CENTER FOR DRUG EVALUATION AND RESEARCH APPLICATION NUMBER: 205755Orig1s000 CLINICAL PHARMACOLOGY AND BIOPHARMACEUTICS REVIEW(S) NDA 205 755, Biophannaceutics Assessment BIOPHARMACEUTICS REVIEW Of?ce of New Drug Quality Assessment Application No.: NDA 205755 (Priority Review) Reviewer: Okpo Eradiri, Division: DOP2 Biopharmaceutics Team Leader: Applicant: Novaltis Angelica Dorantes, Trade Name: . Acting Biopharmaceutics Supervisory Zykadla Capsules, 150 mg Lead: Richard Lostritto, . . . 5 Generic Name: e11t1n1b Capsules, 1- 0 mg . 1/8/2014 Date Indication Treatment of patients with W4) metastatic non small cell 11mg cancer (N SCLC Date of Rev1ew: 3/25/2014 have Formulation/ apsule/ Strength 150 mg Route of . Administration Oral SUBMISSIONS REVIEWED IN THIS DOCUMENT . . Date of Primary Review due in Submissmn Dates Informal/Formal DARRTS Consult 11/27/2013 3/25/2014 Type of Submission: 505 Application (NME) . . - Dissolution method and acceptance criteria ey rev1ew pom 5' - Bridging of three manufacturing sites Page 1 of 15 Reference ID: 3477408 NDA 205755, Biopharmaceutics Assessment TABLE OF CONTENTS I) SUMMARY OF BIOPHARMACEUTICS FINDINGS ....................................................................4 1) Dissolution Method and Acceptance Criterion: ...................................................................... 4 3) Bridging of the sites used to manufacture the clinical batches .............................................. 5 II) RECOMMENDATION ..........................................................................................................5 III) QUESTION BASED REVIEW – BIOPHARMACEUTICS EVALUATION ........................................6 A) GENERAL ATTRIBUTES.....................................................................................................6 1 What are the highlights of the chemistry and physico-chemical properties of the drug substance (e.g. solubility) and formulation of the drug product? .............................................6 Drug Substance............................................................................................................................ 6 Drug Product ............................................................................................................................... 7 2 Is there any information on BCS classification? What claim did the applicant make based on BCS classification? What data are available to support this claim?......................................8 3 What is the proposed dissolution method? .....................................................................8 4 What data are provided to support the adequacy of the proposed dissolution method (e.g medium, apparatus selection, etc.)? .......................................................................................8 5 4.1 Dissolution Medium Selection......................................................................................... 8 4.2 Selection of paddle rotation speed ................................................................................. 9 What information is available to support the robustness (e.g. linearity, accuracy, etc.) of the dissolution methodology? ..............................................................................................10 6 What data are available to support the discriminating power of the method? Is the proposed dissolution method biorelavant? What data are available to support this claim?....10 7 Is the proposed dissolution method acceptable? If not, what are the deficiencies? ........11 Page 2 of 15 Reference ID: 3477408 NDA 205755, Biopharmaceutics Assessment B.2. ACCEPTANCE CRITERIA ..................................................................................................11 8 What are the proposed dissolution acceptance criteria for this product? .......................11 9 What data are available to support these criteria? ........................................................12 10 Are the acceptance criteria satisfactory? If not, what are the recommended criteria? Is the setting of the dissolution acceptance criteria based on data from clinical and registration batches? ..............................................................................................................................13 C) DRUG PRODUCT FORMULATION DEVELOPMENT AND BRIDGING ACROSS PHASES.............14 11 What are the highlights of the drug product formulation development? ........................14 12 Are all the strengths evaluated in the pivotal clinical trials? What data are available to support approval of lower strengths? ...................................................................................15 13 Are there any manufacturing changes implemented (e.g. formulation changes, process changes, site change, etc.) to the clinical trial formulation? What information is available to support these changes?........................................................................................................15 Page 3 of 15 Reference ID: 3477408 NDA 205755, Biopharmaceutics Assessment I) SUMMARY OF BIOPHARMACEUTICS FINDINGS This is a rolling NDA submission with Priority Review status and Breakthrough Therapy designation. The drug molecule, an NME, has been granted an Orphan designation. eritinib (also referred to as LDK378 by the Applicant) is an ATP-competitive inhibitor of anaplastic kinase (ALK) activity. The drug product has been formulated as a single-entity immediate-release 150 mg capsule (hard gelatin capsule, size 00) for oral use. Ceritinib is said to inhibit in-vitro and in-vivo autophosphorylation of ALK, ALK- mediated phosphorylation of signaling proteins, and proliferation of ALK- dependent cancer cells. The drug substance is achiral. The de?nitive clinical study in the NDA is a Phase 1 open-label, dose-escalation study that investigated the safety, PK, PD and anti-nunor activity of the drug in patients with tmnors con?rmed to have genetic abnormalities in ALK. The Application also contains a dose-escalation phase of a study in Japanese patients with tumors characterized by genetic alterations in ALK. The Biopharmaceutics review is focused on the evaluation and acceptability of the following: - Adequacy of the dissolution method; - Adequacy of the proposed dissolution acceptance criterion; - Acceptability of the data supporting the bridging of the sites used to manufactlu?e the clinical batches. The electronic links associated with biopharmaceutics are as follows: Pharnmceutical Development: \\cdsesub 1\evsprod\nda205 75 5\0002\m3 \3 2-bodv- data\3 78-hard-gelatin-capsule-O l\3 202 -pharm-dev\pharmaceutical- Speci?cations Table: \\cdsesub 1\evsprod\nda205 75 5\0002\m3 \3 2-bodv-data\3 flu-drug- prod\ldk3 78-hard-gelatin-capsule-O 1 \3 2p5 -contr-diug-prod\3 2p52-analvt- 1) Dissolution Method and Acceptance Criterion: The Applicant?s proposed dissolution method and acceptance criterion are: USP Medium/Temperature Volume Acceptance Criterion Apparatus/RPM 0.01M min The proposed dissolution method is acceptable. However, the proposed acceptance criterion is not acceptable for the following reasons: Page 4 of 15 Reference ID: 3477408 NDA 205 755, Biophannaceutics Assessment (5) (4) Based on the provided dissolution data, an acceptance criterion of 8% at 15 min is recommended. 2) Bridging of the sites used to manufacture the clinical batches The comparative dissolution data of representative batches from three different manufacturing sites exhibited similar in-vitro drug release pro?les. Bridging of the manufacttu?ing sites has therefore been established and is acceptable. II) RECOMMENDATION The ONDQA/Biophannaceutics team has reviewed NDA 205-755 and its amendments submitted 011 March 4. 2014 and March 21. 2014. and the biopharmaceutics data/infonnation acceptable. In a teleconference 011 March 24. 2014. the Applicant accepted recommended acceptance criterion of 3% at 15 min Therefore. from the Biophannaceutics perspective. APPROVAL is for NDA 205-755 for Ceritinib Capsules. 150 mg. The following dissolution method and acceptance criterion should be implemented for release and on stability: Apparatus/RPM Medium Volume Acceptance Criteria USP Apparatus 2 at 60 0.01M pH 1 900 mL 93% at 15min Okpo Eradiri, Ph. D. Angelica Dorantcs, Biophannaceutics Reviewer Biophannaceutics Team Leader Of?ce of New Drug Quality Assessment Of?ce of New Drug Quality Assessment cc: RLostritto Page 5 of 15 Reference ID: 3477408 NDA 205 755, Biophaimaceutics Assessment QUESTION BASED REVIEW BIOPHARMACEUTICS EVALUATION A) GENERAL ATTRIBUTES 1 What are the highlights of the chemistry and physico-chemical properties of the drug substance solubility) and formulation of the drug product? Drug Substance eritinib is an its chemical Structure is displayed in Figlu?e 1. NH Figure 1: Structural formula of Ceritinib 558.14) eritinib is practically insoluble in water (0.02 mg/mL) but vely soluble in methanol (18 mg/mL). The molecule is also soluble in 0.01M - 0.1M hydrochloric acid. The solubility of ceritinib over the physiologic pH range and in different solvents is presented in Table 1. In general, solubility of ceritinib decreases with increasing pH; the pKa values are 9.7 and 4.1. The diug substance is achiral and non-hygroscopic. M4) Page 6 of 15 Reference ID: 3477408 NDA 205755, Biopharmaceutics Assessment Table l: Solubility of ceritinib in different solvents and at different pH values. Solvent Solubility at Water 0.02 Ethanol 4.3 Methanol 18.3 Isopropanol 3.9 Acetonitrile 3.3 acid 800220021 8.4 Water/Acetonitrile 1:1 2.4 0.1N HCI 11.9 0.01N HCI 5.5 0.001N HCI 0.64 pH 4.5 0.03 pH 5.0 0.04 pH 6.0 0.01 pH 6.8 0.01 pH 8.0 0.003 pH 9.0 0.1 Drug Product The proposed product is f01mulated as a 150 mg size 00 hard gelatin capsule, weighing approximately 493 mg. The composition of the drug product is presented in Table 2. Quantitative composition of Ceritinib Capsules. 150 mg. Ingredient Amount per 150 Function Reference to mg Capsule (mg) standards Capsule fill LDK378 150.00 Active Novartis monograph ingredient Cellulose NF Cellulose USP NF Sodium Starch Glycolate USP NF Magnesium Stearate USP NF Colloidal USP NF Capsule fill weight Empty capsule shell, pre-printed Capsule shell (theoretical weight) Novartis monograph Ink black 2 Total capsule weight (approx.) 493.00 -- Page 7 of 15 Reference ID: 3477408 NDA 205755, Biopharmaceutics Assessment 2 Is there any information on BCS classification? What claim did the applicant make based on BCS classification? What data are available to support this claim? The Applicant performed permeability experiments to compare eritinib to mannitol (a low permeability marker) and propranolol (high permeability marker) and classi?ed the API as a low permeability drug. The drug also exhibits low aqueous solubility at neutral to high pH values. The Applicant therefore designates Ceritinib as a BCS Class 4 compormd. The Applicant also designates eritinib as a substrate for the apical ef?ux transporter, P-glycoprotein. B.1. DISSOLUTION INFORMATION 3 What is the proposed dissolution method? The Applicant?s proposed dissolution method testing conditions can be summarized as follows: Apparatus: USP 2 (Paddle) Medium: 900 mL 0.01M Temperature: 37 i 0.5 Rotation speed: 60 W4) Proposed Spec Sampling Time: 8min Analysis: 4 What data are provided to support the adequacy of the proposed dissolution method (e.g medium, apparatus selection, etc.)? 4.1 Dissolution Medium Selection The following media were used to perform dissolution testing of 150 mg eritinib Capsules (batch 1010000958) in USP 2 (paddle): 0.01M . 4 . A rotatron speed of W) was used 0W) Apparatus 2; in addition, while 60 was used for 00(4) UV was used for the paddle. The Applicant 5 Page(s) has been Withheld in Full as b4 immediately following this page Page 8 of 15 Reference ID: 3477408 NDA 205755, Biopharmaceutics Assessment Based on the data submitted with the original Application and the new data sets provided on March 20, 2014, the most appropriate acceptance criterion for this putative formulation of Ceritinib is: at 15 min. C) DRUG PRODUCT FORMULATION DEVELOPMENT AND BRIDGING ACROSS PHASES 1 1 What are the highlights of the drug product formulation development? No rigorous formulation development has been undertaken by the Applicant at this sta of clinical develoiment of the drui iroduct. The manufacturini irocess involves i Page 14 of 15 Reference ID: 3477408 NDA 205755, Biopharmaceutics Assessment (b) (4) Although several strengths were made and used in the early phase of development, only the 150 mg strength is currently submitted for approval. 12 Are all the strengths evaluated in the pivotal clinical trials? What data are available to support approval of lower strengths? The Applicant is seeking approval for only one dosage strength, 150 mg. 13 Are there any manufacturing changes implemented (e.g. formulation changes, process changes, site change, etc.) to the clinical trial formulation? What information is available to support these changes? There are no manufacturing changes to be assessed in the NDA. However, three different sites (one in the United States and two in Switzerland) were used to manufacture the clinical batches. In the February 18, 2014 IR Letter, the following comment was included to request bridging data for the 3 manufacturing sites: The clinical batches of your proposed drug product were manufactured at two manufacturing sites in Switzerland and the United States. Submit to the NDA, comparative dissolution data of representative batches from the two sites demonstrating similarity in dissolution profiles between the two manufacturing sites. In the March 4, 2014 response, the Applicant submitted dissolution data that demonstrate unequivocally that batches of the product manufactured at the three sites showed similar dissolution profiles (Figure 6). (b) (4) Page 15 of 15 Reference ID: 3477408 --------------------------------------------------------------------------------------------------------This is a representation of an electronic record that was signed electronically and this page is the manifestation of the electronic signature. --------------------------------------------------------------------------------------------------------/s/ ---------------------------------------------------OKPONANABOFA ERADIRI 03/25/2014 ANGELICA DORANTES 03/25/2014 Reference ID: 3477408 Clinical Pharmacology NDA Review Type/Category Brand Name Generic Name Receipt Date Target Action Date PDUFA Date Proposed Indication Dosage Form Route of Administration Dosing Regimen and Strength Applicant OCP Division 0ND Division OCP Reviewers OCP Team Leaders/Secondary Reviewers NDA 205755 DSESUB 1\evsprod\NDA205 75 5\0002 NME (Priority) Granted Breakthrough Therapy and inhan Drug Designation ZYKADIA eritinib (LDK3 78) Pan 1: November 27, 2013 Part 2: December 12, 2013 Pan 3: December 24, 2013 April 17, 2014 August 24, 2014 Treatment of Patients with Metastatic Non-Small Cell Lung Cancer (NSC LC) who Have (m4) (4) 150 mg Capsule Oral 750 mg Once Daily Novartis Phaimaceuticals Division of Clinical Pharmacology Division of Oncology Products 2 Ruby Leong, PhaImD. (Clinical Pharmacology Pengfei Song, (Pharmacometrics Yuzhuo Pan, (Physiologically-Based Phaimacokinetic Modeling) Hong Zhao, (C P) Qi Liu, (PM) Ping Zhao, (PBPK) Table of Contents 1 EXECUTIVE SUMMARY 4 1 . 1 RECOMMENDATIONS 4 1.2 PHASE 4 REQUIREMENTS AND COMMITMENTS 6 1.3 SUMMARY OF IMPORTANT CLINICAL PHARMACOLOGY FINDINGS 7 2 QUESTION-BASED REVIEW 10 2.1 GENERAL ATTRIBUTES 10 2.2 GENERAL CLINICAL PHARMACOLOGY 1 1 2.3 INTRINSIC FACTORS 29 NDA 205755/0000 Reference ID: 3477231 Page 1 of112 2.4 2.5 EXTRINSIC FACTORS.......................................................................................................... 32 GENERAL BIOPHARMACEUTICS ......................................................................................... 42 2.6 ANALYTICAL SECTION....................................................................................................... 45 3 DETAILED LABELING RECOMMENDATIONS ...................................................................... 46 4 APPENDICES ................................................................................................................................ 53 4.1 PHARMACOMETRICS REVIEW ............................................................................................ 53 4.2 PHYSIOLOGICALLY-BASED PHARMACOKINETIC REVIEW ................................................. 96 List of Tables Table 1. Solubility of ceritinib.................................................................................................................... 10 Table 2. Summary of clinical pharmacology and clinical studies .............................................................. 11 Table 3. ΔQTcF for ceritinib at doses of 50 to 750 mg from Cycle 2 Day 1 and beyond .......................... 17 Table 4. ΔQTcF for ceritinib at the 750 mg dose ....................................................................................... 18 Table 5. PK parameters following single doses of ceritinib on cycle 1 day 1 during PK-run-in period .... 22 Table 6. PK parameters following repeat doses of ceritinib on cycle 1 day 8 of dose escalation phase ... 23 Table 7. Steady-state trough concentrations of ceritinib ............................................................................ 24 Table 8. PK parameters following 750 mg ceritinib................................................................................... 25 Table 9. Components of plasma, urine, and feces after a single dose of 750 mg [14C]-ceritinib ............... 28 Table 10. Proportion of patients with ORR or any Grade 3-4 AEs based on renal function ..................... 30 Table 11. Proportion of patients with ORR or any Grade 3-4 AEs based on hepatic function .................. 31 Table 12. Inhibition of [14C]-ceritinib metabolism by selective CYP inhibitors ........................................ 33 Table 13. IC50 and calculated R1 values for ceritinib inhibition of ............................................................. 34 Table 14. Ceritinib induction of mean mRNA and activity levels of CYP enzymes ................................. 35 Table 15. Effect of ceritinib on P-gp-mediated Rho123 efflux (left) ......................................................... 36 Table 16. IC50 and calculated R values for ceritinib inhibition of OATP transporters............................... 36 Table 17. IC50 and calculated unbound Cmax/IC50 values for ceritinib inhibition of OAT and OCT transporters ......................................................................................................................................... 37 Table 18. Transport of ceritinib by OCT1, OAT2, OATP1B1, OATP2B1 ............................................... 38 Table 19. Comparative analysis of ceritinib PK parameters ...................................................................... 40 Table 20. Comparative analysis of ceritinib AUC and Cmax ....................................................................... 41 Table 21. PBPK model simulated ceritinib steady-state AUC and Cmax .................................................... 42 Table 22. Representative high-fat (left) and low-fat meals (right) ............................................................. 43 Table 23. PK parameters of a single dose of ceritinib after a high-fat or a low-fat meal compared to fasted conditions ........................................................................................................................................... 44 Reference ID: 3477231 NDA 205755/0000 Page 2 of 112 Table 24. Summary of accuracy and precision of calibration standards and quality controls used in clinical studies .................................................................................................................................... 46 List of Figures Figure 1. Chemical structure of ceritinib .................................................................................................... 10 Figure 2. Relationship between average observed steady-state trough concentration (Ctrough,ss) and ......... 13 Figure 3. The relationship between average observed Css,trough and Grade 3 or worse (G3+) AEs in patients with ALK-positive tumors who were treated with ceritinib .............................................................. 14 Figure 4. Relationship between average observed Ctrough,ss and individual AEs including (a) G3+ AST elevation, (b) G3+ ALT elevation, (c) G2+ hyperglycemia, and (d) G3+ GI tract AEs in patients with ALK-positive tumors in trial X2101 .......................................................................................... 15 Figure 5. Relationship between average observed Ctrough,ss and proportion of patients with dose reduction (left) and dose interruption (right) in patients with ALK-positive tumors in trial X2101 ................. 16 Figure 6. Concentration-ΔQTcF relationship using a linear mixed effects model ..................................... 18 Figure 7. Mean concentration-time profiles after single doses of ceritinib (day 1) ................................... 21 Figure 8. Mean concentration-time profiles after repeat doses of ceritinib (cycle 2 day 1)....................... 21 Figure 9. Mean radioactive dose recovered in urine and feces .................................................................. 26 Figure 10. The metabolism of ceritinib after a single dose of 750 mg [14C]-ceritinib in 6 healthy subjects ............................................................................................................................................................ 27 Figure 11. Effect of PPI coadministration on ceritinib steady-state exposure ........................................... 39 Reference ID: 3477231 NDA 205755/0000 Page 3 of 112 1 EXECUTIVE SUMNIARY (4) (4) The Applicant seeks accelerated approval of ceritinib for the treatment of patients with ma) metastatic non-small cell 11mg cancer (N SCLC) who have The proposed dosing regimen is 750 mg taken once daily (QD) on an empty stomach at least 2 hours before or 2 hours after food. The ef?cacy and safety of ceritinib was established in an open-label, single-arm study that enrolled patients with metastatic ALK-positive NSC LC who progressed while receiving or were intolerant to crizotinib. The objective response rate (ORR) evaluated by a Blinded Independent Central Review Committee (BRC) was 44% (95% CI: 36, 52) and the median duration of response was 7.1 months (95% CI: 5.6, NE) at the recommended dose of 750 mg ceritinib once daily (n=l63). The most common adverse reactions (incidence 225%) were diarrhea, nausea, vomiting, abdominal pain, fatigue, decreased appetite, and constipation. The most common Grade 3-4 adverse reactions (incidence 25%) were alanine transaminase (ALT) elevation, aspartate transaminase (AST) elevation, lipase (blood) increase, diarrhea, and fatigue. Approximately 60% of patients initiating treatment at 750 mg required at least one dose reduction. Gastrointestinal (GI) disorders were reported in 98% of patients receiving ceritinib and resulted in dose modi?cation in 42% of patients. The Clinical Pharmacology Section of the NDA is supported by food effect, ADME (absorption, distribution, metabolism, excretion), and two drug-drug interaction studies in healthy subjects; single and repeat dose phannacokinetics (PK) studies in cancer patients; and in vitro studies to assess the drug interaction potential of ceritinib with cytochrome P450 (CYP) and transporters. Population pharmacokinetic analyses using data from ALK-positive cancer patients enrolled in the registration trial did not identify clinically important covariates in?uencing ceritinib PK. There were no evident exposure-response (E-R) relationships for effectiveness (OR and progression-free survival Higher exposure appears to be associated with higher incidence and earlier safety events including overall Grade 3-4 adverse events (AEs), time to first dose reduction/interruption, and individual AEs such as elevations and No signi?cant relationships were identified between systemic exposure and gastrointestinal (GI) tract AEs, possibly because the high drug concentration in the GI tract leads to GI tract AEs directly. eritinib prolonged the interval in a concentration dependent manner. 1.1 RECOMMENDATIONS The NDA 205755 is acceptable from a clinical pharmacology perspective provided that the Applicant and the FDA come to an agreement regarding the labeling language and the identified clinical pharmacology trials to be conducted as postrnar?keting requirements. The adequacy of the clinical pharmacology program in the overall drug development plan of ceritinib is summarized in the table below. NDA 205755/0000 Page 4 of 112 Reference ID: 3477231 Drug Development Decision Proposed dosing regimen of 750 mg QD in the fasted state Sufficiently Supported? Yes No Refer to Section 2.2.4.4 Recommendations and Comments PMR: Clinical trial to evaluate the gastrointestinal tolerability, efficacy, and PK of 450 mg ceritinib taken with a meal as compared with that of 750 mg ceritinib taken in the fasted state in metastatic ALK-positive NSCLC patients. Refer to Section 1.2.1. Labeling Recommendation: For severe or intolerable nausea or vomiting or diarrhea despite optimal antiemetic or anti-diarrheal therapy, withhold until improved and resume ceritinib with a 150 mg dose reduction OR take with meals as instructed below: Dose adjustment in patients with organ impairment Yes No Refer to Section 2.3.1 Dose adjustment in patients with comedications that affect the PK of ceritinib Yes No Refer to Section 2.4.2.7 and 2.4.2.8 • If receiving 750 mg daily, reduce the dose to 450 mg daily; • If receiving 600 mg daily, reduce the dose to 300 mg daily. Labeling Recommendation: Dose adjustment is not recommended in patients with mild hepatic impairment. The recommended dose has not been determined in patients with moderate and severe hepatic impairment. PMR: Hepatic impairment study. Refer to Section 1.2.1. Labeling Recommendation: CYP3A Inhibitors and Inducers: Avoid concurrent use of strong CYP3A inhibitors and inducers. If concomitant use of a strong CYP3A inhibitor is unavoidable, reduce the ceritinib dose by approximately one-third, rounded to the nearest 150 mg dosage strength. After discontinuation of a strong CYP3A inhibitor, resume the ceritinib dose that was taken prior to initiating the strong CYP3A4 inhibitor. Acid Reducing Agents: Gastric acid reducing agents (e.g., proton pump inhibitors, H2-receptor antagonists, antacids) may alter the solubility of ceritinib and reduce its bioavailability as ceritinib demonstrates pH-dependent solubility and becomes poorly soluble as pH increases in vitro. PMR: Drug-drug interaction study with gastric acid reducing agents. Refer to Section 1.2.1. Dose adjustment in patients with comedications whose PK are affected by ceritinib Reference ID: 3477231 Yes No Refer to Section 2.4.2.3 Labeling Recommendation: Avoid concurrent use of CYP3A or CYP2C9 substrates with narrow therapeutic indices. If concurrent use of these medications is unavoidable, consider dose reduction of these comedications. PMR: Drug-drug interaction studies with CYP3A and CYP2C9 sensitive substrates. Refer to Section 1.2.1. NDA 205755/0000 Page 5 of 112 1.2 1.2.1 PHASE 4 REQUIREMENTS AND COMMITMENTS Postmarketing Requirements (PMR) The Applicant is required to conduct the following clinical pharmacology studies under postmarketing requirements (PMRs). These PMRs will be included in the Approval letter with milestones agreed upon after negotiation with the Applicant. Drug Development Question Rationale PMR What is the exposure-matched dose of ceritinib taken with food to improve GI tolerability and compliance without compromising efficacy? 98% of patients experienced GI adverse reactions. Ceritinib is proposed to be taken in the fasted state, but taking ceritinib with food may alleviate GI toxicities and improve patients’ compliance. Conduct a clinical trial to evaluate the gastrointestinal tolerability, efficacy, and pharmacokinetics of 450 mg ceritinib taken with a meal as compared with that of 750 mg ceritinib taken in the fasted state in metastatic ALK-positive NSCLC patients. Should the dose of ceritinib be reduced in patients with moderate and severe hepatic impairment? 92% of the administered dose is recovered in the feces, indicating that hepatic elimination is the major elimination pathway. Complete a pharmacokinetic trial to determine the appropriate dose of ceritinib in patients with hepatic impairment. What is the effect of gastric acid reducing agents on the PK of ceritinib? Ceritinib’s solubility is pH-dependent, with a solubility of 11 mg/mL at pH=1 and 0.0002 mg/mL at pH=6.8. Conduct a clinical trial to evaluate if proton pump inhibitors, H2receptor antagonists, and antacids alter the bioavailability of ceritinib. The trial results should allow for a determination on how to dose ceritinib with regard to concomitant gastric acid reducing agents. Final Protocol Submission: To be determined Trial Completion: To be determined Final Report Submission: To be determined Final Protocol Submission: Submitted Trial Completion: January 2016 Final Report Submission: June 2016 Final Protocol Submission: January 2015 Trial Completion: August 2015 Final Report Submission: February 2016 What is the effect of ceritinib on the PK of sensitive CYP3A substrates? Ceritinib inhibits CYP3A4 in vitro (R1 value of 7.0-8.5). Conduct a clinical trial evaluating the effect of repeat doses of ceritinib on the single dose pharmacokinetics of midazolam (CYP3A4 substrate). The results of this clinical trial should allow for a determination on how to dose ceritinib with regard to concomitant sensitive CYP3A4 substrates and CYP3A4 substrates with narrow therapeutic indices. Final Protocol Submission: September 2014 Trial Completion: August 2016 Final Report Submission: February 2017 Reference ID: 3477231 NDA 205755/0000 Page 6 of 112 Drug Development Question What is the effect of ceritinib on the PK of sensitive CYP2C9 substrates? Rationale Ceritinib inhibits CYP2C9 in vitro (R1 value of 6.0). PMR Conduct a clinical trial evaluating the effect of repeat doses of ceritinib on the single dose pharmacokinetics of warfarin (CYP2C9 substrate). The results of this clinical trial should allow for a determination on how to dose ceritinib with regard to concomitant sensitive CYP2C9 substrates and CYP2C9 substrates with narrow therapeutic indices. Final Protocol Submission: September 2014 Trial Completion: August 2016 Final Report Submission: February 2017 Signatures: Reviewer: Ruby Leong, Pharm.D. Division of Clinical Pharmacology V Team Leader: Hong Zhao, Ph.D. Division of Clinical Pharmacology V Reviewer: Pengfei Song, Ph.D. Division of Clinical Pharmacology V PM Secondary Reviewer: Qi Liu, Ph.D. Division of Clinical Pharmacology V Reviewer: Yuzhuo Pan, Ph.D. Division of Pharmacometrics PBPK Secondary Reviewer: Ping Zhao, Ph.D. Division of Pharmacometrics Division Director: NAM Atiqur Rahman, Ph.D. Division of Clinical Pharmacology V Cc: DOP2: DCPV: RPM – K Boyd; MO – G Blumenthal; MTL – S Khozin DDD – B Booth; Office of Clinical Pharmacology Director – I Zineh A mid-cycle Clinical Pharmacology briefing was presented by Ruby Leong, Yuzhuo Pan, and Pengfei Song on February 21, 2014 with the following participants: Brian Booth, Julie Bullock, Vicky Hsu, Sean Khozin, Fang Li, Qi Liu, Nitin Mehrotra, Young-Jin Moon, Olanrewaju Okusanya, NAM Atiqur Rahman, Vikram Sinha, Yaning Wang, Gene Williams, Ping Zhao, and Issam Zineh. 1.3 SUMMARY OF IMPORTANT CLINICAL PHARMACOLOGY FINDINGS Unresolved Dosing Issue with Regard to GI Tolerability The proposed dosing regimen for ceritinib with regard to food is 750 mg daily on an empty stomach at least 2 hours before or 2 hours after food. At the recommended dose under fasted conditions, the majority of patients experienced gastrointestinal (GI) adverse reactions including diarrhea (86%), nausea (80%), vomiting (60%), and abdominal pain (54%). The absolute bioavailability of ceritinib has not been studied but is expected to be low. The food effect study Reference ID: 3477231 NDA 205755/0000 Page 7 of 112 showed that a high-fat, high-calorie meal increased ceritinib AUC by 73% and Cmax by 41%; a low-fat meal increased ceritinib AUC by 58% and Cmax by 43% as compared to fasted conditions. Administration of ceritinib at 750 mg with food may improve GI tolerability and compliance, but could lead to exposure-related toxicities such as AST/ALT elevations, hyperglycemia, and QTc (b) (4) prolongation. The Applicant proposed to conduct a clinical trial to evaluate the effect of a (b) (4) on the PK of ceritinib in healthy subjects based on the hypothesis that the (b) (4) magnitude of increase in ceritinib exposure would be smaller with than that observed with low-fat or high-fat full meals while alleviating GI adverse events. FDA recommends determination of an exposure-matched dose of ceritinib taken with food to improve GI tolerability and compliance in patients without compromising efficacy. A postmarketing trial should be conducted to evaluate the GI tolerability, efficacy, and PK of 450 mg ceritinib taken with a meal as compared with that of 750 mg ceritinib taken in the fasted state in metastatic ALK-positive NSCLC patients. Exposure-Response Relationships No significant exposure-response relationships were identified for the primary efficacy endpoint ORR and secondary efficacy endpoint PFS in ALK-positive NSCLC patients who received prior treatment with an ALK inhibitor. Based on the analysis results, it is unclear whether the plateau of the exposure-efficacy curve has been reached or not. Higher systemic exposure appears to be associated with more frequent and earlier overall Grade 3 or worse (G3+) adverse events (AEs), as well as with higher incidence of individual AEs such as G3+ ALT/AST elevation, and G2+ hyperglycemia. No significant relationships were identified between systemic exposure and gastrointestinal (GI) tract AEs, possibly because the high drug concentration in the GI tract leads to GI tract AEs directly. Higher systemic exposure also appears to be associated with earlier and more frequent dose reductions or dose interruption. Given that the permanent discontinuation due to AEs occurred in only 10% patients, the management of AEs via dose reductions and interruptions is effective in maintaining patients on study drug. Population concentration-QTc analyses using time-matched ECG and PK data from the clinical trial showed that ceritinib prolonged the QTc interval in a concentration dependent manner. Following repeat daily doses of 750 mg ceritinib, large changes (i.e., >20 ms) in the QTc interval were detected at steady-state (Cycle 2 Day 1 and beyond). The largest mean change from baseline was 19.3 ms with the upper bound of the 2-sided 90% confidence interval (CI) of 22.2 ms, observed at Cycle 6 Day 1. ADME and Potential DDIs Following oral administration of ceritinib, Cmax was reached at 4-6 hours, with a terminal halflife (t1/2) of 41 hours. Inter-patient variability following repeat doses of 750 mg ceritinib is 74% for AUC and 76% for Cmax. Ceritinib exhibits nonlinear time-dependent PK, with exposures that are dose proportional after a single dose, but greater than dose proportional after repeat doses in the dose range of 50 to 750 mg. Following ceritinib 750 mg QD, steady state is achieved within Reference ID: 3477231 NDA 205755/0000 Page 8 of 112 15 days with 6-fold accumulation. Apparent clearance (CL/F) at steady-state (33 L/h) is lower than that after a single dose (89 L/h), likely due to auto-inhibition of CYP3A. Ceritinib is primarily metabolized by CYP3A, and is also a reversible and time-dependent inhibitor of CYP3A in vitro. Coadministration of ketoconazole (a strong CYP3A inhibitor) 200 mg twice daily for 14 days and a single dose of 450 mg ceritinib increased ceritinib AUC by 2.9fold and Cmax by 20% in healthy subjects. Coadministration of rifampin (a strong CYP3A inducer) 600 mg daily for 14 days and a single dose of 750 mg ceritinib decreased ceritinib AUC by 70% and Cmax by 44% in healthy subjects. Given that the drug-drug interaction (DDI) studies were conducted using single doses of ceritinib, the effect of strong CYP3A modulators on the steady-state PK of ceritinib in patients is unclear due to nonlinear, time-dependent PK of ceritinib. Physiologically-based pharmacokinetic modeling (PBPK) predicted that ketoconazole can increase the steady-state ceritinib exposure (AUC) by 51%; rifampin can decrease steadystate ceritinib AUC by 67%. Therefore, it is recommended to avoid concomitant use of strong CYP3A inhibitors and inducers given the magnitude of exposure change in the context of the exposure-response curves for safety and effectiveness observed with ceritinib. If avoiding concomitant strong CYP3A inhibitors is not possible, the ceritinib dose should be reduced by approximately one-third, rounded to the nearest 150 mg dosage strength. After discontinuation of a strong CYP3A inhibitor, the ceritinib dose that was taken prior to initiating the strong CYP3A4 inhibitor should be resumed. In addition to CYP3A, ceritinib also inhibits CYP2C9 in vitro. PMRs will be requested to evaluate the effect of repeat doses of ceritinib on the single dose PK of a sensitive CYP3A4 substrate (e.g., midazolam) and CYP2C9 substrate (e.g., warfarin). Given that the aqueous solubility of ceritinib is pH-dependent with lower solubility (<750 mg/250 mL) at higher pH, gastric acid reducing agents may decrease the solubility of ceritinib and subsequently reduce its bioavailability. A PMR will be requested to evaluate the effect of gastric acid reducing agents on the PK of ceritinib. Organ Dysfunction Studies The ADME study showed that hepatic elimination is the major route of elimination (92% of the administered radiolabeled dose was recovered in the feces), indicating that hepatic impairment may increase the systemic exposure of ceritinib. The Applicant is requested to submit the results of an ongoing hepatic impairment study under a PMR to determine the appropriate ceritinib dose given that AEs including AST/ALT elevations, QTc prolongation, and hyperglycemia are exposure-related. No dose adjustment is recommended for mild hepatic impairment because PK (popPK estimated steady-state AUC), effectiveness (ORR), and safety (overall Grade 3-4 AEs) are similar between patients with mild hepatic impairment and patients with normal hepatic function. The ADME study showed that the renal pathway contributed minimally to ceritinib elimination (1.3% of the administered radiolabeled dose was recovered in the urine). Dose adjustments are not recommended for patients with mild or moderate renal impairment because there are no clinically important differences in PK (popPK estimated steady-state AUC), effectiveness Reference ID: 3477231 NDA 205755/0000 Page 9 of 112 (ORR), and safety (overall Grade 3-4 AEs) between patients with mild or moderate renal impairment and patients with normal renal Patients with severe renal impairment were not enrolled in the clinical study. 2 QUESTION-BASED REVIEW 2.1 GENERAL ATTRIBUTES 2.1.1 What are the highlights of the chemistry and physical-chemical properties of the drug substance and the formulation of the drug product? The molecular weight of ceritinib is 558 g/mol (Figure 1). The proposed drug product is available as 150 mg capsules. Figure 1. Chemical structure of ceritinib CH (X. o=s~L=o 0 cu, 3 NH H,c cu, cu3 Ceritinib solubility in aqueous media is pH-dependent with decreased solubility at higher pH (Table Ceritinib is completely soluble at acidic pH=l (dose/250 mL=3 mg/mL), but 55,000- fold less soluble at pH=6.8. Therefore, an in Vivo study to assess the effect of gastric acid reducing agents on the PK of ceritinib is warranted and will be requested as a PMR (refer to Section 1.2.1). Table l. Solubility of ceritinib Solvent Solubility (mg/mL) Water 021 Acidic media (pH=1) 11 Neutral or alkaline media (pH=6.8) 00002? (4) Source: Summary of Biopharmaceutic Studies and Associated Analytical Methods. Table 1-1. Page 5. Ceritinib showed low permeability across Caco-2 cells (refer to Section 2.5.1). The Applicant claims that ceritinib is a BC Class 4 compormd with low permeability and low solubility. 2.1.2 What are the proposed mechanism(s) of action and therapeutic indication(s)? Ceritinib is an inhibitor of anaplastic kinase (ALK). ALK translocation leads to expression of a fusion protein, resulting in aberrant ALK signaling and proliferation of ALK- dependent cancer cells. In the majority of cases, EML4 is the translocation partner for NBA 205755/0000 Page 10 of 112 Reference ID: 3477231 ALK. Ceritinib inhibited EML4-ALK kinase activity in a cell line (H2228) (ICso: 11 nM) in vitro and induced tumor regression in H2228 derived xenografts in mouse and rat. The proposed indication is for the treatment of patients with small cell lung cancer (N SC LC who have 4 . (m metastatic non- (4) 2.1.3 What are the proposed dosage(s) and route(s) of administration? The Applicant?s proposed dosing regimen is 750 mg orally once daily (QD) on an empty stomach at least 2 horu?s before or 2 hours after food. 2.2 GENERAL CLINICAL PHARMACOLOGY 2.2.1 What are the design features of the clinical pharmacology and clinical studies used to support dosing or claims? Table 2 lists the relevant clinical pharmacology and clinical studies included in the application. Table 2. Summary of clinical pharmacology and clinical studies Study Study Study Assessment Dosing regimen Number Design Population A2105 Open label. single Healthy subjects ADME 750 mg [MG-ceritinib oral dose capsule containing 300 pCi radioactiw'ty A2101 Open label. Healthy subjects (n=28) Food effect Single 500 mg dose in the fasted randomized. with FMI state or administered with a high- crossover with 7-day capsule fat. high-calorie meal or low-fat washout meal A2104 Open label. single Healthy subjects (n=l9) DDI Single 450 mg dose and in dose combination with ketoconazole A2106 Open label. single Healthy subjects (n=l9) DDI Single 750 dose and in dose combination with rifampin X1101 Open label dose Japanese cancer patients PK in Single and multiple-dose: 300. escalation and with genetic alterations of Japanese 450. 600 and 750 mg QD expansion phase ALK (n=l9) patients X2101 Open label dose ALK-positive cancer Ef?cacy. Single and multiple-dose: 50. escalation and patients safety. popPK 100. 200. 300. 400. 500. 600. 700 expansion phase (n=59 for dose escalation and 750 mg QD phase: n=245 for expansion phase) ADME: Absorption. distribution. metabolism. excretion: ALK: Anaplastic kinase: DDI: Drug-drug interaction: FMI: Final market image: PK: Phannacokinetics: Population pharmacokinetics: QD: Once daily In addition, ceritinib plasma concentration data from Study X2101 were used to develop a model to assess the potential in?uence of covariates on inter-patient variability in ceritinib PK parameters, and explore exposure-response relationships for ef?cacy and safety endpoints. NDA 205755/0000 Page 11 of 112 Reference ID: 3477231 2.2.2 What is the basis for selecting the response endpoints or biomarkers and how are they measured in clinical pharmacology and clinical studies? The primary efficacy outcome measures of the registration trial X2101 were overall response rate (ORR) according to Response Evaluation Criteria in Solid Tumors (RECIST) as evaluated by investigators and a Blinded Independent Central Review Committee (BIRC). Duration of response (DOR) and progression-free survival were additional outcome measures. 2.2.3 Are the active moieties in the plasma (or other biological fluid) appropriately identified and measured to assess pharmacokinetic parameters and exposureresponse relationships? Yes. Ceritinib was the major component in human plasma after oral administration and it was appropriately identified and measured to assess PK parameters (refer to Section 2.6). 2.2.4 Exposure-response 2.2.4.1 What are the characteristics of the exposure-response relationships (dose-response, concentration-response) for efficacy? The Pharmacometrics review concluded that there were no significant exposure-response (E-R) relationships for the primary efficacy endpoint ORR and for the secondary efficacy endpoint PFS (Figure 2), based on data from Study X2101 that included dose levels ranging from 50 mg to 750 mg in patients with ALK-positive NSCLC (n=167) who have received treatment of an ALK inhibitor (refer to Pharmacometrics review in Section 4.1). Based on the ORR analysis results, it is unclear whether the plateau of the exposure-efficacy curve has been reached, which may be due to two possible reasons: (1) the distribution of the available exposure data (primarily driven by the data from patients who received 750 mg daily dose) may not be enough to adequately characterize the full exposure-efficacy curve for ORR, and (2) the small sample size may limit the robustness for the predicted exposure-efficacy relationship. Furthermore, there was no clear trend observed with PFS curves stratified by Css,trough quartiles. Reference ID: 3477231 NDA 205755/0000 Page 12 of 112 BEST AVAILABLE COPY Figure 2. Relationship between average observed steady-state trough concentration (Cmgm) and proportion of patients with ORR (left) and PFS (right) strati?ed by Cums. quartiles in ALK-positive patients (n=l67) who received prior treatment with an ALK inhibitor 1.0 -- Loan: minute-n Ohlche?t-l?-Df?cn .. ?so. an . I - 04 (1141-2432 ngImL. n=42) a 0.8 06 I I 0.4 Putnam,? of PFS r? Own" Run: 'm ?7116? ?00 "-01 '350 "50 x10 33? 30? 11m (Months) On the left ?gure. the solid black symbols represent the observed ORR per central radiology review in each quartile of average observed Cssmugh for all subjects. regardless of the prognostic factors. The vertical black bars represent the 95% con?dence interval (CI). The solid line and the dashed lines represent the multivariate logistic regression model predicted mean and 95% CI of the probability of ORR by average observed for Caucasian patients with brain metastasis. ECOG status of l. and a baseline sum of longest diameter of 8.1 cm. (Please note that the predictions are not for all patients. therefore do not ?t the solid black symbols which represent the observed ORR in each Cwmugh quartile for all patients. regardless of the prognostic factors. The logistic regression model prediction is generated as follows: numeric covariates at the median. categorical covariates at the highest frequency level). The exposure range in each quartile of is denoted by the horizontal blue line along with the number of responders/total number of patients in each quartile. The right ?gure shows Kaplan-Meier curve of progression free survival (PFS) per central radiology review by Cssmugh quartiles in patients who were previously treated with ALK inhibitors. Source: Phannacometrics review. Figures 1 and 2. 2.2.4.2 What are the Characteristics of the exposure-response relationships (dose-response, concentration-response) for safety? The Pharmacometrics review concluded that higher systemic exposure appears associated with more frequent and earlier overall Grade 3/4 AEs (Figure 3). There were also higher incidences of individual AEs such as Grade 3/4 aspartate aminotransferase (AST) and alanine aminotransferase (ALT) elevations, and Grade 2 or worse with increased ceritinib systemic exposure (Figure 4a, 4b, 4c). Based on these E-R analyses for safety, reducing the dose may decrease certain exposure-related toxicities such as elevations and No signi?cant E-R relationships were identi?ed between Cssmugh and overall Grade 3-4 gastrointestinal (GI) tract AEs (Figure 4d) or Grade 2 or worse diarrhea, possibly because plasma concentration is not a good predictor for GI tract AEs as high drug concentrations in the GI tract may lead to GI tract AEs directly (refer to Pharmacometrics review in Section 4.1). Dose reduction due to severe or intolerable GI toxicity occmred in the clinical trial and is recommended in the proposed product labeling (refer to Section 3). NBA 205755/0000 Page 13 of 112 Reference ID: 3477231 BEST AVAILABLE COPY 1m 00? "women Chum: pm onto (J Figure 3. The relationship between average observed and Grade 3 or worse AEs in patients with ALK-positive tumors who were treated with ceritinib Q- 02012349 Ln=57 - 03 041155443211?an Q6 P'c: olaahenh mm owl-i 3- In m: mundane-3 AE. :g .0 .. 31:57 mm m: 43.55 In the le? ?gure. the solid black symbols represent the observed incidence of 63+ AEs in each quartile of average observed for all subjects. regardless of the prognostic factors. The vertical black bars represent the 95% con?dence interval (CI). The solid and dotted lines represent the multivariate logistic regression model predicted mean and 95% CI of incidence of 63+ AE of concern by average observed Camus? for Caucasian patients with brain metastasis. ECOG status of l. and prior treatment with ALK inhibitors. The exposure range in each quartile of average observed Css,urough is denoted by the horizontal blue line along with the number of patients who experienced AE of concern/total number of patients in each quartile. The right ?gure shows Kaplan-Meier curve of time to ?rst ABS by quartiles of observed average Cssmugh in patients with ALK-positive tumors who were treated with ceritinib. Source: Pharmacometrics review. Figures 3 and 4. NDA 205755/0000 Page 14 of 112 Reference ID: 3477231 BEST AVAILABLE COPY Figure 4. Relationship between average observed Ctrough,ss and individual AEs including (a) G3+ AST elevation, (b) G3+ ALT elevation, (c) G2+ hyperglycemia, and (d) G3+ GI tract AEs in patients with ALK-positive tumors in trial X2101 (a) (b) (c) (d) In each figure above, the solid black symbols represent the observed incidence of AEs of concern in each quartile of average observed Css,trough for all subjects, regardless of the prognostic factors. The vertical black bars represent the 95% confidence interval (CI). The solid and dotted lines represent the multivariate logistic regression model predicted mean and 95% CI of incidence of AE of concern by average observed Css,trough for Caucasian patients with brain metastasis, ECOG status of 1, and prior treatment with ALK inhibitors. The exposure range in each quartile of average observed Css,trough is denoted by the horizontal blue line along with the number of patients who experienced AE of concern/total number of patients in each quartile. Source: Pharmacometrics review, Figures 5, 6, and 7. Results of logistic regression analyses also suggested that higher exposure may be associated with more frequent dose reductions and dose interruptions (Figure 5). Higher exposures also appear to be associated with earlier time to first dose reduction and interruption as shown by Kaplan-Meier analyses (Figure 5). Reference ID: 3477231 NDA 205755/0000 Page 15 of 112 9U 80 r31] 40 Propanlun ornaments mm dus: redunluntm v. '20 Figure 5. Relationship between average observed Chough,SS and proportion of patients with dose reduction (left) and dose interruption (right) in patients with ALK-positive tumors in trial X2101 LOCISUC IEGIESSIDD 0 30 Ell/61' 3516} 42/62 0 l] 25D 500 750 1000 125i] 1500 1750 21100 3350 2500 WOUGMNIITDLJ c: 1 - 02 (5994372 n=62) ?1 . 1* 04(11652432mm62d-w-vH-?Tine (Marlins) Source: Pharmacometrics review. Figures 9 and 10. Proportion ufpancnm with dose adapts) 100 LOGISUC 0 Observed (95$ Cl) 4018-3 431? 1?50 75-0 1'350 1500 2000 ?350 2?50!) 0h served mumssemmea??*l1lL Tlne(Months) In the upper panel. the solid black symbols represent the observed incidence of dose reduction (left. upper) or interruption (right, upper) in each quartile of average observed in for all subjects. regardless of the prognostic factors. The vertical black bars represent the 95% con?dence interval (CI). The solid and dotted lines represent the multivariate logistic regression model predicted mean and 95% CI of incidence of dose reduction or interruption by average observed for Caucasian patients with brain metastasis. ECOG status of and prior treatment with ALK inhibitors. The exposure range in each quartile of average observed is denoted by the horizontal blue line along with the number of patients who experienced dose reduction or interruption/total number of patients in each quartile. Lower panel ?gures show Kaplan-Meier curves of time to ?rst dose reduction (left. lower) and dose interruption (right. lower) by quartiles of average observed Cssmugh in patients with ALK-positive tumors who were treated with ceritinib Reference ID: 3477231 NDA 205755/0000 Page 16 of 1 12 2.2.4.3 Does this drug prolong the QT or QTc interval? In Study X2101, time-matched ECG and PK samples were collected at pre-dose, 4, 8, and 24 hours post-dose on days 1 and 8 of cycle 1, pre-dose on day 1 of cycles 2-6, and end of treatment during the dose escalation phase; pre-dose, 4, 8, and 24 hours post-dose on day 1 of cycle 1, predose on day 1 of cycles 2-6, and end of treatment during the expansion phase. The relationship between ΔQTcF and ceritinib concentrations was determined by pooling data from Cycle 2 Day 1 and beyond in patients who received doses of 50 mg to 750 mg (Table 3). Table 3. ΔQTcF for ceritinib at doses of 50 to 750 mg from Cycle 2 Day 1 and beyond Source: QT-IRT Review, Table 2. N=304 patients. The QT-IRT review concluded that ceritinib prolonged the QTc interval in a concentrationdependent manner based on population concentration-QTc analyses using a linear mixed effects model (Figure 6). Reference ID: 3477231 NDA 205755/0000 Page 17 of 112 Figure 6. Concentration-ΔQTcF relationship using a linear mixed effects model Source: QT-IRT review, Figure 3. Following repeat daily doses of 750 mg ceritinib, large changes (i.e., >20 ms) in the QTc interval were detected at steady-state (Cycle 2 Day 1 and beyond). The largest mean change from baseline was 19.3 ms with the upper bound of the 2-sided 90% confidence interval (CI) of 22.2 ms, observed at Cycle 6 Day 1 (Table 4). Table 4. ΔQTcF for ceritinib at the 750 mg dose Treatment Visit 750 mg Cycle 1 Day 1 Cycle 2 Day 1 Cycle 3 Day 1 Cycle 4 Day 1 Cycle 5 Day 1 Cycle 6 Day 1 Time (h) 4 8 24 0 0 0 0 0 N Mean 236 215 222 216 184 169 149 138 1.0 1.5 2.0 18.7 16.5 17.7 18.7 19.3 90% CI for Mean (-0.6, 2.6) (-0.2, 3.2) (0.3, 3.7) (16.7, 20.6) (14.1, 18.8) (15.3, 20.2) (15.9, 21.4) (16.5, 22.2) Source: QT-IRT Review, Table 1. N=255 patients. It is recommended in the labeling for patients with QTc interval >500 ms on at least 2 separate ECGs, to withhold ceritinib until QTc interval <481 ms or recovery to baseline; if baseline QTc is >481 msec, resume with a 150 mg dose reduction. Reference ID: 3477231 NDA 205755/0000 Page 18 of 112 2.2.4.4 Is the dose and dosing regimen selected by the applicant consistent with the known relationship between dose-concentration-response, and is there any unresolved dosing or administration issue? The Pharmacometrics review concluded that based on currently available data, the proposed dose of 750 mg QD is acceptable. The results of E-R analyses for efficacy did not show a clear relationship between systemic exposure and ORR or PFS (refer to Section 2.2.4.1). However, the ORR of 44% (95% CI: 36, 52) and median duration of response of 7.1 months (95% CI: 5.6, NE) suggested that the proposed dose of 750 mg is effective. Higher systemic exposure is associated with more frequent and earlier overall Grade 3-4 AEs, and higher incidence of individual AEs including Grade 3-4 AST/ALT elevations and Grade 2 or worse hyperglycemia. Higher systemic exposure is also associated with earlier and more frequent dose reductions or dose interruptions (refer to Section 2.2.4.2). Given that permanent discontinuation due to AEs occurred in only 10% of patients, the proposed dose of 750 mg with management of AEs via dose reductions or interruptions appears acceptable. Although the current proposed dosing regimen is acceptable, there may be alternative ways to administer ceritinib that may lead to better tolerability and compliance. At the recommended dose of 750 mg QD, the majority of patients experienced GI adverse reactions including diarrhea (86%), nausea (80%), vomiting (60%), and abdominal pain (54%). No significant E-R relationships were identified between systemic exposure and overall Grade 3/4 GI tract AEs or Grade 2 or worse diarrhea, possibly because systemic exposure is not a good predictor for GI tract AEs as high drug concentrations in the GI tract may lead to GI tract AEs directly. The absolute bioavailability of ceritinib has not been studied, but is expected to be low. The food effect study showed that a high-fat, high-calorie meal increased ceritinib AUC by 73% and Cmax by 41%; a low-fat meal increased ceritinib AUC by 58% and Cmax by 43% as compared to fasted conditions (refer to Section 2.5.3). Administration of ceritinib 750 mg with food may improve GI tolerability and compliance, but could lead to exposure-related toxicities such as AST/ALT elevations, hyperglycemia, and QTc prolongation (refer to Sections 2.2.4.2 and 2.2.4.3). To address this unresolved issue, the Applicant proposed to conduct a clinical trial to evaluate the (b) (4) effect of a on the PK of ceritinib in healthy subjects based on the hypothesis that the magnitude of increase in ceritinib exposure would be smaller with (b) (4) concomitant intake of a than that observed with low-fat or highfat full meals while alleviating GI adverse events. FDA recommends a postmarketing trial be conducted to evaluate the GI tolerability, efficacy, and PK of 450 mg ceritinib taken with a meal as compared with that of 750 mg ceritinib taken in the fasted state in metastatic ALK-positive NSCLC patients. Given an expected increase of exposures by 58% to 73% when ceritinib is administered with meals, exposures following a 450 mg dose of ceritinib taken with a meal would be similar to exposures following 750 mg ceritinib taken in the fasted state. Additional unresolved dosing and administration issues with ceritinib to be addressed under PMRs include potential dose adjustment in patients with hepatic impairment and in patients receiving acid reducing agents. Reference ID: 3477231 NDA 205755/0000 Page 19 of 112 2.2.5 What are the PK characteristics of the drug? 2.2.5.1 What are the single dose and multiple dose PK parameters? The single dose PK of ceritinib have been evaluated in four clinical pharmacology studies in healthy subjects (A2101, A2104, A2105, A2106), in patients in Study X2101, and in Japanese patients (X1101). The multiple-dose PK of ceritinib has been evaluated in patients in Study X2101, and in Japanese patients (X1101). Ceritinib exhibits nonlinear time-dependent PK, with exposures that are dose proportional after a single dose, but greater than dose proportional after repeat doses in the dose range of 50 to 750 mg. Following ceritinib 750 mg QD, steady state is achieved by approximately 15 days with 6fold accumulation after three weeks. Apparent clearance (CL/F) at steady-state (33 L/h) is lower than that after a single dose (89 L/h), likely due to auto-inhibition of CYP3A. The terminal halflife (t1/2) at the 750 mg dose is 41 hours. During the dose escalation phase in Study X2101, intensive PK samples were collected at predose, 0.5, 1, 2, 3, 4, 6, 8, 24, 48, 72 hours post-dose during PK run-in; at pre-dose, 0.5, 1, 2, 3, 4, 6, 8, 24 hours post-dose on cycle 1 day 8; at pre-dose on day 15 of cycles 1 and 2; day 1 of cycle 3 and subsequent cycles; at pre-dose, 1, 2, 4, 6, 8, and 24 hours post-dose on cycle 2 day 1 in only those patients enrolled in the 50, 100, and 200 mg dose levels. During the expansion phase, intensive PK samples were collected at pre-dose, 1, 2, 4, 6, 8, 24 hours post-dose on day 1 of cycles 1 and 2; pre-dose on day 15 of cycles 1 and 2 and day 1 of cycles 3 and 4. Mean concentration-time profiles after a single dose of 50 to 750 mg ceritinib are shown in Figure 7 and after 750 mg QD doses in Figure 8. Reference ID: 3477231 NDA 205755/0000 Page 20 of 112 Figure 7. Mean concentration-time profiles after single doses of ceritinib (day 1) Source: Summary of Clinical Pharmacology Studies, Figure 2-2, Page 22. Figure 8. Mean concentration-time profiles after repeat doses of ceritinib (cycle 2 day 1) Source: Summary of Clinical Pharmacology, Figure 2-3, Page 23. Reference ID: 3477231 NDA 205755/0000 Page 21 of 112 Single dose and multiple dose PK parameters of ceritinib were determined using noncompartmental analysis and summarized in Table 5 and Table 6, respectively. Table 5. PK parameters following single doses of ceritinib on cycle 1 day 1 during PK-run-in period of dose escalation phase Source: Summary of Clinical Pharmacology, Table 2-1, Page 24. Reference ID: 3477231 NDA 205755/0000 Page 22 of 112 Table 6. PK parameters following repeat doses of ceritinib on cycle 1 day 8 of dose escalation phase Source: Summary of Clinical Pharmacology, Table 2-2, Page 25. Reference ID: 3477231 NDA 205755/0000 Page 23 of 112 Steady-state trough concentrations of ceritinib by dose are summarized in Table 7. Table 7. Steady-state trough concentrations of ceritinib Dose Cycle 2 day 1 pre-dose ceritinib Average steady-state trough concentrations (ng/mL) concentrations (ng/mL) 50 16.2 15.1 (11.2) n=1 n=2 100 30.3 43.6 (50.0) n=1 n=2 200 88.0 (300) 149 (213) n=2 n=3 300 418 (119) 414 (113) n=3 n=3 400 466 (92.4) 463 (77.5) n=10 n=14 500 524 (35.3) 680 (50.7) n=7 n=8 600 646 (88.6) 733 (57.6) n=7 n=9 700 417 (267) 1231 (34.2) n=2 n=3 750 773 (66.0) 871 (46.5) n=207 n=203 Source: apkconc.xpt; values are for geometric mean (CV% of geometric mean) where applicable Average steady-state trough concentration is defined as the geometric mean of all evaluable steady-state trough concentrations for each patient. Reference ID: 3477231 NDA 205755/0000 Page 24 of 112 PK parameters following 750 mg ceritinib on day 1 of cycles 1 and 2 of the expansion phase are summarized in Table 8. Table 8. PK parameters following 750 mg ceritinib on day 1 of cycles 1 and 2 of the expansion phase Source: Summary of Clinical Pharmacology, Table 2-3, Page 26. 2.2.5.2 How does the PK of the drug and its major active metabolites in healthy volunteers compare to that in patients? After oral administration of ceritinib capsules, median Tmax ranged from 4 to 6 hours in patients across 400 to 750 mg dose groups, and approximately 6 to 8 hours in healthy subjects. The mean terminal t1/2 ranged from 31 to 41 hours across 400 to 750 mg dose groups in patients, and 36 to 48 hours across 450 to 750 mg dose groups in healthy subjects. 2.2.5.3 What are the characteristics of drug absorption? Following oral administration of ceritinib, Cmax was reached approximately 4-6 hours under fasting conditions. Administration of a single 500 mg dose of ceritinib with a high-fat, highcalorie meal in healthy subjects resulted in a 73% increase in AUC and 41% increase in Cmax as compared with fasted conditions while a low-fat meal increased AUC by 58% and Cmax by 43% as compared to fasted conditions. An absolute bioavailability study was not conducted. Based on the mean percentage of the dose recovered as metabolites in the excreta in the ADME study [A2105], a lower limit on the extent of oral absorption is estimated to be approximately 25%. Given that apparent passive permeability of ceritinib is low and ceritinib is a substrate for P-glycoprotein (P-gp), ceritinib Reference ID: 3477231 NDA 205755/0000 Page 25 of 112 absorption is likely limited. 2.2.5.4 What are the characteristics of drug distribution? The geometric mean apparent volume of distribution (Vd/F) ranged from 1990 to 6230 L across 400 to 750 mg dose groups. Ceritinib is 97.2% bound to human plasma proteins (type of plasma proteins, albumin or alpha glycoprotein, is unspecified) independent of concentration over the concentration range of 50 to 10,000 ng/mL [Study 900777]. The mean blood to plasma ratio in human blood in vitro is 1.35. 2.2.5.5 Does the mass balance study suggest renal or hepatic as the major route of elimination? The ADME study (A2105) in 6 healthy subjects who received a single dose of 750 mg [14C]ceritinib with blood samples collected up to 336 hours and urine and feces samples collected up to 384 hours post-dose, suggests hepatic as the major route of elimination. The mean recovery of the administered dose was 92.3% (68% as unchanged parent compound) in the feces and 1.3% in the urine (Figure 9). The major circulating component in plasma was unchanged ceritinib, which constituted 81.6% of plasma radioactivity. Figure 9. Mean radioactive dose recovered in urine and feces after a single dose of 750 mg [14C]-ceritinib in 6 healthy subjects Source: A2105 Final Study Report, Figure 11-10, Page 69 2.2.5.6 What are the characteristics of drug metabolism? Hepatic oxidative metabolism of ceritinib is primarily mediated by CYP3A (>90%) as shown in an in vitro study using human liver microsomes [Study R0900839]. The metabolic pathway in humans is shown in Figure 10. Reference ID: 3477231 NDA 205755/0000 Page 26 of 112 Figure 10. The metabolism of ceritinib after a single dose of 750 mg [14C]-ceritinib in 6 healthy subjects Source: A2105 Final Study Report, Figure 11-9, Page 66 After a single oral 750 mg dose of [14C]-ceritinib, plasma samples were collected at pre-dose, 0.5, 1, 1.5, 2, 3, 4, 6, 10, 12, 24, 48, 72, 96, 120, 144, 168, 192, 216, 240, 264, 288, 312, and 336 hours post-dose in 6 healthy subjects [Study A2105]. Unchanged [14C]-ceritinib accounted for 81.6% of plasma radioactivity. Eleven metabolites were identified in plasma, with no metabolite contributing >2.3% to the mean radioactivity AUC or >5.8% to the radioactivity AUC in any individual subject (Table 9). Unchanged [14C]-ceritinib accounted for 68% of the recovered dose in the feces with ten metabolites identified (no metabolite contributing >6.5% of the administered dose) (Table 9). Urine was not profiled for metabolites due to the low levels of radioactivity recovered in urine. Ceritinib (parent compound) accounted for >90% of AUC0-24h of total drug related material in patients (Study X2101). Reference ID: 3477231 NDA 205755/0000 Page 27 of 112 Table 9. Components of plasma, urine, and feces after a single dose of 750 mg [14C]-ceritinib in healthy subjects (n=6) Mean % of total AUC in plasma (Range) Mean % of dose in feces (Range) Parent 81.6 (75.9-89.1) 68.0 (59.7-79.2) M23.6 1.2 (0.5-2.5) 3.9 (1.6-5.3) M26.9 1.1 (0.3-1.6) - M27.6 1.4 (0.1-2.7) - M30.4 - 2.1 (1.0-2.8) M32.9 1.9 (0.8-5.8) 1.8 (1.3-2.4) M33.4 1.6 (0.5-2.7) 1.4 (0.7-3.5) M34.5 1.9 (0.1-4.5) 1.2 (0.6-1.7) M35.8 1.8 (0.8-2.9) 6.5 (4.5-7.6) M46.1 2.3 (1.5-3.7) 1.6 (0.9-2.8) M46.6 1.7 (0.4-5.1) 1.1 (0.6-1.7) M48.8 1.7 (1.1-2.2) 1.4 (0.9-2.1) M52.0 2.0 (0.7-4.0) 2.2 (1.1-3.7) Source: A2105 Final Study Report, Table 11-8, Page 61; Table 11-9, Page 65. 2.2.5.7 What are the characteristics of drug excretion? The ADME study (A2105) showed that 92.3% (68% unchanged parent compound) of the [14C]ceritinib oral dose was recovered in the feces and 1.3% in the urine. Non-metabolic elimination such as biliary excretion and gastrointestinal secretion in humans cannot be ruled out. Biliary excretion of unchanged ceritinib accounted for 34.9% of the intravenous dose in bile-duct cannulated rats, while direct GI secretion accounted for 12.1% of the dose, indicating that 50% of the clearance in rats is mediated through non-metabolism pathways [Study 900773a]. Elimination Apparent clearance (CL/F) at steady-state (33 L/h) is lower than that after a single dose (89 L/h), likely due to auto-inhibition of CYP3A. The terminal half-life (t1/2) at the 750 mg dose is 41 hours. 2.2.5.8 Based on PK parameters, what is the degree of linearity or non-linearity based in the dose-concentration relationship? Ceritinib exhibits nonlinear, time-dependent PK in humans. After a single dose, exposures are dose proportional in the range of 50 to 750 mg with an estimated slope of 0.99 (90% CI: 0.68, 1.30) for AUC0-24h and 0.97 (90% CI: 0.65, 1.29) for Cmax using a power model. After repeat doses of 50 to 750 mg, exposures (Ctrough on Cycle 2 Day 1) are greater than dose proportional with an estimated slope of 1.47 (90% CI: 1.10, 1.84). Reference ID: 3477231 NDA 205755/0000 Page 28 of 112 2.2.5.9 How do the PK parameters change with time following chronic dosing? The mean accumulation ratio for AUC0-24h is 4.7 (day 8/day 1 of cycle 1) and 6.2 (day 1 of cycle 2/cycle 1). Refer to Table 5 and Table 6 for single and multiple dose PK parameters, respectively. 2.2.5.10 What is the inter- and intra-subject variability of the PK parameters in volunteers and patients and what are the major causes of variability? The inter-subject variability as estimated by geometric mean CV% in healthy subjects administered single oral doses of 450 to 750 mg was determined to be 42-74% for AUCinf and 35-94% for Cmax. Inter-patient variability in steady state AUC and Cmax at the dose of 750 mg is 74% and 76%, respectively. The population PK (popPK) analysis assessed the influence of covariates including body weight, gender, race, age, hepatic function (baseline ALT, total bilirubin, albumin), renal function (baseline eGFR), ECOG performance status, prior crizotinib treatment, and concomitant medications (CYP3A inhibitors and inducers, pH-elevating agents). Body weight and baseline albumin were found to be statistically significant covariates impacting CL/F of ceritinib. However, the effects of these covariates on ceritinib PK were not considered to be clinically important. 2.3 2.3.1 INTRINSIC FACTORS What intrinsic factors influence exposure and/or response, and what is the impact of any differences in exposure on effectiveness or safety responses? No formal studies have been conducted to assess the effect of age, race, weight, height, or organ dysfunction on exposure and response to ceritinib. The Applicant’s popPK analysis, verified by our pharmacometrics review, did not identify clinically important effects of body weight, age, gender, mild and moderate renal impairment, and mild hepatic impairment as covariates on clearance or volume of distribution of ceritinib (refer to Pharmacometrics review in Section 4.1). Relationship between Age and Exposure The popPK analysis showed that age (<65 years, ≥65 years) is not a statistically significant covariate influencing ceritinib PK. Relationship between Gender and Exposure The popPK analysis showed that gender is not a statistically significant covariate influencing ceritinib PK. Exposures of ceritinib in women (n=170) were 12% higher than that observed in men (n=132), which is not considered to be a clinically important difference requiring dose modification. Relationship between Race and Exposure The effect of race on ceritinib PK was evaluated in Caucasian (n=197, 65.2%), Asian (n=95, 31.5%), Black (n=4, 1.3%), and other races (n=6, 2.0%). The popPK analysis suggested that exposures were 10% higher in Asians as compared to non-Asians, most likely explained by the Reference ID: 3477231 NDA 205755/0000 Page 29 of 112 lower body weight in Asians (mean body weight for Asians is 61 kg versus non-Asians of 72 kg). This difference in exposure is not considered to be a clinically important and does not require dose modification. Relationship between Weight and Exposure PopPK analyses identified body weight as a significant covariate influencing ceritinib PK. Patients with lower body weight (<60 kg) had 20% higher systemic exposures and those with higher body weight (>80 kg) had 15% lower exposures as compared to patients with body weight of 60-80 kg. However, these differences are not considered to be clinically important. Relationship between Renal Impairment and Exposure The Applicant did not conduct a dedicated renal impairment study. Renal function was not retained as a covariate in the final population PK model (Figure 12). The AUCss in patients with mild or moderate renal impairment were predicted to have a 1.09-fold (90% CI: 0.97-1.25) and 1.19-fold (90% CI: 0.95-1.49) increase in AUCss compared to patients with normal renal function. These potential PK differences are not considered to be clinically important because additional analyses suggested that baseline renal function has no apparent influence on effectiveness (ORR) and safety (any Grade 3-4 AEs) (Table 10). Therefore, no dose adjustment in patients with mild and moderate renal impairment is recommended. Patients with severe renal impairment were excluded from enrollment in the clinical studies. A severe renal impairment study is not recommended as renal excretion is a minor route of elimination (1.3% of the orally administered dose was recovered in the urine) and based on discussion with the Clinical reviewer, lung cancer does not typically metastasize to the kidneys and the patient population is not intrinsically renally compromised. Table 10. Proportion of patients with ORR or any Grade 3-4 AEs based on renal function Normal Renal Function Mild Renal Impairment Moderate Renal Impairment Any Grade 3-4 AEs1 113/160 (71%) 60/79 (76%) 12/16 (75%) ORR2 58/108 (54%) 25/46 (54%) 6/9 (67%) 1 2 ITT Safety Population, data cut-off of October 31, 2013, n=255 NSCLC patients with prior ALK inhibitor who received 750 mg ceritinib, data cut-off of October 31, 2013, n=163 Relationship between Hepatic Impairment and Exposure Because ADME study results showed that 92% of the oral administered dose was recovered in the feces, hepatic impairment is likely to increase the systemic exposure of ceritinib. Therefore, a PMR will be requested to submit the results from the ongoing single dose hepatic impairment study in otherwise healthy subjects. PopPK analyses showed similar ceritinib exposures between patients with mild hepatic impairment (total bilirubin ≤ULN and AST >ULN or total bilirubin >1-1.5 × ULN and AST any value, n=48) and patients with normal hepatic function (total Reference ID: 3477231 NDA 205755/0000 Page 30 of 112 bilirubin ≤ULN and AST ≤ULN, n=254). In addition, the data suggested that baseline hepatic function (mild hepatic impairment) has no apparent influence on effectiveness (ORR) and safety (any Grade 3-4 AEs) (Table 11). Dose adjustment in patients with mild hepatic impairment is not recommended. Patients with moderate and severe hepatic impairment were excluded from enrollment in the clinical study and dose recommendations have not been determined in patients with moderate and severe hepatic impairment. Table 11. Proportion of patients with ORR or any Grade 3-4 AEs based on hepatic function Normal Hepatic Function Mild Hepatic Impairment Any Grade 3-4 AEs1 155/215 (73%) 30/40 (75%) ORR2 74/136 (54%) 15/27 (56%) 1 ITT Safety Population, data cut-off of October 31, 2013, n=255 NSCLC patients with prior ALK inhibitor who received 750 mg ceritinib, data cut-off of October 31, 2013, n=163 2 2.3.2 Based upon what is known about exposure-response relationships and their variability and the groups studied, healthy volunteers vs. patients vs. specific populations, what dosage regimen adjustments, if any, are recommended for each of these groups? No clinically important PK differences have been identified in specific patient populations. Therefore, no dosage regimen adjustments are currently recommended for specific patient populations. The Applicant is requested to conduct a hepatic impairment study under a PMR to determine the recommended dose in patients with moderate or severe hepatic impairment (refer to Section 1.2.1). 2.3.2.1 Elderly Age was not identified as a significant covariate influencing ceritinib PK based on a popPK analysis which included patients 22-80 years of age. 2.3.2.2 Pediatric A study has not been conducted in pediatric patients. As ceritinib received orphan drug designation for the treatment of patients with NSCLC that is ALK-positive, a study in pediatric patients is not required for this indication. 2.3.2.3 Gender The popPK analysis showed that gender is not a statistically significant covariate influencing ceritinib PK. Exposures of ceritinib in women (n=170) were 12% higher than that observed in men (n=132), which is not considered to be a clinically important difference requiring dose modification based on gender. Reference ID: 3477231 NDA 205755/0000 Page 31 of 112 2.3.2.4 Race/Ethnicity The popPK analysis in Caucasian (n=197, 65.2%), Asian (n=95, 31.5%), Black (n=4, 1.3%), and other races (n=6, 2.0%) showed that exposures were 10% higher in Asians as compared to nonAsians, most likely explained by the lower body weight in Asians (mean body weight for Asians is 61 kg versus non-Asians is 72 kg). This difference in exposure is not considered to be clinically important and does not require dose modification. 2.3.2.5 Renal Impairment Refer to Section 2.3.1. PopPK analyses showed 9% and 19% higher ceritinib exposures in patients with mild renal impairment (CLcr 60 to <90 mL/min, n=97) and moderate renal impairment (CLcr 30 to <60 mL/min, n=22), respectively, as compared to those with normal renal function (CLcr ≥90 mL/min, n=183). These PK differences are not considered to be clinically important because data suggested that baseline renal function has no apparent influence on effectiveness (ORR) and safety (any Grade 3-4 AEs); therefore patients with mild and moderate renal impairment do not require dose modification. 2.3.2.6 Hepatic Impairment Refer to Section 2.3.1. PopPK analyses showed that ceritinib exposures were similar between patients with mild hepatic impairment (total bilirubin ≤ULN and AST >ULN or total bilirubin >1-1.5 × ULN and AST any value, n=254) and patients with normal hepatic function (total bilirubin ≤ULN and AST ≤ ULN, n=48). In addition, data suggested that baseline mild hepatic impairment has no apparent influence on effectiveness (ORR) and safety (any Grade 3-4 AEs). No dose modification in patients with mild hepatic impairment is recommended. The Applicant is requested to conduct a clinical trial under a PMR to determine the appropriate ceritinib dose in patients with moderate and severe hepatic impairment. 2.3.2.7 What pregnancy and lactation use information is there in the application? The proposed labeling states that ceritinib can cause fetal harm when administered to a pregnant woman based on its mechanism of action, and lists ceritinib under pregnancy category D. It is not known whether ceritinib is present in human milk. The proposed labeling states that (b) (4) (b) (4) discontinue nursing taking into account the importance of the drug to the mother. 2.4 2.4.1 EXTRINSIC FACTORS What extrinsic factors (drugs, herbal products, diet, smoking, and alcohol use) influence dose-exposure and/or dose-response and what is the impact of any differences in exposure on response? The effect of extrinsic factors including concomitant medications (CYP3A inhibitors and inducers) were evaluated in vivo and prior crizotinib treatment was evaluated using popPK analyses. Reference ID: 3477231 NDA 205755/0000 Page 32 of 112 (b) (4) 2.4.2 Drug-drug interactions? 2.4.2.1 Is there an in vitro basis to suspect in vivo drug-drug interactions? Yes. See below. 2.4.2.2 Is the drug a substrate of CYP enzymes? Yes. CYP3A4 was identified as the primary CYP enzyme (>90%) responsible for the hepatic oxidative metabolism of ceritinib in human liver microsomes (Table 12). Table 12. Inhibition of [14C]-ceritinib metabolism by selective CYP inhibitors Source: Report R0900839, Table 6-7, Page 25. 2.4.2.3 Is the drug an inhibitor and/or an inducer of CYP enzymes? Inhibitor Yes. Ceritinib reversibly inhibits CYP2A6, CYP2B6, CYP2C9, CYP2D6, and CYP3A4 in vitro. Ceritinib is also time-dependent inhibitor of CYP3A4/5 but not CYP1A2, CYP2C9, or CYP2D6 at ceritinib concentrations of up to 50 μM. The potential of ceritinib to inhibit CYP enzymes was evaluated using human liver microsomes [Study 900796]. As shown by the R1 values >1.1 calculated assuming clinical concentrations (maximal steady-state concentration of 1010 ng/mL or 1.8 μM) (Table 13), ceritinib reversibly inhibits CYP2A6, CYP2B6, CYP2C9, CYP2D6, and CYP3A4 in vitro. In vivo studies to evaluate the drug interaction potential of ceritinib with coadministered drugs that are CYP3A4 and CYP2C9 sensitive substrates or those with narrow therapeutic indices are warranted and will be requested as PMRs. If these in vivo studies show no drug interaction, in vivo evaluation of a CYP2D6 substrate will not be needed because the R1 value for CYP2D6 is smaller than those for CYP3A4 and CYP2C9 and only slightly >1.1. An in vivo study to evaluate the drug interaction potential of ceritinib with coadministered drugs that are CYP2A6 substrates is not necessary because CYP2A6 is not among those CYP enzymes requiring routine assessment as recommended in the 2012 Drug Interaction draft FDA Guidance for Industry. Reference ID: 3477231 NDA 205755/0000 Page 33 of 112 Table 13. IC50 and calculated R1 values for ceritinib inhibition of CYP activities in human liver microsomes CYP Enzyme Substrate ceritinib IC50 (unbound) µM Ki value (unbound) µM R1 value (1+I/Ki) CYP1A2 Phenacetin > 100 (2.5) - 1.0 CYP2A6 Coumarin 5 (1.5) 0.03 (0.009) 61.0 CYP2B6 Bupropion 2 (0.3) 5.3 (0.780) 1.3 CYP2C8 Paclitaxel 25 (0.6) - 1.1 CYP2C8 Amodiaquine 2 (0.6) 16.7 (4.86) 1.1 CYP2C9 Diclofenac 2 (0.6) 0.24 (0.0701) 8.5 CYP2C19 S-mephenytoin 70 (1.8) - 1.0 CYP2D6 Bufuralol 20 (2.9) - 1.2 CYP2E1 Chlorzoxazone 30 (4.4) - 1.1 CYP3A4/5 Midazolam 0.2 (0.06) 0.16 (0.0469) 12.3 CYP3A4/5 Testosterone 0.2 (0.06) - 19.0 Calculation of R1 values based on maximal steady-state concentration of 1010 ng/mL or 1.8 μM [I] Ki assumed to be IC50/2 for competitive inhibition for those CYP enzymes without an experimental Ki value Inducer Yes. There was dose-dependent induction of CYP3A4 mRNA. The potential of ceritinib (0.252.5 µM) to induce CYP enzymes was evaluated using human hepatocytes from three donors [Study 200856]. Results showed that there was no induction of CYP1A2, CYP2B6, or CYP2C9 mRNA (<4-fold) relative to vehicle control (Table 14). There was dose-dependent induction of CYP3A4 mRNA with 3.2-, 8.7-, and 6.1-fold change in the three human hepatocyte lots, respectively. Ceritinib did not induce CYP1A2 or CYP2B6 activity (<2-fold) relative to vehicle control. Marginal induction of CYP2C9 activity (2.3-fold) was observed at 1 µM in one of the three human hepatocyte lots. Marginal induction of CYP3A activity (2.1-fold) was observed at 2.5 μM in one of the human hepatocyte lots. The inconsistency in magnitude of change between CYP3A mRNA and activity levels may be due to ceritinib-mediated time-dependent inhibition of CYP3A [Study R0700696]. Reference ID: 3477231 NDA 205755/0000 Page 34 of 112 Table 14. Ceritinib induction of mean mRNA and activity levels of CYP enzymes CYP1A2 CYP2B6 CYP2C9 CYP3A4 Ceritinib (μM) mRNA Activity mRNA Activity mRNA Activity mRNA Activity 0.25 1.01 1.15 1.02 1.05 1.12 1.27 1.27 0.98 1.0 1.14 1.22 1.17 1.20 1.77 2.73 1.43 1.15d 2.5 1.07 1.01 0.96 0.92 1.18 1.10 6.03b 1.33c Known positive 32.1 31.7 10.2 4.65 2.34 3.50 31.1 8.81 inducera a Omeprazole 50 μM for CYP1A2, phenobarbital 1000 μM for CYP2B6 and CYP2C9, rifampin 10 μM for CYP3A4 b Mean of 3.24-, 8.74-, and 6.12-fold change relative to vehicle control in the three human hepatocyte lots c Mean of 0.936-, 0.983-, 2.06-fold change relative to vehicle control in the three human hepatocyte lots d Mean of 1.75-, 1.24-, and 2.31-fold change relative vehicle control in the three human hepatocyte lots The potential of PXR transcriptional activation of the CYP3A4 promoter by ceritinib was evaluated in Study R0700559. Ceritinib was not an activator of PXR in vitro up to 1 μM (the highest concentration tested in which the cells were viable). However, maximal steady-state concentrations in humans (1.8 μM) exceed the concentration of ceritinib used in this in vitro study. 2.4.2.4 Is the drug an inhibitor and/or an inducer of transporters? Ceritinib is not a P-gp (P-glycoprotein) inhibitor or BCRP (breast cancer resistance protein) inhibitor. Ceritinib inhibited OATP1B1 (organic anion transporting polypeptide 1B1) or OATP1B3 activity (organic anion transporting polypeptide 1B3) by 31.8% and 24.1%, respectively. Ceritinib inhibited OAT1 (organic anion transporter) activity by 16.3% and did not inhibit OAT3 activity. Ceritinib inhibited OCT2 (organic cation transporter) activity by 35.4% and did not inhibit OCT1 activity. There are no in vitro studies to evaluate ceritinib as an inducer of transporters. Inhibitor P-gp and BCRP: The potential for ceritinib to inhibit P-gp and BCRP was evaluated by assessing the effect of increasing concentrations of ceritinib (0.0-1.5 μM) on the efflux of probe fluorescent substrates for P-gp and BCRP, Rhodamine 123 (Rho123) and Bodipy FL prazosin (BDP), respectively in vitro [Study R0900229]. Maximal steady-state concentrations in humans (1.8 μM) slightly exceed the highest ceritinib concentration of 1.5 μM tested due to low solubility of ceritinib. Given that the efflux of probe substrates Rho123 and BDP did not appear to decrease considerably with increasing concentrations of ceritinib as measured by fluorescence (Table 15), an in vivo DDI study with a P-gp or BCRP substrate is considered unnecessary. Reference ID: 3477231 NDA 205755/0000 Page 35 of 112 Table 15. Effect of ceritinib on P-gp-mediated Rho123 efflux (left) and BCRP-mediated BDP efflux (right) CsA: Cyclosporin A, positive control inhibitor of P-gp FTC: Fumitremorgin C, positive control inhibitor of BCRP Source: Report R0900229, Table 6-1, Page 16. Source: Report R0900229, Table 6-2, Page 16. OATP1B1 and OATP1B3: The potential for ceritinib to inhibit OATP1B1 or OATP1B3 transporters was evaluated by assessing the effect of increasing concentrations of ceritinib (0.055 μM) on the accumulation of the OATP substrate, [3H]estradiol-17ß-glucuronide, and percent inhibition of OATP activity in human embryonic kidney (HEK) cells in vitro [Study R1300170]. Ceritinib inhibited OATP1B1 and OATP1B3 activity by 31.8% and 24.1%, respectively. The IC50 values of OATP1B1 or OATP1B3 inhibition were determined to be >5 μM (concentrations were limited up to 5 μM due to low solubility of ceritinib). Given that the calculated R-value is <1.25 (Table 16), an in vivo DDI study with a sensitive substrate of OATP1B1 or OATP1B3 is considered unnecessary. Table 16. IC50 and calculated R values for ceritinib inhibition of OATP transporters Substrate Ceritinib IC50 (μM) Cmaxa/IC50 R valueb (1+ (fu × Iin,max/IC50) OATP1B1 Estradiol 17β-Dglucuronide >5 < 0.36 1.06 OATP1B3 Estradiol 17β-Dglucuronide >5 < 0.36 1.06 Transporter Plasma maximal steady-state concentration (Cmax) of 1010 ng/mL or 1.8 μM Iin,max is the estimated maximum inhibitor concentration at the inlet to the liver and is equal to Cmax + (ka × Dose × Fa Fg/Qh). Fa Fg (fraction of the dose of inhibitor which is absorbed) = 1; ka (absorption rate constant of the inhibitor) = 0.1 min-1; Qh (estimated hepatic blood flow) = 1500 mL/min. a b OAT1 and OAT3: The potential for ceritinib to inhibit organic anion transporters (OAT1 and OAT3) was evaluated by assessing the effect of increasing concentrations of ceritinib (0.05-5 μM) on the accumulation of the OAT1 substrate ([3H]cidofovir) and the OAT3 substrate ([3H]estrone-3-sulfate) and percent inhibition of OAT activity in HEK cells in vitro [Study R1200913]. Ceritinib inhibited OAT1 activity by 16.3% and did not inhibit OAT3 activity. The Reference ID: 3477231 NDA 205755/0000 Page 36 of 112 IC50 value of OAT1 inhibition was determined to be >5 μM (concentrations were limited up to 5 μM due to low solubility of ceritinib). Given that the unbound Cmax/IC50 for OAT1 is <0.1 and ceritinib did not inhibit OAT3 (Table 17), an in vivo DDI study with a sensitive substrate of OAT1 or OAT3 is considered unnecessary. Table 17. IC50 and calculated unbound Cmax/IC50 values for ceritinib inhibition of OAT and OCT transporters Transporter Substrate Ceritinib IC50 (μM) Unbound Cmaxa/IC50 OAT1 Cidofovir >5 < 0.01 OAT3 Estrone-3-sulfate NDb - OCT1 MPP+ NDb - OCT2 Metformin >5 < 0.01 a b Cmax of 1.8 μM; 3% unbound ceritinib - unbound Cmax of 0.05 μM Not determined; ceritinib did not inhibit transporter in vitro OCT1 and OCT2: The potential for ceritinib to inhibit OCT1 and OCT2 was evaluated by assessing the effect of increasing concentrations of ceritinib (0.05-5 μM) on the accumulation of the OCT1 substrate ([3H]MPP+) and the OCT2 substrate ([14H]metformin) and percent inhibition of OCT activity in HEK cells in vitro [Study R1300023]. Ceritinib inhibited OCT2 activity by 35.4% and did not inhibit OCT1 activity. The IC50 value of OCT2 inhibition was determined to be >5 μM (concentrations were limited up to 5 μM due to low solubility of ceritinib). Given that the unbound Cmax/IC50 for OCT2 is <0.1 and ceritinib did not inhibit OCT1 (Table 17), an in vivo DDI study with a sensitive substrate of OCT1 or OCT2 is considered unnecessary. Inducer There are no in vitro studies to evaluate ceritinib as an inducer of transporters. 2.4.2.5 Are there other metabolic/transporter pathways that may be important? Ceritinib is a P-gp and BCRP substrate in vitro; however their contribution to ceritinib absorption is considered limited. Ceritinib is not a substrate for the hepatic uptake transporters OCT1, OAT2, OATP1B1, and OATP2B1. Ceritinib is a P-gp substrate in vitro as the net flux ratio is ≥2 (182 at 3.0 μM and 19.5 at 14 μM ceritinib) in Caco-2 cells and LY335979 (a P-gp inhibitor) reduced the efflux ratio (>50%) to approximately 6 at both 3.0 and 14 μM [Study R1000083]. Sensitivity analyses of effective permeability (Peff) were conducted during the PBPK review to determine P-gp contribution to oral absorption of ceritinib. If the intestinal efflux transporter P-gp significantly contributed to the oral absorption of ceritinib, an inhibition of P-gp would result in increased Peff of ceritinib. Sensitivity analyses showed that an increase in Peff led to a similar increase of AUC and Cmax with a 50% increase in Peff resulting in 34% and 36% increase in ceritinib AUC and Cmax, respectively, and a 2-fold increase in Peff resulting in 61% and 66% increase in ceritinib AUC and Reference ID: 3477231 NDA 205755/0000 Page 37 of 112 Cmax, respectively. The in vivo DDI study showed that ketoconazole (a strong CYP3A4/P-gp inhibitor) coadministration resulted in an increase of ceritinib AUC and Cmax that was not in the same magnitude (2.8-fold in AUC and 22% in Cmax). These simulation and in vivo results suggest that ketoconazole did not increase Peff or affect intestinal efflux transport of ceritinib. Given that the observed effect of ketoconazole on ceritinib exposure is most likely due to CYP3A inhibition and not P-gp inhibition, an in vivo DDI study with a P-gp inhibitor is considered unnecessary. Ko143 (a BCRP inhibitor) had a smaller effect on the reduction of efflux ratio to 121 at 3 μM ceritinib, indicating that an in vivo DDI study with a BCRP inhibitor is not necessary. Ceritinib is not a substrate for the hepatic uptake transporters OCT1, OAT2, OATP1B1 and OATP2B1 as shown by <2-fold uptake in cells expressing transporters as compared to control cells [Study R1000482] (Table 18). Table 18. Transport of ceritinib by OCT1, OAT2, OATP1B1, OATP2B1 Source: Report R1000482, Table 5-2, Page 13. 2.4.2.6 Does the label specify co-administration of another drug and, if so, has the interaction potential between these drugs been evaluated? No, ceritinib is recommended as monotherapy for this marketing application. 2.4.2.7 What other co-medications are likely to be administered to the target population? Concomitant medications used by ≥20% of the 304 patients (NSCLC, n=290 and non-NSCLC, n=14) treated with any dose of ceritinib in the clinical trial included anti-diarrheal therapy (primarily loperamide, 55%), antiemetics (metoclopramide, 37%), serotonin (5HT3) antagonists (ondansetron, 38%), glucocorticoids (primarily dexamethasone and prednisone, 39%), opioids (primarily oxycodone and morphine, 37%), proton pump inhibitors (PPI, primarily omeprazole and pantoprazole, 30%), acetaminophen (33%), laxatives (primarily lactulose, polyethylene glycol, and magnesium oxide, 24%), benzodiazepines (primarily lorazepam, 22%), and fluoroquinolones (primarily levofloxacin, 22%). Coadministration of gastric acid reducing agents such as PPIs may lead to decreased drug absorption, leading to decreased effectiveness of ceritinib as ceritinib exhibits pH-dependent Reference ID: 3477231 NDA 205755/0000 Page 38 of 112 solubility (refer to Section 2.1.1). Concomitant administration of acid reducing agents (e.g., H2receptor antagonists and PPIs) was significantly associated with a decrease in the absorption rate (ka) of ceritinib in the popPK analysis, while exposures were similar in patients who received PPIs as compared to that in patients who did not receive PPIs (Figure 11). However, these analyses are limited and should be interpreted in the context that the timing of coadministration of a PPI was not taken into consideration in the PK sampling plan. Given that approximately 30% of patients in the clinical trial received PPIs, an appropriate dosing strategy with regard to gastric acid reducing agents should be determined. A PMR will be requested to conduct a clinical trial to determine how to dose ceritinib with regard to concomitant gastric acid reducing agents (refer to Section 1.2.1). Figure 11. Effect of PPI coadministration on ceritinib steady-state exposure (AUC on cycle 2 day 1) 2.4.2.8 Are there any in vivo drug-drug interaction studies that indicate the exposure alone and/or exposure-response relationships are different when drugs are co-administered? Yes. Coadministration of ketoconazole (a strong CYP3A inhibitor) 200 mg twice daily for 14 days and a single dose of 450 mg ceritinib increased ceritinib AUC by 2.9-fold (90% CI: 2.5, 3.3) and Cmax by 22% (90% CI: 7.0%, 39%) in 19 healthy subjects. Coadministration of rifampin (a strong CYP3A inducer) 600 mg QD for 14 days and a single dose of 750 mg ceritinib decreased ceritinib AUC by 70% (90% CI: 61%, 77%) and Cmax by 44% (90% CI: 24%, 59%) in 19 healthy subjects. DDI with Strong CYP3A Modulators The effect of ketoconazole (a strong CYP3A inhibitor) on ceritinib PK was evaluated in a single dose, two period crossover study in 19 healthy subjects under fasted conditions (Study 2104). Coadministration of ketoconazole 200 mg twice daily for 14 days with a single dose of 450 mg ceritinib increased ceritinib AUC by 2.9-fold and Cmax by 22% as compared to ceritinib alone (Table 19). Clearance (CL/F) of ceritinib decreased by 65% from 78.1 L/h to 27.1 L/h with coadministration of ketoconazole. Reference ID: 3477231 NDA 205755/0000 Page 39 of 112 Table 19. Comparative analysis of ceritinib PK parameters on day 1 (without ketoconazole) and day 18 (with ketoconazole) (n=19) Geometric Mean Ratio* (90% CI) Exposure parameter Ceritinib alone (450 mg) Ceritinib (450 mg) with ketoconazole (200 mg BID for 14 days) Geometric Mean Ratio* (90% CI) AUC0-336h (ng∙hr/mL) 5640 (44.3) 16300 (47.0) 2.86 n=18 n=19 (2.45-3.34) 5760 (43.4) 16600 (47.2) 2.86 n=18 n=19 (2.46-3.33) 133 (34.9) 164 (40.3) 1.22 n=18 n=19 (1.07-1.39) AUC0-inf (ng∙hr/mL) Cmax (ng/mL) * ceritinib 450 mg alone vs. ceritinib 450 mg and ketoconazole 200 mg BID for 14 days The effect of rifampin (a strong CYP3A inducer) on ceritinib PK was evaluated in a single dose, two period crossover study in 19 healthy subjects under fasted conditions (Study 2106). Coadministration of rifampin 600 mg QD for 14 days with a single dose of 750 mg ceritinib decreased ceritinib AUC by 70% and Cmax by 44% (Table 20). Clearance (CL/F) of ceritinib increased by 3.3-fold from 70.6 L/h to 234 L/h with coadministration of rifampin. Reference ID: 3477231 NDA 205755/0000 Page 40 of 112 Table 20. Comparative analysis of ceritinib AUC and Cmax on day 1 (without rifampin) and day 21 (with rifampin) (n=19) Geometric Mean Ratio* (90% CI) Exposure parameter Ceritinib alone (750 mg) Ceritinib (750 mg) with rifampin (600 mg QD for 14 days) Geometric Mean Ratio* (90% CI) AUC0-168h (ng∙hr/mL) 10100 (74.2) 3130 (87.8) 0.31 n=17 n=17 (0.24-0.40) 10600 (72.1) 3210 (85.4) 0.30 n=17 n=17 (0.23-0.39) 219 (93.6) 122 (85.1) 0.56 n=17 n=17 (0.41-0.76) AUC0-inf (ng∙hr/mL) Cmax (ng/mL) * ceritinib 750 mg alone vs. ceritinib 750 mg and rifampin 600 mg QD for 14 days Given that the drug-drug interaction studies were conducted using single doses of ceritinib, the magnitude of the effect of strong CYP3A modulators on the steady-state PK of ceritinib after repeat dosing in patients is unknown due to nonlinear, time-dependent PK of ceritinib. PBPK modeling predicted that ketoconazole can increase the steady-state ceritinib exposure (AUC) by 51% and that rifampin can decrease steady-state AUC by 67%. Given the magnitude of exposure change in the context of the exposure-response curves for safety and effectiveness observed with ceritinib (refer to Section 2.2.4.1 and 2.2.4.2), the following recommendations are provided with regard to concomitant use of strong CYP3A inhibitors: • Avoid concomitant use of strong CYP3A inhibitors with ceritinib. • If avoiding concomitant strong CYP3A inhibitors is not possible, the ceritinib dose should be reduced by approximately one-third, rounded to the nearest 150 mg dosage strength. After discontinuation of the strong CYP3A inhibitor, the ceritinib dose that was taken prior to initiating the strong CYP3A4 inhibitor should be resumed. With regard to concomitant administration of strong CYP3A inducers, the recommendation is avoidance of concurrent use of strong CYP3A inducers. In order to compensate for the 70% decrease in exposure caused by a strong CYP3A inducer, approximately 8-9 capsules would have to be administered to achieve a dose of 1275 mg ceritinib, which is likely not feasible for patients. DDI with Moderate CYP3A Modulators PBPK modeling predicted that fluconazole, a moderate CYP3A4 inhibitor, can increase ceritinib exposure by 37% and efavirenz, a moderate CYP3A4 inducer, can decrease ceritinib exposure by 43% (Table 21). A clinical trial to further evaluate the effect of a moderate CYP3A4 inhibitor or Reference ID: 3477231 NDA 205755/0000 Page 41 of 112 inducer on ceritinib PK is considered unnecessary given the magnitude of predicted changes in steady-state exposures after coadministration with a moderate CYP3A4 inhibitor and inducer. It is not recommended to restrict the concomitant use of moderate CYP3A4 inhibitors and inducers. Table 21. PBPK model simulated ceritinib steady-state AUC and Cmax following coadministration of ceritinib 750 mg daily with CYP3A modulators Dosing of CYP3A modulators Sponsor’s model AUC Ratio Cmax Ratio (90% CI) (90% CI) FDA’s model AUC Ratio Cmax Ratio (90% CI) (90% CI) Strong CYP3A inhibitor ketoconazole 200 mg twice daily 1.35 (1.27, 1.42) 1.32 (1.26, 1.39) 1.51 (1.43, 1.59) 1.47 (1.40, 1.54) Moderate CYP3A inhibitor fluconazole 200 mg once daily 1.15 (1.12, 1.19) 1.14 (1.11, 1.17) 1.37 (1.31, 1.42) 1.32 (1.28, 1.37) Strong CYP3A inducer rifampin 600 mg once daily 0.51 (0.45, 0.57) 0.54 (0.49, 0.60) 0.33 (0.30, 0.36) 0.37 (0.34, 0.40) Moderate CYP3A inducer efavirenz 600 mg once daily NA NA 0.57 (0.52, 0.62) 0.61 (0.56, 0.65) Source: PBPK review, Table A6. 2.4.2.9 Is there a known mechanistic basis for pharmacodynamic drug-drug interactions? No. Pharmacodynamic data were not collected in the clinical study. 2.4.2.10 Are there any unresolved questions related to metabolism, active metabolites, metabolic drug interactions or protein binding? The Applicant is requested to conduct DDI studies with CYP3A and CYP2C9 sensitive substrates and gastric acid reducing agents under PMRs. Refer to Section 1.2.1. 2.4.3 What issues related to dose, dosing regimens, or administration are unresolved and represent significant omissions? Refer to Section 2.2.4.4 on unresolved dosing and administration issues with regard to GI intolerability. 2.5 2.5.1 GENERAL BIOPHARMACEUTICS Based on BCS principles, in what class is this drug and formulation? What solubility, permeability and dissolution data support this classification? The Applicant classifies ceritinib as BCS class 4 based on data showing that ceritinib has low permeability in Caco-2 cells [Study R1000083] and poor aqueous solubility (refer to Section 2.1.1). The apparent passive permeability of ceritinib in the presence of LY335979 (a P-gp Reference ID: 3477231 NDA 205755/0000 Page 42 of 112 inhibitor) was 0.27 × 10-5 cm∙min-1 at 3.0 µM and 0.98 × 10-5 cm∙min-1 at 14.0 µM, which were lower than that of the low permeability marker mannitol (4.0 × 10-5 cm∙min-1). 2.5.2 What is the relative bioavailability of the proposed to-be-marketed formulation to the clinical trial formulation? The proposed drug product has been available as 25, 50, and 150 mg capsules during ceritinib development. The 150 mg capsule is the only capsule strength intended for commercialization. There were minor differences between the 150 mg capsules used in the clinical trials and the commercial product as listed below: • Compounds for starting materials • Color of capsules (b) (4) • processes • Manufacturing sites These differences are not expected to alter the bioavailability of the to-be-marketed capsule. Refer to the Biopharmaceutics review. 2.5.3 What is the effect of food on the bioavailability (BA) of the drug from the dosage form? What dosing recommendation should be made, if any, regarding administration of the product in relation to meals or meal types? The effect of food on ceritinib PK was evaluated in a single 500 mg dose, randomized, threetreatment, two period crossover study (A2101) with a 14-day washout period in 28 healthy subjects who received the to-be-marketed formulation of 150 mg capsules plus one 50 mg clinical trial capsule. Given that ceritinib exhibits linear PK in the dose range of 50 to 750 mg following a single dose, the use of a single 500 mg dose that is lower than the clinical dose of 750 mg appears reasonable. The high-fat meal consisted of approximately 1000 total calories with a composition of 50% fat, 35% carbohydrates, and 15% protein as recommended in the Food Effect FDA Guidance for Industry; the low-fat meal consisted of approximately 330 total calories with a composition of 25% fat, 60% carbohydrates, and 15% protein. Refer to Table 22 for representative high- and low-fat meals. Table 22. Representative high-fat (left) and low-fat meals (right) Source: A2101 Final Study Report, Tables 7-3 and 7-4, Page 260 Reference ID: 3477231 NDA 205755/0000 Page 43 of 112 At the recommended dose of 750 mg QD, the majority of patients experienced GI adverse reactions. Taking ceritinib with food may improve GI tolerability, but would increase exposures. Administration of a single 500 mg dose of ceritinib with a high-fat, high-calorie meal in healthy subjects resulted in a 73% increase in AUC and 41% increase in Cmax with no effect on tmax as compared with fasted conditions; a low-fat meal increased AUC by 58% and Cmax by 43% and decreased tmax by a median of 2 hours as compared to fasted conditions (Table 23). Table 23. PK parameters of a single dose of ceritinib after a high-fat or a low-fat meal compared to fasted conditions Geometric Mean (%CV) Geometric Mean Ratio (90% CI) High-fat highLow-fat lowFasted High-fat/fed Low-fat/fed calorie meal calorie meal (n=27) (n=14) (n=14) AUCinf 12002 (32) 10941 (23) 6929 (42) 1.73 (1.46, 2.05) 1.58 (1.34, 1.86) AUC0-168h 11419 (32) 10564 (23) 6650 (42) 1.71 (1.45, 2.03) 1.59 (1.35, 1.87) Cmax 225 (29) 229 (20) 160 (44) 1.41 (1.18, 1.68) 1.43 (1.21, 1.71) Source: Study 2101 final study report, Table 11-3, Page 48; Listing 16.2.6-1.1, Pages 555-558 PK Parameter In order to determine an exposure-matched dose of ceritinib taken with food to improve GI tolerability and compliance without compromising efficacy, a postmarketing trial will be requested to evaluate the GI tolerability, efficacy, and PK of 450 mg of ceritinib taken with a meal as compared with that of 750 mg ceritinib taken in the fasted state in metastatic ALKpositive NSCLC patients. Refer to Section 2.2.4.4. 2.5.4 When would a fed BE study be appropriate and was one conducted? Not applicable. 2.5.5 How do dissolution conditions and specifications ensure in vivo performance and quality of the product? Refer to Biopharmaceutics review. 2.5.6 If different strength formulations are not bioequivalent based on standard criteria, what clinical safety and efficacy data support the approval of various strengths of the to-be-marketed product? Not applicable as only one strength of 150 mg capsules will be marketed. Refer to CMC review. 2.5.7 If the NDA is for a modified release formulation of an approved immediate product without supportive safety and efficacy studies, what dosing regimen changes are necessary, if any, in the presence or absence of PK-PD relationship? Not applicable. Reference ID: 3477231 NDA 205755/0000 Page 44 of 112 2.5.8 If unapproved products or altered approved products were used as active controls, how is BE to the ‘to-be-marketed’ product? What is the basis for using either in vitro or in vivo data to evaluate BE? Not applicable. 2.5.9 What other significant, unresolved issues in relation to in vitro dissolution of in vivo BA and BE need to be addressed? None. 2.6 2.6.1 ANALYTICAL SECTION How are the active moieties identified and measured in the plasma and the other matrices? Ceritinib was measured in human plasma using a validated liquid chromatography-mass spectrometry (LC-MS/MS) method [Validated Report No. R1100240, R1100240-01]. In the mass balance study, total radioactivity in blood, plasma, urine, and feces were measured by liquid scintillation counting. 2.6.2 Which metabolites have been selected for analysis and why? Concentrations of ceritinib metabolites were not measured as there were no major circulating metabolites. Eleven metabolites were identified in plasma, with no metabolite contributing >2.3% to the mean radioactivity AUC or >5.8% to the radioactivity AUC in any individual subject. 2.6.3 For all moieties measured is free, bound or total measured? Given that ceritinib is 97.2% bound to human plasma proteins, total plasma concentrations were measured. 2.6.4 What bioanalytical methods are used to assess concentrations? Ceritinib concentrations in human plasma were measured by LC-MS/MS using validated methods [Validated Report No. R1100240, R1100240-01]. 2.6.4.1 What is the range of the standard curve? How does it relate to the requirements for clinical studies? What curve fitting techniques are used? The standard curve was generated using seven calibration samples in the concentration range of 1.0 ng/mL (LLOQ) to 500 ng/mL and weighted (1/x2) linear regression. A 100-fold dilution of human plasma samples has been validated. This standard curve range was adequate for the purposes of determining plasma concentrations of ceritinib in the clinical studies. 2.6.4.2 What are the lower and upper limits of quantification? See Section 2.6.4.1. 2.6.4.3 What are the accuracy, precision and selectivity at these limits? The mean %bias and %CV of calibration standards and quality controls for validation of the Reference ID: 3477231 NDA 205755/0000 Page 45 of 112 bioanalytical method were ≤15%, and are acceptable based on the 2013 FDA Guidance for Industry Bioanalytical Method Validation. Acceptance criteria for QC samples in each run were met (%bias within ±15% of the nominal concentration for at least 2/3 of QC samples and QC samples at a minimum of three concentrations) (Table 24). Table 24. Summary of accuracy and precision of calibration standards and quality controls used in clinical studies Calibration standards Report No. Quality controls Study Mean accuracy (%Bias) Mean precision (%CV) Mean accuracy (%Bias) Mean precision (%CV) RCLDK378X2101 X2101a -6.4-3.6 a 0.9-6.6 a -1.2-0.8 b 2.7-7.4 b RCLDK378A2101 A2101 -5.6-5.0 4.7-10.4 -8.0-2.4 8.6-11.3 RCLDK378A2104 A2104 -5.0-4.8 2.5-7.5 -4.8-1.2 2.3-6.8 RCLDK378A2105 A2105 -2.8-1.8 1.0-4.9 -2.0 to -0.3) 1.4-4.8 RCLDK378A2106 A2106 -5.2-3.0 2.5-4.3 -4.8-0.0 2.5-4.4 a Source: DMPK RCLDK378X2101 Report Tables 5-6, 5-7, 5-8, 5-9, 5-10 for data on Cs. Data from Cs obtained on five different instruments. b Source: DMPK RCLDK378X2101 Report Tables 5-11, 5-12, 5-13, 5-14, 5-15 for data on QCs. Data from QCs obtained on five different instruments. The selectivity/specificity of the method was established using blank plasma from six sources. Selectivity was ensured at the LLOQ. 2.6.4.4 What is the sample stability under the conditions used in the study? (long-term, freezethaw, sample-handling, sample transport, autosampler) The stability of ceritinib in human plasma under the following conditions is summarized: • Short-term stability in spiked human plasma: 24 hours at room temperature. • Post-preparative stability in extracts on the autosampler set at 10°C: 137 hours. • Long-term stability in incurred human plasma: 21 months at ≤ -65°C, 39 days at ≤ -15°C. • Three freeze-thaw cycles ≤ -70°C at concentrations of 3.0 and 400 ng/mL. • Stability in solution stored between 2 and 8°C: 13 months. 2.6.4.5 What is the QC sample plan? Four QC samples at concentrations of 3.0 ng/mL (low, 3 × LLOQ), 25.0 ng/mL (mid), 250 ng/mL (mid), and 400 ng/mL (high) were prepared in duplicate in each run. 3 DETAILED LABELING RECOMMENDATIONS Only relevant clinical pharmacology sections are included. The Agency’s suggested changes to the proposed labeling are shown in underline blue text and removal of content shown by red strikethroughs. Of note, the Agency’s labeling modifications have not been agreed upon by the Applicant as of the date of this review. Reference ID: 3477231 NDA 205755/0000 Page 46 of 112 2 DOSAGE AND ADMINISTRATION 2.1 (mo?Dosing and Administration The dose of ZYKADIA is 750 mg (mo-orally once daily -1n1til disease progression or unacceptable toxicitv. Take ZYKADIA Wt least 2 hours before (?mg 2 hours after mmfood [see (?w-Clinical Phannacology A dose has not been determined for patients with moderate to severe hepatic impainnent [see Use in Speci?c Populations 2.2 Dose Modi?cations Table 1: ZYKADIA Dose Interruption. Reduction. or Discontinuation Recommendations (hm Criteria ZYKADIA Dosing Severe or intolerable nausea or Withhold initil improved then: vomiting or diarrhea despite optimal . . . . Resume ZYKADLA with a 150 mg dose reduction OR antiemenc or ant1-dia1rheal therapv Take with meals as instructed below: 0 Ifreceiving 750 111g dailv. reduce the dose to 450mg dailv; Ifreceiving 600 mg dailx. reduce the dose to 300 mg dailx. 2.3 Dose Modi?cation for Strong CYP3A4 Inhibitors Avoid concmrent use of strong CYP3A inhibitors during treatment with ZYKADIA [see Drug Interactions (7.1) and Clinical Pharmacology . If concomitant use of a strong CYP3A inhibitor is unavoidable. reduce ZYKADIA dose bv approximatelv one-third. r01mded to the nearest 150 mg dosage strength. After discontinuation of a strong CYP3A inhibitor. resume ZYKADIA dose that was taken prior to initiating the strong CYP3A4 inhibitor. 7 DRUG INTERACTIONS 7.1 Effect of Other Drugs on Ceritinib- Ceritinib is primarilv metabolized bv CYP3A4 and is a substrate of the ef?ux transporter P-glvcoprotein gP-gpz. Strong CYP3A Inhibitors Ketoconazole (a strong inhibitor) increased the svstemic exposure of ceritinib (bx-o -[see Clinical Pharmacologv Avoid concurrent (mm?use of strong CYP3A inhibitors during treatment with ZYKADLA: If concomitant use of strong CYP3A inhibitors including certain antivirals ritonavir). ?no-macrolide antibiotics telithromycin). antifungals ge.g.. ketoconazole). and nefazodone i_s tniavoidable. reduce ZYKADIA dose by approximatelv one-third. rounded to the nearest 150 mg dosage strength. After discontinuation of a strong CYP3A inhibitor, resmne ZYKADLA dose that was taken prior NDA 205755/0000 Page 47 of 112 Reference ID: 3477231 to initiatin the stron CYP3A4 inhibitor. Avoid grapefruit and grapefruit juice as they may inhibit CYP3A. Strong CYP3A Inducers in, and St. John?s Wort) __a_mect ?rower mu Cerium? Ceritinib inhibits CYP3A and CYP2C9 .t clinicaF concentrations [see Clinical Pharmacology Avoid concmrent use of CYP3A an CYP2C9 substrates known to have narrow ther eutic indices or substrates rimaril metabolized CYP3A and CYP2C9 durin_ treatment with ZYKADIA. Ifuse of these medications is unavoidable consider dos a reduction o_f -CYP3A substrates with narrow therapeutic indices alfentanil clos rine dih droer_otamine er_otamine fentan pimozide,- narrow therapeutic indices quinidine, sirolimusa tacrolimus) and CYP2C9 substrates phenytoin, warfarin). 8 USE IN SPECIFIC POPULATIONS 8.6 Hepatic Impairment As ceritinib is eliminated n'marily via the liver, atients with atic airment ma have increased ex sure Dose a lustment Is not recommentoe or atients wr m1 I t-tatic talrment to. . 1n andAST or total bilirubin >1.0 to 1.5 ULN and any based on results of the mpulation pharmacokinetic NDA 205755I0000 Page 48 of 112 Reference ID: 3477231 analysis [see Clinical Pharmacology A recommended dose has not been determined for patients with mo ate to severe at'c t. 12.2 Pharmacodynamics Cardiac Electroph?iology Serial ECGs were collected following a single dose and at steady-state to evaluate the effect of ceritinib on the QT interval in an own-label, dose-escalation and ex ansion stud . A total of 304 atients were treated with ZYKADIA doses ran from 50 to 750 with 255 atients treated with ZYKADIA 750 . One of 304 atients were found to have corrected the Fridericia method >500 msec and 16 atients 5.3% had an increase from baseline >60 msec. A central tendenc anal is of the data at stead -state demonstrated that the hi est er bound of the two-sided 90% CI for was 22 msec at ZYKADIA 750 harmacokinetic/pharmacodmamic analysis suggested concentration-dependent interval prolongation [see Warnings and Precautions Based on central review of ECG data, 2 of 304 patients had_ bradycardia de?ned as less than 50 beats er minute. Brad cardia was reported as an adverse drug reaction in 3"/o of patients in Study 1; 12.3 Pharmacokinetics Absorption After sin oral administration of ZYKADIA in atients eak- plasma levels (Cum) of ceritinib were I achieved a_t approximately 4 to 6 hours, and AUC and increased dose promrtionally over 50 to 750 . ea soute roav a tyo ZYKADIA asnot been determined. Following ZYKADIA 750 mg once daily dosing, steady-state was reached by approximately 15 days with a geometric mean accumulation ratio of 6.2 after 3 weeks. S?temic exmsure increased in a ggeater than dose proportional manner after repeat doses of 50 to 750 mg once daily. Systemic exposure 'o_f ceritinib m'mcreased when administered with-. A hi -fat meal increased ceritinib AUC low-fat meal increased ceritinib AUC 58% and 43% as co ared with the fasted state [see Dose Modi cations (2.2) NDA 205755I0000 Page 49 of 112 Reference ID: 3477231 Distribution Ceritinib is 97% bound to human 11 cut 0 concentration is 4230 followin a sm ose ZYKADIA 750 n_1ga Ceritinib also has a slight preferential distribution to red blood cells, relative to lasma, with a mean in vitro blood-to- lasma ratio of 1.35. Elimination Following 3 single dose of 750 ZYKADIA the geometric mean 41 hours in atients. ear PK over time. geometric mean apparent earance ceritinib was lower at steady-state (33. 2 a?er 750 mg daily -dosing than after a single 750 mg- dose (88.5 Metabolism. In vitro studies demonstrated that CYP3A was the major ewe involved in the metabolic clearance of ceritinib. Following oral administration of a single 750 mg radiolabeled ceritinib dose, ceritinib as the arent co und was the main circulatin co onent 82% in human plasma. Excretion: Followin oral administration of a sin 750 radiolabeled cerith dose 92.3% of the administered dose was recovered in the feces 68% as unchanged parent compound) while] of the administered NDA 205755I0000 Page 50 of 112 Reference ID: 3477231 garment may ve increased exp_osure. A pharmacokinelic tn'al in patients with hgpatic gp airment has not been conducted. Based on a ppoulation phaimacokinetic anal?is of 48 patients with mild atic airment total bilirubin and AST or total bilirubin 0 to 1.5 i1_npairm ailment [see Use in Speci?c Populations (8. Renal Im airment: -A 2 CO In 111230611115 W1 ren Qpan?ment not en NDA 205755I0000 Page 51 of 112 Reference ID: 3477231 conducted as ceritinib elimination via the kidney is lo_w of a single oral administered dose). Based 011 a A population pharmacokinetic analysis of 97 patients with mild renal impairment (CLcr 60 to <90 mL/min). 22 patients with moderate renal impairment (CLcr 30 to <60 mL/niin) and 183 atients with normal renal ?mction 290 mL/min . ceritinib ex osures were similar in atients with mild and moderate renal impairment and normal renal Patients with severe renal impairment (CLcr <30 mL/min) were not included in the clinical trial. Pediatrics: No trials have been conducted to evaluate the pharmacokinetics of ceritinib in pediatric patients. (hm Drug Interactions Effect of Strong CYP3A Inhibitors on Ceritinib: 111 vitro studies show that ceritinib is a substrate of CYP3A. Coadministration of a single 450 mg ZYKADIA dose with ketoconazole (a strong CYP3A inhibitor) 200 mg twice dailv for 14 davs increased ceritinib AUC (90% CI) bv 29-fold (2.5. 3.3) and 9% (90% C1) by 1.2-fold (1.1. 1.4) in 19 healthv subjects [see Drug Interactions The steady- state AUC of ceritinib at reduced doses after coadrninistration with ketoconazole 200 mg twice dailv for 14 days was predicted bv simulations to be similar to the steadv?state AUC of ceritinib alone [see Dose Modi?cation for Concurrent Use with Strong YP3A4 Inhibitors Effect of Strong YP3A Inducers on eritinib.? oadministration of a single 750 mg ZYKADIA dose with rifampin (a strong CYP3A inducer) 600 mg dailv for 14 days decreased ceritinib AUC (90% CI) by 70% 77%) and cm): (90% CI) by 44% 59%) in 19 healthv subiects [see Drug Interactions 7 . I ect Ceritinib on CYP Substrates: Based 011 in vitro data ceritinib 111a inhibit CYP3A and CYP2C9 at clinical concentrations [see Drug Interactions Time-dependent inhibition of CYP3A was also observed. Effect of Transporters on eritinib Disposition: eritinib is a substrate of cf?ux transporter P-gp. but is not a substrate of Breast Cancer Resistance Protein (BC RP). Multidrug Resistance Protein MRP2 . Or anic Cation Trans orter' . Or anic Anion Trans orter OAT2 . or Or anic Anion Transporting Polypeptide l) in vitro. Drugs that inhibit P-gp rnav increase ceritinib concentrations. ect Ceritinib on Trans orters: Based 011 in vitro data. ceritinib does not inhibit a ical ef?ux transporters. P-gp. BC RP. or MRP2. hepatic uptake transporters and OATP1B3. renal organic anion uptake transporters and OAT3. or organic cation uptake transporters and OCT2 at clinical concentrations. Effect of Gastr 1c Acid Reducing Agents on Ceritinib. Gastric acid ireducing agents g. proton prunp bioavailabilitv as ceritinib demonstrates pH- dependent solubilitv and becomes poorlv as pH increases in vitro. A dedicated study has not been conducted to evaluate the effect of gastric acid reducing agents 011 the bioavailabilitv of ceritinib. NDA 205755/0000 Page 52 of 112 Reference ID: 3477231 4 APPENDICES 4.1 PHARMACOMETRICS REVIEW NDA 205755/0000 Page 53 of 112 Reference ID: 3477231 OFFICE OF CLINICAL PHARMACOLOGY: PHARMACOMETRIC REVIEW PHARMACOMETRIC REVIEW NDA Submission Type Submission Date Generic Name Applicant Primary Pharmacometric Reviewer Secondary Pharmacometric Reviewer Clinical Division 205755 Original 12/24/2013 Ceritinib Novartis Pengfei Song, Ph.D. Qi Liu, Ph.D. Division of Oncology Products 2 (DOP2) Table of Contents 1 SUMMARY OF FINDINGS.......................................................................................................... 56 1.1 Background ....................................................................................................................... 56 1.2 Key Review Questions ...................................................................................................... 56 1.2.1 Are there significant exposure-response relationships for efficacy? .............. 56 1.2.2 Are there significant exposure-response relationships for safety?.................. 59 1.2.3 Based on population PK analyses, what covariates affect the systemic exposure of ceritinib? ...................................................................................... 70 1.2.4 Is the proposed dosing regimen acceptable for the accelerated approval? ..... 72 1.3 Recommendations ............................................................................................................. 72 1.4 1.5 2 Post Marketing Requirements or Commitments ............................................................... 72 Label Statements ............................................................................................................... 72 APPLICANT’S ANALYSES......................................................................................................... 75 2.1 Population Pharmacokinetic Analysis............................................................................... 75 2.1.1 2.1.2 2.1.3 2.2 Exposure-Response Analyses ........................................................................................... 82 2.2.1 2.2.2 2.2.3 3 Datasets ........................................................................................................... 75 Methods ........................................................................................................... 75 Results ............................................................................................................. 77 Methods ........................................................................................................... 83 Results ............................................................................................................. 83 The Applicant’s Conclusions .......................................................................... 92 REVIEWER’S ANALYSES .......................................................................................................... 93 3.1 Reference ID: 3477231 Exposure-Response Analysis ............................................................................................ 93 3.1.1 Objectives ........................................................................................................ 93 NDA 205755/0000 Page 54 of 112 3.1.2 3.1.3 4 Methods ........................................................................................................... 93 Datasets ........................................................................................................... 94 3.1.4 Results ............................................................................................................. 94 LISTING OF ANALYSES CODES AND OUTPUT FILES ........................................................ 95 Reference ID: 3477231 NDA 205755/0000 Page 55 of 112 1 SUMMARY OF FINDINGS 1.1 BACKGROUND Ceritinib (also known as LDK378) is an anaplastic lymphoma kinase (ALK) inhibitor. The applicant seeks an accelerated approval of ceritinib for the treatment of patients with NSCLC who have received crizotinib. The efficacy and safety evaluations of ceritinib are based on the registration Phase 1 trial LDK378X2101, in which the proposed dosing regimen of ceritinib [750 mg once daily (QD)] induced a high overall response rate (ORR) [44% (36%, 52%)] of early (median time to first response of approximately 6 weeks) and durable responses (median 7.1 months) in patients with ALK-positive NSCLC who were previously treated with an ALK inhibitor. In the registration trial, Grade 3 or worse adverse events were observed in 73% patients. Dosing interruptions occurred in 69% of patients and dose reductions occurred in 59% of patients for a median dose intensity of 636 mg/day (range: 226.7 mg/day to 750 mg/day). The rate of adverse events resulting in permanent discontinuation was 10%. The main purpose of this pharmacometric review is to evaluate the appropriateness of the proposed dosing regimen by addressing the following key questions: 1.2 KEY REVIEW QUESTIONS 1.2.1 Are there significant exposure-response relationships for efficacy? No. No significant exposure-response relationships were identified for the efficacy endpoints per central radiology review including the primary efficacy endpoint ORR and secondary efficacy endpoint progression free survival (PFS). Based on the ORR analysis results, it is unclear whether the plateau of the exposure-efficacy curve has been reached or not, due to the following two possible reasons: (1) the distribution of the available exposure data (primarily driven by the data for patients in the 750 mg dose group) may not be enough to adequately characterize the full exposure-efficacy curve, and (2) the small sample size may limit the robustness for the predicted exposure-efficacy relationship (Figure 1). Furthermore, there is no clear separation of PFS curves stratified by Css, trough quartiles (Figure 2). Exposure-response analyses were conducted, using the average observed Css, trough of ceritinib at dose levels ranging from 50 mg to 750 mg QD as exposure endpoint, for ORR (N=167) and progression free survival (PFS) (N=167) in patients with ALK-positive NSCLC who were previously treated with an ALK inhibitor in Trial LDK378X2101. The average observed Css, trough used for the efficacy (ORR, PFS) analyses is defined as the geometric mean of all evaluable steady-state trough concentrations (0-6 hours before dose) of each patient between Day15 of Cycle 1 and the minimum of last day of study drug (on or prior to the cut-off date) and date of progression/death (for patients who progressed/died) or date of last adequate tumor assessment (for patients who did not progress/die). Overall response rate (ORR) Exposure-response analyses were conducted, using logistic regression model with the average observed Css, trough of ceritinib at dose levels ranging from 50 mg to 750 mg QD as systemic exposure endpoint, for ORR (N=167) in patients with ALK-positive NSCLC who were previously treated with an ALK inhibitor in Trial LDK378X2101. Reference ID: 3477231 NDA 205755/0000 Page 56 of 112 No significant exposure-response relationships were identified for the primary efficacy endpoint ORR (P = 0.32), after controlling prognostic factors including ECOG status, race, brain metastasis, and baseline sum of longest diameter. Based on the ORR analysis results, it is unclear whether the plateau of the exposure-response curve has been reached or not, because (1) the distribution of available exposure data (primarily driven by the data for patients in the 750 mg dose group) may not be enough to adequately characterize the full exposure-efficacy curve, and (2) the small sample size limited the robustness for the predicted exposure-efficacy relationship (Figure 1). Figure 1. The relationship between the average observed Css, trough of ceritinib at dose levels ranging from 50 mg to 750 mg QD and overall response rate (ORR) per central radiology review in ALK-positive NSCLC patients who were previously treated with an ALK inhibitor in trial LDK378X2101. The solid black symbols represent the observed ORR per central radiology review in each quartile of average observed Css, trough for all subjects. The vertical black bars represent the 95% confidence interval (CI). The solid line and the dashed lines represent the logistic regression model predicted mean and 95% CI of the probability of ORR by average observed Css, trough (P = 0.32) for Caucasian patients with brain metastasis, ECOG status of 1, and a baseline sum of longest diameter of 8.1 cm (Please note that the predictions are not for all patients, therefore do not fit the solid black symbols which represent the observed ORR in each Css, trough quartile for all patients, regardless of the prognostic factors. The logistic regression model prediction is generated as follows: numeric covariates at the median, categorical covariates at the highest frequency level). The exposure range in each quartile of Css, trough is denoted by the horizontal blue line along with the number of responders/total number of patients in each quartile. * Note: The prognostic factors that were used in the logistic regression model were based on discussion with clinical and clinical pharmacology review team. Reference ID: 3477231 NDA 205755/0000 Page 57 of 112 Progression free survival (PFS) Exposure-response analyses were conducted, using the average observed SS, trough of ceritinib at dose levels ranging from exposure endpoint, for progression free survival (PFS) in patients with ALK-positive NSC LC who were previously treated with an ALK inhibitor in Trial LDK378X2101. Kaplan-Meier methodology was applied to the PPS using the average observed Cssmugh quartiles as a strati?cation factor. No clear exposure-PPS relationship was identi?ed. - - on- 02(566-809 Time(Months) Figure 2. Kaplan-Meier curve of progression ?ee survival (PFS) per central radiology review by SS, trough quartiles at dose levels ranging from ceritinib in ALK-positive NSC LC patients who were previously treated with an ALK inhibitor DA 205755/0000 Page 58 of 112 Reference ID: 3477231 1.2.2 Are there significant exposure-response relationships for safety? Yes. Results of exposure-response analyses suggested that higher systemic exposure is associated with more frequent (Figure 3) and earlier (Figure 4) overall Grade 3 or worse (G3+) adverse events (AEs), as well as with higher incidence of individual AEs such as G3+ alanine aminotransferase (ALT) elevation (upper, Figure 6), G3+ aspartate aminotransferase (AST) elevation (lower, Figure 6), and Grade 2 or worse (G2+) hyperglycemia (Figure 7). No significant relationships were identified between systemic exposure and overall G3+ gastrointestinal tract AEs (Figure 5) or individual AEs such as G2+ diarrhea (upper, Figure 8), possibly because systemic exposure is not a good predictor for GI tract AEs (i.e., the high concentration of drug in the GI tract may lead to GI tract AEs directly). Similarly to the relationship between the systemic exposure and incidence of G3+ AEs, higher systemic exposure is associated with more frequent and earlier dose reductions (Figure 9) or dose interruptions (Figure 9). Time to first dose change (dose reduction, dose interruption) was also evaluated (Figure 10). Higher systemic exposure is associated with earlier and more frequent dose reductions or dose interruption. Given that the permanent discontinuation due to AEs occurred in only 10% patients, the management of AEs via dose reductions and interruptions is effective in maintaining patients on study drug. Exposure-response analyses for safety were conducted using the average observed Css, trough of ceritinib as systemic exposure endpoint in patients with ALK-positive tumors who were treated with ceritinib (50-750 mg QD) in Trial LDK378X2101 (N=275). The average observed Css, trough used for the safety analyses is defined as the geometric mean of all evaluable steady-state trough concentrations (0-6 hours before dose) of each patient after Day 15 of Cycle 1 and before the occurrence of the safety events of interest (including AEs, dose reduction, dose interruption, etc). For patients without the safety events of concern, the Css, trough was calculated as the geometric mean of all evaluable steady-state trough concentrations (0-6 hours before dose) up to and on last day of study drug (on or prior to the cut-off date). The sample size may vary for each specific analysis because some patients may experience the safety event of concern before the availability of evaluable steady state trough concentrations. Reference ID: 3477231 NDA 205755/0000 Page 59 of 112 Exposure-response analyses for overall Grade 3 or worse (G3+) AEs Exposure-response analyses with Css, trough as the exposure variable were conducted for overall G3+ AEs using logistic regression model. Higher exposures is associated with higher incidence of overall G3+ AEs (P = 0.002), after controlling prognostic factors including prior ALK inhibitor treatment, ECOG status, race, and brain metastasis. Figure 3. The relationship between average observed Css, trough and Grade 3 or worse (G3+) AEs in patients with ALK-positive tumors who were treated with ceritinib (50750 mg QD). The solid black symbols represent the observed incidence of G3+ AEs in each quartile of average observed Css,trough for all subjects, regardless of the prognostic factors. The vertical black bars represent the 95% confidence interval (CI). The solid and dotted lines represent the logistic regression model predicted mean and 95% CI of incidence of G3+ AEs by average observed Css, trough (P = 0.002) for Caucasian patients with brain metastasis, ECOG status of 1, and prior treatment with ALK inhibitors. The exposure range in each quartile of average observed Css,trough is denoted by the horizontal blue line along with the number of patients who experienced G3+ AEs/total number of patients in each quartile. * Note: The prognostic factors that were used in the logistic regression model were based on discussion with clinical and clinical pharmacology review team. Reference ID: 3477231 NDA 205755/0000 Page 60 of 112 Time to ?rst AEs Time to ?rst AE was estimated based on Kaplan-Meier methodology using observed C55, trough quartiles as a strati?cation factor in patients with ALK-positive tumors who were treated with ceritinib (50-750 mg QD). The results suggest that higher exposures (Quartile 4) may be associated with earlier and more frequent AEs than lower exposures (Quartile 1), with no marked differences between the two intennediate concentration quartile ranges (Quartiles 2 and 3). It is noted that the prognostic factors may not be balanced across the quartiles (Table 10.0 PM) othade 3MB. 04 r?r 03 (849?1155 ngmL. n=57) 01 (14.0-612 n=57) Q2 (612-849 nymL. n=57) 04 (1155-2432 ngImL. n=58) Figure 4. Kaplan-Meier curve of time to first Grade 3 or worse ABS by quartiles of observed average C55, trough in patients with ALK-positive tumors who were treated with ceritinib (50 - 750 mg QD) Table 1. Summary of covariates and exposure for patients per quartile of average observed trough concentration at steady-state Css,trough for AE analyses Reference ID: 3477231 Parameters Quartile 1 Quartile 2 Quartile 3(n=57) Quartile 4 (n=58) Cu 1.34. Median [Range ](ng/mL) [14.0, 607] [612.844] [848, 1146] [1155. 2432] Mean:SD (ng/rnL) 399: 163 731166 1006:94 1382:237 Baseline Age (YearBaseline weight (kg) 73.7 15.7 71.4 14.9 67.0 14.5 60.6: 10.6 Baseline Body Mass 25.6 :48 24.7:39 24.0:3.9 22.5:2.9 Patients with prior 5856 7056 68% 6635 Fe male 5836 46 35 61 35 6235 Race Asian 735 16 36 41% 3635 Black 035 0?76 035 5% Caucacian 8956 7956 44% 4735 Other 035 056 233 2% ECOG Status 0 1956 2356 33% 34% 8156 77 56 67% 6633 Brain Metastasis 3756 5156 45% 5533 NBA 205755/0000 Page 61 of 112 Grade 3 or worse (G3+) GI tract AEs Exposure-response analyses using average observed Css,trough as the systemic exposure variable were conducted to evaluate whether the high systemic exposure is related to the higher incidence of G3+ GI tract AEs. Results suggested that no apparent trend was identified (P = 0.86), possibly because systemic exposure of ceritinib is not a good predictor for GI tract AEs (i.e., the locally high concentration of drug in the GI tract leads to GI tract AEs directly). Figure 5. The relationship between average observed Css, trough and Grade 3 or worse (G3+) GI tract AEs in patients with ALK-positive tumors who were treated with ceritinib (50-750 mg QD). The solid black symbols represent the observed incidence of G3+ GI tract AEs in each quartile of Css, trough for all subjects, regardless of the prognostic factors. The vertical black bars represent the 95% confidence interval (CI). The solid line and the area between dotted lines represent the logistic regression model predicted mean and 95% CI for incidence of G3+ AEs by average observed Css, trough (P = 0.86) for Caucasian patients with brain metastasis, ECOG status of 1, and prior treatment with ALK inhibitors. The exposure range in each quartile of Css, trough is denoted by the horizontal blue line along with the number of patients who experienced G3+ GI tract AEs /total number of patients in each quartile. Reference ID: 3477231 NDA 205755/0000 Page 62 of 112 Individual AEs The most common individual adverse reactions (incidence of at least 25%) were diarrhea, nausea, vomiting, abdominal pain, fatigue, decreased appetite, and constipation. The most common G3+ adverse reactions (incidence of at least 5%) were diarrhea, fatigue, alanine transaminase (ALT) elevation, aspartate transaminase (AST) elevation, hyperglycemia, and lipase (blood) increase. Exposure-response analyses were conducted for ALT or AST elevation, GI tract AEs, hyperglycemia, diarrhea, and fatigue using multivariate logistic regression models. Results suggested that increasing systemic exposure is associated with higher incidence of G3+ ALT elevation (P = 0.002) (upper, Figure 6), G3+ AST elevation ( P = 0.02) (lower, Figure 6), G2+ hyperglycemia (P = 0.004) (Figure 7) after adjusting with prognostic factors including prior ALK inhibitor treatment, ECOG status, race, and brain metastasis. However, no apparent trends were identified for G2+ diarrhea (P = 0.11) (upper, Figure 8) and G2+ fatigue (P = 0.92) (lower, Figure 8) with Css,trough of ceritinib. Reference ID: 3477231 NDA 205755/0000 Page 63 of 112 Grade 3+ ALT or AST elevation Multivaraite logistic regression model were conducted using average observed Css,trough as the exposure to evaluate whether the high systemic exposure is related to the higher incidence of G3+ ALT or AST elevation. Results suggested that increasing systemic exposure is associated with higher incidence of G3+ ALT elevation (P = 0.002) (upper, Figure 6), and G3+ AST elevation ( P = 0.02) (lower, Figure 6), after controlling race, ECOG status, brain metastasis, and prior ALK inhibitor treatment. BEST AVAILABLE COPY Figure 6. The relationship between average observed Css, trough and Grade 3 or worse (G3+) ALT (upper) or AST (lower) elevation in patients with ALK-positive tumors who were treated with ceritinib (50-750 mg QD). The solid black symbols represent the observed incidence of G3+ ALT or AST elevation in each quartile of Css, trough for all subjects, regardless of the prognostic factors. The vertical black bars represent the 95% confidence interval (CI). The solid line and the area between dotted lines represent the logistic regression model predicted mean and 95% CI for incidence of G3+ ALT elevation (P =0.002) or G3+ AST elevation (P = 0.02) by the average observed Css, trough for Caucasian patients with brain metastasis, ECOG status of 1, and prior treatment with ALK inhibitors. The exposure range in each quartile of Css, trough is denoted by the horizontal blue line along with the number of patients who experienced G3+ ALT or AST elevation /total number of patients in each quartile. Reference ID: 3477231 NDA 205755/0000 Page 64 of 112 Grade 2 or worse hyperglycemia Multivariate logistic regression model were conducted using average observed Css,trough as the exposure to evaluate whether the high systemic exposure is related to the higher incidence of G2+ hyperglycemia. Results suggested that increasing systemic exposure is associated with higher incidence of G2+ hyperglycemia (P = 0.004), after controlling race, ECOG status, brain metastasis, and prior ALK inhibitor treatment. Figure 7. The relationship between observed average Css, trough and Grade 2 or worse (G2+) hyperglycemia in patients with ALK-positive tumors who were treated with ceritinib (50750 mg QD). The solid black symbols represent the observed incidence of G2+ hyperglycemia in each quartile of Css, trough for all subjects, regardless of the prognostic factors. The vertical black bars represent the 95% confidence interval (CI). The solid line and the area between dotted lines represent the logistic regression model predicted mean and 95% CI for the incidence of G2+ hyperglycemia by the average observed Css, trough (P = 0.004) for Caucasian patients with brain metastasis, ECOG status of 1, and prior treatment with ALK inhibitors. The exposure range in each quartile of the average observed Css, trough is denoted by the horizontal blue line along with the number of patients who experienced G2+ hyperglycemia/total number of patients in each quartile. Reference ID: 3477231 NDA 205755/0000 Page 65 of 112 Grade 2 or worse diarrhea/fatigue Multivariate logistic regression model were conducted using average observed Css,trough as the systemic exposure to evaluate whether the high systemic exposure is related to the higher incidence of G2+ diarrhea or fatigue. Results suggested that increasing systemic exposure is not significantly associated with higher incidence of G2+ diarrhea (P = 0.11) or fatigue (P = 0.92), after controlling race, ECOG status, brain metastasis, and prior ALK inhibitor treatment. BEST AVAILABLE COPY Figure 8. The relationship between average observed Css, trough and Grade 2 or worse (G2+) diarrhea (upper) or fatigue (lower) in patients with ALK-positive tumors who were treated with ceritinib (50-750 mg QD). The solid black symbols represent the observed incidence of G2+ diarrhea/fatigue in each quartile of average observed Css, trough for all subjects, regardless of the prognostic factors. The vertical black bars represent the 95% confidence interval (CI). The solid line and the area between dotted lines represent the logistic regression model predicted mean and 95% CI for the incidence of G2+ diarrhea (P = 0.11) or fatigue (P = 0.92) by the average observed Css, trough for Caucasian patients with brain metastasis, ECOG status of 1, and prior treatment with ALK inhibitors. The exposure range in each quartile of Css,trough is denoted by the horizontal blue line along with the number of patients who experienced Grade 2+ diarrhea or fatigue /total number of patients in each quartile. Reference ID: 3477231 NDA 205755/0000 Page 66 of 112 Dose adjustments Dosing interruptions occurred in 69% of patients and dose reductions occurred in 59% of patients. The rate of adverse events resulting in permanent discontinuation was 10%. Dose reduction or interruption for adverse reactions of diarrhea, nausea, and vomiting was needed in 16%, 20%, and 16% of patients, respectively. Similarly to the relationship between the systemic exposure and incidence of Grade 3+ AEs, higher systemic exposure is associated with more frequent and earlier dose reductions (upper, Figure 9 ) (P < 0.0001) or dose interruptions (lower, Figure 9) (P=0.06). Time to first dose change (dose reduction, dose delay) was estimated based on Kaplan-Meier methodology using Css,trough quartile as a stratification factor and including all ALK-positive cancer patients. Results suggested that higher systemic exposure appears to be associated with earlier and more frequent dose reduction (upper, Figure 10) or dose interruption (lower, Figure 10). Given that the permanent discontinuation due to AEs occurred in only 10% patients, the management of AEs via dose reductions and study drug interruptions is effective to maintain patients on study drug. Reference ID: 3477231 NDA 205755/0000 Page 67 of 112 Exposure-response analyses for dose reduction and dose interruption (delay) Multvariate logistic regression model were conducted using average observed Css,trough as the systemic exposure endpoint to evaluate whether the high systemic exposure is related to dose reduction and dose interruption. Results suggested that increasing systemic exposure may be associated with higher incidence of dose reduction (P< 0.0001) and dose interruption (P = 0.06), after controlling race, ECOG status, brain metastasis, and prior ALK inhibitor treatment. BEST AVAILABLE COPY Figure 9. The relationship between observed average Css, trough and the probability of dose reduction (upper) or dose delay (lower) in patients with ALK-positive tumors who were treated with ceritinib (50-750 mg QD). The solid black symbols represent the observed proportion of patients who experienced dose reduction (upper) or delay (lower) in each quartile of Css, trough for all subjects, regardless of the prognostic factors. The vertical black bars represent the 95% confidence interval (CI). The solid line and the area between dotted lines represent the logistic regression model predicted mean and 95% CI for the probability of dose reduction (p < 0.0001) or dose delay (P = 0.06) by the average observed Css,trough for Caucasian patients with brain metastasis, ECOG status of 1, and prior treatment with an ALK inhibitor. The exposure range in each quartile of Css, trough is denoted by the horizontal blue line along with the number of patients who experienced dose reduction or dose delay /total number of patients in each quartile. Reference ID: 3477231 NDA 205755/0000 Page 68 of 112 Kaplan-Meier analyses for time to ?rst dose change Time to ?rst dose change was evaluated separately for dose reduction and dose interruption, based on Kaplan-Meier methodology using the average observed Cssmugh quartiles as a strati?cation factor in patients with ALK-positive tlunors who were treated with ceritinib (50-750 mg QD). Results suggested that higher systemic exposure appear to be associated with earlier and more frequent dose reduction (upper, Figure 10) or dose interruption (lower, Figure 10). 01 (140-599 non-IL. n=ea) 02 (599-372 nohnL n42) - 03 (872-1165 nan. n=62) 04 (1165-2432 0:62) . 1.0 a" I I'-I-I- 1k. A 3? pp m-urn-P1Tll'le?h??s) . I d- a 3 I'hl 8 8 . o" I l:Lll. ?1 4H- n. I jamTme(Morlhs) Figure 10. Kaplan-Meier curve of time to ?rst dose reduction (upper), dose interruption (lower) by quartiles of average observed SS, trough in patients with ALK- positive tumors who were treated with ceritinib (50-750 mg QD) DA 205755/0000 Page 69 of 112 Reference ID: 3477231 1.2.3 Based on population PK analyses, what covariates affect the systemic exposure of ceritinib? Based on the applicant’s population PK analyses, no covariates (including age, gender, race, and body weight, etc) are considered clinically important. Therefore, no dose adjustments are needed for these covariates. The final population PK model of ceritinib was described by a one-compartment model with delayed first-order absorption and time-dependent elimination by allowing clearance to decrease stepwise over time. The final population PK model contained effect of body weight and baseline albumin on apparent clearance CL/F, body weight on apparent volume of distribution V/F, and concomitant use of H2 receptor antagonists (H2RA) or proton pump inhibitors (PPI) on the absorption rate constant (ka) (Figure 11). Among all the significant covariates included in the final model, body weight had the largest effect on ceritinib PK. The incorporation of these covariates led to the slight decrease in inter-individual variability (IIV) for CL/F (from 36.0% to 33.9%) and ka (from 86.4% to 84.0%) compared to the base model. The proposed flat dose of 750 mg once daily is acceptable, as the exploratory analyses suggested that body size has no clinically relevant impact on PK. The predicted AUCss in patients with lower body weight (< 60 kg) was 1.20-fold (90% CI: 1.05-1.37) higher than the reference population (body weight 60-80 kg). The AUCss in patients with higher body weight (> 80 kg) was 0.85-fold (90% CI: 0.73-0.98) lower. The effect on the systemic exposure of LDK378 of baseline liver function (classified based on the NCI-ODWG criteria) was also evaluated. The AUCss of LDK378 in patients with mild hepatic impairment (N=48) was similar to the AUCss in patients with normal hepatic function (N=254). Renal function was not retained as a covariate in the final population PK model (Figure 12). The AUCss in patients with mild or moderate renal impairment were predicted to have a 1.09-fold (90% CI: 0.97-1.25) and 1.19-fold (90% CI: 0.95-1.49) increase in AUCss compared to patients with normal renal function. Reference ID: 3477231 NDA 205755/0000 Page 70 of 112 Figure 11. Fold change of LDK378 steady-state exposure (AUCss) relative to reference group based on the final population pharmacokinetic model Figure 12. Effect of renal function on LDK378 steady-state exposure (Cmin,ss and AUCss) based on the final population pharmacokinetic model Reference ID: 3477231 NDA 205755/0000 Page 71 of 112 1.2.4 Is the proposed dosing regimen acceptable for the accelerated approval? Yes. Based on currently available data, the proposed starting dose of 750 mg QD is acceptable from a clinical pharmacology perspective. At this dose, ceritinib induces a high overall response rate (ORR) [44% (36%, 52%)] of early (median time to first response of approximately 6 weeks) and durable responses [median 7.1 months] in patients with ALK-positive NSCLC who were previously treated with an ALK inhibitor. The results of exposure-analyses for efficacy did not show a clear relationship between systemic exposure and primary efficacy endpoints ORR or secondary efficacy endpoint PFS. However, the high ORR rate and durable response at the proposed dose suggested that the proposed dose is efficacious. Higher systemic exposure appears associated with more frequent and earlier overall Grade 3/4 AEs, as well as with higher incidence of individual AEs such as Grade 3/4 alanine aminotransferase (ALT) elevation, Grade 3/4 aspartate aminotransferase (AST) elevation, and Grade 2/3/4 hyperglycemia. Higher systemic exposure also appears to be associated with earlier and more frequent dose reductions or dose interruptions. Given that the permanent discontinuation due to AEs occurred in only 10% patients, the management of AEs via dose reductions and interruptions is effective in maintaining patients on study drug for as long as it is clinically indicated. 1.3 RECOMMENDATIONS The pharmacometric reviewer finds that the NDA205755 is acceptable from a clinical pharmacology perspective, provided that a satisfactory agreement is reached between the Applicant and the Agency regarding the labeling language. 1.4 POST MARKETING REQUIREMENTS OR COMMITMENTS None. 1.5 LABEL STATEMENTS Only relevant clinical pharmacology sections are included. Reference ID: 3477231 NDA 205755/0000 Page 72 of 112 Applicant?s Proposed Language FDA proposed 8.6 Hepatic Impairment 8.6 Hepatic Impairment [see Clinical Pharmacology 12.3 Pharmacokinetics 12.3 Pharmacokinetics Speci?c Populations Speci?c Populations for patients with moderate or severe hepatic impairment [see Use in Speci?c Populations (8. 6) . NDA 205755I0000 Page 73 of 112 Reference ID: 3477231 (b) (4) impairment has not been conducted. Based on a population pharmacokinetic analysis of 48 patients with mild hepatic impairment (total bilirubin ≤ULN and AST >ULN or total bilirubin >1.0 to 1.5 × ULN and any AST) and 254 patients with normal hepatic function (total bilirubin ≤ULN and AST ≤ULN), ceritinib exposures were similar in patients with mild hepatic impairment and normal hepatic function. The pharmacokinetics of ceritinib have not been studied in patients with moderate or severe hepatic impairment [see Use in Specific Populations (8.6)]. Renal impairment: A pharmacokinetic trial in patients with renal impairment has not been conducted as ceritinib elimination via the kidney is low (1.3% of a single oral administered dose). Based on a population pharmacokinetic analysis of 97 patients with mild renal impairment (CLcr 60 to <90 mL/min), 22 patients with moderate renal impairment (CLcr 30 to <60 mL/min) and 183 patients with normal renal function (≥90 mL/min), ceritinib exposures were similar in patients with mild and moderate renal impairment and normal renal function. Patients with severe renal impairment (CLcr <30 mL/min) were not included in the clinical trial. Reference ID: 3477231 NDA 205755/0000 Page 74 of 112 2 APPLICANT’S ANALYSES The applicant performed population PK analyses to identify significant factors affecting ceritinib PK in a study report entitled “Population pharmacokinetics of LDK378 in adult patients with tumors characterized by genetic abnormalities in anaplastic lymphoma kinase (ALK)”. The applicant also performed exposure-response analyses for efficacy and safety data submitted in the original NDA submission on December 24, 2013. In response to the FDA clinical pharmacology information requests, the applicant submitted additional exposure-response analyses results on February 14, 19, 20, and 25 in response to the FDA clinical pharmacology information requests. The key findings from the Applicant’s analyses are summarized below: 2.1 POPULATION PHARMACOKINETIC ANALYSIS The primary objective of the population PK analysis was describe the pharmacokinetics (PK) of ceritinib in adult patients with tumors characterized by genetic abnormalities in ALK using a population PK approach and to investigate the effects of intrinsic and extrinsic factors that may affect the PK of ceritinib in this population. 2.1.1 Datasets The population PK analysis was performed with 4406 ceritinib concentration values from 302 patients in the registration study LDK378X2101. The study design, study population, and timing of blood samples are summarized in Table 3-1. 2.1.2 Methods The population PK analysis was performed using the first-order conditional estimation with interaction (FOCEi) as implemented in NONMEM software (Version VII, level 2.0) compiled with Intel Fortran Compiler (Version 11.1) on the MODESIM high performance computing Reference ID: 3477231 NDA 205755/0000 Page 75 of 112 environment. Diagnostic graphics and post-processing of NONMEM output were performed using the S-Plus software (Version 8.1, TIBCO Software Inc., Palo Alto, CA) and R (Version 2.13.2, http://www.r-project.org/index.html). The population PK model was developed following 3 steps: base model, full model with covariates, and final model, as briefly described below. Base model development The base model for ceritinib consists of the following sub-models: • Structural model Three alternative models were evaluated in an attempt to adequately but parsimoniously capture the time- and concentration-dependent auto-inhibition of LDK378: 1) PK-enzyme turnover model 2) first-order inhibition model 3) time-dependent elimination model with ceritinib apparent oral clearance (CL/F) stepwise decreasing after C1D8 and C2D1 Based on the performance of the above three models in regards of successful NONMEM run, Schwartz’s Bayesian Criterion (SBC), diagnostic plots, and over-parameterization evaluation, the third model was selected for the full model and final model development. Given that the steady-state of ceritinib was achieved by approximately C2D1 and full PK concentration-time profiles were collected on C1D8 and C2D1, in the third model, LDK378 CL/F from C1D8 inclusive to C2D1 (CL/F1) was parameterized with CL/F prior to C1D8 (CL/F0), and the fraction coefficient chgA (Equation 8). CL/F from C2D1 inclusive onwards (CL/F2) was parameterized with CL/F0 and the fraction coefficients chgA and chgB (Equation 9). The fraction coefficients were described using a logit model (Equations 10 and 11) to ensure the posterior individual estimates of these coefficients were constrained from 0 to 1. • Random effects model Inter-individual variability (IIV) in model parameters was modeled as lognormal distribution: where θi is the value of a compartment model parameter for the ith individual, θTV is the typical value of the model parameter, ηi denotes the inter-individual random effect accounting for the ith individual’s deviation from the typical value, and ηi ~ N(0, ω2) is a realization of a normally Reference ID: 3477231 NDA 205755/0000 Page 76 of 112 distributed random variable with zero mean and variance ω2. The IIV is reported as approximate percent coefficient of variation (%CV), calculated as: • Residual error model Residual variability was described by a combined proportional and additive error model: where Yij denotes the observed ceritinib concentration for the ith individual at time tj, Ŷij denotes the corresponding predicted concentration based on the PK model, w is the standard deviation of ceritinib concentration in plasma, and εij denotes the intraindividual (residual) random variable, which is assumed to have zero mean and variance 1. W is parameterized in terms of proportional (θprop) and additive (θadd) components. Full model development Full model development proceeded by a heuristic, minimal backwards elimination process. First, all PK-covariate relationships listed in Table 4-1 were included in the full model as they were considered to be relevant based on clinical and pharmacological judgment and collinearity of covariates. Then some of these PK-covariate relationships were eliminated if the estimates of covariate parameters appeared to be very weak and/or pharmacological implausible. Final model development The final model was developed from the full model by stepwise backward elimination. Covariate effects were subjected to a backward elimination algorithm using the likelihood ratio test (based on differences in the NONMEM objective function values, ΔOFV) to assess the significance of their effects when excluded one at a time. 2.1.3 Results The population PK of ceritinib was described by a one-compartment model with delayed firstorder absorption and time-dependent elimination that could be described by auto-inhibition or more heuristically but more parsimoniously, by allowing clearance to decrease stepwise over time. The final model contained effect of body weight and baseline albumin on apparent clearance CL/F, body weight on apparent volume of distribution V/F, and concomitant use of H2 receptor antagonists (H2RA) or proton pump inhibitors (PPI) on the absorption rate constant (ka). Model-based simulation results showed that body weight has the largest effect on ceritinib exposure. The AUCss in patients with body weight <60 kg was estimated to be 1.20-fold higher than in the reference population with body weight 60-80 kg (90% prediction interval (PI): 1.051.37). However, this magnitude of exposure increase was not considered to be clinically relevant. Reference ID: 3477231 NDA 205755/0000 Page 77 of 112 Although renal function was not retained as a covariate in the final model, model-based simulation identified a mild trend of increase in ceritinib exposure with decrease in baseline creatinine clearance. The AUCss in patients with mild or moderate renal impairment were predicted to have a 1.09-fold (90%PI: 0.97-1.25) and 1.19-fold (90%PI: 0.95-1.49) increase in AUCss compared to patients with normal renal function. Reference ID: 3477231 NDA 205755/0000 Page 78 of 112 BEST AVAILABLE COPY Figure 13: Predicted versus observed goodness-of-fit plots for the final PopPK model Reference ID: 3477231 NDA 205755/0000 Page 79 of 112 Figure 14: Plot of inter-individual random effects versus covariates on CL/F in the final PK model Reference ID: 3477231 NDA 205755/0000 Page 80 of 112 Figure 15: Plot of inter-individual random effects versus covariates on V/F in the final PK model Figure 16: Plot of inter-individual random effects of covariates on Ka in the final PK model Reference ID: 3477231 NDA 205755/0000 Page 81 of 112 Reviewer’s comment: The population PK analysis followed a reasonable model selection and optimization process. The applicant’s population PK analysis is acceptable. 2.2 EXPOSURE-RESPONSE ANALYSES Objectives: The objectives of the exposure-response analyses were to explore the exposureresponse relationships for safety and efficacy, as well as exposure-dose change relationships. Reference ID: 3477231 NDA 205755/0000 Page 82 of 112 2.2.1 Methods Exposure-response analyses based on data from Study LDK378X2101 with a cut-off date of October 31, 2013 were conducted to explore the exposure-efficacy relationship of ceritinib in ALK-positive NSCLC patients, and the exposure-safety and exposure-dose change relationships of ceritinib in ALK-positive malignancies (NSCLC and non-NSCLC) patients using observed and predicted average steady-state trough concentrations. Efficacy endpoints included overall response rate (ORR) and progression-free survival (PFS) by investigator assessment for NSCLC patients in the efficacy analysis set (EAS), and duration of response (DOR) and time to first response by investigator assessment for NSCLC patients in the EAS with a confirmed response. Safety endpoints included all grade 3/4 AEs and grade 3/4 gastrointestinal AEs for selected preferred terms in the GI disorders system organ class. For safety endpoints, newly occurring grade 2/3/4 hepatic toxicity (AST, ALT, total bilirubin) and grade 2/3/4 hyperglycemia based on laboratory parameters were also assessed. All analyses were conducted separately for all ceritinib dose groups combined and for the ceritinib 750 mg dose group alone. Ceritinib dose change assessments included time to the first dose reduction, time to first dose interruption, and time to study drug discontinuation. All analyses were conducted separately for all ceritinib dose groups combined and for the ceritinib 750 mg dose group alone. The plasma PK exposure measure chosen for the exposure-response analyses was average observed steady-state trough concentration (Ctrough,ss_obs), The average observed Css, trough used for the efficacy (ORR and PFS) analyses is defined as the geometric mean of all evaluable steadystate trough concentrations (0-6 hours before dose) of each patient between Day15 of Cycle 1 and the minimum of last day of study drug (on or prior to the cut-off date) and date of progression/death (for patients who progressed/died) or date of last adequate tumor assessment (for patients who did not progress/die).. Exposure-response analyses for safety were conducted using the average observed Css, trough of ceritinib at dose levels ranging from 50 mg to 750 mg QD as exposure endpoint in all ALKpositive patients in Trial LDK378X2101. The average observed Css, trough used for the safety analyses is defined as the geometric mean of all evaluable steady-state trough concentrations (0-6 hours before dose) of each patient after Day 15 of Cycle 1 and before the occurrence of the safety event of interest (including AEs, dose reduction, dose interruption, etc). For patients without the safety events of interest, the Css, trough was calculated as the geometric mean of all evaluable steady-state trough concentrations (0-6 hours before dose) up to and on last day of study drug (on or prior to the cut-off date). In addition, population PK posthoc estimates of steady-state trough concentration (Ctrough,ss_pred) was also used as exposure endpoint for analyses. Ctrough,ss_pred is defined as the geometric mean of trough concentrations predicted from the population PK model at the nominal time of 0 hour on Day 1 of each cycle (starting from Cycle 2) up to eight months after the first dose of study drug. 2.2.2 Results Note: Only results using Ctrough,ss_obs as exposure endpoint were summarized below. Results using Ctrough,ss_pred as exposure endpoint were not shown due to their high similarity to those of Reference ID: 3477231 NDA 205755/0000 Page 83 of 112 Ctrough,ss_obs. Exposure-safety relationship Grade 3/4 AEs When the analysis was conducted with all dose groups combined, the logistic regression analysis revealed a positive association between G3+ AEs and ceritinib Ctrough,ss (observed), after adjusting for the other covariates (Figure 2-1). Ctrough,ss and race were identified as significant predictors of G3+ AEs, indicating that besides exposure, race may potentially have an impact on G3+ AEs. With a 200 ng/ml increase in ceritinib Ctrough,ss_obs and Ctrough,ss_pred, the estimated odds ratios of having a grade 3/4 AE is 1.32 (95% CI: 1.11, 1.57) and 1.35 (95% CI: 1.14, 1.60), respectively, in the presence of other covariates. Grade 3/4 gastrointestinal AEs There was no apparent association between ceritinib Ctrough,ss (observed and predicted) and grade 3/4 GI events. The number of events was low and similar across exposure quartiles. Reference ID: 3477231 NDA 205755/0000 Page 84 of 112 BEST AVAILABLE COPY Grade 2/3/4 hepatic elevations Newly occurring grade 2/3/4 ALT elevations appeared to be more frequent in patients with higher Ctrough,ss (observed and predicted). Similar trends were observed for grade 2/3/4 AST elevations (Figure 2-7 and Figure 2-8). Logistic regression analyses revealed a positive association between newly occurring grade 2/3/4 AST or ALT elevations and Ctrough,ss (observed and predicted), after adjusting for the other covariates. Ctrough,ss (observed and predicted) was identified as the only significant predictor of grade 2/3/4 ALT/AST elevations. With a 200 ng/ml increase in ceritinib Ctrough,ss_obs and Ctrough,ss_pred, the estimated odds ratio of having ALT elevations are 1.42 (95% CI: 1.21- 1.68) and 1.50 (95% CI: 1.27-1.76), respectively, in the presence of other covariates. The corresponding estimated odds ratio for AST elevations are 1.32 (95% CI: 1.13- 1.54) and 1.38 (95% CI: 1.181.60), respectively. Reference ID: 3477231 NDA 205755/0000 Page 85 of 112 BEST AVAILABLE COPY Grade 2/3/4 hyperglycemia There appeared to be a positive association between newly occurring grade 2/3/4 hyperglycemia based on laboratory parameters and ceritinib average Ctrough,ss (observed and predicted). With a 200 ng/ml increase in ceritinib Ctrough,ss_obs and Ctrough,ss_pred, the estimated odds ratio of having grade 2/3/4 hyperglycemia is 1.42 (95% CI: 1.13-1.78) and 1.39 (95% CI: 1.11-1.74), respectively. Ctrough,ss (observed and predicted) and age were identified as significant predictors of grade 2/3/4 hyperglycemia, indicating that besides exposure, age may potentially have an impact on this safety parameter. Time to first grade 3/4 AE The results suggest that higher exposures (≥Q3) are associated with earlier and more frequent grade 3/4 AEs than lower exposures (