Results The ITT population was utilized for all circadian rhythm related outcomes: the primary endpoint of entrainment, the cortisol entrainment endpoint, and the aMT6s baseline secretion rate Reference ID: 3409480 NDA 205-677, Tasimelteon Page 21 analyses. The Analysis Population was used for analysis of all other endpoints including the step-down primary and all other efficacy analyses. Table 4 presents a summary of the results of the primary efficacy endpoints. Table 4: Summary of Primary Efficacy Endpoints (ITT and Analysis Population) Source: Table 14 of sponsor’s Clinical Study Report Entrainment was defined as having a post-baseline τ value less than 24.1 and a 95% CI that included 24.0. The proportion of Non-24 patients who were entrained after tasimelteon treatment during the Randomization Phase was statistically significantly greater than the proportion of Non-24 patients who were entrained after placebo treatment (% difference = 17.4; p = 0.0171). The Clinical Response Rate was defined as the coincident demonstration of entrainment of the aMT6s rhythm and a score of ≥3 on the N24CRS. The proportion of patients who were entrained (aMT6s) and had a clinical response rate (N24CRS) ≥3 after tasimelteon treatment during the Randomization Phase was statistically significantly greater than the proportion of patients who were entrained and had a clinical response rate ≥3 after placebo treatment (% difference = 23.7; p = 0.0028). Reference ID: 3409480 NDA 205-677, Tasimelteon Page 22 3.2.1.7 Sponsor’s Secondary Efficacy Results Table 5 presents a summary of the results of the secondary efficacy endpoints during the Randomization Phase. Please note that all p-values in this table are nominal p-values. Table 5: Summary of Secondary Efficacy Endpoints (ITT and Analysis Population) Source: Table 15 of sponsor’s Clinical Study Report Reference ID: 3409480 NDA 205-677, Tasimelteon Page 23 3.2.2 PROTOCOL VP-VEC-162-3203 (RESET STUDY) 3.2.2.1 Study Objectives The primary objective of this study was to demonstrate the maintenance of effect of tasimelteon to entrain circadian rhythms in patients with Non-24 as measured by urinary aMT6s. The secondary objectives of this study were the following: • To demonstrate the maintenance of effect of tasimelteon to treat Non-24 as measured by the time to relapse with relapse defined as a 45 minute or greater decrement in the weekly average subjective nighttime total sleep time (nTST) compared to the Run-in Phase; • To demonstrate the maintenance of effect of tasimelteon to entrain circadian rhythms in patients with Non-24 as assessed by urinary cortisol; • To demonstrate the maintenance of effect of tasimelteon in patients with Non-24 as measured by the proportion of patients with non-entrainment and an average of 30 minutes or greater decrement of nTST compared to the Run-in Phase; • To demonstrate the maintenance of effect of tasimelteon on subjective nTST in patients with Non-24, as assessed by the change from the Run-in Phase in the average nTST; • To demonstrate the maintenance of effect of tasimelteon on subjective nTST in patients with Non-24 as assessed by the change from the Run-in Phase in the lower quartile of days of nTST (LQ-nTST); • To demonstrate the maintenance of effect of tasimelteon on subjective daytime total sleep duration (dTSD) in patients with Non-24, as assessed by the change from the Run-in Phase in the average dTSD; • To demonstrate the maintenance of effect of tasimelteon on subjective dTSD in patients with Non-24, as assessed by the change from the Run-in Phase in the upper quartile of days of dTSD (UQ-dTSD); • To demonstrate the maintenance of effect of tasimelteon on the midpoint of sleep time (MoST); and • To assess symptoms of withdrawal after a minimum of 3 months of tasimelteon treatment assessed by the Benzodiazepine Withdrawal Symptom Questionnaire (BWSQ). 3.2.2.2 Study Design This was a multicenter, randomized withdrawal, double-masked, placebo-controlled, parallelgroup study designed to evaluate the long-term maintenance effect and safety of 20 mg of tasimelteon versus placebo in patients with Non-24. Patients who met the entrance criteria and who had previously participated in, or were screened for, Study VP-VEC-162-3201 were eligible to participate. The study had 2 phases: a Pre-Randomization Phase (consisting of an Open-label tasimelteon Run-in Phase [approximately 6 weeks] and a τ Estimation Phase [approximately 6 weeks]), and a Randomized Withdrawal Phase (8 weeks). Twenty patients (tasimelteon 10; placebo 10) were randomized in a ratio of 1:1 to receive tasimelteon (20 mg/day) or placebo during the Randomized Withdrawal Phase. This study was conducted at 18 investigative sites in the US. Figure 4 presents an overview of the study design. Reference ID: 3409480 NDA 205-677, Tasimelteon Page 24 Figure 4: Study Design Source: Figure 1 of sponsor’s Clinical Study Report 3.2.2.3 Efficacy Measures Efficacy measurements included assessments of urinary aMT6s, time to relapse, urinary cortisol, nighttime sleep, daytime naps, and timing of sleep. The primary efficacy endpoint was the following: • The proportion of non-entrainment of the circadian melatonin rhythm as measured by urinary aMT6s. The secondary efficacy endpoints were the following: • Time to relapse with relapse defined as a 45 minute or greater decrement in the weekly average of nTST; • The proportion of non-entrainment of the circadian rhythm as measured by cortisol; • Proportion of patients who were non-entrained and had an average of 30 minutes or greater decrement of subjective nTST compared to the Run-in Phase; • The change in average nTST; • The change in average LQ-nTST; • The change in average dTSD; • The change in average UQ-dTSD; and • The change in average MoST. 3.2.2.4 Statistical Analysis Plan The following analysis plan was excerpted from sponsor’s Statistical Analysis Plan. The following analysis populations were defined for this study: • Intent-to-Treat (ITT) Population: The ITT Population included all randomized patients who had τ calculated post-randomization. Reference ID: 3409480 NDA 205-677, Tasimelteon Page 25 • Safety Population for the Tasimelteon Run-in Phase: The Safety Population for the tasimelteon Run-in Phase included all patients who entered the tasimelteon Run-in Phase and received at least one dose of study drug. The primary efficacy endpoint was defined as the proportion of patients who become nonentrained to a 24-hour day after randomization to tasimelteon or placebo. Hence, the primary null hypothesis was that no difference existed in the proportion of patients with “non-entrained” circadian rhythms between patients receiving tasimelteon and patients receiving placebo. Barnard’s Exact Test was used to test the null hypothesis in the ITT population. The Fisher’s Exact Test was used as the sensitivity analysis method. Circadian cycle time to first relapse event was analyzed via Kaplan-Meier product-limit survival curve estimates and an un-stratified log-rank test for treatment group comparison. Patients who did not meet the criteria for a relapse event (relapse event being defined as a 45 minute decrement in their nTST) during the Randomized Withdrawal Phase were censored at the cycle time of the discontinuation or completion of the study. The cortisol non-entrainment and the patients who were non-entrained as measured by aMT6s and had an average of 30 minutes or greater decrement of subjective nTST compared to the Runin Phase, were summarized and analyzed in the same manner as the primary endpoint in the ITT Population. For the analysis of continuous efficacy variables, descriptive statistics were presented by treatment group for all patients in the ITT population. Treatment groups were compared using an analysis of covariance (ANCOVA) model with the terms of treatment group and the corresponding efficacy value in the tasimelteon Run-in Phase as a covariate. 3.2.2.5 Patient Disposition, Demographic and Baseline Characteristics Patients Disposition Figure 5 presents the disposition of all patients. The whole Pre-Randomization Phase including the tasimelteon Run-in Phase and the τ Estimation Phase was referred to as the “Run-in Phase.” Of the 58 patients who were screened for this study, 1 patient failed screening and was not enrolled in the Run-in Phase. A total of 57 patients received tasimelteon during the Run-in Phase; 37 of those patients failed the Run-in Phase and were not enrolled in the Randomized Withdrawal Phase. An equal number of patients were randomized into each treatment group of the Randomized Withdrawal Phase: 10 patients in the placebo group and 10 patients in the tasimelteon group. All randomized patients completed the Randomized Withdrawal Phase. Reference ID: 3409480 NDA 205-677, Tasimelteon Page 26 Figure 5: Flow Diagram of Patient Disposition (All Patients) Source: Figure 2 of sponsor’s Clinical study Report Datasets Analyzed Data sets analyzed for the Run-in Phase and the Randomized Withdrawal Phase are presented in Table 6. In the Run-in Phase, all 57 patients received at least one dose of study drug and were included in the Safety Population. In the Randomized Withdrawal Phase, 20 patients received at Reference ID: 3409480 NDA 205-677, Tasimelteon Page 27 least one dose of study drug and were included in the Safety Population. The ITT Population included 20 randomized patients who had a τ value calculated post-randomization. Table 6: Analysis Population Summary -- All Enrolled Patients Source: Table 11 of sponsor’s Clinical Study Report Demographic and Other Baseline Characteristics Table 7 summarizes demographics and baseline characteristics for all enrolled patients in the Run-in Phase and the ITT Population in the Randomized Withdrawal Phase. It seems that there were no major differences between the placebo and tasimelteon treatment groups in regard to demographic and baseline characteristics in the Randomized Withdrawal Phase. Reference ID: 3409480 NDA 205-677, Tasimelteon Page 28 Table 7: Demographic and Baseline Characteristics – All Enrolled Patients and ITT population Reference ID: 3409480 NDA 205?677, Tasimelteon Page 29 Pie-t Randomized ?i'ithdran'al Phase: Total Erind?mizedl Placebo Iasimelteou Total (3:57} 5:37} (9:10) (9:10) (9:20) 11 [1 1 Ethnicity Hispanic or 3 11:13.3} 1 {13.3) 2 {13.3} 2 Latino Ker Hispanic 3: 3? (133.3] 91:93.3} 9 {93.3) 13 (93.3) 55 (93.5) Latino ?'eight at screening (kgMean 31.13 {14.213} 33.?3 {13.323} 33.21 (13.332) 33.43 (14.333) 31.93 {14.2?3j 3333' mass index at screening [kg-ink} 3? 13 13 23 5? Mean 29.35 (4.325} 23.34 (3.559} 29.24 {3.252) 23.34 29.13 Average during run?in phase (hoursMean 4.259 5.344 {1.3333} 5.512 (3.9393) 5.423 (1.1412) (1.4313) Median 4.232 5.23? 5.353 5.32? 4.999 Average during run?in phase (hoursMean 1.533 {1.2?31] 3.994 {3.9325} 3.953 {1.1355} 3.9?5 (3.9324) 1.293 (1.1355) Median 1.153 3.391 3.339 3.391 3.993 Average of during run?in phase {hoursMean 5.399 {3.9224} 3.331 (1.323?) 3.312 (3.3335) 3.333 (3.9223) 3.255 (1.313?) Median 3.359 3.943 3.919 3.933 3.293 Average 31' during run-in phase {heurs} 33 13 13 23 53 Mean 3.33? {3.3345} 3.353 {3.2339} 3.343 (3.44?9} 3.343 (3.3339) 3.533 (3.5331) Median 3.355 3.313 3.233 3.293 3.323 Reference ID: 3409480 NDA 205-677, Tasimelteon Page 30 Source: Table 13 of sponsor’s Clinical Study Report 3.2.2.6 Sponsor’s Primary Efficacy Results The ITT population was utilized for all primary and secondary efficacy endpoints. Table 8 presents a summary of the results of the primary efficacy endpoint. The primary efficacy endpoint was the proportion of non-entrainment of the circadian melatonin rhythm as measured by urinary aMT6s. Non-entrainment was defined as having a post-baseline τ value ≥24.1 or the lower bound of the 95% CI >24.0. The proportion of Non-24 patients who became non-entrained to a 24-hour day after randomization to tasimelteon was statistically significantly less than the proportion of Non-24 patients who became non-entrained after randomization to placebo treatment (% difference = -70.0; p = 0.0026). Reference ID: 3409480 NDA 205-677, Tasimelteon Page 31 Table 8: Summary of Primary Efficacy Endpoint (ITT Population) Source: Table 14 of sponsor’s Clinical Study Report 3.2.2.7 Sponsor’s Secondary Efficacy Results Table 9 presents a summary of the results of the secondary efficacy endpoints. Tasimelteon was superior to placebo in the proportion of patients who were non-entrained (as measured by cortisol), the time (circadian and actual) to first relapse event, and sleep/wake parameters (nTST, LQ-nTST, dTSD, UQ-dTSD, and MoST). Please note that all p-values in this table are nominal p-values. APPEARS THIS WAY ON ORIGINAL Reference ID: 3409480 NDA 205-677, Tasimelteon Page 32 Table 9: Summary of Secondary Efficacy Endpoints (ITT Population) Source: Table 15 of sponsor’s Clinical Study Report 3.2.3 REVIEWER’S ANALYSIS For both Study 3201 and Study 3203, no agreement was reached between the Agency and the sponsor regarding the primary efficacy endpoint(s), analysis populations and analysis methods. The analyses presented in this section were conducted by this reviewer and they are post-hoc exploratory analyses without multiplicity adjustment. 3.2.3.1 Summary of Important Events Related to Statistical Analysis in Study 3201 Table 10 presents the summary of important events that occurred in Study 3201. In the original protocol, the primary endpoint proposed by the sponsor was nTST and the proposed sample size was 160 patients, based on the postulated mean treatment difference of 39 minutes and standard deviation of 66 minutes. In Amendment 6 submitted to the Agency, the sample size was changed Reference ID: 3409480 NDA 205-677, Tasimelteon Page 33 from 160 to 100 patients, based on the new postulated mean treatment difference of 30 minutes and standard deviation of 45 minutes. In Amendment 9, the primary endpoint was changed to entrainment and the sample size was reduced to 84 patients. At the time of Amendment 9 (May 21, 2012), 95% of the patients were randomized and 56% of the patients completed the study. Amendment 11 was dated on December 11, 2012 and the trial data was unblinded on December 12, 2012. It is unclear to the Agency how much these changes might have impacted the trial results. Table 10: Summary of Important Events Related to Statistical Analysis in Study 3201 Study 3201 Primary Sample Date Note Important Events Endpoint Size postulated mean Original 5/24/2010 nTST 160 difference =39 mins and std=66 mins First Patient Enrolled 8/25/2010 postulated mean Amendment 6 8/8/2011 nTST 100 difference =30 mins and std=45 mins 80/84 patients (95%) randomized; Amendment 9 5/21/2012 Entrainment 84 47/84 patients (56%) completed Last Patient Completed 10/29/2012 Amendment 11 Data Unblinding Source: Reviewer’s Analysis 12/11/2012 12/12/2012 Entrainment Entrainment 84 84 3.2.3.2 Summary of Statistical Issues Study 3201 There are three main statistical issues for this study: Primary Efficacy endpoint(s) Sponsor’s primary efficacy analyses were based on a melatonin-based biomarker endpoint “entrainment”. However, throughout the development program for tasimelteon, the Agency repeatedly disagrees that this biomarker is adequate to assess the efficacy for this indication and requires clinically meaningful primary endpoint(s). No agreement was reached between the Agency and the sponsor regarding the primary endpoint(s). The Agency has decided that the efficacy evaluation for tasimelteon should be based on the following clinical endpoints: LQnTST, UQ-dTSD, MoST, CGIC, nTST and dTSD. Analysis Populations Reference ID: 3409480 NDA 205-677, Tasimelteon Page 34 For Study 3201, 84 patients were randomized (n=84). The ITT Population defined by the sponsor included all patients randomized into the study that had τ calculated post-randomization (thereafter referred as Sponsor ITT, n=78). The Analysis Population defined by the sponsor included all patients in the Sponsor ITT population that had at least 70% of 1 circadian cycle of nTST data reported during each screening and post-randomization (thereafter referred as Sponsor Analysis Population, n=72). The sponsor’s ITT and Analysis Population were selected after the randomization and they are non-randomized subsets. Six patients in the Randomized Population without τ calculated post-randomization were excluded from the Sponsor ITT and six patients in the Sponsor ITT were excluded from the Sponsor Analysis Population due to less than 70% of 1 circadian cycle of nTST data reported during screening and/or post-randomization. Thus, 12 patients in total were excluded from the 84 patients randomized. Please refer to Table 2 for Analysis Population Summary. In sponsor’s efficacy analyses, the Sponsor ITT population was utilized for all circadian rhythm related outcomes and the Analysis Population was used for analyses of all other endpoints including the step-down primary and all other efficacy analyses. However, the 84 randomized patients all took study medication and had at least one baseline and postbaseline assessment. Based on the intent-to-treat principal, all the 84 patients should be included in the ITT population (n=84). This ITT population is a randomized population. The efficacy analyses conducted by this reviewer are based on ITT population. In Section 3.2.3.4, this reviewer will present information regarding the 12 patients excluded from the efficacy analyses by the sponsor and explain why these 12 patients should be included in the efficacy analyses. Analysis Methods In sponsor’s efficacy analyses for the clinical endpoints, ANCOVA model was used which includes baseline value as a covariate and treatment group and pooled sites as factors. However, due to small sample size, non-normal distribution of the data, and some heterogeneity in variances, this reviewer thinks permutation ANCOVA is more appropriate than ANCOVA. In addition, in sponsor’s ANCOVA analysis for Study 3201, study site was included as a factor, but the Study Report shows that the randomization was not stratified by study site. Normally, if the randomization isn’t stratified by study site, in order to comply with the trial design, the site is not necessarily included in the analysis model. Therefore, this reviewer thinks permutation ANCOVA without site is more appropriate than ANCOVA with site. Study 3203 For this study, 20 randomized patients all received the study medication and completed the study, thus all 20 patients were included in the ITT population. The issues related to primary endpoint and analysis method are similar to Study 3201, but please note that for Study 3203, the randomization was not stratified by site and site was not a factor in sponsor’s ANCOVA analysis. 3.2.3.3 Graphic Presentation of Clinical Endpoints In order to visually examine the distribution of the clinical endpoints, this reviewer generated histograms for each of the clinical endpoints in both studies. The following figures present the Cumulative Distribution Functions (CDFs) and the histograms for baseline, postbaseline and changes from baseline for each of the clinical endpoints in each study. Reference ID: 3409480 NDA 205-677, Tasimelteon Page 35 Figure 6: Histograms and CDF for LQ-nTST (Study 3201, FDA ITT, n=84) Source: Reviewer’s Analysis *: For LQ-nTST, larger values indicate longer nighttime total sleep time and positive change from baseline shows improvement in nighttime total sleep time. Reference ID: 3409480 NDA 205-677, Tasimelteon Page 36 Figure 7: Histograms and CDF for UQ-dTSD (Study 3201, FDA ITT, n=84) Source: Reviewer’s Analysis *: For UQ-dTSD, smaller values indicate shorter daytime total sleep duration and negative change from baseline shows improvement in daytime total sleep duration. Reference ID: 3409480 NDA 205-677, Tasimelteon Page 37 Figure 8: Histograms and CDF for MoST (Study 3201, FDA ITT, n=84) Source: Reviewer’s Analysis *: For MoST, an individual who has late afternoon naps and early morning awakenings would have a small MoST value or a negative number, and a positive change from baseline indicates improvement in MoST. Reference ID: 3409480 NDA 205-677, Tasimelteon Page 38 Figure 9: Histograms and CDF for CGIC (Study 3201, FDA ITT, n=84) Source: Reviewer’s Analysis *: CGI-C is a 7-point rating scale: 1 = very much improved; 2 = much improved; 3 = minimally improved; 4 = no change; 5 = minimally worse; 6 = much worse; or 7 = very much worse. APPEARS THIS WAY ON ORIGINAL Reference ID: 3409480 NDA 205-677, Tasimelteon Page 39 Figure 10: Histograms and CDF for nTST (Study 3201, FDA ITT, n=84) Source: Reviewer’s Analysis *: For nTST, larger values indicate longer nighttime total sleep time and positive change from baseline shows improvement in nighttime total sleep time. Reference ID: 3409480 NDA 205-677, Tasimelteon Page 40 Figure 11: Histograms and CDF for dTSD (Study 3201, FDA ITT, n=84) Source: Reviewer’s Analysis *: For dTSD, smaller values indicate shorter daytime total sleep duration and negative change from baseline shows improvement in daytime total sleep duration. Reference ID: 3409480 NDA 205-677, Tasimelteon Page 41 Figure 12: Histograms and CDF for LQ-nTST (Study 3203, n=20) Source: Reviewer’s Analysis *: For LQ-nTST, larger values indicate longer nighttime total sleep time and positive change from baseline shows improvement in nighttime total sleep time. Reference ID: 3409480 NDA 205-677, Tasimelteon Page 42 Figure 13: Histograms and CDF for UQ-dTSD (Study 3203, n=20) Source: Reviewer’s Analysis *: For UQ-dTSD, smaller values indicate shorter daytime total sleep duration and negative change from baseline shows improvement in daytime total sleep duration. Reference ID: 3409480 NDA 205-677, Tasimelteon Page 43 Figure 14: Histograms and CDF for MoST (Study 3203, n=20) Source: Reviewer’s Analysis *: For MoST, an individual who has late afternoon naps and early morning awakenings would have a small MoST value or a negative number, and a positive change from baseline indicates improvement in MoST. Reference ID: 3409480 NDA 205-677, Tasimelteon Page 44 Figure 15: Histograms and CDF for nTST (Study 3203, n=20) Source: Reviewer’s Analysis *: For nTST, larger values indicate longer nighttime total sleep time and positive change from baseline shows improvement in nighttime total sleep time. Reference ID: 3409480 NDA 205-677, Tasimelteon Page 45 Figure 16: Histograms and CDF for dTSD (Study 3203, n=20) Source: Reviewer’s Analysis *: For dTSD, smaller values indicate shorter daytime total sleep duration and negative change from baseline shows improvement in daytime total sleep duration. 3.2.3.4 The Twelve Patients Excluded from Efficacy Analysis by Sponsor for Study 3201 Sponsor’s efficacy analyses for clinical endpoints were based on Analysis Population. The Analysis Population included all patients in the Sponsor ITT population that had at least 70% of 1 circadian cycle of nTST data reported during each screening and post-randomization. Based on this definition, 12 patients in the Randomized Population were excluded from efficacy analysis for clinical endpoints. Table 11 presents the cycle length, % of one circadian cycle of nTST at baseline and postbaseline for the 12 patients. Among these 12 patients, 10 patients had more than 50% of 1 circadian cycle of nTST at baseline and 6 patients had more than 50% of 1 circadian cycle of nTST at postbaseline. It seems that the data are not extremely sparse for these 12 patients and this reviewer thinks they should be included in the efficacy analysis. Reference ID: 3409480 NDA 205-677, Tasimelteon Page 46 Table 11: The 12 Patients Excluded from Efficacy Analysis (Study 3201) Subject ID Treatment Group Cycle Length % Cycle Base % Cycle Postbaseline (b) (6) PLACEBO 32 275.0 53.1 PLACEBO 54 148.1 13.0 PLACEBO 54 94.4 11.1 PLACEBO 58 36.2 129.3 PLACEBO 59 113.6 27.1 PLACEBO 79 65.8 160.8 PLACEBO 81 60.5 145.7 PLACEBO 96 49.0 79.2 VEC-162 72 70.8 2.8 VEC-162 85 116.5 23.5 VEC-162 94 66.0 107.4 VEC-162 Source: Reviewer’s Analysis 97 84.5 36.1 The table below (Table 12) provides the descriptive statistics for the 12 patients excluded from the Sponsor Analysis Population versus the 72 patients in the Sponsor Analysis Population. It appears that the inclusion of the 12 patients neither consistently strengthen nor consistently weaken the treatment effect for different clinical endpoints. Reference ID: 3409480 NDA 205-677, Tasimelteon Page 47 Table 12: Descriptive Statistics for the 12 Patients Excluded from Sponsor Analysis Population Std Std Endpoint Treatment Flag* N Mean Median Min. Max. Dev Error PLACEBO 1. LQ-nTST VEC-162 PLACEBO 2. UQ-dTSD VEC-162 PLACEBO 3. MoST VEC-162 PLACEBO 4. CGIC VEC-162 PLACEBO 5. nTST VEC-162 PLACEBO 6. dTSD VEC-162 N 8 0.56 0.42 1.92 0.68 -1.71 3.06 Y 34 0.32 0.22 0.71 0.12 -0.99 2.11 N 4 -0.03 0.46 1.69 0.85 -2.44 1.42 Y 38 0.92 0.98 1.04 0.17 -1.29 3.96 N 8 -0.33 -0.53 1.20 0.43 -1.78 2.15 Y 34 -0.41 -0.31 0.60 0.10 -1.87 0.90 N 4 -1.54 -1.58 2.25 1.13 -3.58 0.60 Y 38 -0.69 -0.45 0.91 0.15 -3.03 0.68 N 8 0.13 -0.52 1.56 0.55 -1.06 3.44 Y 34 0.32 0.22 0.47 0.08 -0.36 1.94 N 4 0.37 0.45 0.33 0.16 -0.09 0.67 Y 38 0.51 0.32 0.65 0.11 -0.29 2.35 N 2 2.25 2.25 0.35 0.25 2.00 2.50 Y 33 3.41 3.50 1.03 0.18 1.00 6.00 Y 36 2.63 2.75 1.17 0.19 1.00 5.00 N 8 -0.00 -0.10 0.58 0.21 -0.78 0.89 Y 34 0.40 0.39 0.51 0.09 -0.34 1.40 N 4 -0.21 0.15 1.40 0.70 -2.16 1.02 Y 38 0.71 0.65 0.87 0.14 -1.91 2.61 N 8 -0.05 -0.19 0.55 0.20 -0.65 1.22 Y 34 -0.24 -0.11 0.39 0.07 -1.28 0.34 N 4 -0.65 -0.61 0.91 0.45 -1.54 0.16 Y 38 -0.31 -0.16 0.45 0.07 -2.20 0.18 Source: Reviewer’s Analysis *Flag: N=the patient isn’t in the Sponsor Analysis Population; Y=the patient is in the Sponsor Analysis Population. 3.2.3.5 Results of ANCOVA Analysis for Clinical Endpoints on ITT Population The three tables below (Table 13, Table 14 and Table 15) present the results of ANCOVA analysis on ITT population for clinical endpoints. In sponsor’s ANCOVA analysis for Study Reference ID: 3409480 NDA 205-677, Tasimelteon Page 48 3201, baseline was included as a covariate if applicable and the pooled sites as a factor (variable name: SITEGR3). However, since the randomization wasn’t stratified by site, this reviewer thinks it isn’t necessary to include site as a factor in the analysis model. Based on the results of ANCOVA analysis without site, it seems that for both studies the nominal p-values were statistically significant or marginally significant for LQ-nTST, UQ-dTSD, MoST, CGIC and dTSD. The nominal p-value for nTST wasn’t statistically significant in either of the studies. Table 13: LS Mean in Hours and P-value from ANCOVA with Site (Study 3201, ITT, n=84) Endpoints Placebo Tasimelteon P-value LQ-nTST 0.29 0.89 0.0232 UQ-dTSD -0.28 -0.84 0.0031 MoST 0.22 0.56 0.0229 1 CGIC 3.38 2.62 0.0104 nTST 0.32 0.65 0.0658 dTSD -0.14 -0.39 0.0026 Source: Reviewer’s Analysis 1 :For CGIC, n=71 since CGIC was missing for 13 patients. *: For LQ-nTST, MoST and nTST, larger values indicate better outcome; for UQ-dTSD, CGIC and dTSD, smaller values indicate better outcome. Table 14: LS Mean in Hours and P-value from ANCOVA without Site (Study 3201, ITT, n=84) Endpoints Placebo Tasimelteon P-value LQ-nTST 0.37 0.83 0.0510 UQ-dTSD -0.36 -0.81 0.0118 MoST 0.25 0.54 0.0366 CGIC1 3.34 2.63 0.0080 nTST 0.35 0.60 0.1149 -0.18 -0.36 0.0166 dTSD Source: Reviewer’s Analysis 1 :For CGIC, n=71 since CGIC was missing for 13 patients. *: For LQ-nTST, MoST and nTST, larger values indicate better outcome; for UQ-dTSD, CGIC and dTSD, smaller values indicate better outcome. Table 15: LS Mean in Hours and P-value from ANCOVA without Site (Study 3203, ITT, n=20) Endpoints Placebo Tasimelteon P-value LQ-nTST -1.23 -0.11 0.0233 UQ-dTSD 0.83 -0.16 0.0266 MoST 0.83 -0.16 0.0266 nTST -0.74 -0.20 0.1315 dTSD 0.30 -0.05 0.0547 Source: Reviewer’s Analysis *: For LQ-nTST, MoST and nTST, larger values indicate better outcome; for UQ-dTSD and dTSD, smaller values indicate better outcome. Reference ID: 3409480 NDA 205-677, Tasimelteon Page 49 3.2.3.6 Permutation ANCOVA Test for Clinical Endpoints for ITT Population The following two tables (Table 16 and Table 17) provide the descriptive statistics for clinical endpoints. It seems that there is some heterogeneity in the variances of the clinical endpoints. Furthermore, due to small sample size and non-normal distribution of the clinical endpoints, this reviewer thinks permutation ANCOVA would be more appropriate than ANCOVA. Table 16: Descriptive Statistics for Clinical Endpoints (Study 3201, ITT, n=84) Planned Std Endpoint n Mean Variance Min. Treatment Dev Max. LQnTST PLACEBO 42 0.37 1.02 1.04 -1.71 3.06 VEC-162 42 0.83 1.13 1.27 -2.44 3.96 UQdTSD PLACEBO 42 -0.39 0.73 0.54 -1.87 2.15 VEC-162 42 -0.77 1.08 1.18 -3.58 0.68 PLACEBO 42 0.28 0.77 0.60 -1.06 3.44 VEC-162 42 0.50 0.63 0.39 -0.29 2.35 PLACEBO 35 3.34 1.04 1.08 1.00 6.00 VEC-162 36 2.63 1.17 1.36 1.00 5.00 PLACEBO 42 0.33 0.54 0.30 -0.78 1.40 VEC-162 42 0.62 0.95 0.90 -2.16 2.61 PLACEBO 42 -0.20 0.43 0.18 -1.28 1.22 MoST CGIC nTST dTSD VEC-162 42 -0.34 0.50 0.25 -2.20 0.18 Source: Reviewer’s Analysis *: For LQ-nTST, MoST and nTST, larger values indicate better outcome; for UQ-dTSD, CGIC and dTSD, smaller values indicate better outcome. Reference ID: 3409480 NDA 205-677, Tasimelteon Page 50 Table 17: Descriptive Statistics for Clinical Endpoints (Study 3203, ITT, n=20) Std Endpoint Planned Treatment N Mean Variance Min. Dev LQ-nTST UQ-dTSD MoST nTST dTSD Max. PLACEBO 10 -1.21 1.18 1.40 -3.42 -0.06 VEC-162 10 -0.13 0.80 0.64 -1.66 0.91 PLACEBO 10 0.82 1.30 1.68 -0.18 4.38 VEC-162 10 -0.15 0.48 0.23 -1.19 0.34 PLACEBO 10 -0.10 0.61 0.38 -1.42 0.72 VEC-162 10 0.16 0.64 0.41 -1.00 1.29 PLACEBO 10 -0.74 0.74 0.55 -2.45 0.07 VEC-162 10 -0.20 0.75 0.56 -2.11 0.37 PLACEBO 10 0.29 0.56 0.31 -0.20 1.81 VEC-162 10 -0.05 0.23 0.05 -0.59 0.23 Source: Reviewer’s Analysis *: For LQ-nTST, MoST and nTST, larger values indicate better outcome; for UQ-dTSD, CGIC and dTSD, smaller values indicate better outcome. The results of permutation ANCOVA are shown in the following three tables (Table 18, Table 19 and Table 20). For each permutation test, the number of repetition is 100,000. The seeds were 3201/3203, 11, 14, 2013 and 100000 since the study number is 3201/3203 and Advisory Committee meeting was scheduled on Nov. 14, 2013. Table 18: Permutation ANCOVA with Site (Study 3201, ITT, n=84) Permutation ANCOVA ANCOVA Endpoint with Site Seed=3201 Seed=11 Seed=14 Seed=2013 Seed=100000 LQ-nTST 0.0232 0.0236 0.0244 0.0246 0.0242 0.0244 UQ-dTSD 0.0031 0.0032 0.0033 0.0028 0.0030 0.0029 MoST 0.0229 0.0218 0.0219 0.0207 0.0213 0.0219 CGIC 0.0104 0.0138 0.0146 0.0146 0.0138 0.0145 nTST 0.0658 0.0684 0.0663 0.0689 0.0677 0.0677 dTSD 0.0026 0.0024 0.0023 0.0020 0.0025 0.0023 Source: Reviewer’s Analysis Reference ID: 3409480 NDA 205-677, Tasimelteon Page 51 Table 19: Permutation ANCOVA without Site (Study 3201, ITT, n=84) Permutation ANCOVA ANCOVA Endpoint without Site Seed=3201 Seed=11 Seed=14 Seed=2013 Seed=100000 LQ-nTST 0.0510 0.0516 0.0519 0.0528 0.0518 0.0527 UQ-dTSD 0.0118 0.0112 0.0117 0.0108 0.0114 0.0112 MoST 0.0366 0.0361 0.0365 0.0363 0.0365 0.0371 0.0080 CGIC 0.0083 0.0083 0.0080 0.0077 0.0082 nTST 0.1149 0.1168 0.1164 0.1163 0.1169 0.1174 dTSD 0.0166 0.0154 0.0162 0.0146 0.0155 0.0157 Source: Reviewer’s Analysis Table 20: Results of Permutation ANCOVA without Site (Study 3203, ITT, n=20) Permutation ANCOVA ANCOVA Endpoint without Site Seed=3203 Seed=11 Seed=14 Seed=2013 Seed=100000 LQ-nTST 0.0233 0.0226 0.0227 0.0234 0.0237 0.0235 UQ-dTSD 0.0266 0.0069 0.0064 0.0074 0.0070 0.0067 MoST 0.0108 0.0063 0.0067 0.0075 0.0068 0.0064 nTST 0.1315 0.1532 0.1530 0.1546 0.1548 0.1523 dTSD 0.0547 0.0224 0.0229 0.0234 0.0224 0.0235 Source: Reviewer’s Analysis It appears that the results of permutation ANCOVA without site is fairly close to those of ANCOVA without site. For LQ-nTST, UQ-dTSD, MoST, CGIC, and dTSD, the nominal pvalues are statistically significant or marginally significant for both studies. The p-value for nTST isn’t statistically significant for either of the studies. 3.3 Evaluation of Safety Please read Dr. Jillapalli’s review for safety assessment. 4 FINDINGS IN SPECIAL/SUBGROUP POPULATIONS 4.1 4.1.1 Gender, Race, Age and Geographic Region STUDY 3201 (SET STUDY) The following four tables (Table 21, Table 22, Table 23, Table 24) present the results of subgroup analysis by sex, race, age group (<50 and >=50) and country (Germany and US). It seems that the point estimates of treatment effects are all in the right direction for sex, age group and country for all clinical endpoints. For race, the point estimates of treatment effect for LQ-nTST, MoST and Reference ID: 3409480 NDA 205-677, Tasimelteon Page 52 CGIC aren’t in the right direction for non-white patients. However, this doesn’t raise concerns since there are only 14 non-white patients in this study. Table 21: Subgroup Analysis by Sex (Study 3201, ITT, n=84) Endpoint Sex Treatment N Mean F 1. LQ-nTST M F 2. UQ-dTSD M F 3. MoST M F 4. CGIC M F 5. nTST M F 6. dTSD M Median Std Dev Std Error PLACEBO 17 0.22 0.04 0.70 0.17 VEC-162 18 0.76 0.70 1.19 0.28 PLACEBO 25 0.47 0.34 1.20 0.24 VEC-162 24 0.89 1.04 1.09 0.22 PLACEBO 17 -0.57 -0.49 0.52 0.13 VEC-162 18 -0.74 -0.37 0.99 0.23 PLACEBO 25 -0.27 -0.08 0.84 0.17 VEC-162 24 -0.79 -0.65 1.17 0.24 PLACEBO 17 0.25 0.22 0.68 0.16 VEC-162 18 0.41 0.28 0.63 0.15 PLACEBO 25 0.31 0.22 0.84 0.17 VEC-162 24 0.56 0.35 0.63 0.13 PLACEBO 15 3.63 4.00 0.88 0.23 VEC-162 15 2.60 3.00 0.99 0.25 PLACEBO 20 3.13 3.00 1.12 0.25 VEC-162 21 2.64 2.00 1.31 0.28 PLACEBO 17 0.33 0.35 0.56 0.14 VEC-162 18 0.45 0.47 0.94 0.22 PLACEBO 25 0.32 0.13 0.54 0.11 VEC-162 24 0.75 0.75 0.96 0.20 PLACEBO 17 -0.32 -0.19 0.44 0.11 VEC-162 18 -0.38 -0.16 0.57 0.13 PLACEBO 25 -0.12 -0.08 0.41 0.08 VEC-162 24 -0.31 -0.17 0.45 0.09 Source: Reviewer’s Analysis *: For LQ-nTST, MoST and nTST, larger values indicate better outcome; for UQ-dTSD, CGIC and dTSD, smaller values indicate better outcome. Reference ID: 3409480 NDA 205-677, Tasimelteon Page 53 Table 22: Subgroup Analysis by Race (Study 3201, ITT, n=84) Endpoint Race Treatment N Mean NON-WHITE 1. LQ-nTST WHITE NON-WHITE 2. UQ-dTSD WHITE NON-WHITE 3. MoST WHITE NON-WHITE 4. CGIC WHITE NON-WHITE 5. nTST WHITE NON-WHITE 6. dTSD WHITE Median Std Dev Std Error PLACEBO 8 1.52 1.77 1.19 0.42 VEC-162 6 0.57 0.55 1.37 0.56 PLACEBO 34 0.10 0.21 0.77 0.13 VEC-162 36 0.88 1.01 1.10 0.18 PLACEBO 8 -0.64 -0.25 0.79 0.28 VEC-162 6 -0.65 -0.40 0.95 0.39 PLACEBO 34 -0.33 -0.34 0.72 0.12 VEC-162 36 -0.79 -0.56 1.12 0.19 PLACEBO 8 0.75 0.50 1.23 0.44 VEC-162 6 0.53 0.52 0.35 0.14 PLACEBO 34 0.17 0.21 0.59 0.10 VEC-162 36 0.49 0.32 0.66 0.11 PLACEBO 6 2.92 3.00 1.16 0.47 VEC-162 6 3.17 3.75 1.57 0.64 PLACEBO 29 3.43 3.50 1.02 0.19 VEC-162 30 2.52 2.50 1.07 0.20 PLACEBO 8 0.58 0.54 0.44 0.16 VEC-162 6 0.73 0.70 1.04 0.42 PLACEBO 34 0.27 0.12 0.55 0.10 VEC-162 36 0.60 0.59 0.95 0.16 PLACEBO 8 -0.28 -0.17 0.34 0.12 VEC-162 6 -0.34 -0.15 0.39 0.16 PLACEBO 34 -0.18 -0.12 0.45 0.08 VEC-162 36 -0.34 -0.18 0.52 0.09 Source: Reviewer’s Analysis *: For LQ-nTST, MoST and nTST, larger values indicate better outcome; for UQ-dTSD, CGIC and dTSD, smaller values indicate better outcome. Reference ID: 3409480 NDA 205-677, Tasimelteon Page 54 Table 23: Subgroup Analysis for Age Group (Study 3201, ITT, n=84) Endpoint Age Group Treatment N Mean Median <50 1. LQ-nTST >=50 <50 2. UQ-dTSD >=50 <50 3. MoST >=50 <50 4. CGIC >=50 <50 5. nTST >=50 <50 6. dTSD >=50 Std Dev Std Error PLACEBO 18 0.30 0.14 1.17 0.28 VEC-162 17 1.19 1.01 1.12 0.27 PLACEBO 24 0.42 0.33 0.91 0.19 VEC-162 25 0.59 0.60 1.08 0.22 PLACEBO 18 -0.54 -0.51 0.77 0.18 VEC-162 17 -1.29 -1.10 1.13 0.28 PLACEBO 24 -0.28 -0.31 0.70 0.14 VEC-162 25 -0.42 -0.20 0.91 0.18 PLACEBO 18 0.38 0.21 1.02 0.24 VEC-162 17 0.80 0.67 0.79 0.19 PLACEBO 24 0.21 0.22 0.53 0.11 VEC-162 25 0.29 0.25 0.38 0.08 PLACEBO 15 3.40 3.50 0.87 0.22 VEC-162 14 2.43 1.75 1.38 0.37 PLACEBO 20 3.30 3.25 1.17 0.26 VEC-162 22 2.75 3.00 1.02 0.22 PLACEBO 18 0.24 0.27 0.53 0.12 VEC-162 17 0.84 0.76 0.85 0.21 PLACEBO 24 0.39 0.27 0.56 0.11 VEC-162 25 0.47 0.48 1.00 0.20 PLACEBO 18 -0.33 -0.19 0.47 0.11 VEC-162 17 -0.53 -0.34 0.59 0.14 PLACEBO 24 -0.10 -0.11 0.38 0.08 VEC-162 25 -0.21 -0.09 0.40 0.08 Source: Reviewer’s Analysis *: For LQ-nTST, MoST and nTST, larger values indicate better outcome; for UQ-dTSD, CGIC and dTSD, smaller values indicate better outcome. Reference ID: 3409480 NDA 205-677, Tasimelteon Page 55 Table 24: Subgroup Analysis for Country (Study 3201, ITT, n=84) Endpoint Country Treatment N Mean Median Germany 1. LQ-nTST US Germany 2. UQ-dTSD US Germany 3. MoST US Germany 4. CGIC US Germany 5. nTST US Germany 6. dTSD US Std Dev Std Error PLACEBO 4 -0.99 -1.14 0.74 0.37 VEC-162 4 1.27 1.44 0.70 0.35 PLACEBO 38 0.51 0.33 0.94 0.15 VEC-162 38 0.79 0.76 1.16 0.19 PLACEBO 4 0.20 -0.26 1.35 0.68 VEC-162 4 -0.53 -0.49 0.70 0.35 PLACEBO 38 -0.45 -0.34 0.64 0.10 VEC-162 38 -0.80 -0.45 1.12 0.18 PLACEBO 4 -0.18 -0.09 0.66 0.33 VEC-162 4 0.72 0.82 0.84 0.42 PLACEBO 38 0.33 0.22 0.78 0.13 VEC-162 38 0.48 0.33 0.61 0.10 PLACEBO 2 4.50 4.50 0.71 0.50 VEC-162 4 1.75 1.75 0.29 0.14 PLACEBO 33 3.27 3.50 1.02 0.18 VEC-162 32 2.73 3.00 1.19 0.21 PLACEBO 4 -0.12 -0.26 0.68 0.34 VEC-162 4 0.93 0.96 0.45 0.22 PLACEBO 38 0.37 0.39 0.52 0.08 VEC-162 38 0.59 0.55 0.99 0.16 PLACEBO 4 -0.09 -0.16 1.03 0.51 VEC-162 4 -0.25 -0.18 0.33 0.17 PLACEBO 38 -0.21 -0.12 0.34 0.06 VEC-162 38 -0.35 -0.16 0.52 0.08 Source: Reviewer’s Analysis *: For LQ-nTST, MoST and nTST, larger values indicate better outcome; for UQ-dTSD, CGIC and dTSD, smaller values indicate better outcome. 4.1.2 STUDY 3203 (RESET STUDY) There were 20 patients (10 patients per group) randomized in Study 3203. Due to small sample size, the results of the subgroup analysis for this study aren’t presented in this review. Reference ID: 3409480 NDA 205-677, Tasimelteon Page 56 4.2 Other Special/Subgroup Populations No other subgroups were analyzed. 5 SUMMARY AND CONCLUSIONS 5.1 Statistical Issues and Collective Evidence This original NDA submission includes two pivotal efficacy studies, Study 3201 (SET Study) and Study 3203 (RESET Study). Study 3201 was a multicenter, randomized, double-masked, placebo-controlled, parallel study designed to evaluate the efficacy and safety of 20 mg of tasimelteon versus placebo in patients suffering from Non-24. Eighty-four patients (tasimelteon 42; placebo 42) were randomized to receive tasimelteon (20 mg/day) or placebo. Of these 84 patients who were randomized to study drug, 62 (73.8%) patients completed the Randomization Phase and 22 (26.2%) patients discontinued early. This study was conducted at investigative sites in the US and Germany. The study began with a Pre-Randomization Phase (~5-6 weeks) and was followed by either a Randomization Phase (~26 weeks) or an Open-Label Extension Phase (~26 weeks). Study 3203 was a multicenter, randomized withdrawal, double-masked, placebo-controlled, parallel group study designed to evaluate the long-term maintenance effect and safety of 20 mg of tasimelteon versus placebo in patients with Non-24. Every patient who enrolled in Study 3203 had previously been screened in 3201. Patients who met the inclusion criteria and who had previously participated in, or were screened for Study 3201, were eligible to participate. The study had 2 phases: a Pre-Randomization Phase (consisting of an Open-label tasimelteon Run-in Phase [~6 weeks] and a τ Estimation Phase [~ 6 weeks]), and a Randomized Withdrawal Phase (~8 weeks). Twenty patients in the US were randomized into the study and they all completed the Randomized Withdrawal Phase. For both Study 3201 and Study 3203, no agreement was reached between the Agency and the sponsor regarding the primary efficacy endpoint(s), analysis populations and analysis methods. The Agency has decided that the primary efficacy evaluation for tasimelteon should be based on the clinical endpoints, instead of a melatonin-based biomarker proposed by the sponsor. The sections below present sponsor’s efficacy analyses, summary of important events related to statistical analysis in Study 3201, summary of statistical issues, and this reviewer’s efficacy analyses. 5.1.1 SPONSOR’S EFFICACY ANALYSES Study 3201: The primary efficacy endpoints were the following: • The entrainment of the circadian melatonin rhythm as measured by urinary aMT6s (Entrainment is a melatonin-based biomarker and is defined as having a post-baseline τ value less than 24.1 and a 95% CI that included 24.0. τ is circadian period.) Reference ID: 3409480 NDA 205-677, Tasimelteon Page 57 • (Step-down primary endpoint) The Clinical Response rate corresponding to individuals who had both entrainment of the aMT6s rhythm and a score of ≥3 on the Non-24 Clinical Response Scale (N24CRS). N24CRS is a 4-item scale that includes the Lower Quartilenighttime Total Sleep Time (LQ-nTST), Upper Quartile-daytime Total Sleep Duration (UQ-dTSD), Midpoint of Sleep Time (MoST), and Clinician Global Impression-Change (CGI-C) assessments. The proportion of Non-24 patients who were entrained after tasimelteon treatment during the Randomization Phase was statistically significantly greater than the proportion of Non-24 patients who were entrained after placebo treatment (% difference = 17.4; p = 0.0171). The proportion of patients who were entrained (aMT6s) and had a clinical response rate (N24CRS) ≥3 after tasimelteon treatment during the Randomization Phase was statistically significantly greater than the proportion of patients who were entrained and had a clinical response rate ≥3 after placebo treatment (% difference = 23.7; p = 0.0028). Study 3203 The primary efficacy endpoint was the proportion of non-entrainment of the circadian melatonin rhythm as measured by urinary aMT6s. The proportion of Non-24 patients who became nonentrained to a 24-hour day after randomization to tasimelteon was statistically significantly less than the proportion of Non-24 patients who became non-entrained after randomization to placebo treatment (% difference = -70.0; p = 0.0026). 5.1.2 SUMMARY OF IMPORTANT EVENTS RELATED TO STATISTICAL ANALYSIS IN STUDY 3201 In the original protocol, the primary endpoint proposed by the sponsor was nTST and the proposed sample size was 160 patients, based on the postulated mean treatment difference of 39 minutes and standard deviation of 66 minutes. In Amendment 6 submitted to the Agency, the sample size was changed from 160 to 100 patients, based on the new postulated mean treatment difference of 30 minutes and standard deviation of 45 minutes. In Amendment 9, the primary endpoint was changed to entrainment and the sample size was reduced to 84 patients. At the time of Amendment 9 (May 21, 2012), 95% of the patients were randomized and 56% of the patients completed the study. Amendment 11 was dated on December 11, 2012 and the trial data was unblinded on December 12, 2012. It is unclear to the Agency how much these changes might have impacted the trial results. 5.1.3 SUMMARY OF STATISTICAL ISSUES Study 3201 There are three main statistical issues for this study: Primary Efficacy endpoint(s) Sponsor’s primary efficacy analyses were based on a melatonin-based biomarker endpoint “entrainment”. However, throughout the development program for tasimelteon, the Agency repeatedly disagrees that this biomarker is adequate to assess the efficacy for this indication and requires clinically meaningful primary endpoint(s). No agreement was reached between the Agency and the sponsor regarding the primary endpoint(s). The Agency has decided that the Reference ID: 3409480 NDA 205-677, Tasimelteon Page 58 primary efficacy evaluation for tasimelteon should be based on the following clinical endpoints: LQ-nTST, UQ-dTSD, MoST, CGIC, nTST and dTSD. Analysis Populations For Study 3201, 84 patients were randomized (n=84). The ITT Population defined by the sponsor included all patients randomized into the study that had τ calculated post-randomization (thereafter referred as Sponsor ITT, n=78). The Analysis Population defined by the sponsor included all patients in the Sponsor ITT population that had at least 70% of 1 circadian cycle of nTST data reported during each screening and post-randomization (thereafter referred as Sponsor Analysis Population, n=72). The sponsor’s ITT and Analysis Population were selected after the randomization and they are non-randomized subsets. Six patients in the Randomized Population without τ calculated post-randomization were excluded from the Sponsor ITT and six patients in the Sponsor ITT were excluded from the Sponsor Analysis Population due to less than 70% of 1 circadian cycle of nTST data reported during screening and/or post-randomization. Thus, 12 patients in total were excluded from the 84 patients randomized. In sponsor’s efficacy analyses, the Sponsor ITT population was utilized for all circadian rhythm related outcomes and the Analysis Population was used for analyses of all other endpoints including the step-down primary and all other efficacy analyses. However, the 84 randomized patients all took study medication and had at least one baseline and postbaseline assessment. Based on the intent-totreat principal, all the 84 patients should be included in the ITT population (n=84). The ITT population is a randomized population. The efficacy analyses conducted by this reviewer are based on ITT population. In Section 3.2.3.4, this reviewer will present information regarding the 12 patients excluded from the efficacy analyses by the sponsor and explain why these 12 patients should be included in the efficacy analyses. Analysis Methods In sponsor’s efficacy analyses for the clinical endpoints, ANCOVA model was used which includes baseline value as a covariate and treatment group and pooled sites as factors. However, due to small sample size, non-normal distribution of the data, and some heterogeneity in variances, this reviewer thinks permutation ANCOVA is more appropriate than ANCOVA for analyzing the clinical endpoints. In addition, in sponsor’s ANCOVA analysis for Study 3201, study site was included as a factor, but the Study Report shows that the randomization was not stratified by study site. Normally, if the randomization isn’t stratified by study site, in order to comply with the trial design, the site is not necessarily included in the analysis model. Therefore, this reviewer thinks permutation ANCOVA without site is more appropriate than ANCOVA with site. Study 3203 For this study, 20 randomized patients all received the study medication and completed the study, thus all 20 patients were included in the ITT population. The issues related to primary endpoint and analysis method are similar to Study 3201, but please note that for Study 3203, the randomization was not stratified by site and site was not a factor in sponsor’s ANCOVA analysis. Reference ID: 3409480 NDA 205-677, Tasimelteon Page 59 5.1.4 REVIEWER’S EFFICACY ANALYSES For both Study 3201 and Study 3203, no agreement was reached between the Agency and the sponsor regarding the primary efficacy endpoint(s), analysis populations and analysis methods. The analyses presented below were conducted by this reviewer and they are post-hoc exploratory analyses without multiplicity adjustment. In order to visually examine the distribution of the clinical endpoints, this reviewer generated histograms for each of the clinical endpoints in both studies. Please refer to Section 3.2.3.3 for details. For Study 3201, sponsor’s efficacy analyses for clinical endpoints were based on Analysis Population. The Analysis Population included all patients in the Sponsor ITT population that had at least 70% of 1 circadian cycle of nTST data reported during each screening and postrandomization. Based on this definition, 12 patients in the Randomized Population were excluded from the efficacy analysis for clinical endpoints. However, among these 12 patients, 10 patients have more than 50% of 1 circadian cycle of nTST at baseline and 6 patients have more than 50% of 1 circadian cycle of nTST at postbaseline. It seems that the data are not extremely sparse for these 12 patients and this reviewer thinks they should be included in the efficacy analysis. It appears that the inclusion of the 12 patients neither consistently strengthen nor consistently weaken the treatment effect for different clinical endpoints. Please refer to Section 3.2.3.4for details. This reviewer conducted ANCOVA analysis on ITT for clinical endpoints. In sponsor’s ANCOVA analysis for Study 3201, baseline was included as a covariate if applicable and the pooled sites as a factor (variable name: SITEGR3). However, since the randomization wasn’t stratified by site, this reviewer thinks it isn’t necessary to include site as a factor in the analysis model. Based on the results of ANCOVA analysis without site, it seems that for both studies the nominal p-values were statistically significant or marginally significant for LQ-nTST, UQdTSD, MoST, CGIC and dTSD. The nominal p-value for nTST wasn’t statistically significant in either of the studies. Please refer to Section 3.2.3.5 for details. Because of small sample size, non-normal distribution of the clinical endpoints, and some heterogeneity in the variances of the clinical endpoints, this reviewer thinks permutation ANCOVA would be more appropriate than ANCOVA. It appears that the results of permutation ANCOVA without site is fairly close to those of ANCOVA without site. For LQ-nTST, UQdTSD, MoST, CGIC, and dTSD, the nominal p-values are statistically significant or marginally significant for both studies. The p-value for nTST isn’t statistically significant for either of the studies. Please refer to Section 3.2.3.6 for details. This reviewer performed subgroup analysis by sex, race, age group (<50 and >=50) and country (Germany and US) for Study 3201. It seems that the point estimates of treatment effects are all in the right direction for sex, age group and country for all clinical endpoints. For race, the point estimates of treatment effect for LQ-nTST, MoST and CGIC aren’t in the right direction for nonwhite patients. However, this doesn’t raise concerns since there are only 14 non-white patients in this study. The subgroup analysis for study 3203 wasn’t performed due to small sample size (n=20). Please refer to Section 4.1 for details. Reference ID: 3409480 NDA 205-677, Tasimelteon Page 60 5.2 Conclusions and Recommendations This original NDA submission includes two pivotal efficacy studies, Study 3201 (SET Study) and Study 3203 (RESET Study). For both studies, no agreement was reached between the Agency and the sponsor regarding the primary efficacy endpoint(s), analysis populations and analysis methods. The Agency has decided that the primary efficacy evaluation for tasimelteon should be based on the clinical endpoints. According to this reviewer’s post-hoc exploratory analyses without multiplicity adjustment, five out of the six clinical endpoints yield nominal pvalues less than 0.05 or approximately 0.05; thus, the two studies appear to suggest that tasimelteon 20 mg may be beneficial for Non-24 Hour Disorder in Totally Blind Individuals based on all the clinical endpoints except for night Total Sleep Time (nTST, the original primary endpoint). However, the results of these two studies should be interpreted with caution, since they are based on post-hoc exploratory analysis without multiplicity adjustment and thus it is unknown whether the overall type I error is properly controlled. APPEARS THIS WAY ON ORIGINAL Reference ID: 3409480 --------------------------------------------------------------------------------------------------------This is a representation of an electronic record that was signed electronically and this page is the manifestation of the electronic signature. --------------------------------------------------------------------------------------------------------/s/ ---------------------------------------------------JINGYU J LUAN 11/19/2013 KUN JIN 11/19/2013 I concur with the review, KOOROS MAHJOOB 11/21/2013 I concur with the review. Reference ID: 3409480 US. Department of Health and Human Services Food and Drug Administration Center for Drug Evaluation and Research Of?ce of Pharmacoepidemiology and Statistical Science Of?ce of Biostatistics STATISTICAL REVIEW AND EVALUATION NDA: Drug Name: Indication: Applicant: Date: Review Priority: Biometrics Division: Statistical Reviewer: Concurring Reviewer: Medical Division: Toxicologist: Project Manager: Keywords: Reference ID: 3401357 CARCINOGENICITY STUDY 205677 20mg capsules (VEC-162, a melatonin agonist) (Olphan Drug) Treatment of Non-24-Hom? Sleep-Wake Disorder in the Totally Blind Vanda Pharmaceuticals, Inc., Washington, DC CRO: ww Submitted 31 May 2013 Assigned to Reviewer: 26 August 2013 Priority Division 6 Steve Thomson Team Leader: Karl Lin, Ph. D. Nelu?ology Products Reviewer: Melissa Banks-Muckenfuss, Team Leader: Lois Freed, Cathleen Michaloski, BSN, MPH Carcinogenicity, Cox regression, Kaplan-Meier product limit, Suwival analysis, Trend test NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. Table of Contents 1. EXECUTIVE SUMMARY ....................................................................................................................................3 1.1. CONCLUSIONS AND RECOMMENDATIONS ............................................................................................................3 1.2. BRIEF OVERVIEW OF THE STUDIES ......................................................................................................................7 1.3. STATISTICAL ISSUES AND FINDINGS ....................................................................................................................8 1.3.1. Statistical Issues ..........................................................................................................................................8 1.3.2. Statistical Findings ....................................................................................................................................14 2. INTRODUCTION .................................................................................................................................................14 2.1. OVERVIEW .........................................................................................................................................................14 2.2. DATA SOURCES .................................................................................................................................................14 3. STATISTICAL EVALUATION ..........................................................................................................................14 3.1. EVALUATION OF EFFICACY ................................................................................................................................14 3.2. EVALUATION OF SAFETY ...................................................................................................................................14 3.2.1 STUDY VEC-162: CARCINOGENICITY STUDY BY ORAL GAVAGE ADMINISTRATION TO CD RATS FOR 104 WEEKS. ....................................................................................................................................................................14 3.2.2 STUDY VEC-162: CARCINOGENICITY STUDY BY ORAL GAVAGE ADMINISTRATION TO CD-1 MICE FOR 104 WEEKS. ....................................................................................................................................................................22 4. FINDINGS IN SPECIAL/SUBGROUP POPULATIONS................................................................................25 5. SUMMARY AND CONCLUSIONS...................................................................................................................26 5.1. STATISTICAL ISSUES AND COLLECTIVE EVIDENCE.............................................................................................26 5.2. CONCLUSIONS AND RECOMMENDATIONS ..........................................................................................................26 APPENDICES: ..........................................................................................................................................................27 APPENDIX 1. SURVIVAL ANALYSIS ...........................................................................................................................27 APPENDIX 2. FDA POLY-K TUMORIGENICITY ANALYSIS .........................................................................................31 APPENDIX 3. REFERENCES ........................................................................................................................................50 2 Reference ID: 3401357 NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. 1. EXECUTIVE SUMMARY According to the reports provided by Contract Research Organization, this submission was intended to assess the carcinogenic potential of daily administration of tasimelteon, labeled as VEC-162 in the studies, a melatonin agonist being developed for treatment of sleep disorders, when administered orally by gavage to both rats and CD-1 mice for 104 weeks. Both studies (b) (4) were conducted by The description of the studies is taken from (b) (4) the final reports. 1.1. Conclusions and Recommendations This submission summarizes the results from both a standard rat study and a mouse study. The Sponsor’s report summarizes the design of the rat study as follows: “The carcinogenic potential of VEC-162 (a melatonin agonist) was assessed over a period of 102 weeks in male and 104 weeks in female Crl:CD® (SD)IGS BR rats by oral gavage administration. Three groups, each comprising 65 males and 65 females, received VEC-162 by gavage at doses of 20, 100 or 250 mg/kg/day. A similarly constituted Control group received the vehicle (100% polyethylene glycol-400) at the same volume-dose.” (page 10 of rat report) Similarly, for the mouse study: “The carcinogenic potential of VEC-162 (a melatonin agonist) was assessed over a period of 104 weeks in Crl: CD-1™ (ICR) BR mice by oral administration. Three groups, each comprising 66 males and 66 females, received VEC-162 by gavage at doses of 30, 100 or 300 mg/kg/day. A similarly constituted Control group received the vehicle (100% polyethylene glycol-400) at the same volume-dose. A further 39 males and 39 females were allocated to each group and were used to provide blood samples for toxicokinetic evaluation.” (page 10 of mice report) As noted in Section 1.3.1.2, the Kaplan-Meier (product-limit) estimate of the survivor function is a valuable graphic to represent survival. Survival curves for each gender in each species are given in Appendix 1. Summary incidence of death tables are presented on pages 18 and 19 and 24 of this report. In male rats the Kaplan-Meier estimated survival curves of the four dose groups show overall lower survival in the low dose group while the medium dose group has generally the highest survival, with survival in the high dose and vehicle crossing, but mostly bounded between these survival curves. This explains the results in the statistical tests of differences in survival among male rats given below: Table 1. Statistical Significances of Tests of Homogeneity and Trend in Survival in Rats Males Females Hypotheses Logrank Wilcoxon Logrank Wilcoxon Homogeneity over all four groups 0.0223 0.0082 0.5557 0.3796 No Trend over all four groups 0.1225 0.2021 0.2271 0.2051 No difference between high dose and vehicle 0.4005 0.7710 0.1343 0.3311 3 Reference ID: 3401357 NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. Thus, the tests of homogeneity in survival over doses in male rats are statistically significant at the usual 0.05 level (Logrank p = 0.0233, Wilcoxon p = 0.0082). However, from the Kaplan-Meier curve, it seems difficult to attribute this result to any particular dose related effect. Rather results are consistent with the lack of evidence for any particular statistically significant dose related trend (Logrank p = 0.1225, Wilcoxon p = 0.2021) and with the lack of any consistent dominance of the survival curves for the high dose and vehicle (Logrank p = 0.4005, Wilcoxon p = 0.7710). In female rats, things are rather different. Whether analyzing trend among all four dose groups or the just the simple comparison of the High dose and vehicle, no tests were statistically significant (i.e., all 6 p ≥ 0.1343). While absence of proof is not proof of absence, the lack of evidence for such differences in survival in female rats is consistent with the hypotheses of no differences or trends. Table 2. Statistical Significances of Tests of Homogeneity and Trend in Survival in Mice Hypotheses Males Females Logrank Wilcoxon Logrank Wilcoxon Homogeneity over all four groups 0.6054 0.6502 0.1549 0.0948 No Trend over all four groups 0.7949 0.5793 0.9095 0.9852 No difference between high dose and vehicle 0.4564 0.3617 0.7822 0.6885 In male mice, from the survival curve shown in Figure A.1.3 in Appendix 1, the vehicle eventually slightly dominates the survival curves in the other groups, but this is not sufficient to result in any statistically significant test of homogeneity, dose related trend, or difference between the high dose and control (i.e., all 6 p ≥ 0.3617). From Figure A.1.4, in female mice it does seem that the vehicle, low, and high dose groups are eventually intertwined, but with higher survival than the medium dose group. As the Wilcoxon statistic is more sensitive to later differences, this is sufficient to result in a test of overall homogeneity that is arguable somewhat close to significance (Wilcoxon p = 0.0948), but less so when weighting earlier differences (Logrank p = 0.1549). However, there is no particular evidence of dose related trend or heterogeneity between the high dose and control (all 4 remaining p ≥ 0.6885). Note that a large number of tumors are typically identified in the analysis of neoplasms, implying a large number of statistical tests. The problem of adjusting for the multiplicity of statistical tests is discussed in Section 1.3.1.4, below. Following the frequentist paradigm, when interpreting significance levels (i.e., p-values), this reviewer would recommend using the Haseman-Lin-Rahman (HLR) rules to adjust for the multiplicity of tests. That is, when testing for trend over dose and the difference between the highest dose group with a control group, to control the overall Type I error rate for the joint tests in a two species submission to roughly 10%, one compares the unadjusted significance level of the trend test to 0.005 for common tumors and 0.025 for rare tumors, and the pairwise test between the high dose and control to 0.01 for common tumors and 0.05 for rare tumors. Using these adjustments for other tests, like testing the comparisons between the Low and Medium dose groups versus vehicle can be expected to increase the overall type I error rate to some value above the nominal rough 10% 4 Reference ID: 3401357 NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. level, possibly considerably higher than the nominal 10% rate. Note that if one tests trend and pairwise differences jointly, a slightly different set of bounds is used. Table 3 below, shows the tumors that had at least one non-multiplicity adjusted test of trend in rats that was statistically significant at or moderately close to a 0.05 level. Table 3. Potentially Statistically Significant Neoplasms in Rats Sex/ Incidence Significance Levels organ/ Veh Low Med High ptrend phigh pmed ____tumor_________________________________________________________vsVeh vsVeh Male Rats LIVER # Evaluated 65 65 65 65 Adj. # at Risk 43.7 36.1 49.9 43.2 HEPATOCELLULAR ADENOMA 2 2 6 8 .0123 .0444 .1802 Adj. # at Risk 43.7 36.1 50.7 43.2 Hepatocellular Adenoma/Carcinoma 2 2 8 8 .0145 .0444 .0746 PANCREAS # Evaluated 65 65 65 65 Adj. # at Risk 42.9 35.7 49.2 42.8 ISLET CELL ADENOMA 1 5 5 2 .6541 .5000 .1411 Adj. # at Risk 42.9 36.1 49.3 43.0 Islet Cell Adenoma/Carcinoma 2 6 7 2 .7843 .6921 .1212 PITUITARY # Evaluated 63 65 65 64 Adj. # at Risk 47.6 45.6 55.0 46.9 ADENOMA, PARS DISTALIS 19 28 30 23 .4822 .2361 .1104 SKIN # Evaluated 65 65 65 65 Adj. # at Risk 43.1 35.7 50.3 43.4 FIBROMA 1 0 5 3 .0920 .3080 .1404 Adj. # at Risk 43.1 35.7 50.3 43.9 Fibroma/Sarcoma NOS/Fibrosarcoma 1 0 5 4 .0407 .1800 .1404 TESTES # Evaluated 65 65 65 65 Adj. # at Risk 43.3 35.7 49.2 43.2 INTERSTITIAL (LEYDIG) CELL ADENOMA 3 1 4 6 .0589 .2417 .5735 Female Rats LIVER # Evaluated Adj. # at Risk HEPATOCELLULAR ADENOMA MAMMARY # Evaluated Adj. # at Risk MAMMARY ADENOCARCINOMA OVARIES # Evaluated Adj. # at Risk SERTOLIFORM TUBULAR ADENOMA Adj. # at Risk Thecal Cell Tmr/Serto.Tub.Adenoma UTERINE CERVIX # Evaluated Adj. # at Risk SQUAMOUS CELL CARCINOMA .6222 .6222 .0692 .0878 .0296 1 1 .9135 65 65 65 65 45.4 39.6 40.5 40.4 2 1 6 7 .0087 .0539 .0980 .8511 64 65 64 65 47.6 42.4 43.5 44.8 10 10 12 18 .0146 .0356 .3134 .4864 .0546 .2066 . . .0525 .2066 .4643 . .0568 .2126 . . 64 65 64 63 45.2 39.6 39.3 38.2 0 0 0 2 45.2 39.6 39.4 38.2 0 0 1 2 64 65 64 64 45.2 39.6 39.3 39.3 0 0 0 2 5 Reference ID: 3401357 plow vsVeh NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. Table 3. (cont.) Potentially Statistically Significant Neoplasms in Rats Sex/ Incidence Significance Levels organ/ Veh Low Med High ptrend phigh pmed ____tumor_________________________________________________________vsVeh vsVeh Female Rats (cont.) UTERUS # Evaluated 64 65 64 64 Adj. # at Risk 45.3 39.7 39.3 40.6 ENDOMETRIAL ADENOCARCINOMA 2 2 2 11 .0002 .0035 .6360 Adj. # at Risk 45.3 39.7 39.3 40.6 Endo. Adenoma/-Adenocarcinoma 3 2 2 11 .0005 .0102 .7722 Adj. # at Risk 45.5 39.6 39.5 39.9 SQUAMOUS CELL CARCINOMA 1 0 1 3 .0543 .2558 .7160 Uterus(w/uterine cervix) # Evaluated 64 65 64 64 Adj. # at Risk 45.3 39.7 39.3 40.6 ENDOMETRIAL ADENOCARCINOMA 2 2 2 11 .0002 .0035 .6360 Adj. # at Risk 45.3 39.7 39.3 40.6 Endometrial Adenoma/-Adenocarcinoma 3 2 2 11 .0005 .0102 .7722 Addj. # at Risk 45.5 39.6 39.5 40.8 SQUAMOUS CELL CARCINOMA 1 0 1 5 .0059 .0765 .7160 plow vsVeh .6360 .7722 1 .6360 .7722 1 All the tumors in male rats listed above would be classified as common (incidence > 1%). Using the Haseman-Lin-Rahman rules described in 1.3.1.5, in male rats no test of trend or pairwise difference between the high dose and control was significant either when testing trend or pairwise difference, although the joint tests of liver tumors would be equivocal. In female rats, in the table above, sertoliform tubular adenoma, plus pooled thecal cell tumor and sertoliform tubular adenoma, both of the ovaries, and squamous cell carcinoma of the uterine cervix would be classified as rare tumors (but not in the uterus or pooled uterus and cervix), the remainder as common. Tumor incidence based on combining the uterus with the uterine cervix only affects the results for squamous cell carcinoma. Going down the table of organ tumor combinations, the test of trend in hepatocellular adenoma of the liver in female rats would be close to the multiplicity adjusted level of statistical significance (p = 0.0087 ≈ 0.005). The corresponding tests comparing the high dose to vehicle might be classified as close to significant when testing jointly, but not seperately ( p = 0.0539 ≈ 0.05, but > 0.01). The tests of trend in endometrial adenocarcinoma and pooled adenoma and adenocarcinoma of the uterus (and thus the uterus with uterine cervix) would both be categorized as statistically significant (p = 0.0002, 0.0005 < 0.005, respectively), while the corresponding tests between the high dose and vehicle were statistically significant or close ( p = 0.0035 < 0.01 < 0.05 and 0.0102 ≈ 0.01 < 0.05). Strictly following the HLR rules and specifying that squamous cell carcinoma of the uterus with cervix is a common tumor, we would conclude that both the test of trend and the pairwise comparison are close to statistical significance (p = 0.0059 ≈ 0.005 amd 0.0765 ≈ 0.05). The test of trend would No other tests of carcinogenicity between dose groups achieved the multiplicity adjusted levels of significance. Table 4, below, shows similar results in mice. In mice the only test of trend that met the multiplicity adjusted limits for statistical significane was in male mice, namely in adenomas in the caecum or colon or duodenum (i.e the intestine) ( p = 0.0134 < 0.025), while the corresponding pairwise test was close to significance (p = 0.1025 ≈  for testing hypotheses jointly (please see Section 1.3.1.5). Note, however, that there is a bit of a conundrum in that 6 Reference ID: 3401357 NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. strictly following the vehicle incidence to determine whether or not a tumor can be classified as common or rare, the test of trend in the more complete intestine (adding the jejunum) and pooling adenoma and adenocarcinoma would not be statistically significant (p = 0.0114 > 0.005). Table 4. Potentially Statistically Significant Neoplasms in Mice Sex/ Incidence Significance Levels organ/ Veh Low Med High ptrend phigh pmed ____tumor_________________________________________________________vsVeh vsVeh Male Mice Caecum+Colon+Duodenum # Evaluated 66 66 66 66 Adj. # at Risk 50.1 45.8 46.5 45.1 Adenoma 0 0 0 3 .0134 .1025 . Caecum+Colon+Duodenum+Jejunum # Evaluated 66 66 66 66 Adj. # at Risk 50.6 45.8 46.5 45.4 Adenoma/Adenocarcinoma 2 0 0 5 .0114 .1764 1 DUODENUM # Evaluated 66 65 65 66 Adj. # at Risk 50.1 44.9 46.5 45.1 ADENOMA 0 0 0 2 .0582 .2217 . Female Mice UTERINE CERVIX # Evaluated Adj. # at Risk ENDOMETRIAL POLYP 66 64 65 66 48.6 44.3 39.8 48.1 0 0 1 2 .0652 .2474 .4483 plow vsVeh . 1 . . Complete results of statistical poly-k tests of tumor trend and differences between dose groups are given in Tables A.2.3 through A.2.6 in Appendix 2. 1.2. Brief Overview of the Studies Two studies were submitted: VEC-162: Carcinogenicity Study by Oral Gavage Administration to CD Rats for 104 Weeks, and, VEC-162: Carcinogenicity Study by Oral Gavage Administration to CD-1 Mice for 104 Weeks. These studies were designed to assess the carcinogenic potential of daily administration of VEC-162, a melatonin agonist being developed for treatment of sleep disorders, when administered orally by gavage to both rats and CD-1 mice for about 104 weeks. The dose groups in each study were labeled in this report as the Low, Medium, and High dose groups, respectively, plus the Vehicle control group. 7 Reference ID: 3401357 NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. 1.3. Statistical Issues and Findings 1.3.1. Statistical Issues In this section, several issues, typical of statistical analyses of these studies, are considered. These issues include comments on the multiple housing of animals, details of the survival analyses, tests on tumorigenicity, multiplicity of tests on neoplasms, and the validity of the designs. 1.3.1.1. Multiple Housing and Dosing of Animals: The Sponsor’s report indicates that unless reduced by mortality or isolation, study animals were housed 5 per cage in the main rat study and 3 per cage in the main mice study. Social interaction is important for the welfare of the animals. However, an argument could be made that housing animals together should be treated as part of the treatment of the experiment, and thus the appropriate unit of analysis would be the cage of the animals, not the individual animal. This would effectively reduce the study sample size and thus decrease power. Further, carcinogenicity tendencies could be communicated across animals in that competition for food, fighting, or other within cage effects could cause positive or negative correlations in response. When animals in a cage are dosed together or at the same time, variations in dosing across occasions might also induce positive or negative correlations. Thus, it is possible that within treatment estimated variances may be too large or too small, resulting in conservative or anti-conservative tests (in terms of Type I error). Unless it has been clearly shown that tumor incidence is independent of cage, from a purely statistical analysis point of view, this reviewer would generally recommend single housing and dosing of animals. 1.3.1.2. Survival Analysis: The survival analyses in rats presented here are based on both the log rank test and the Wilcoxon test comparing survival curves. The log rank tests tend to put higher weight on later events, while the Wilcoxon test tends to weight events more equally, and thus is more sensitive to earlier differences in survival. The logrank test is most powerful when the survival curves track each other, and thus the hazards, i.e., the conditional probability of the event in the next infinitesimal interval, would be roughly proportional. Note the logrank test seems to be the test usually recommended by statisticians, and is the test used by the Sponsor. Both tests are used in the FDA analysis of mortality. Appendix 1 reviews the specific FDA animal survival analyses in more detail. The results of the Sponsor’s analyses are summarized in Sections 3.2.1.1. and 3.2.2.1. Arguably the best pictorial representation of survival is a plot of the survival function, S(t), defined as the proportion of subjects surviving to at least time t (i.e. greater than or equal to time t). A natural estimate of this is defined for event times t1 < t2 < t3 < . . . < tN, with dj deaths in the nj animals at risk in time tj (animals that die or leave the study earlier are not at risk). Then for times j=1 to N, the Kaplan-Meier estimated survival curve is defined as the cumulative product at the nth point as S(tn) = ∏ (1 – tj/nj) for j=1,…,n ≤ N. 8 Reference ID: 3401357 NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. 1.3.1.3. Multiplicity of Tests on Survival: Using both the logrank and Wilcoxon tests, for each gender in each species there are 6 tests of survival differences. Assuming tests were performed at the usual 0.05 level, and the tests were stochastically independent, but there were actually absolutely no differences in survival across groups (so one would hope no tests would be statistically significant), the probability of at least one statistically significant result in each species by gender combination was about 0.265. These bounds assume that the tests are independent, which they clearly are not, but these values can give some idea of the possible price paid for the multiplicity of hypothesis tests in the statistical frequentist paradigm. 1.3.1.4. Tests on Neoplasms: The data sets requested for the analysis of rodent carcinogenicity studies are supposed to include a record for each animal organ combination that was not evaluated. If a number of the animals are not examined, but the proportions of animals showing the tumor under study in each treatment group is roughly the same as in the subset of animals actually reported the calculated p-values will generally be too large, i.e., results will be less statistically significant than they should be, possibly much less. If we can assume the process that determines whether or not a tumor is analyzed in each specific tumor is random, it is perhaps appropriate to consider such endpoints to be both analyzed AND have the tumor. Ignoring these possible problems, the Sponsor’s analyses of tumorigenicity are based on Peto et al (1966) methods, based on logrank for fatal and mortality independent tumors and a Mantel-Haenzel analysis of incidental tumors stratified by time of detection. Among other problems, this Peto style analysis requires accurate determination of whether a tumor is fatal or incidental. The FDA analysis is based on a modification of the Cochran-Armitage test of trend in mortality (please see Bailer & Portier, 1988, Bieler & Williams, 1993). Inspecting a large number of studies, Bailer and Portier noted that survival time seemed to fit a Weibull probability distribution, generally with a shape parameter of between 1 and 5, with 3 a typical value. With tmax denoting the maximal time to terminal sacrifice and tobs the time to detection of the tumor in the animal, they proposed weighting the animal by (tobs/tmax)k, so that an animal that survives for say 52 weeks in 104 week study without the tumor being analyzed is counted as (1/2)k of an animal in the risk set for that tumor. For k = 3, that means that particular animal would count as 1/8 of an animal. Further, the k = 3 specification seems to represent tumor incidence where some animals are perhaps more sensitive and respond earlier to the insult than the remining animals. Under this structure time to incidence would tend to follow a cubic expression. Thus an animal with the specific tumor being studied or who survives to terminal sacrifice without the tumor will be given a weight of 1 when counting the number of animals at risk. However, animals that die early without the tumor are down weighted when counting the number of animals in the risk set for that specific tumor. With differential mortality, this can mean a substantial reduction in the size of that risk set. Note this seems to be an appropriate adjustment 9 Reference ID: 3401357 NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. for dose groups that are terminated early. The report of the Society of Toxicological Pathology “town hall” meeting in June 2001 recommeded the use of this poly-k modification of the socalled Cochran-Armitage tests of trend over the corresponding Peto tests used by the Sponsor. The computed significance levels are based on small sample exact permutation tests of tumor incidence. In the tumor incidence tables the effective size of the risk set for each tumor is listed in the row labeled “Adjusted # at risk ”, and seems to be a more appropriate denominator when comparing incidence rates than the simple unadjusted number evaluated. 1.3.1.5. Multiplicity of Tests on Neoplasms: Testing dose related treatment differences for each species by gender by organ by tumor combination involves a large number of comparisons. Current FDA practice is based on the Haseman-Lin-Rahman multiplicity adjustments. The Haseman-Lin-Rahman rules are based on the original multiplicity adjustment of Haseman (1983) and extended by Lin and Rahman with various simulations. Based on his extensive experience with such analyses, for pairwise tests in a two species study comparing control to the High dose group, Haseman (1983) claimed that for a roughly 0.10 (10%) overall false positive error rate, rare tumors should be tested at a 0.05 (5%) level, and common tumors (with a historical control incidence greater than 1%) at a 0.01 level. Lin & Rahman (1998) proposed a further p-value adjustment for tests of trend. That is, for a roughly 0.10 (10%) overall false positive error rate in tests of trend, rare tumors should be tested at a 0.025 (2.5%) level and common tumors at a 0.005 (0.5%) level. Other specifications are presented in the Table 4 below. This approach is intended to balance both Type I error and Type II error (i.e., the error of concluding there is no evidence of a relation to tumorgenicity when there actually is such a relation). The proposed Haseman-Lin-Rahman bounds are taken from Guidance for Industry Statistical Aspects of the Design, Analysis, and Interpretation of Chronic Rodent Carcinogenicity Studies of Pharmaceuticals, (HHS, 2013). The bounds on the right in table 5, below, are grouped so that the last four columns correspond to testing either trend or pairwise comparison between the high dose and control seperately. The previous four columns ( columns 2-5), correspond to testing both overall trend and pairwise tests between the high dose and control together. Within each group there is a column giving the corresponding bounds for a two species study and another column for a one species study. In this analysis we follow the usual practice of testing parameters separately, so the bounds in the leftmost column are used. The observed tumor incidence in the vehicle group is used to decide if a tumor is classified rare or common. 10 Reference ID: 3401357 NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. Table 5. Recommended Multiplicity Adjusted Bounds on Significance Levels Common Tumor Rare Tumor Testing trend or pairwise (not both) Two Species One Species Trend Pairwise Trend Pairwise 0.005 0.01 0.01 0.025 0.025 0.05 0.05 0.10 Joint testing of trend and pairwise Two Species One Species Trend Pairwise Trend Pairwise 0.005 0.05 0.01 0.05 0.025 0.10 0.05 0.10 The significance levels of the pairwise tests between the vehicle control, and Low and Medium dose groups are also provided in the tumor analysis tables below. Following the HLR rules, adding these comparisons can be expected to increase the overall type I error rate to some level above the usual rough 10% level, possibly considerably larger. Again, because of the possibility of genetic drift and for convenience the vehicle group is used to determine if the tumor is classified as rare or common. 1.3.1.6. Validity of the Designs: When determining the validity of designs there are two key points: 1) adequate drug exposure 2) tumor challenge to the tested animals. 1) is related to whether or not sufficient animals survived long enough to be at risk of forming late-developing tumors and 2) is related to the Maximum Tolerated Dose (MTD), designed to achieve the greatest likelihood of tumorigenicity. Lin and Ali (2006), quoting work by Haseman, have suggested that in standard laboratory rodent species, a survival rate of about 25 animals, out of 50 or more animals (i.e., 50%), between weeks 80-90 of a two-year study may be considered a sufficient number of survivors as well as one measure of adequate exposure. From tables 14 on page 18, 15 on page 19, and 20 and 21 on page 24, as a percentage of the High dose group animals that survived to week 91, this criterion does seem to be not quite met in female rats (Males: 58.5% and Females: 44.6%), but is exceded in male rats and mice (Males: 60.6% and Females: 65.2%). The mean weight values used to derive differences and ratios in the following tables were taken directly from the Sponsor’s reports ( both labeled Table 4, Body Weight group mean values (g), pages 71-78 in rats and 53-60 in mice). The mean change from baseline is that reported by the Sponsor and is calculated from the weight changes of individual animals surviving the specified period. 11 Reference ID: 3401357 NDA 205677 Tasimelteon Vanda Pharmaceuticals. Inc. Table 6. Mean Weights and Changes (in g) in Male Rats Dose Dose Week Change change Group mg/kg/day 0 1021 from relative to Baseline vehicle 1. Vehicle Medium 100 167 710 545 95% 4. Higll 250 172 682 513 90% 1Control group reached 20 Slu'vivors at Week 102. so all groups were sacri?ced. Table 7. Mean Weights and Changes (in g) in Female Rats Dose Dose Week Change change Group mg/kg/day 0 104 from relative to Baseline vehicle 1. Vehicle 149 503 354 108% 3. Medium 100 144 453 309 94% 4. High 2501 144 445 297 90% IDosing stopped at Week 101. Table 8. Mean Weights and Changes (in g) in Male Mice Dose Dose Week Change change Group mg/kg/ day 0 104 from relative to Baseline vehicle 1. Vehicle 0 34.2 49.2 14.9 2. LOW 30 34.2 48.8 13.9 93% 3. Medium 100 33.6 46.7 12.7 85% 4. High 300 34.3 44.9 11.1 74% Table 9. Mean Weights and Changes (in g) in Female Mice Dose Dose Week Change change Group mg/kg/day 0 71 from relative to Baseline vehicle 1. Vehicle 0 25.7 39.1 13.7 2. LOW 30 25.7 40.0 14.6 107% 3. Medimn 100 25.9 40.6 15.2 111% 4. High 300 25.9 40.0 14.0 102% hu, euto, and Ward (1981), citing earlier work by Sontag et a1. (1976) recommend that the MTD ?is taken as ?the highest dose that causes no more than a 10% weight decrement as compared to the appropriate control groups, and does not produce mortality, clinical signs of toxicity, or pathologic lesions (other than those that may be related to a neoplastic response) that would be predicted to shorten the animal?s natlu?al life span? From Table 8 above. it seems that the weight criterion is exceded in the high dose male mice, as well as possibly in the mediiun 12 Reference ID: 3401357 NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. dose, but as indicated in Tables 6, 7, and 9, not in rats or female mice. Further, table 9 in female mice indicates thar all three treatment groups gained weight relative to the vehicle control. This may be evidence that the dose was under the MTD. The Sponsor reports than in the rat study food consumption was not affected by treatment, while in mice food consumption was low from Week 16 for males and females receiving the High dose, 300 mg/kg/day. Again from 2) above, excess mortality not associated with any tumor or sacrifice in the higher dose groups might suggest that the MTD was exceeded. This suggests that a useful way to assess whether or not the MTD was achieved is to measure early mortality not associated with any identified tumor. If this is high in the higher dose groups it suggests that animals tend to die before having time to develop tumors. Table 10, below, displays the number of animals in each dose group that died of a natural death or moribund sacrifice, but did not show any tumors (i.e., the “Event”): Table 10. Natural Death with No Identified Tumor in Rats (Male/Female) Males Event No event Females Event No event 1.Vehicle 14 51 3 62 2. Low 22 43 5 60 3. Medium 13 52 5 60 4. High 16 49 9 56 It is apparent that there is no dose related trend in dying without tumor in male rats. Further, there is no particular evidence of the more general hypothesis of heterogeneity in this event among the dose groups (chi-square p = 0.2615, Fisher exact p = 0.2741). Due the relative rarity of events, there is no overwhelming evidence of heterogeneity in female rats (chi-square p = 0.2869, Fisher exact p = 0.3304). However the more powerful test of dose related trend, i.e. the Cochran-Armitage test is statistically significant (p = 0.0324), although not overwhelmingly so. Table 11. Natural Death with No Identified Tumor in Mice (Male/Female) Males Event No event Females Event No event 1.Vehicle 16 50 8 58 2. Low 15 51 11 55 3. Medium 16 50 19 47 4. High 16 50 13 53 It is clear that no statistical test of treatment group differences is necessary in male mice. However, for completeness, the tests of heterogeneity over dose group in the event are highly not statistically significant (chi-square p = 0.9959, Fisher exact p = 1.0). There is more evidence of a dose related effect in female mice, although results are not statistically significant at the usual 0.05 level (chi-square p = 0.0981, Fisher exact p = 0.1068, Cochrane-Armitage p = 0.1940). Like the other observations above, this requires the expertise of the toxicologist, but these tests 13 Reference ID: 3401357 NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. may provide evidence that the MTD was not exceeded in male rats or mice, but is a bit more debatable in in female rats, and may not have even been achieved in female mice. 1.3.2. Statistical Findings Please see Section 1.1 above. 2. INTRODUCTION 2.1. Overview Results from studies in Crl:CD® (SD)IGS BR rats and Crl: CD-1™ (ICR) BR mice were submitted to assess the carcinogenic potential of tasimelteon. 2.2. Data Sources Two SAS data sets, both labeled tumor.sas7bdat, were translated from SAS transport files, both labeled tumor.xpt in each study. These followed the usual Biostatistics requested data format. The toxicologist provided a reasonably extensive guide for combining organs and especially associated tumors. 3. STATISTICAL EVALUATION 3.1. Evaluation of Efficacy NA 3.2. Evaluation of Safety More detailed results on the study are presented below. 3.2.1 VEC-162: Carcinogenicity Study by Oral Gavage Administration to CD Rats for 104 Weeks. STUDY DURATION: Male rats 102 weeks, Females 104 weeks. DOSING STARTING DATE: 10 April 2006. TERMINAL SACRIFICE: Males Week 102, Females Week 104. NECROPSY COMPLETED: Males 25-27 March 2008 Females: 7-9 April 2008 STUDY ENDING DATE (Final Report dated): 27 August 2009. RAT STRAIN: Crl:CD® (SD)IGS BR Rats. ROUTE: Daily Oral gavage 14 Reference ID: 3401357 NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. Animals were dosed once daily by oral gavage. Gross aspects of the study designs for the main study animals are summarized in Table 12 be1ow: Table 12. Design of Rat Study (dose volume 2.5 mL/kg) Treatment Group 1. Vehicle 2 2. Low 3. Medium 4. High # Main study animals (# TK1 animals)/gender 65 ( 10) 65 ( 10) 65 ( 10) 65 ( 10) Nominal Dose (mg/kg/day) Nominal Dosing Concentration (mg/mL) 0 20 100 250 0 8 40 100 1 Toxicokinetic phase animals began dosing during Week 1 of the carcinogenicity phase, were sampled in Weeks 6 and 26 and then terminated. In Week 96, samples were also taken from three male and three female Main Study animals with the lowest animal number. 2 100% polyethylene glycol-400 (PEG400). The Sponsor summarizes study conduct as follows: “The carcinogenic potential of VEC162 (a melatonin agonist) was assessed over a period of 102 weeks in male and 104 weeks in female Crl:CD® (SD)IGS BR rats by oral gavage administration. Three groups, each comprising 65 males and 65 females, received VEC-162 by gavage at doses of 20, 100 or 250 mg/kg/day. A similarly constituted Control group received the vehicle (100% polyethylene glycol-400) at the same volume-dose.” (page 10 of rat report) “For males, the number of survivors in Group 1 (Control) fell to 20 during Week 102. In accordance with the FDA sacrifice criteria, the entire sex was terminated in Week 103. The duration of treatment for males is therefore reported as 102 weeks. “For females, the number of survivors in Group 4 (250 mg/kg/day) fell to 20 in Week 101. In accordance to the FDA sacrifice criteria, dosing was suspended for this group/sex for the remainder of the study. Females in Groups 1, 2 and 3 completed the scheduled 104 weeks of treatment. The duration of treatment for females is therefore reported as 104 weeks.” (page 13 of rat report) After acclimation, animals were approximately five to six weeks old at first and their bodyweights were reported to be in the range of 136 to 211g for males and 109 to 174g for females. During the study animals were housed in groups of five animals of the same sex in the main study. Food and water were available ad libitum. The Sponsor states that detailed physical examinations were made on all animals twice weekly for the first month, then weekly for the next three months, and then biweekly and finally monthly. Body weights were recorded weekly for the first 16 weeks, beginning approximately one week before initiation of dosing, and every 4 weeks thereafter. Group mean weight changes were calculated from the weight changes of individual animals surviving the specified period. 15 Reference ID: 3401357 NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. Dosing was justified as follows: “The doses used in this study (0, 20, 100 and 250 mg/kg/day) were selected by the Sponsor with reference to previous work with this compound performed based on findings from oneand six-month toxicity studies. At the highest dose of 500 mg/kg/day in the six month study, one female was killed in moribund condition after three doses due to severe clinical signs, including seizures and laboured respiration. In addition, the incidence and severity of progressive nephropathy were increased at 500 mg/kg/day. In the onemonth toxicity study, doses of 100 or 400 mg/kg/day caused hyaline droplet nephropathy. After a one-month recovery period, the kidney lesions were still detectable in males. Based on these findings, 250 mg/kg/day was selected as the high dose for this carcinogenicity study. The low and mid doses (20 and 100 mg/kg/day) were selected based on appropriateness of the dosing intervals to establish a dose-response.” (page 12 of rat report) 3.2.1.1 Sponsor’s Results and Conclusions This section will present a summary of the Sponsor’s analysis on survivability and tumorigencity in rats. Survival analysis: The Sponsor summarized mortality results as follows: “A total of 162 males and 180 females died or were killed prematurely during the treatment period. In Week 101 the number of female survivors in the high dose group (250 mg/kg/day) fell to 20 and dosing was suspended for this group/sex until the end of the treatment period. Females of the other groups completed the scheduled 104-week treatment period. In Week 102 the number of male survivors in the control group fell to 20 and, consequently, the entire sex was terminated early. “The group distribution of the deaths (i.e. up to Week 102 in males and Week 104 in females) is presented below and there was no treatment-related effect on mortality.” (page 36 of report) Table 13. Sponsor Table: Group distributions of mortality Group/sex 1M 2M 3M 4M 1F 2F Dose (mg/kg/day) 0 20 100 250 0 20 Premature deaths 45 46 33 38 43 45 Percentage survival 31 29 49 42 34 31 3F 100 44 32 4F 250 48 26 “Statistical analysis revealed that the [logrank] trend test was not statistically significant when all groups were included in the analysis (p=0.126 and 0.232 for males and females, respectively). Only the pair-wise comparison of mortality in males receiving100 mg/kg/day was statistically significant (p=0.035).” (page 36 of rat report) Note that this comparison corresponds to a intermediate dose group and was not tested in the FDA analysis. Tumorigenicity analysis: The Sponsor used a standard Peto analysis as described in Section 1.3.1.4. Results are summarized below: 16 Reference ID: 3401357 NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. “Males “Liver … For benign hepatocellular adenoma, the trend test was statistically significant when all groups were included in the analysis (p=0.010). Upon exclusion of the 250 mg/kg/day treated group the trend test was no longer significant (p=0.082). The pairwise comparison of the control group with the 250 mg/kg/day treated group was statistically significant (p=0.045). “For benign hepatocellular adenoma and malignant hepatocellular combined, the trend test was statistically significant when all groups were included in the analysis (p=0.011). Upon exclusion of the 250 mg/kg/day treated group the trend test was still significant (p=0.018). The pairwise comparison of the control group with the 250 mg/kg/day treated group was statistically significant (p=0.045). “Pituitary (pars distalis) …. For benign adenoma, the trend test was not statistically significant when all groups were included in the analysis (p=0.753). The pairwise comparison of the control group with the 20 mg/kg/day treated group was statistically significant (p=0.017). “Females “Liver … For benign hepatocellular adenoma, the trend test was statistically significant when all groups were included in the analysis (p=0.005). Upon exclusion of the 240 mg/kg/day treated group the trend test was still significant (p=0.017). None of the pairwise tests were statistically significant. “Mammary areas … For malignant adenocarcinoma, the trend test was statistically significant when all groups were included in the analysis (p=0.008). Upon exclusion of the 240 mg/kg/day treated group the trend test was no longer significant (p=0.221). The pairwise comparison of the control group with the 240 mg/kg/day treated group was statistically significant (p=0.020). “Ovaries … For benign sertoliform tubular adenoma, the trend test was statistically significant when all groups were included in the analysis (p=0.039). None of the pairwise tests were statistically significant. “Uterus … For malignant endometrial adenocarcinoma, the trend test was statistically significant when all groups were included in the analysis (p<0.001). Upon exclusion of the 240 mg/kg/day treated group the trend test was no longer significant (p=0.537). The pairwise comparison of the control group with the 240 mg/kg/day treated group was statistically significant (p=0.003). 17 Reference ID: 3401357 NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. “For benign endometrial adenoma and malignant endometrial adenocarcinoma combined, the trend test was statistically significant when all groups were included in the analysis (p<0.001). Upon exclusion of the 240 mg/kg/day treated group the trend test was no longer significant (p=0.670). The pairwise comparison of the control group with the 240 mg/kg/day treated group was statistically significant (p=0.008). “For malignant squamous cell carcinoma, the trend test was statistically significant when all groups were included in the analysis (p=0.041). Upon exclusion of the 240 mg/kg/day treated group the trend test was no longer significant (p=0.499). None of the pairwise tests were statistically significant. “Uterine cervix … For malignant squamous cell carcinoma, the trend test was statistically significant when all groups were included in the analysis (p=0.046). None of the pairwise tests were statistically significant.” (pages 427-428 of rat report) 3.2.1.2 FDA Reviewer's Results This section will present the current Agency findings on survival and tumorigenicity in male and female rats. Survival analysis: Kaplan-Meier plots comparing treatment groups in both studies are given in Appendix 1, along with more details of the analysis. The following tables (Table 14 for male rats, Table 15 for female rats) summarize the mortality results for the dose groups. The data were grouped for the specified time period, and present the number of deaths during the time interval over the number at risk at the beginning of the interval. The percentage cited is the percent that survived at the end of the interval. Table 14. Summary of Period 1.Vehicle 0-52 3/651 95.4%2 53-78 15/62 72.3% 79-91 15/47 49.2% 92-104 12/32 30.8% Terminal 20 105 1 2 Male Rats Mortality (dose/kg/day) 2.Low 3.Medium 4.High 12/65 3/65 8/65 81.5% 95.4% 87.7% 16/53 7/62 12/57 56.9% 84.6% 69.2% 11/37 13/55 7/45 40.0% 64.6% 58.5% 7/26 10/42 11/38 29.2% 49.2% 41.5% 19 32 27 number deaths / number at risk per cent survival to end of period. 18 Reference ID: 3401357 NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. Table 15. Summary of Female Rats Mortality (dose/kg/day) Period 1.Vehicle 2.Low 3.Medium 4.High 1 7/65 9/65 7/65 0-52 2/65 89.2% 86.2% 89.2% 96.9%2 53-78 11/63 15/58 11/56 13/58 80.0% 66.2% 69.2% 69.2% 79-91 11/52 10/43 9/45 16/45 63.1% 50.8% 55.4% 44.6% 92-102 19/41 13/33 15/36 12/29 33.8% 30.8% 32.3% 26.2% Terminal 22 20 21 17 102-104 1 2 number deaths / number at risk per cent survival to end of period. As described in Appendix 1, the Kaplan-Meier (product-limit) estimate of the survivor function is probably the best figure to represent survival. Curves for each gender in each species are given in Appendix 1. Summary incidence of death tables are presented on pages 18 and 24 of this report. In male rats the Kaplan-Meier estimated survival curves of the four dose groups show overall lower survival in the low dose group while the medium dose group has generally the highest survival, with survival in the high dose and vehicle crossing, but mostly bounded between these survival curves. These descriptions explain the results in the tests of differences in survival given below: Table 16. Statistical Significances of Tests of Homogeneity and Trend in Survival in Rats Hypotheses Males Females Logrank Wilcoxon Logrank Wilcoxon Homogeneity over all four groups 0.0223 0.0082 0.5557 0.3796 No Trend over all four groups 0.1225 0.2021 0.2271 0.2051 No difference between high dose and vehicle 0.4005 0.7710 0.1343 0.3311 The tests of homogeneity in survival over doses in male rats are statistically significant at the usual 0.05 level (Logrank p = 0.0233, Wilcoxon p = 0.0082). From the Kaplan-Meier curve, it seems difficult to attribute this result to any particular dose related effect, but does correspond to treatment differences no clearly related to dose. This seems to be consistent with the lack of any statistically significant dose related trend (Logrank p = 0.1225, Wilcoxon p = 0.2021) or any significant difference between the the high dose and vehicle (Logrank p = 0.4005, Wilcoxon p = 0.7710). In female rats, when dealing with all four dose groups things are quite different. Whether analyzing among all four dose groups or the comparison of the High dose and vehicle were not statistically significance (i.e., all 6 p ≥ 0.1343). While absence of proof is not proof of absence, the lack of evidence for such differences in survival is consistent with the hypotheses of no differences or trends. 19 Reference ID: 3401357 NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. Tumorigenicity analysis: Table 17 below, a repeat of Table 3 above, and a slight reduction in Table A.2.1 below, shows those tumors that had at least one non-multiplicity adjusted test that was statistically significant at a about a 0.05 level. Table 17. Potentially Statistically Significant Neoplasms in Rats Sex/ Incidence Significance Level organ/ Veh Low Med High ptrend phigh pmed ____tumor_________________________________________________________vsVeh vsVeh Male Rats LIVER # Evaluated 65 65 65 65 Adj. # at Risk 43.7 36.1 49.9 43.2 HEPATOCELLULAR ADENOMA 2 2 6 8 .0123 .0444 .1802 Adj. # at Risk 43.7 36.1 50.7 43.2 Hepatocellular Adenoma/Carcinoma 2 2 8 8 .0145 .0444 .0746 PANCREAS # Evaluated 65 65 65 65 Adj. # at Risk 42.9 35.7 49.2 42.8 ISLET CELL ADENOMA 1 5 5 2 .6541 .5000 .1411 PITUITARY # Evaluated 63 65 65 64 Adj. # at Risk 47.6 45.6 55.0 46.9 ADENOMA, PARS DISTALIS 19 28 30 23 .4822 .2361 .1104 SKIN # Evaluated 65 65 65 65 Adj. # at Risk 43.1 35.7 50.3 43.9 Fibroma/Sarcoma NOS/Fibrosarcoma 1 0 5 4 .0407 .1800 .1404 Female Rats LIVER # Evaluated Adj. # at Risk HEPATOCELLULAR ADENOMA MAMMARY # Evaluated Adj. # at Risk MAMMARY ADENOCARCINOMA OVARIES # Evaluated Adj. # at Risk SERTOLIFORM TUBULAR ADENOMA Adj. # at Risk Thecal Cell Tmr/Serto.Tub.Adenoma UTERINE CERVIX # Evaluated Adj. # at Risk SQUAMOUS CELL CARCINOMA UTERUS # Evaluated Adj. # at Risk ENDOMETRIAL ADENOCARCINOMA Adj. # at Risk Endo. Adenoma/-Adenocarcinoma .6222 .6222 .0692 .0296 1 65 65 65 65 45.4 39.6 40.5 40.4 2 1 6 7 .0087 .0539 .0980 .8511 64 65 64 65 47.6 42.4 43.5 44.8 10 10 12 18 .0146 .0356 .3134 .4864 .0546 .2066 . . .0525 .2066 .4643 . .0568 .2126 . . .0002 .0035 .6360 .6360 .0005 .0102 .7722 .7722 64 65 64 63 45.2 39.6 39.3 38.2 0 0 0 2 45.2 39.6 39.4 38.2 0 0 1 2 64 65 64 64 45.2 39.6 39.3 39.3 0 0 0 2 64 65 64 45.3 39.7 39.3 2 2 2 45.3 39.7 39.3 3 2 2 20 Reference ID: 3401357 plow vsVeh 64 40.6 11 40.6 11 NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. Table 17. (cont.) Potentially Statistically Significant Neoplasms in Rats Sex/ Incidence Significance Level organ/ Veh Low Med High ptrend phigh pmed ____tumor_________________________________________________________vsVeh vsVeh Female Rats (cont.) Uterus(w/uterine cervix) # Evaluated 64 65 64 64 Adj. # at Risk 45.3 39.7 39.3 40.6 ENDOMETRIAL ADENOCARCINOMA 2 2 2 11 .0002 .0035 .6360 Adj. # at Risk 45.3 39.7 39.3 40.6 Endometrial Adenoma/-Adenocarcinoma 3 2 2 11 .0005 .0102 .7722 Addj. # at Risk 45.5 39.6 39.5 40.8 SQUAMOUS CELL CARCINOMA 1 0 1 5 .0059 .0765 .7160 plow vsVeh .6360 .7722 1 All the tumors in male rats listed above would be classified as common (incidence > 1%). Using the Haseman-Lin-Rahman rules described in 1.3.1.5, in male rats no test of trend or pairwise difference between the high dose and control was significant either when testing trend or pairwise difference, although the joint tests of liver tumors would be equivocal. In female rats, in the table above, sertoliform tubular adenoma, plus pooled thecal cell tumor and sertoliform tubular adenoma, both of the ovaries, and squamous cell carcinoma of the uterine cervix would be classified as rare tumors (but not in the uterus or pooled uterus and cervix), the remainder as common. Tumor incidence based on combining the uterus with the uterine cervix only affects the results for squamous cell carcinoma. Going down the table of organ tumor combinations, the test of trend in hepatocellular adenoma of the liver in female rats would be close to the multiplicity adjusted level of statistical significance (p = 0.0087 ≈ 0.005). The corresponding tests comparing the high dose to vehicle might be classified as close to significant when testing jointly, but not seperately ( p = 0.0539 ≈ 0.05, but > 0.01). The tests of trend in endometrial adenocarcinoma and pooled adenoma and adenocarcinoma of the uterus (and thus the uterus with uterine cervix) would both be categorized as statistically significant (p = 0.0002, 0.0005 < 0.005, respectively), while the corresponding tests between the high dose and vehicle were statistically significant or close ( p = 0.0035 < 0.01 < 0.05 and 0.0102 ≈ 0.01 < 0.05). Strictly following the HLR rules and specifying that squamous cell carcinoma of the uterus with cervix is a common tumor, we would conclude that both the test of trend and the pairwise comparison are close to statistical significance (p = 0.0059 ≈ 0.005 amd 0.0765 ≈ 0.05). The test of trend would No other tests of carcinogenicity between dose groups achieved the multiplicity adjusted levels of significance. Complete results of statistical poly-k tests of tumor trend and differences between dose groups in male rats are given in Tables A.2.3 and A.2.4, in Appendix 1. 21 Reference ID: 3401357 NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. 3.2.2 VEC-162: Carcinogenicity Study by Oral Gavage Administration to CD1 Mice for 104 Weeks. STUDY DURATION: 104 weeks. DOSING STARTING DATE: 27 April 2006. TERMINAL SACRIFICE: Week 104. NECROPSY COMPLETED: 2 May 2008 STUDY ENDING DATE (Final Report dated): 15 July 2009. MICE STRAIN: Crl: CD-1™ (ICR) BR Mice. ROUTE: Daily oral gavage The drug vehicle is 100% polyethylene glycol-400 (PEG400). Animals were dosed once daily by oral gavage. Gross aspects of the study designs for the main study animals are summarized in Table 18 be1ow: Table 18. Design of Mice Study (dose volume 4 mL/kg) Treatment Group 1. Vehicle 2 2. Low 3. Medium 4. High # Main study animals (# TK1 animals)/gender Nominal Dose (mg/kg/day) 66 (39) 66 (39) 66 (39) 66 (39) 0 30 100 300 Nominal Dosing Concentration (mg/mL) 0 7.5 25 75 1 Toxicokinetic phase animals began dosing during Week 1, were samples during Weeks 4 and 26, then terminated during Week 26. 2 100% polyethylene glycol-400. The Sponsor summarizes the study conduct as follows: “The carcinogenic potential of VEC-162 (a melatonin agonist) was assessed over a period of 104 weeks in Crl: CD-1™ (ICR) BR mice by oral administration. Three groups, each comprising 66 males and 66 females, received VEC-162 by gavage at doses of 30, 100 or 300 mg/kg/day. A similarly constituted Control group received the vehicle (100% polyethylene glycol-400) at the same volume-dose. A further 39 males and 39 females were allocated to each group and were used to provide blood samples for toxicokinetic evaluation.” (page 10 of mice report) Animals were approximately five to six weeks old at first dosing. During the study animals were housed in groups of three animals of the same sex. Food and water were available ad libitum. The Sponsor states that detailed physical examinations were made on all animals “daily during the first week of treatment, twice weekly during Weeks 2 to 4 (middle and end of each week), once weekly during Weeks 5 to 13, once every two weeks during Weeks 14 to 52 and once every four weeksthereafter.” (page 20 of mice report) 22 Reference ID: 3401357 NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. 3.2.1.1 Sponsor’s Results and Conclusions This section will present a summary of the Sponsor’s analysis on survivability and tumorigencity in rats. Survival analysis: The Sponsor summarized mortality results as follows: “A total of 133 males and 162 females died or were killed prematurely during the treatment period. The group distribution of the deaths is presented below and there was no treatment related effect on mortality.” (page 33) Table 19. Sponsor Table: Group distribution of mortality Group/sex 1M 2M 3M 4M 1F 2F 3F Dose (mg/kg/day) 0 30 100 300 0 30 100 Premature deaths 30 38 32 33 37 41 47 Percentage survival 55 42 52 50 44 38 30 4F 300 38 42 “Statistical analysis revealed that the trend test was not statistically significant when all groups were included in the analysis (p=0.802 and 0.897 for males and females, respectively). None of the pair-wise comparisons was statistically significant.” (page 33 of report) Note that this table is consistent with tables 20 and 21 reported in the FDA analysis in Section 3.2.1.2, below. Tumorigenicity analysis: “There were no neoplastic changes that were attributed to treatment.” (page 36 of mouse report) At least to the extent that there were no statistically significant, multiplicity adjusted results, the statisistical analysis below is consistent with this conclusion. 3.2.1.2 FDA Reviewer's Results This section will present the current Agency findings on survival and tumorigenicity in male and female rats. Survival analysis: Kaplan-Meier plots comparing treatment groups in both studies are given in Appendix 1, along with more details of the analysis. The following tables (Table 20 for male mice, Table 21 for female mice) summarize the mortality results for the dose groups. The data were grouped for the specified time period, and present the number of deaths during the time interval over the number at risk at the beginning of the interval. The percentage cited is the percent survived at the end of the interval. 23 Reference ID: 3401357 NDA 205677 Tasimelteon Table 20. Summary of Period 1.Vehicle 0-52 6/661 90.9%2 53-78 5/60 83.3% 79-91 8/55 71.2% 92-104 11/47 54.5% Terminal 36 105 1 2 Vanda Pharmaceuticals, Inc. Male Mice Mortality (dose/kg/day) 2.Low 3.Medium 4.High 4/66 9/66 8/66 93.9% 86.4% 87.9% 13/62 10/57 10/58 74.2% 71.2% 72.7% 7/49 4/47 8/48 63.6% 65.2% 60.6% 14/42 9/43 7/40 42.4% 51.5% 50.0% 28 34 33 number deaths / number at risk per cent survival to end of period. Table 21. Summary of Female Mice Mortality (dose/kg/day) Period 1.Vehicle 2.Low 3.Medium 4.High 1 0-52 3/66 3/66 9/66 2/66 95.5%2 95.5% 86.4% 97.0% 53-78 9/63 12/63 13/57 17/64 81.8% 77.3% 66.7% 71.2% 79-91 9/54 10/51 14/44 4/47 68.2% 62.1% 45.5% 65.2% 92-102 16/45 16/41 10/30 15/43 43.9% 37.9% 30.3% 42.4% Terminal 29 25 20 28 102-104 1 2 number deaths / number at risk per cent survival to end of period. A summary of the results of tests of dose effect on survival in mice are presented below: Table 22. Statistical Significances of Tests of Homogeneity and Trend in Survival in Mice Hypotheses Males Females Logrank Wilcoxon Logrank Wilcoxon Homogeneity over all four groups 0.6054 0.6502 0.1549 0.0948 No Trend over all four groups 0.7949 0.5793 0.9095 0.9852 No difference between high dose and vehicle 0.4564 0.3617 0.7822 0.6885 In male mice, the survival curve for the vehicle eventually slightly dominates the survival curves in the other groups, but this is not sufficient to result in any statistically significant test of homogeneity, dose related trend, or difference between the high dose and control (all 6 p ≥ 0.3617). In female mice it does seem that the vehicle, low, and high dose groups are eventually 24 Reference ID: 3401357 NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. intertwined, but with higher survival than the medium dose group. As the Wilcoxon statistic is more sensitive to later differences, this is sufficient to result in test of overall homogeneity that is arguable somewhat close to significance (Wilcoxon p = 0.0948), but less so when weighting earlier difference (Logrank p = 0.1549). However, there is no strong evidence of dose related trend or heterogeneity between the high dose and control (all four p ≥ 0.6885). Tumorigenicity analysis: Table 23 below, a repeat of Table 4 above and a synopsis of Table A.2.2 below, shows the tumors that had at least one non-multiplicity adjusted test that was statistically significant at a 0.05 level. Table 23. Potentially Statistically Significant Neoplasms in Mice Sex/ Incidence Significance Levels organ/ Veh Low Med High ptrend phigh pmed ____tumor_________________________________________________________vsVeh vsVeh Male Mice Caecum+Colon+Duodenum # Evaluated 66 66 66 66 Adj. # at Risk 50.1 45.8 46.5 45.1 Adenoma 0 0 0 3 .0134 .1025 . Caecum+Colon+Duodenum+Jejunum # Evaluated 66 66 66 66 Adj. # at Risk 50.6 45.8 46.5 45.4 Adenoma/Adenocarcinoma 2 0 0 5 .0114 .1764 1 DUODENUM # Evaluated 66 65 65 66 Adj. # at Risk 50.1 44.9 46.5 45.1 ADENOMA 0 0 0 2 .0582 .2217 . Female Mice UTERINE CERVIX # Evaluated Adj. # at Risk ENDOMETRIAL POLYP 66 64 65 66 48.6 44.3 39.8 48.1 0 0 1 2 .0652 .2474 .4483 plow vsVeh . 1 . . In mice the only test that met the multiplicity adjusted limits for statistical significane were adenomas in the caecum or colon or duodenum (i.e the intestine) of male mice ( p = 0.0134 < 0.025), while the corresponding pairwise test was close to significance (p = 0.1025 ≈ . Note, however, if we consider pooled adenoma and adenocarcinoma of those organs plus the jejunum to be rare (ignoring the incidence in the vehicle group) the dose realed trend test, but not the pairwise comparison with control, would also be classified as statistically significant. Complete results of statistical poly-k tests of tumor trend and differences between dose groups in mice are given in Tables A.2.5 and A.2.6 in Appendix 2. 4. FINDINGS IN SPECIAL/SUBGROUP POPULATIONS NA 25 Reference ID: 3401357 NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. 5. SUMMARY AND CONCLUSIONS 5.1. Statistical Issues and Collective Evidence Please see Section 1.3 above. 5.2. Conclusions and Recommendations Please see section 1.1. APPEARS THIS WAY ON ORIGINAL 26 Reference ID: 3401357 NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. APPENDICES: Appendix 1. Survival Analysis Simple summary life tables in mortality are presented in the report (Tables 14, 15, 20, and 21, above). These curves, defined across study groups for each species by gender combination, are displayed below in Figures A.1.1 through A.1.4. The plots include 95% confidence intervals around each survival curve (colored area around each curve). These plots are also supported by tests of homogeneity in survival over the treatment groups. The statistical significance levels (i.e., p-values) are provided in Tables A.1.1. and A.1.2, below. One might note that the log rank tests place greater weight on later events, while the Wilcoxon test tends to weight them more equally, and thus places more weight on differences in earlier events than does the log rank test. Table A.1.1 Statistical Significances of Tests of Homogeneity and Trend in Survival in Rats Hypotheses Males Females Logrank Wilcoxon Logrank Wilcoxon Homogeneity over all four groups 0.0223 0.0082 0.5557 0.3796 No Trend over all four groups 0.1225 0.2021 0.2271 0.2051 No difference between high dose and vehicle 0.4005 0.7710 0.1343 0.3311 The tests of homogeneity in survival over doses in male rats are statistically significant at the usual 0.05 level (Logrank p = 0.0233, Wilcoxon p = 0.0082). From the Kaplan-Meier curve in Figure A.1.1, it seems clear that this is due to the overall lower survival in the low dose group while the medium dose group has generally the highest survival, with survival in the high dose and vehicle crossing, but mostly bounded between these curves. It seems difficult to attribute these results to any particular dose related effect. This seems to be consistent with the lack of any statistically significant dose related trend (Logrank p = 0.1225, Wilcoxon p = 0.2021) or any statistically significant difference between the the high dose and vehicle (Logrank p = 0.4005, Wilcoxon p = 0.7710). In female rats, the Kaplan-Meier curves suggest that through most of the study the vehicle has highest survival, with the other curves rather intertwined. However, these apparent differences were not statistically significant. Whether analyzing homogeneity or trend among all four dose groups or the comparison of the High dose and vehicle, no tests were statistically significance (i.e., all 6 p ≥ 0.1343). While absence of proof is not proof of absence, the lack of evidence for such differences in survival is consistent with the hypotheses of no dose related differences or trends. The Kaplan-Meier survival curves in male and female rats rats are presented next. 27 Reference ID: 3401357 NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. Figure A.1.1 Kaplan-Meier Survival Curves for Male Rats Figure A.1.2 Kaplan-Meier Survival Curves for Female Rats 28 Reference ID: 3401357 NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. The summary of the results of tests of dose effect on survival in mice are presented below: Table A.1.2 Statistical Significances of Tests of Homogeneity and Trend in Survival in Mice Hypotheses Males Females Logrank Wilcoxon Logrank Wilcoxon Homogeneity over all four groups 0.6054 0.6502 0.1549 0.0948 No Trend over all four groups 0.7949 0.5793 0.9095 0.9852 No difference between high dose and vehicle 0.4564 0.3617 0.7822 0.6885 In male mice, the survival curve for the vehicle eventually slightly dominates the survival curves in the other groups, but this is not sufficient to result in any statistically significant test of homogeneity, dose related trend, or difference between the high dose and control (all 6 p ≥ 0.3617). In female mice it does seem that the vehicle, low, and high dose groups are eventually intertwined, but with higher survival than the medium dose group. As the Wilcoxon statistic is more sensitive to later differences, this is sufficient to result in test of overall homogeneity that is arguable somewhat close to significance (Wilcoxon p = 0.0948), but less so when weighting earlier difference (Logrank p = 0.1549). However, there is no strong evidence of dose related trend or heterogeneity between the high dose and control (all 4 p ≥ 0.6885). Figure A.1.3 Kaplan-Meier Survival Curves for Male Mice 29 Reference ID: 3401357 NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. The survival curves in female mice is given below: Figure A.1.4 Kaplan-Meier Survival Curves for Female Mice 30 Reference ID: 3401357 NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. Appendix 2. FDA Poly-k Tumorigenicity Analysis The poly-k test, here with k=3, modifies the original Cochran-Armitage test to adjust for differences in mortality (please see Bailer & Portier, 1988, Bieler & Williams, 1993). The tests used here are small sample exact permutation tests of tumor incidence. When there were no tumors of the specific type being analyzed in either column of the 2x2 table corresponding to a pairwise comparison an argument could be made that the p-value for this test should be 1.0. However, largely for readability, in the tables below these p-values are considered as missing (i.e., corresponding to a null test), denoted by a period “.”. Note that the StatXact program used for these analyses adjusts for the variance, which would be 0. Then the significance levels of the test statistics are based on the result of a division by 0, i.e., undefined, and hence StatXact codes these p-values as missing. For each gender by organ the number of animals microscopically analyzed is presented first. Note that indicating an organ was not examined requires a specification in the data (please see section 2.2 above). It is possible that this specification could be missing in some of this data. Then the number of animals at risk could be inflated, and the proportion of animals with tumor would be artificially decreased. Thus, as discussed in Section 1.5 above, for some of these organs it is possibly more appropriate to define the actual endpoint used in the statistical analysis be the condition of being microscopically analyzed AND show the tumor. This does have problems unless treatment groups are not treated equally except for actual treatment. The entry for each tumor is preceded by the adjusted number of animals at risk for that endpoint. It seems clear that an animal that dies early without having displaying that endpoint reduces the size of the risk set for that getting that particular endpoint. The poly-k test down weights such animals, and as discussed in Section 1.3.1.4, above, the sum of these poly-k weights seems to be a better estimate of the number of animals at risk of getting that tumor than the simple number of animals analyzed. This sum is given in the row labeled “Adjusted # at risk ”. Tumor incidence is presented next, with the significance levels of the tests of trend, and the results of pairwise tests between the high and medium dose groups versus vehicle. The next row continues with the pvalues of the pairwise test between the low and vehicle dose groups and the p-values between the vehicle dose group and high dose group with water, respectively. For these analyses, incidence in the water only group is used to assess background tumor incidence, and thus whether a tumor is considered to be rare (background incidence <1%) or common. Note that a tumor is only classified as rare if the vehicle group shows none of that particular tumor. To adjust for the multiplicity of tests the so-called Haseman-Lin-Rahman (HLR) rules discussed in Section 1.3.1.5 are often applied. The question here is whether we want to control type I error over both studies when testing trend and the pairwise difference between the high dose and control jointly or separately. That is, when testing for both trend and pairwise difference together, to control the overall Type I error rate to roughly 10% one compares the unadjusted significance level of the trend test to 0.005 for common tumors and 0.025 for rare tumors, and the pairwise test to 0.05 for common tumors and 0.10 for rare tumors. To compare them separately we use the same bounds for the test of trend, but for the pairwise test use 0.01 31 Reference ID: 3401357 NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. and 0.05, respectively. Using these adjustments for other tests, like testing the comparisons between the low and medium dose groups versus vehicle can be expected to increase the overall type I error rate to some value above the nominal rough 10% level, possibly considerably higher than the nominal 10% rate. Table A.2.1 below, shows the tumors that had at least one non-multiplicity adjusted test in rats that was statistically significant at least at a 0.05 level Table A.2.1. Potentially Statistically Significant Neoplasms in Rats Sex/ Incidence Significance Levels organ/ Veh Low Med High ptrend phigh pmed ____tumor_________________________________________________________vsVeh vsVeh Male Rats LIVER # Evaluated 65 65 65 65 Adj. # at Risk 43.7 36.1 49.9 43.2 HEPATOCELLULAR ADENOMA 2 2 6 8 .0123 .0444 .1802 Adj. # at Risk 43.7 36.1 50.7 43.2 Hepatocellular Adenoma/Carcinoma 2 2 8 8 .0145 .0444 .0746 PANCREAS # Evaluated 65 65 65 65 Adj. # at Risk 42.9 35.7 49.2 42.8 ISLET CELL ADENOMA 1 5 5 2 .6541 .5000 .1411 Adj. # at Risk 42.9 36.1 49.3 43.0 Islet Cell Adenoma/Carcinoma 2 6 7 2 .7843 .6921 .1212 PITUITARY # Evaluated 63 65 65 64 Adj. # at Risk 47.6 45.6 55.0 46.9 ADENOMA, PARS DISTALIS 19 28 30 23 .4822 .2361 .1104 SKIN # Evaluated 65 65 65 65 Adj. # at Risk 43.1 35.7 50.3 43.4 FIBROMA 1 0 5 3 .0920 .3080 .1404 Adj. # at Risk 43.1 35.7 50.3 43.9 Fibroma/Sarcoma NOS/Fibrosarcoma 1 0 5 4 .0407 .1800 .1404 TESTES # Evaluated 65 65 65 65 Adj. # at Risk 43.3 35.7 49.2 43.2 INTERSTITIAL (LEYDIG) CELL ADENOMA 3 1 4 6 .0589 .2417 .5735 Female Rats LIVER # Evaluated Adj. # at Risk HEPATOCELLULAR ADENOMA MAMMARY # Evaluated Adj. # at Risk MAMMARY ADENOCARCINOMA OVARIES # Evaluated Adj. # at Risk SERTOLIFORM TUBULAR ADENOMA Adj. # at Risk Thecal Cell Tmr/Serto.Tub.Adenoma .6222 .6222 .0692 .0878 .0296 1 1 .9135 65 65 65 65 45.4 39.6 40.5 40.4 2 1 6 7 .0087 .0539 .0980 .8511 64 65 64 65 47.6 42.4 43.5 44.8 10 10 12 18 .0146 .0356 .3134 .4864 .0546 .2066 . . .0525 .2066 .4643 . 64 65 64 63 45.2 39.6 39.3 38.2 0 0 0 2 45.2 39.6 39.4 38.2 0 0 1 2 32 Reference ID: 3401357 plow vsVeh NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. Table A.2.1. (cont.) Potentially Statistically Significant Neoplasms in Rats Sex/ Incidence Significance Levels organ/ Veh Low Med High ptrend phigh pmed ____tumor_________________________________________________________vsVeh vsVeh Female Rats (cont.) UTERINE CERVIX # Evaluated 64 65 64 64 Adj. # at Risk 45.2 39.6 39.3 39.3 SQUAMOUS CELL CARCINOMA 0 0 0 2 .0568 .2126 . UTERUS # Evaluated 64 65 64 64 Adj. # at Risk 45.3 39.7 39.3 40.6 ENDOMETRIAL ADENOCARCINOMA 2 2 2 11 .0002 .0035 .6360 Adj. # at Risk 45.3 39.7 39.3 40.6 Endo. Adenoma/-Adenocarcinoma 3 2 2 11 .0005 .0102 .7722 Adj. # at Risk 45.5 39.6 39.5 39.9 SQUAMOUS CELL CARCINOMA 1 0 1 3 .0543 .2558 .7160 Uterus(w/uterine cervix) # Evaluated 64 65 64 64 Adj. # at Risk 45.3 39.7 39.3 40.6 ENDOMETRIAL ADENOCARCINOMA 2 2 2 11 .0002 .0035 .6360 Adj. # at Risk 45.3 39.7 39.3 40.6 Endometrial Adenoma/-Adenocarcinoma 3 2 2 11 .0005 .0102 .7722 Addj. # at Risk 45.5 39.6 39.5 40.8 SQUAMOUS CELL CARCINOMA 1 0 1 5 .0059 .0765 .7160 plow vsVeh . .6360 .7722 1 .6360 .7722 1 Note that all the tumors in male rats would be classified as common (incidence > 1%). Using the Haseman-Lin-Rahman rules for separate tests A, in male rats no test of trend or pairwise difference between the high dose and control was statistically significant, although the tests for hepatocellular adenoma (trend p = 0.0123 ≈ 0.005, pairwise p = 0.0444 ≈ 0.01) and pooled adenoma and carcinoma (trend p = 0.0145 ≈ 0.005, pairwise p = 0.0444 ≈ 0.01) of the liver were all somewhat close to significance. In female rats, in the table above, sertoliform tubular adenoma, plus pooled thecal cell tumor and sertoliform tubular adenoma, both of the ovaries, and squamous cell carcinoma of the uterine cervix would be classified as rare tumors (but not in the uterus or pooled uterus and cervix), the remainder as common. For tumor incidence combining the uterus with the uterine cervix only affects the results for squamous cell carcinoma. The test of trend in hepatocellular adenoma of the liver would be close to the multiplicity adjusted level of statistical significance (p = 0.0087 ≈ 0.005). The corresponding tests comparing the high dose to vehicle might be classified as close to significance ( p = 0.0539 ≈ 0.01). The tests of trend in endometrial adenocarcinoma and pooled adenoma and adenocarcinoma of the uterus (and thus the uterus with uterine cervix) would both be categorized as statistically significant (p = 0.0002, 0.0005 < 0.005, respectively), while the corresponding tests between the high dose and vehicle were statistically significant or close ( p = 0.0035 < 0.01 < 0.05 and 0.0102 ≈ 0.01 < 0.05). Strictly following the HLR rules and specifying that squamous cell carcinoma of the uterus with cervix is a common tumor, we would conclude that both the test of trend and the pairwise comparison are close to statistical significance (p = 0.0059 ≈ 0.005 amd 0.0765 ≈ 0.01). No other tests of carcinogenicity between dose groups achieved the multiplicity adjusted levels of significance. Table A.2.2 below, shows the tumors that had at least one non-multiplicity adjusted test in mice that was statistically significant at a 0.05 level. In mice the only test that met the 33 Reference ID: 3401357 NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. multiplicity adjusted limits for statistical significane were adenomas in the caecum or colon or duodenum (i.e the intestine) of male mice ( p = 0.0134 < 0.025), while the corresponding pairwise test was close to significance (p = 0.1025 ≈ . Note, however, if we consider pooled adenoma and adenocarcinoma of those organs plus the jejunum to be rare (ignoring the incidence in the vehicle group) the dose realed trend test, but not the pairwise comparison with control, would also be classified as statistically significant. Table A.2.2. Potentially Statistically Significant Neoplasms in Mice Sex/ Incidence Significance Levels organ/ Veh Low Med High ptrend phigh pmed ____tumor_________________________________________________________vsVeh vsVeh Male Mice Caecum+Colon+Duodenum+Jejunum # Evaluated 66 66 66 66 Adj. # at Risk 50.6 45.8 46.5 45.4 Adenoma/Adenocarcinoma 2 0 0 5 .0114 .1764 1 Caecum+Colon+Duodenum # Evaluated 66 66 66 66 Adj. # at Risk 50.1 45.8 46.5 45.1 Adenoma 0 0 0 3 .0134 .1025 . DUODENUM # Evaluated 66 65 65 66 Adj. # at Risk 50.1 44.9 46.5 45.1 ADENOMA 0 0 0 2 .0582 .2217 . LUNGS + BRONCHI # Evaluated 66 66 66 66 Adj. # at Risk 51.4 50.3 46.8 47.4 BRONCHIOLOALVEOLAR ADENOMA 19 27 14 18 .7390 .5404 .8218 Female Mice UTERINE CERVIX # Evaluated Adj. # at Risk ENDOMETRIAL POLYP 66 64 65 66 48.6 44.3 39.8 48.1 0 0 1 2 .0652 .2474 .4483 plow vsVeh 1 . . .0680 . Complete Incidence tables are provided below: Table A.2.3. Tumor Incidence and Results of Tests in Male Rats Organ/ Incidence Significance Levels tumor Veh Low Med High ptrend phigh pmed __________________________________________________________________vsVeh vsVeh ABDOMEN # Evaluated 2 1 0 1 Adj. # at Risk 2.0 0.8 0.0 0.6 MALIGNANT SCHWANNOMA 1 0 0 0 1 . . Adj. # at Risk 1.5 1.0 0.0 0.6 MESOTHELIOMA 1 1 0 0 1 . . ADIPOSE TISSUE # Evaluated 7 4 2 1 Adj. # at Risk 5.2 2.7 1.7 0.8 FIBROMA 0 0 1 0 .1250 . .1667 34 Reference ID: 3401357 plow vsVeh . . . NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. Table A.2.3 (cont.). Tumor Incidence and Results of Tests in Male Rats Organ/ Incidence Significance Levels tumor Veh Low Med High ptrend phigh pmed __________________________________________________________________vsVeh vsVeh ADRENALS # Evaluated 65 65 64 65 Adj. # at Risk 43.3 36.1 49.3 42.8 CORTICAL ADENOMA 2 2 2 2 .5308 .6831 .7394 Adj. # at Risk 43.4 35.7 48.7 42.8 MALIGNANT PHAEOCHROMOCYTOMA 2 0 0 0 1 1 1 Adj. # at Risk 44.6 36.0 49.8 43.4 PHAEOCHROMOCYTOMA 9 6 10 5 .8488 .9219 .6045 Adj. # at Risk 45.1 36.0 49.8 43.4 Pheochromocytoma [B&M] 11 6 10 5 .9137 .9681 .7634 BRAIN # Evaluated 65 65 65 65 Adj. # at Risk 43.1 36.5 49.9 42.8 ASTROCYTOMA 1 1 1 0 .8467 1 .7843 Adj. # at Risk 43.1 37.4 49.9 42.8 Astrocytoma/Oligodendroglioma 1 2 1 1 .6370 .7471 .7843 Adj. # at Risk 42.9 35.7 49.2 42.8 GRANULAR CELL TUMOUR 0 0 0 1 .2500 .5000 . Adj. # at Risk 42.9 36.6 49.2 42.8 OLIGODENDROGLIOMA 0 1 0 1 .3121 .5000 . H-POIETIC TUMOUR # Evaluated 65 65 65 65 Adj. # at Risk 42.9 36.2 49.6 43.2 HISTIOCYTIC SARCOMA 0 1 1 1 .3151 .5059 .5385 Adj. # at Risk 43.7 35.7 49.2 43.8 MALIGNANT LYMPHOMA 2 0 0 1 .6085 .8794 1 Adj. # at Risk 43.1 35.7 49.2 43.7 MYELOID CELL LEUKAEMIA 1 0 0 1 .4430 .7529 1 HEAD # Evaluated 2 0 1 2 Adj. # at Risk 2.0 0.0 0.0 1.1 SQUAMOUS CELL CARCINOMA, ZYMBAL 2 0 0 1 1 . . JEJUNUM # Evaluated 54 53 59 52 Adj. # at Risk 36.7 30.8 46.9 36.3 ADENOMA 0 0 1 0 .5541 . .5610 KIDNEYS # Evaluated 65 65 65 65 Adj. # at Risk 42.9 35.7 49.2 42.8 TUBULAR CARCINOMA 0 1 0 0 .7500 . . LIVER # Evaluated 65 65 65 65 Adj. # at Risk 43.7 36.1 49.9 43.2 HEPATOCELLULAR ADENOMA 2 2 6 8 .0123 .0444 .1802 Adj. # at Risk 42.9 35.7 50.0 42.8 HEPATOCELLULAR CARCINOMA 0 0 2 0 .5225 . .2872 Adj. # at Risk 43.7 36.1 50.7 43.2 Hepatocellular Adenoma/Carcinoma 2 2 8 8 .0145 .0444 .0746 LN MESENTERIC # Evaluated 65 64 65 65 Adj. # at Risk 42.9 35.5 49.9 42.8 HAEMANGIOMA 2 0 3 1 .5436 .8795 .5735 LUNGS + BRONCHI # Evaluated 65 65 65 65 Adj. # at Risk 43.0 35.7 49.2 42.8 BRONCHIOLOALVEOLAR ADENOMA 1 0 0 0 1 1 1 35 Reference ID: 3401357 plow vsVeh .6222 1 .7440 .8573 .7069 .4431 . .4615 .4615 1 1 . . .4545 .6222 . .6222 1 1 NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. Table A.2.3. (cont.) Tumor Incidence and Results of Tests in Male Rats Organ/ Incidence Significance Levels tumor Veh Low Med High ptrend phigh pmed __________________________________________________________________vsVeh vsVeh MAMMARY # Evaluated 65 65 65 65 Adj. # at Risk 42.9 35.7 49.2 42.8 MAMMARY ADENOCARCINOMA 1 0 0 0 1 1 1 Adj. # at Risk 43.8 36.6 49.2 42.8 MAMMARY FIBROADENOMA 3 1 0 0 .9963 1 1 Adj. # at Risk 43.0 35.7 49.2 43.1 SARCOMA NOS 1 0 0 1 .4430 .7529 1 PANCREAS # Evaluated 65 65 65 65 Adj. # at Risk 42.9 35.7 49.2 42.8 ACINAR CELL ADENOMA 0 1 1 2 .1264 .2470 .5385 Adj. # at Risk 42.9 36.1 49.3 43.0 ISLET CELL ADENOMA 1 5 5 2 .6541 .5000 .1411 Adj. # at Risk 42.9 35.7 49.2 42.8 ISLET CELL CARCINOMA 1 1 2 0 .8142 1 .5582 Adj. # at Risk 42.9 36.1 49.3 43.0 Islet Cell Adenoma/Carcinoma 2 6 7 2 .7843 .6921 .1212 Adj. # at Risk 43.2 35.7 49.2 42.8 MIXED CELL ADENOMA 1 0 0 0 1 1 1 PARATHYROIDS # Evaluated 61 60 62 60 Adj. # at Risk 40.4 33.6 48.3 39.3 ADENOMA 2 0 2 1 .5661 .8751 .7574 PINNAE # Evaluated 1 1 3 2 Adj. # at Risk 1.0 1.0 3.0 1.0 LEIOMYOMA 0 1 0 0 1 . . PITUITARY # Evaluated 63 65 65 64 Adj. # at Risk 47.6 45.6 55.0 46.9 ADENOMA, PARS DISTALIS 19 28 30 23 .4822 .2361 .1104 Adj. # at Risk 41.7 35.7 49.2 42.3 ADENOMA, PARS INTERMEDIA 1 1 0 0 .9408 1 1 PREPUTIAL GLANDS # Evaluated 65 65 63 65 Adj. # at Risk 43.0 35.7 48.4 42.9 SQUAMOUS CELL PAPILLOMA 1 0 0 1 .4409 .7530 1 SALIVARY GLANDS # Evaluated 65 62 64 65 Adj. # at Risk 42.9 35.7 49.0 42.8 MALIGNANT SCHWANNOMA 0 1 0 0 .7500 . . SKELETAL MUSCLE # Evaluated 65 65 65 65 Adj. # at Risk 42.9 35.7 49.2 42.8 HAEMANGIOMA 0 0 0 1 .2500 .5000 . Adj. # at Risk 42.9 35.7 49.2 42.8 OSTEOSARCOMA 0 0 1 0 .5417 . .5385 36 Reference ID: 3401357 plow vsVeh 1 .9179 1 .4545 .0692 .7057 .0878 1 1 . .0296 .7123 1 .4545 . . NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. Table A.2.3. (cont.) Tumor Incidence and Results of Tests in Male Rats Organ/ Incidence Significance Levels tumor Veh Low Med High ptrend phigh pmed __________________________________________________________________vsVeh vsVeh SKIN # Evaluated 65 65 65 65 Adj. # at Risk 42.9 35.7 49.2 42.8 BASAL CELL CARCINOMA 1 0 1 0 .7914 1 .7897 Adj. # at Risk 42.9 35.7 49.2 42.8 BASAL CELL TUMOUR 0 0 0 1 .2500 .5000 . Adj. # at Risk 42.9 35.7 49.2 42.8 Basal Cell Tumor/Carcinoma 1 0 1 1 .4223 .7530 .7897 Adj. # at Risk 43.1 35.7 50.3 43.4 FIBROMA 1 0 5 3 .0920 .3080 .1404 Adj. # at Risk 43.1 35.7 50.3 43.9 Fibroma/Sarc.NOS/fibrosarcoma 1 0 5 4 .0407 .1800 .1404 Adj. # at Risk 43.7 36.5 49.6 43.4 KERATOACANTHOMA 8 2 3 5 .6269 .8861 .9858 Adj. # at Risk 42.9 35.7 49.2 43.3 SARCOMA NOS 0 0 0 1 .2544 .5059 . Adj. # at Risk 42.9 35.7 49.2 42.8 SEBACEOUS CELL ADENOMA 0 0 1 0 .5417 . .5385 Adj. # at Risk 43.1 36.8 49.2 42.8 SQUAMOUS CELL PAPILLOMA 1 3 1 0 .9298 1 .7843 Adj. # at Risk 43.9 37.6 49.6 43.4 Sq. Cell Papillpma/Keratocanthoma 9 5 4 5 .8459 .9288 .9808 Adj. # at Risk 43.1 35.7 49.2 42.8 TRICHIOEPITHELIOMA 1 0 0 0 1 1 1 SPINAL C. CERV. # Evaluated 65 65 65 65 Adj. # at Risk 42.9 36.6 49.2 42.8 ASTROCYTOMA 0 1 0 0 .7515 . . STOMACH # Evaluated 64 64 65 65 Adj. # at Risk 42.6 35.6 49.2 43.1 SQUAMOUS CELL CARCINOMA 0 0 0 1 .2544 .5059 . Systemic # Evaluated 65 65 65 65 Adj. # at Risk 43.4 36.4 49.2 42.8 Schwannoma [M] 1 1 0 0 .9371 1 1 TAIL # Evaluated 3 2 2 5 Adj. # at Risk 2.0 2.0 2.0 4.9 FIBROSARCOMA 0 0 0 1 .5000 .6667 . TESTES # Evaluated 65 65 65 65 Adj. # at Risk 43.3 35.7 49.2 43.2 INTERSTITIAL (LEYDIG) CELL ADENOMA 3 1 4 6 .0589 .2417 .5735 THYMUS # Evaluated 60 62 63 62 Adj. # at Risk 40.1 33.2 47.5 41.0 THYMOMA (EPITHELIAL) 1 0 0 0 1 1 1 37 Reference ID: 3401357 plow vsVeh 1 . 1 1 1 .9846 . . .2435 .8787 1 .4615 . .7069 . .9135 1 NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. Table A.2.3. (cont.) Tumor Incidence and Results of Tests in Male Rats Organ/ Incidence Significance Levels tumor Veh Low Med High ptrend phigh pmed __________________________________________________________________vsVeh vsVeh THYROIDS # Evaluated 65 64 64 65 Adj. # at Risk 43.9 35.9 49.0 43.7 C-CELL ADENOMA 5 2 1 4 .5090 .7583 .9915 Adj. # at Risk 42.9 35.7 49.0 42.8 C-CELL CARCINOMA 0 0 0 1 .2500 .5000 . Adj. # at Risk 43.9 35.9 49.0 43.7 C-cell Adenoma/Carcinoma. 5 2 1 5 .3379 .6308 .9915 Adj. # at Risk 42.9 35.7 49.0 42.8 FOLLICULAR CELL ADENOMA 0 0 1 0 .5417 . .5385 Adj. # at Risk 42.9 35.7 49.0 42.8 FOLLICULAR CELL CARCINOMA 0 0 0 1 .2500 .5000 . Adj. # at Risk 42.9 35.7 49.0 42.8 Foll. Cell Adenoma/Carcinoma. 0 0 1 1 .2081 .5000 .5385 plow vsVeh .9070 . .9070 . . . Table A.2.4. Tumor Incidence and Results of Tests in Female Rats Organ/ Incidence Significance Levels tumor Veh Low Med High ptrend phigh pmed __________________________________________________________________vsVeh vsVeh ABDOMEN # Evaluated 0 1 2 4 Adj. # at Risk 0.0 1.0 1.9 2.9 MESOTHELIOMA 0 1 0 0 1 . . Adj. # at Risk 0.0 0.5 2.0 2.9 SARCOMA NOS 0 0 1 0 1 . . ADRENALS # Evaluated 65 65 65 64 Adj. # at Risk 45.6 39.7 40.3 37.5 CORTICAL ADENOMA 1 2 1 0 .8501 1 .7227 Adj. # at Risk 46.2 39.6 40.2 37.5 MALIGNANT PHAEOCHROMOCYTOMA 2 0 0 0 1 1 1 Adj. # at Risk 47.6 39.6 40.9 38.6 PHAEOCHROMOCYTOMA 11 1 3 3 .9032 .9888 .9914 Adj. # at Risk 48.5 39.6 40.9 38.6 Pheochromocytoma [B&M] 13 1 3 3 .9530 .9960 .9971 BRAIN # Evaluated 65 65 65 65 Adj. # at Risk 45.3 39.6 40.2 38.5 ASTROCYTOMA 0 0 0 1 .2346 .4578 . FEMUR INC. JOINT # Evaluated 65 65 65 65 Adj. # at Risk 45.3 39.6 40.3 38.4 OSTEOMA 0 0 1 0 .4815 . .4706 H-POIETIC TUMOUR # Evaluated 65 65 65 65 Adj. # at Risk 46.2 39.6 40.2 38.5 MALIGNANT LYMPHOMA 1 0 0 1 .4130 .7031 1 Adj. # at Risk 45.3 39.6 40.4 38.4 MYELOID CELL LEUKAEMIA 0 0 1 0 .4815 . .4706 HEAD # Evaluated 1 1 0 0 Adj. # at Risk 0.9 1.0 0.0 0.0 SQUAMOUS CELL CARCINOMA, ZYMBA 0 1 0 0 1 . . Adj. # at Risk 1.0 0.3 0.0 0.0 SQUAMOUS CELL CARCINOMA-ORAL C 1 0 0 0 1 . . 38 Reference ID: 3401357 plow vsVeh . . .4459 1 .9997 .9999 . . 1 . . . NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. Table A.2.4. (cont.) Tumor Incidence and Results of Tests in Female Rats Organ/ Incidence Significance Levels tumor Veh Low Med High ptrend phigh pmed __________________________________________________________________vsVeh vsVeh HEART # Evaluated 65 65 65 65 Adj. # at Risk 45.3 39.6 40.2 38.7 MALIGNANT SCHWANNOMA 0 0 0 1 .2346 .4578 . JEJUNUM # Evaluated 61 63 61 63 Adj. # at Risk 44.4 39.5 38.8 38.2 LEIOMYOSARCOMA 0 0 1 0 .4780 . .4634 LIVER # Evaluated 65 65 65 65 Adj. # at Risk 45.4 39.6 40.5 40.4 HEPATOCELLULAR ADENOMA 2 1 6 7 .0087 .0539 .0980 LN MESENTERIC # Evaluated 65 65 65 63 Adj. # at Risk 45.4 39.6 40.2 37.9 HAEMANGIOMA 1 1 0 0 .9231 1 1 MAMMARY # Evaluated 64 65 64 65 Adj. # at Risk 52.2 50.7 46.5 48.3 Adenoma/Adenocarc./Fibroad. 33 35 26 33 .4178 .3649 .8178 Adj. # at Risk 47.6 42.4 43.5 44.8 MAMMARY ADENOCARCINOMA 10 10 12 18 .0146 .0356 .3134 Adj. # at Risk 45.6 39.7 39.6 39.2 MAMMARY ADENOMA 2 2 2 4 .1372 .2721 .6360 Adj. # at Risk 49.9 48.5 43.0 44.0 MAMMARY FIBROADENOMA 25 30 17 22 .7415 .5774 .9056 MESENTRY # Evaluated 0 0 0 1 Adj. # at Risk 0.0 0.0 0.0 1.0 LEIOMYOSARCOMA 0 0 0 1 1 . . Mesentry+Jejunum # Evaluated 65 65 65 65 Adj. # at Risk 45.3 39.6 40.3 39.0 Leiomyosarcoma 0 0 1 1 .1743 .4643 .4706 OVARIES # Evaluated 64 65 64 63 Adj. # at Risk 45.2 39.6 39.3 38.2 SERTOLIFORM TUBULAR ADENOMA 0 0 0 2 .0546 .2066 . Adj. # at Risk 45.2 39.6 39.4 37.9 THECAL CELL TUMOUR 0 0 1 0 .4750 . .4643 Adj. # at Risk 45.2 39.7 39.3 37.9 TUBULAR ADENOMA 0 1 0 0 .7188 . . Adj. # at Risk 45.2 39.6 39.4 38.2 Thecal Cell Tmr/Serto.Tub.Adenoma 0 0 1 2 .0525 .2066 .4643 PANCREAS # Evaluated 65 65 65 64 Adj. # at Risk 45.6 39.8 40.2 38.4 ISLET CELL ADENOMA 2 2 0 0 .9746 1 1 PARATHYROIDS # Evaluated 61 63 62 60 Adj. # at Risk 43.6 38.6 39.0 34.6 ADENOMA 1 0 0 0 1 1 1 39 Reference ID: 3401357 plow vsVeh . . .8511 .7160 .3123 .4864 .6360 .1747 . . . . .4643 . .6360 1 NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. Table A.2.4. (cont.) Tumor Incidence and Results of Tests in Female Rats Organ/ Incidence Significance Levels tumor Veh Low Med High ptrend phigh pmed __________________________________________________________________vsVeh vsVeh PITUITARY # Evaluated 65 65 65 64 Adj. # at Risk 60.9 57.7 58.6 53.1 ADENOMA, PARS DISTALIS 50 52 50 46 .5063 .4032 .4300 Adj. # at Risk 45.3 39.6 40.2 38.1 CARCINOMA, PARS DISTALIS 0 0 0 1 .2346 .4578 . Adj. # at Risk 60.9 57.7 58.6 53.2 Pars Dist. Adenoma/Carcinoma 50 52 50 47 .3764 .2951 .4300 SKELETAL MUSCLE # Evaluated 64 65 65 64 Adj. # at Risk 45.3 39.6 40.2 38.4 SARCOMA NOS 1 0 0 0 1 1 1 SKIN # Evaluated 65 65 65 65 Adj. # at Risk 45.3 39.6 40.3 39.0 BASAL CELL TUMOUR 0 0 1 1 .1705 .4578 .4706 Adj. # at Risk 45.3 39.6 40.2 38.6 FIBROMA 0 0 0 1 .2346 .4578 . Adj. # at Risk 45.3 39.6 40.3 38.4 KERATOACANTHOMA 0 0 1 0 .4815 . .4706 Adj. # at Risk 45.3 39.8 40.2 38.4 SEBACEOUS CELL ADENOMA 0 1 0 0 .7222 . . Adj. # at Risk 45.3 39.6 40.3 38.8 SQUAMOUS CELL PAPILLOMA 0 0 1 1 .1705 .4578 .4706 Adj. # at Risk 45.3 39.6 40.3 38.8 Sq. Cell Papilloma/Keratocanth 0 0 2 1 .1801 .4578 .2185 Adj. # at Risk 45.4 39.6 40.2 38.4 TRICHIOEPITHELIOMA 1 0 0 0 1 1 1 SPLEEN # Evaluated 65 65 65 64 Adj. # at Risk 45.6 39.6 40.2 38.4 HAEMANGIOSARCOMA 1 0 0 0 1 1 1 Systemic # Evaluated 65 65 65 65 Adj. # at Risk 45.7 39.6 40.2 38.4 Hemangioma/Hemangiosarcoma 2 1 0 0 .9796 1 1 Adj. # at Risk 45.3 39.6 41.1 38.4 Schwannoma [M] 0 0 1 0 .4847 . .4767 THYMUS # Evaluated 64 61 64 63 Adj. # at Risk 44.4 36.4 39.6 37.6 THYMOMA (LYMPHOID) 0 1 1 0 .5967 . .4699 THYROIDS # Evaluated 65 65 65 65 Adj. # at Risk 46.3 41.5 40.3 39.2 C-CELL ADENOMA 4 4 2 2 .8000 .8563 .8635 Adj. # at Risk 45.4 39.6 40.2 38.5 C-CELL CARCINOMA 1 0 0 1 .4152 .7091 1 Adj. # at Risk 46.4 41.5 40.3 39.2 C-cell Adenoma/Carcinoma. 5 4 2 3 .7122 .8070 .9203 Adj. # at Risk 45.4 39.6 40.2 38.5 FOLLICULAR CELL ADENOMA 1 0 0 1 .4152 .7091 1 TONGUE # Evaluated 65 65 65 65 Adj. # at Risk 45.3 39.7 40.2 38.4 SQUAMOUS CELL PAPILLOMA 0 1 0 0 .7222 . . 40 Reference ID: 3401357 plow vsVeh .1588 . .1588 1 . . . .4643 . . 1 1 .8511 . .4500 .5765 1 .6968 1 .4643 NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. Table A.2.4. (cont.) Tumor Incidence and Results of Tests in Female Rats Organ/ Incidence Significance Levels tumor Veh Low Med High ptrend phigh pmed __________________________________________________________________vsVeh vsVeh URINARY BLADDER # Evaluated 64 65 65 64 Adj. # at Risk 45.8 39.6 40.2 38.4 TRANSITIONAL CELL PAPILLOMA 1 0 0 0 1 1 1 UTERINE CERVIX # Evaluated 64 65 64 64 Adj. # at Risk 45.2 39.6 39.8 38.6 GRANULAR CELL TUMOUR 0 1 1 1 .2649 .4578 .4643 Adj. # at Risk 45.7 39.6 39.3 38.4 POLYP 1 0 0 0 1 1 1 Adj. # at Risk 45.7 39.6 39.3 39.3 Polyp/Squamous Cell Carcinoma 1 0 0 2 .1442 .4459 1 Adj. # at Risk 45.2 39.6 39.3 39.3 SQUAMOUS CELL CARCINOMA 0 0 0 2 .0568 .2126 . UTERUS # Evaluated 64 65 64 64 Adj. # at Risk 45.3 39.7 39.3 40.6 ENDOMETRIAL ADENOCARCINOMA 2 2 2 11 .0002 .0035 .6360 Adj. # at Risk 45.2 39.6 39.3 38.4 ENDOMETRIAL ADENOMA 1 0 0 0 1 1 1 Adj. # at Risk 46.5 39.7 39.3 39.6 ENDOMETRIAL POLYP 5 4 1 5 .3860 .5206 .9786 Adj. # at Risk 45.3 39.7 39.3 40.6 Endo. Adenoma/-Adenocarc. 3 2 2 11 .0005 .0102 .7722 Adj. # at Risk 45.2 39.6 39.3 38.5 LEIOMYOMA 0 0 0 1 .2360 .4578 . Adj. # at Risk 45.2 39.6 40.1 38.4 MALIGNANT SCHWANNOMA 0 0 1 0 .4815 . .4706 Adj. # at Risk 45.5 39.6 39.5 39.9 SQUAMOUS CELL CARCINOMA 1 0 1 3 .0543 .2558 .7160 Uterus(w/cervix) # Evaluated 64 65 64 64 Adj. # at Risk 45.3 39.7 39.3 40.6 ENDOMETRIAL ADENOCARCINOMA 2 2 2 11 .0002 .0035 .6360 Adj. # at Risk 45.2 39.6 39.3 38.4 ENDOMETRIAL ADENOMA 1 0 0 0 1 1 1 Adj. # at Risk 46.5 39.7 39.3 39.6 ENDOMETRIAL POLYP 5 4 1 5 .3860 .5206 .9786 Adj. # at Risk 45.3 39.7 39.3 40.6 Endo. Adenoma/-Adenocarc. 3 2 2 11 .0005 .0102 .7722 Adj. # at Risk 45.2 39.6 39.8 38.6 GRANULAR CELL TUMOUR 0 1 1 1 .2649 .4578 .4643 Adj. # at Risk 45.2 39.6 39.3 38.5 LEIOMYOMA 0 0 0 1 .2360 .4578 . Adj. # at Risk 45.2 39.6 40.1 38.4 MALIGNANT SCHWANNOMA 0 0 1 0 .4815 . .4706 Adj. # at Risk 45.7 39.6 39.3 38.4 POLYP 1 0 0 0 1 1 1 Adj. # at Risk 45.7 39.6 39.3 39.3 Polyp/Squamous Cell Carcinoma 1 0 0 2 .1442 .4459 1 Adj. # at Risk 45.5 39.6 39.5 40.8 SQUAMOUS CELL CARCINOMA 1 0 1 5 .0059 .0765 .7160 41 Reference ID: 3401357 plow vsVeh 1 .4643 1 1 . .6360 1 .6684 .7722 . . 1 .6360 1 .6684 .7722 .4643 . . 1 1 1 NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. Table A.2.4. (cont.) Tumor Incidence and Results of Tests in Female Rats Organ/ Incidence Significance Levels tumor Veh Low Med High ptrend phigh pmed __________________________________________________________________vsVeh vsVeh VAGINA # Evaluated 63 65 63 63 Adj. # at Risk 44.4 40.1 38.3 36.9 GRANULAR CELL TUMOUR 0 1 0 0 .7215 . . Adj. # at Risk 44.4 39.6 38.3 36.9 POLYP 0 0 0 1 .2293 .4500 . plow vsVeh .4762 . Table A.2.5. Tumor Incidence and Results of Tests in Male Mice Organ/ Incidence Significance Levels tumor Veh Low Med High ptrend phigh pmed __________________________________________________________________vsVeh vsVeh ADIPOSE TISSUE # Evaluated 4 6 4 0 Adj. # at Risk 3.9 4.8 3.2 0.0 HAEMANGIOMA 1 1 0 0 .9333 . 1 Adj. # at Risk 3.9 4.3 3.2 0.0 HIBERNOMA 0 1 0 0 .7000 . . Adj. # at Risk 3.9 4.2 3.3 0.0 OSTEOSARCOMA 0 0 1 0 .3000 . .5000 ADRENALS # Evaluated 66 66 66 64 Adj. # at Risk 50.1 45.8 46.5 44.6 CORTICAL ADENOMA 1 1 0 1 .5076 .7197 1 Adj. # at Risk 50.1 46.1 47.2 44.4 SUBCAPSULAR CELL ADENOMA 4 2 2 0 .9752 1 .8831 BONE # Evaluated 13 10 6 7 Adj. # at Risk 12.5 7.5 5.4 4.7 OSTEOMA 0 1 0 0 .5714 . . CAECUM # Evaluated 66 66 64 65 Adj. # at Risk 50.6 45.8 46.0 45.1 ADENOCARCINOMA 1 0 0 0 1 1 1 Adj. # at Risk 50.1 45.8 46.0 45.1 ADENOMA 0 0 0 1 .2432 .4737 . Adj. # at Risk 50.6 45.8 46.0 45.1 Adenoma/Adenocarcinoma 1 0 0 1 .4283 .7256 1 COLON # Evaluated 66 66 66 66 Adj. # at Risk 50.1 45.8 46.5 45.4 ADENOCARCINOMA 1 0 0 1 .4263 .7256 1 Adj. # at Risk 50.1 45.8 46.5 45.1 ADENOMA 0 0 0 1 .2419 .4737 . Caecum+Colon+Duodenum+Jejunum # Evaluated 66 66 66 66 Adj. # at Risk 50.6 45.8 46.5 45.4 Adenoma/Adenocarcinoma 2 0 0 5 .0114 .1764 1 Caecum+Colon+Duodenum # Evaluated 66 66 66 66 Adj. # at Risk 50.1 45.8 46.5 45.1 Adenoma 0 0 0 3 .0134 .1025 . Caecum+Colon+Jejunum # Evaluated 66 66 66 66 Adj. # at Risk 50.1 45.8 46.5 45.4 Adenocarcinoma 1 0 0 2 .1457 .4601 1 42 Reference ID: 3401357 plow vsVeh .8571 .5714 . .7256 .8777 .3684 1 . 1 1 . 1 . 1 NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. Table A.2.5. (cont.) Tumor Incidence and Results of Tests in Male Mice Organ/ Incidence Significance Levels tumor Veh Low Med High ptrend phigh pmed __________________________________________________________________vsVeh vsVeh DUODENUM # Evaluated 66 65 65 66 Adj. # at Risk 50.1 44.9 46.5 45.1 ADENOMA 0 0 0 2 .0582 .2217 . EPIDIDYMIDES # Evaluated 66 66 66 66 Adj. # at Risk 50.1 45.8 46.6 45.1 INTERSTITIAL (LEYDIG) ADENOMA 0 0 1 0 .4892 . .4792 GALL BLADDER # Evaluated 59 58 61 59 Adj. # at Risk 45.6 42.5 46.0 43.4 PAPILLOMA 0 2 2 0 .7390 . .2527 H-POIETIC TUMOUR # Evaluated 66 66 66 66 Adj. # at Risk 50.1 45.8 46.5 45.4 HISTIOCYTIC SARCOMA 0 1 0 1 .2956 .4737 . Adj. # at Risk 51.2 47.4 48.0 46.4 LYMPHOMA 5 4 5 3 .7070 .8298 .5902 Adj. # at Risk 50.1 47.2 47.4 45.8 MYELOID CELL LEUKAEMIA 0 3 1 1 .5208 .4737 .4845 HARDERIAN GLANDS # Evaluated 66 66 65 66 Adj. # at Risk 51.9 47.3 46.9 48.5 ADENOMA 12 7 14 7 .7911 .9177 .2954 JEJUNUM # Evaluated 66 65 65 66 Adj. # at Risk 50.1 44.9 46.2 45.1 ADENOCARCINOMA 0 0 0 1 .2432 .4737 . KIDNEYS # Evaluated 66 66 66 66 Adj. # at Risk 50.1 46.4 46.5 45.1 SCHWANNOMA 0 1 0 0 .7326 . . Adj. # at Risk 50.1 45.8 46.5 45.1 TUBULAR ADENOMA 0 1 1 1 .2781 .4737 .4792 LACHRYMAL GLDS # Evaluated 66 66 66 66 Adj. # at Risk 50.1 45.8 46.5 45.3 ADENOMA 0 0 0 1 .2419 .4737 . LIVER # Evaluated 66 66 66 66 Adj. # at Risk 50.1 45.8 46.5 45.1 HAEMANGIOMA 0 3 0 0 .8823 . . Adj. # at Risk 50.1 45.8 46.5 45.1 HEPATOBLASTOMA 0 0 1 0 .4892 . .4792 Adj. # at Risk 50.2 46.2 46.6 45.3 HEPATOCELLULAR ADENOMA 4 3 4 5 .2393 .4326 .5947 Adj. # at Risk 50.1 45.9 46.5 45.1 HEPATOCELLULAR CARCINOMA 0 1 0 1 .2956 .4737 . Adj. # at Risk 50.2 46.4 46.6 45.3 Hepatocellular Adenoma/Carcinoma 4 4 4 6 .1793 .3045 .5947 LN MESENTERIC # Evaluated 66 65 66 65 Adj. # at Risk 50.1 45.1 46.5 44.8 HAEMANGIOMA 0 0 1 0 .4865 . .4792 43 Reference ID: 3401357 plow vsVeh . . .2302 .4737 .7140 .1100 .9100 . .4792 .4737 . .1025 . .7463 .4737 .5947 . NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. Table A.2.5. (cont.) Tumor Incidence and Results of Tests in Male Mice Organ/ Incidence Significance Levels tumor Veh Low Med High ptrend phigh pmed __________________________________________________________________vsVeh vsVeh LUNGS + BRONCHI # Evaluated 66 66 66 66 Adj. # at Risk 51.4 50.3 46.8 47.4 BRONCHIOLOALVEOLAR ADENOMA 19 27 14 18 .7390 .5404 .8218 Adj. # at Risk 50.9 47.1 48.0 45.4 BRONCHIOLOALVEOLAR CARCINOMA 8 5 11 5 .6675 .8390 .2710 Adj. # at Risk 52.3 51.7 48.3 47.7 Bronchioalv. Adenoma/Carcinoma 26 31 23 22 .8040 .6978 .6584 PANCREAS # Evaluated 66 66 65 66 Adj. # at Risk 50.1 45.8 45.6 45.1 ISLET CELL ADENOMA 1 0 0 0 1 1 1 PARATHYROIDS # Evaluated 56 52 48 51 Adj. # at Risk 41.6 37.4 35.1 37.0 CHIEF CELL ADENOMA 0 1 0 0 .7267 . . PAWS # Evaluated 3 0 0 2 Adj. # at Risk 1.9 0.0 0.0 2.0 SQUAMOUS CELL PAPILLOMA 0 0 0 2 .3333 .3333 . PINNAE # Evaluated 3 3 3 3 Adj. # at Risk 2.5 2.2 0.9 1.2 SQUAMOUS CELL CARCINOMA 0 1 0 0 .6000 . . PREPUTIAL GLANDS # Evaluated 66 63 66 66 Adj. # at Risk 50.1 44.1 46.5 45.2 HAEMANGIOMA 0 0 0 1 .2432 .4737 . PROSTATE # Evaluated 66 66 66 66 Adj. # at Risk 50.1 45.8 46.5 45.1 ADENOMA 1 0 0 0 1 1 1 Pinnae+Skin+Paws # Evaluated 66 66 66 66 Adj. # at Risk 50.3 45.8 46.5 46.4 Squamous Cell Carc./-Pap./Kerato. 1 1 0 2 .2281 .4684 1 SALIVARY GLANDS # Evaluated 66 66 66 66 Adj. # at Risk 50.1 45.8 46.5 45.1 ADENOMA 1 0 0 0 1 1 1 SEMINAL VESICLES # Evaluated 66 66 66 66 Adj. # at Risk 50.1 45.8 46.5 45.1 SARCOMA NOS 0 0 1 0 .4892 . .4792 SKIN # Evaluated 66 66 66 66 Adj. # at Risk 50.7 45.8 47.1 45.1 FIBROSARCOMA 1 0 2 0 .6753 1 .4766 Adj. # at Risk 50.3 45.8 46.5 45.1 KERATOACANTHOMA 1 0 0 0 1 1 1 SPLEEN # Evaluated 66 66 66 66 Adj. # at Risk 50.1 45.8 46.5 45.1 HAEMANGIOMA 0 0 1 1 .1779 .4737 .4792 STOMACH # Evaluated 66 66 66 66 Adj. # at Risk 50.1 45.8 46.5 45.1 ADENOMA 0 0 1 0 .4892 . .4792 44 Reference ID: 3401357 plow vsVeh .0680 .8584 .1834 1 .4744 . .5000 . 1 .7256 1 . 1 1 . . NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. Table A.2.5. (cont.) Tumor Incidence and Results of Tests in Male Mice Organ/ Incidence Significance Levels tumor Veh Low Med High ptrend phigh pmed __________________________________________________________________vsVeh vsVeh TESTES # Evaluated 66 66 66 66 Adj. # at Risk 50.1 45.8 47.2 45.1 HAEMANGIOMA 0 0 1 0 .4920 . .4845 Adj. # at Risk 50.1 45.8 46.7 45.1 INTERSTITIAL (LEYDIG) CELL ADENOMA 1 1 3 2 .2695 .4601 .2776 THORAX # Evaluated 0 2 1 3 Adj. # at Risk 0.0 0.9 0.3 1.4 MESOTHELIOMA 0 0 0 1 1 . . Adj. # at Risk 0.0 1.7 0.3 0.5 OSTEOSARCOMA 0 1 0 0 1 . . THYROIDS # Evaluated 65 66 66 66 Adj. # at Risk 49.1 46.4 46.5 45.1 FOLLICULAR CELL ADENOMA 0 1 0 1 .2982 .4787 . plow vsVeh . .7256 . . .4842 Table A.2.6. Tumor Incidence and Results of Tests in Female Mice Organ/ Incidence Significance Levels tumor Veh Low Med High ptrend phigh pmed __________________________________________________________________vsVeh vsVeh ADRENALS # Evaluated 65 66 66 65 Adj. # at Risk 47.7 45.7 39.3 46.0 PHAEOCHROMOCYTOMA 0 0 0 1 .2599 .4946 . Adj. # at Risk 47.7 45.7 39.3 46.0 SUBCAPSULAR CELL ADENOMA 0 0 0 1 .2599 .4946 . Adj. # at Risk 47.7 45.7 39.3 46.0 SUBCAPSULAR CELL CARCINOMA 1 0 0 0 1 1 1 Adj. # at Risk 47.7 45.7 39.3 46.0 Subcaps. Cell Adenoma/Carcinom 1 0 0 1 .4533 .7473 1 BONE # Evaluated 7 4 5 0 Adj. # at Risk 5.5 4.0 3.5 0.0 LEIOMYOMA 0 0 1 0 .2727 . .3750 CAECUM # Evaluated 65 66 64 65 Adj. # at Risk 47.9 45.7 37.7 46.7 LEIOMYOMA 0 1 0 0 .7314 . . CLITORAL GLANDS # Evaluated 63 63 61 63 Adj. # at Risk 47.3 44.0 38.3 44.2 CARCINOMA 0 0 1 0 .4740 . .4471 Caecum+Jejunum # Evaluated 66 66 66 66 Adj. # at Risk 48.6 46.6 39.3 47.0 Leiomyoma/Leiomyosarcoma 0 2 0 0 .7906 . . GALL BLADDER # Evaluated 63 61 59 57 Adj. # at Risk 46.8 43.0 35.8 42.5 PAPILLOMA 1 0 1 0 .7141 1 .6806 GENERAL COMMENTS # Evaluated 0 1 1 0 Adj. # at Risk 0.0 1.0 0.4 0.0 CARCINOMA 0 1 0 0 1 . . 45 Reference ID: 3401357 plow vsVeh . . 1 1 . .4891 . .2368 1 . NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. Table A.2.6. (cont.) Tumor Incidence and Results of Tests in Female Mice Organ/ Incidence Significance Levels tumor Veh Low Med High ptrend phigh pmed __________________________________________________________________vsVeh vsVeh H-POIETIC TUMOUR # Evaluated 66 66 66 66 Adj. # at Risk 50.1 47.1 40.0 48.5 HISTIOCYTIC SARCOMA 5 4 1 5 .4009 .6033 .9727 Adj. # at Risk 51.7 51.6 43.1 49.7 LYMPHOMA 12 14 15 10 .7448 .7312 .1627 Adj. # at Risk 49.2 46.5 39.3 47.0 MYELOID CELL LEUKAEMIA 1 1 0 0 .9270 1 1 HARDERIAN GLANDS # Evaluated 65 66 66 66 Adj. # at Risk 48.4 45.7 39.3 47.0 ADENOCARCINOMA 0 1 0 0 .7303 . . Adj. # at Risk 49.5 47.6 39.5 47.2 ADENOMA 6 7 3 4 .8010 .8240 .8542 Adj. # at Risk 49.5 47.7 39.5 47.2 Adenoma/Adenocarcinoma 6 8 3 4 .8364 .8240 .8542 JEJUNUM # Evaluated 64 66 64 65 Adj. # at Risk 47.2 45.7 38.0 47.0 ADENOMA 0 0 0 1 .2614 .4946 . Adj. # at Risk 47.2 46.6 38.0 46.8 LEIOMYOSARCOMA 0 1 0 0 .7345 . . LIVER # Evaluated 66 66 66 66 Adj. # at Risk 48.6 45.7 39.3 47.0 HAEMANGIOMA 0 1 0 0 .7303 . . Adj. # at Risk 48.6 46.0 39.3 47.0 HAEMANGIOSARCOMA 0 1 0 0 .7318 . . Adj. # at Risk 48.6 45.8 39.4 47.0 HEPATOCELLULAR ADENOMA 0 1 1 0 .6096 . .4483 LUNGS + BRONCHI # Evaluated 66 65 66 66 Adj. # at Risk 51.3 45.9 41.5 50.4 BRONCHIOLOALVEOLAR ADENOMA 16 9 9 14 .4313 .7217 .8941 Adj. # at Risk 49.0 45.2 41.6 47.5 BRONCHIOLOALVEOLAR CARCINOMA 4 4 5 4 .5123 .6309 .3994 Adj. # at Risk 51.7 46.1 43.8 50.9 Bronchioalveolar Adenoma/Carcinoma 20 13 14 18 .4408 .7048 .8116 MAMMARY # Evaluated 66 66 66 66 Adj. # at Risk 49.9 46.3 40.8 47.0 Adenocanthoma/Adenocarcinoma 3 3 6 1 .8274 .9334 .1521 Adj. # at Risk 49.5 46.3 39.5 47.0 MAMMARY ADENOACANTHOMA 1 1 3 1 .5167 .7366 .2273 Adj. # at Risk 48.9 45.8 40.6 47.0 MAMMARY ADENOCARCINOMA 2 2 3 0 .9092 1 .4132 Adj. # at Risk 48.9 45.7 39.3 47.0 MAMMARY FIBROADENOMA 1 0 0 0 1 1 1 46 Reference ID: 3401357 plow vsVeh .7247 .4103 .7366 .4839 .4674 .3545 . .4946 .4839 .4894 .4839 .9340 .6059 .9121 .6307 .7366 .6668 1 NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. Table A.2.6. (cont.) Tumor Incidence and Results of Tests in Female Mice Organ/ Incidence Significance Levels tumor Veh Low Med High ptrend phigh pmed __________________________________________________________________vsVeh vsVeh OVARIES # Evaluated 65 66 66 66 Adj. # at Risk 47.9 45.7 39.3 47.0 CYSTADENOMA 1 0 1 0 .7313 1 .7042 Adj. # at Risk 47.9 45.7 39.3 47.0 Cystadenoma/Tubular Adenoma 2 0 1 0 .8618 1 .8416 Adj. # at Risk 47.6 45.8 39.3 47.0 GRANULOSA CELL TUMOUR 0 2 0 0 .7948 . . Adj. # at Risk 47.7 45.7 39.3 47.0 HAEMANGIOMA 1 0 1 0 .7313 1 .7042 Adj. # at Risk 47.6 45.7 39.3 47.0 LEIOMYOMA 1 0 0 0 1 1 1 Adj. # at Risk 48.7 45.7 39.3 47.2 LUTEOMA 7 0 0 3 .6659 .9512 1 Adj. # at Risk 48.7 45.8 39.3 47.2 Luteoma/Granulosa Cell Tumor 7 2 0 3 .7958 .9512 1 Adj. # at Risk 47.6 45.7 39.3 47.0 TUBULAR ADENOMA 1 0 0 0 1 1 1 PANCREAS # Evaluated 66 66 66 66 Adj. # at Risk 48.6 45.8 39.3 47.0 HAEMANGIOMA 0 1 0 0 .7303 . . Adj. # at Risk 48.6 46.0 39.4 47.0 ISLET CELL ADENOMA 2 1 1 0 .9111 1 .8368 PITUITARY # Evaluated 66 66 66 65 Adj. # at Risk 48.6 45.8 39.4 46.8 ADENOMA, PARS DISTALIS 1 3 1 2 .4437 .4839 .6985 SALIVARY GLANDS # Evaluated 66 64 65 65 Adj. # at Risk 48.6 44.5 38.9 47.5 ADENOCARCINOMA 0 0 0 1 .2655 .4947 . SKELETAL MUSCLE # Evaluated 66 66 66 66 Adj. # at Risk 48.6 45.9 39.3 47.0 FIBROSARCOMA 0 1 0 0 .7303 . . Adj. # at Risk 48.6 46.6 39.9 47.3 SARCOMA NOS 0 1 1 2 .1200 .2421 .4483 Adj. # at Risk 48.6 46.8 39.9 47.3 Sarcoma NOS/Fibrosarcoma 0 2 1 2 .1991 .2421 .4483 SKIN # Evaluated 66 66 66 66 Adj. # at Risk 48.8 45.7 39.3 47.0 BASAL CELL CARCINOMA 1 0 0 0 1 1 1 Adj. # at Risk 49.2 45.7 39.7 47.0 FIBROSARCOMA 1 0 1 0 .7256 1 .6928 Adj. # at Risk 49.3 45.7 39.3 47.0 HAEMANGIOSARCOMA 1 0 0 0 1 1 1 Adj. # at Risk 48.6 45.9 39.3 47.0 KERATOACANTHOMA 0 1 0 0 .7303 . . Adj. # at Risk 48.6 45.7 39.3 47.3 SARCOMA NOS 0 0 0 1 .2626 .4947 . Adj. # at Risk 49.2 45.7 39.7 47.3 Sarcoma NOS/Fibrosarcoma 1 0 1 1 .3841 .7421 .6928 47 Reference ID: 3401357 plow vsVeh 1 1 .2365 1 1 1 .9806 1 .4839 .8667 .2843 . .4839 .4894 .2368 1 1 1 .4839 . 1 NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. Table A.2.6. (cont.) Tumor Incidence and Results of Tests in Female Mice Organ/ Incidence Significance Levels tumor Veh Low Med High ptrend phigh pmed __________________________________________________________________vsVeh vsVeh SPLEEN # Evaluated 66 66 66 66 Adj. # at Risk 48.6 46.2 39.3 47.0 HAEMANGIOMA 0 2 0 0 .7906 . . Adj. # at Risk 48.6 45.7 39.3 47.0 HAEMANGIOSARCOMA 0 1 0 0 .7303 . . STOMACH # Evaluated 66 66 66 66 Adj. # at Risk 48.6 45.7 39.3 47.8 SQUAMOUS CELL CARCINOMA 0 0 0 1 .2626 .4947 . Adj. # at Risk 48.6 45.7 39.8 47.0 SQUAMOUS CELL PAPILLOMA 0 0 1 0 .4775 . .4483 Adj. # at Risk 48.6 45.7 39.8 47.8 Sq.Cell Papilloma/Carcinoma 0 0 1 1 .1829 .4947 .4483 Skin + Tail # Evaluated 66 66 66 66 Adj. # at Risk 48.6 45.7 39.3 47.0 Sarcoma NOS/Fibrosarcoma 0 0 0 1 .2584 .4894 . Systemic # Evaluated 66 66 66 66 Adj. # at Risk 49.2 46.2 39.4 47.1 HAEMANGIOMA 4 4 2 1 .9361 .9688 .8370 Adj. # at Risk 49.8 46.0 39.3 47.4 HAEMANGIOSARCOMA 2 2 0 2 .4840 .6756 1 Adj. # at Risk 50.4 46.5 39.4 47.6 Hemangioma/Hemangiosarcoma 6 6 2 3 .8759 .9050 .9372 TAIL # Evaluated 4 1 3 6 Adj. # at Risk 3.5 1.0 1.5 3.8 FIBROSARCOMA 0 0 0 1 .4286 .5000 . Adj. # at Risk 3.5 1.0 1.5 3.8 HAEMANGIOMA 1 0 0 0 1 1 1 UTERINE CERVIX # Evaluated 66 64 65 66 Adj. # at Risk 48.6 44.3 39.8 48.1 ENDOMETRIAL POLYP 0 0 1 2 .0652 .2474 .4483 Adj. # at Risk 48.6 45.2 39.0 47.0 ENDOMETRIAL STROMAL SARCOMA 0 1 0 0 .7288 . . Adj. # at Risk 48.6 45.2 39.8 48.1 Endo.Polyp/Stromal Cell Sarc. 0 1 1 2 .1251 .2474 .4483 Adj. # at Risk 49.2 44.3 39.0 47.0 LEIOMYOSARCOMA 1 0 0 0 1 1 1 Adj. # at Risk 48.6 44.9 39.0 47.0 SCHWANNOMA 0 1 0 0 .7273 . . 48 Reference ID: 3401357 plow vsVeh .2368 .4839 . . . . .6066 .6672 .5597 . . . .4839 .4839 1 .4783 NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. Table A.2.6. (cont.) Tumor Incidence and Results of Tests in Female Mice Organ/ Incidence Significance Levels tumor Veh Low Med High ptrend phigh pmed __________________________________________________________________vsVeh vsVeh UTERUS # Evaluated 66 66 66 66 Adj. # at Risk 48.6 45.7 39.3 47.0 ENDOMETRIAL ADENOCARCINOMA 1 0 1 0 .7284 1 .6985 Adj. # at Risk 48.8 46.1 39.3 47.0 ENDOMETRIAL POLYP 4 3 1 2 .7787 .8882 .9537 Adj. # at Risk 48.6 45.9 39.7 47.0 ENDOMETRIAL STROMAL CELL SARCO 0 1 1 0 .6096 . .4483 Adj. # at Risk 48.8 46.2 39.7 47.0 Endo.Polyp/Stromal Cell Sarco 4 4 2 2 .8196 .8882 .8435 Adj. # at Risk 49.1 45.7 39.4 47.1 HAEMANGIOMA 2 0 1 1 .5585 .8711 .8321 Adj. # at Risk 49.1 45.7 39.3 47.4 HAEMANGIOSARCOMA 1 0 0 2 .1674 .4842 1 Adj. # at Risk 48.7 45.7 39.3 47.0 LEIOMYOMA 1 1 1 2 .2573 .4839 .6985 Adj. # at Risk 48.7 45.7 39.3 47.0 MALIGNANT SCHWANNOMA 1 0 0 0 1 1 1 Uterus(w/cervix) # Evaluated 66 64 65 66 Adj. # at Risk 48.6 44.3 39.0 47.0 ENDOMETRIAL ADENOCARCINOMA 1 0 1 0 .7282 1 .6914 Adj. # at Risk 48.8 44.7 39.8 48.1 ENDOMETRIAL POLYP 4 3 2 4 .4404 .6428 .8435 Adj. # at Risk 48.6 44.5 39.4 47.0 ENDOMETRIAL STROMAL CELL SARCO 0 1 1 0 .6111 . .4483 Adj. # at Risk 48.6 45.2 39.0 47.0 ENDOMETRIAL STROMAL SARCOMA 0 1 0 0 .7288 . . Adj. # at Risk 48.6 45.2 39.8 48.1 Endo.Polyp/Stromal Cell Sarc. 0 1 1 2 .1251 .2474 .4483 Adj. # at Risk 48.8 44.8 39.4 47.0 Endo.Polyp/Stromal Cell Sarco 4 4 2 2 .8261 .8882 .8435 Adj. # at Risk 49.1 44.3 39.0 47.1 HAEMANGIOMA 2 0 1 1 .5612 .8711 .8262 Adj. # at Risk 49.1 44.3 39.0 47.4 HAEMANGIOSARCOMA 1 0 0 2 .1708 .4842 1 Adj. # at Risk 48.7 44.4 39.0 47.0 LEIOMYOMA 1 1 1 2 .2602 .4839 .6914 Adj. # at Risk 49.2 44.3 39.0 47.0 LEIOMYOSARCOMA 1 0 0 0 1 1 1 Adj. # at Risk 48.7 44.3 39.0 47.0 MALIGNANT SCHWANNOMA 1 0 0 0 1 1 1 Adj. # at Risk 48.6 44.9 39.0 47.0 SCHWANNOMA 0 1 0 0 .7273 . . 49 Reference ID: 3401357 plow vsVeh 1 .7645 .4839 .6186 1 1 .7363 1 1 .7451 .4783 .4839 .4839 .5928 1 1 .7305 1 1 .4783 NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. Appendix 3. References Bailer, A. and Portier, C. (1988), “Effects of Treatment-Induced Mortality on Tests for Carcinogenicity in Small Samples”, Biometrics, 44, 4, 417-431. Bieler, G.S., and Williams, R.L. (1993), “Ratio Estimates, the Delta Method, and Quantal Response Tests for Increased Carcinogenicity”, Biometrics, 49, 4, 793-801. Chu, K.C., Ceuto, C., and Ward, J.M. (1981), “Factors in the Evaluation of 200 National Cancer Institute Carcinogen Bioassays”, Journal of Toxicology and Environmental Health, 8, 251-280. Greaves, P. (2007), “Neoplasia of Adrenal Medulla,” In: Histopathology of Preclinical Toxicity Studies. 3rd edition, pp 819. Oxford, UK: Academic Press, Elsevier Inc. Haseman, J. K. (1983), “A Reexamination of False-positive Rates for Carcinogenicity Studies”, Fundamental and Applied Toxicology, 3, 334-339. Jara, A. (2007), “Applied Bayesian Non- and Semi-parametric Inference using DPpackage”, Rnews, 7, 3, 17-26. Lin, K. K. and Ali, M.W. (2006), “Statistical Review and Evaluation of Animal Tumorigenicity Studies”, Statistics in the Pharmaceutical Industry, Third Edition, edited by C.R. Buncher and J.Y. Tsay, Marcel Dekker, Inc. New York. Lin, K. K. and Rahman, M.A. (1998), “Overall False Positive Rates in Tests for Linear Trend in Tumor Incidence in Animal Carcinogenicity Studies of New Drugs”, Journal of Biopharmaceutical Statistics. 8, 1, 1-15. McConnell, E.E., Solleveld, H.A., Swenberg, J.A., and Boorman, G.A. (1986), “Guidelines for Combining Neoplasms for Evaluation of Rodent Carcinogenesis Studies”, Journal of the National Cancer Institute. 76, 283-289. Parola, A, and Jacobs, A. (2010). “Combining Tumors for Statistical Analysis”, online FDA handout. Peto, R., Pike, M.C., Day, N.E., Gray, R.G., Lee, P.N., Parrish, S., Peto, J., Richards, S., and Wahrendorf, J. (1980). “Guidelines for sample sensitive significance tests for carcinogenic effects in long-term animal experiments”, IARC Monographs on the Evaluation of Carcinogenic Risk of Chemicals to Humans, supplement 2: Long term and Short term Screening Assays for Carcinogens: A Critical Appraisal, International Agency for Research Against Cancer, 311-426. 50 Reference ID: 3401357 NDA 205677 Tasimelteon Vanda Pharmaceuticals, Inc. R Development Core Team (2009). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL http://www.R-project.org. Rahman, M.A. and Lin, K.K. (2008), “A Comparison of False Positive Rates of Peto and Poly-3 Methods for Long Term Carcinogenicity Data Analysis Using Multiple Comparison Adjustment Method Suggested by Lin and Rahman”, Journal of Biopharmaceutical Statistics. 18, 949-958. STP Peto Working Group (2002), “Statistical Methods for Carcinogenicity Studies”, Toxicologic Pathology. 30 (3), 403-414. U.S. Department of Health and Human Services (2013), Guidance for Industry Statistical Aspects of the Design, Analysis, and Interpretation of Chronic Rodent Carcinogenicity Studies of Pharmaceuticals (DRAFT GUIDANCE), Center for Drug Evaluation and Research, Food and Drug Administration 51 Reference ID: 3401357 --------------------------------------------------------------------------------------------------------This is a representation of an electronic record that was signed electronically and this page is the manifestation of the electronic signature. --------------------------------------------------------------------------------------------------------/s/ ---------------------------------------------------STEVEN F THOMSON 11/04/2013 Statistical Carcinogenicity Review KARL K LIN 11/05/2013 Concur with review Reference ID: 3401357 STATISTICS FILING CHECKLIST FOR A NEW NDA/BLA NDA Number: 205-677 Applicant: Vanda Drug Name: Tasimelteon NDA/BLA Type: priority Stamp Date: 5/31/2013 On initial overview of the NDA/BLA application for RTF: Content Parameter Yes Index is sufficient to locate necessary reports, tables, data, etc. 2 X ISS, ISE, and complete study reports are available (including original protocols, subsequent amendments, etc.) x Safety and efficacy were investigated for gender, racial, and geriatric subgroups investigated (if applicable). x Data sets in EDR are accessible and do they conform to applicable guidances (e.g., existence of define.pdf file for data sets). 4 NA Comments x 1 3 No IS THE STATISTICAL SECTION OF THE APPLICATION FILEABLE? __Yes______ If the NDA/BLA is not fileable from the statistical perspective, state the reasons and provide comments to be sent to the Applicant. Please identify and list any potential review issues to be forwarded to the Applicant for the 74day letter. Content Parameter (possible review concerns for 74day letter) Yes No NA Comment Designs utilized are appropriate for the indications requested. Review issue Endpoints and methods of analysis are specified in the protocols/statistical analysis plans. Interim analyses (if present) were pre-specified in the protocol and appropriate adjustments in significance level made. DSMB meeting minutes and data are available. Appropriate references for novel statistical methodology (if present) are included. Review issue Safety data organized to permit analyses across clinical trials in the NDA/BLA. Investigation of effect of dropouts on statistical analyses as described by applicant appears adequate. Reference ID: 3327050 X x x Review issue STATISTICS FILING CHECKLIST FOR A NEW NDA/BLA Jingyu (Julia) Luan, Ph.D. Reviewing Statistician Date Kun Jin, Ph.D. Supervisor/Team Leader Date Reference ID: 3327050 --------------------------------------------------------------------------------------------------------This is a representation of an electronic record that was signed electronically and this page is the manifestation of the electronic signature. --------------------------------------------------------------------------------------------------------/s/ ---------------------------------------------------JINGYU J LUAN 06/18/2013 KUN JIN 06/18/2013 Reference ID: 3327050