May 
  9, 
  2016 
  
 
  

Docket 
  ID 
  No. 
  EPA-­‐HQ-­‐ORD-­‐2013-­‐0430 
  
 
  
 
  

Dr. 
  Marry 
  Ross 
  
Deputy Director, National Center for Environmental Assessment
US Environmental Protection Agency
Washington, DC
Dear Dr. Ross: 
  
 
  
I 
  am 
  submitting 
  these 
  comments 
  because 
  I 
  was 
  surprised 
  and 
  shocked 
  when 
  I 
  read 
  EPA’s 
  
cancer 
  weight-­‐of-­‐evidence 
  (WOE) 
  conclusion 
  for 
  hexahydro-­‐1,3,5-­‐trinitro-­‐1,3,5-­‐triazine 
  
(RDX). 
  It 
  is 
  obvious 
  to 
  me 
  that 
  the 
  WOE 
  determination 
  for 
  RDX 
  is 
  inconsistent 
  with 
  both 
  the 
  
data 
  that 
  are 
  summarized 
  in 
  the 
  IRIS 
  Draft 
  Toxicological 
  Review 
  of 
  RDX 
  (EPA/635/R-­‐
15/034a) 
  and 
  with 
  the 
  cancer 
  descriptors 
  provided 
  in 
  EPA’s 
  2005 
  Guidelines 
  for 
  
Carcinogen 
  Risk 
  Assessment. 
   
   
  
 
  
The 
  Executive 
  Summary 
  of 
  the 
  IRIS 
  draft 
  review 
  of 
  RDX 
  concludes 
  “results 
  from 
  animal 
  
studies 
  provide 
  suggestive 
  evidence 
  of 
  carcinogenic 
  potential 
  for 
  RDX 
  based 
  on 
  positive 
  
trends 
  in 
  liver 
  and 
  lung 
  tumor 
  incidence 
  in 
  experimental 
  animals.” 
  This 
  statement 
  is 
  
incomplete 
  since 
  there 
  are 
  significant 
  increases 
  in 
  tumor 
  incidence 
  in 
  addition 
  to 
  the 
  
positive 
  trends 
  in 
  RDX 
  exposed 
  animals. 
   
  In 
  addition, 
  while 
  acknowledging 
  two 
  separate 
  
organ 
  sites 
  for 
  tumor 
  induction 
  by 
  RDX, 
  the 
  statement 
  neglects 
  to 
  note 
  that 
  carcinogenic 
  
effects 
  were 
  observed 
  in 
  two 
  species. 
  As 
  indicated 
  on 
  page 
  1-­‐73 
  of 
  Section 
  1, 
  the 
  data 
  for 
  
RDX 
  matches 
  the 
  descriptor 
  likely 
  to 
  be 
  carcinogenic 
  to 
  humans 
  because 
  it 
  “induced 
  dose-­‐
related 
  increases 
  in 
  tumors 
  in 
  two 
  species 
  (mouse 
  and 
  rat), 
  in 
  both 
  sexes, 
  and 
  at 
  two 
  sites 
  
(liver 
  and 
  lung).” 
  In 
  recognizing 
  the 
  multiple 
  site 
  carcinogenicity 
  of 
  RDX, 
  the 
  dose-­‐response 
  
data 
  for 
  liver 
  and 
  lung 
  tumors 
  in 
  female 
  B6C3F1 
  mice 
  from 
  the 
  study 
  by 
  Lish 
  et 
  al. 
  (1984) 
  
were 
  considered 
  to 
  be 
  reliable 
  for 
  estimating 
  carcinogenic 
  risk 
  from 
  oral 
  exposure 
  to 
  RDX 
  
(see 
  Section 
  2.3, 
  pages 
  1-­‐25 
  to 
  1-­‐31); 
  furthermore, 
  no 
  uncertainties 
  were 
  listed 
  for 
  the 
  
reported 
  liver 
  or 
  lung 
  tumor 
  responses 
  that 
  were 
  used 
  to 
  derive 
  the 
  cancer 
  risk 
  value 
  (see 
  
Table 
  2.8). 
   
  
 
  
Regarding 
  the 
  liver 
  tumor 
  effects 
  of 
  RDX, 
  the 
  draft 
  document 
  notes 
  that 
  “an 
  increased 
  
incidence 
  of 
  liver 
  tumors 
  was 
  observed 
  in 
  one 
  chronic 
  mouse 
  study 
  (Lish 
  et 
  al., 
  1984) 
  and 
  
one 
  of 
  two 
  chronic 
  rat 
  studies 
  (Levine 
  et 
  al., 
  1983b).” 
  In 
  female 
  mice 
  from 
  the 
  Lish 
  et 
  al. 
  
study 
  there 
  were 
  significant 
  increases 
  in 
  the 
  trend 
  and 
  incidences 
  of 
  hepatocellular 
  
adenomas 
  or 
  carcinomas, 
  while 
  in 
  male 
  mice 
  the 
  incidence 
  of 
  hepatocellular 
  adenomas 
  or 
  
carcinomas 
  was 
  higher 
  in 
  the 
  two 
  high 
  dose 
  groups 
  (42.4% 
  and 
  48.1%) 
  compared 
  to 
  
controls 
  (31.7%). 
  These 
  findings 
  show 
  consistency 
  for 
  the 
  liver 
  tumor 
  response 
  in 
  male 
  
and 
  female 
  mice. 
  After 
  the 
  reanalysis 
  of 
  the 
  liver 
  lesions 
  in 
  female 
  mice 
  (Parker 
  et 
  al., 
  
2006), 
  the 
  trend 
  was 
  still 
  significant. 
  The 
  RDX 
  draft 
  document 
  failed 
  to 
  note 
  in 
  the 
  text 
  or 
  in 
  
Table 
  1-­‐13 
  that 
  the 
  incidence 
  of 
  hepatocellular 
  adenomas 
  or 
  carcinomas 
  was 
  also 
  
significantly 
  increased 
  in 
  the 
  two 
  high 
  dose 
  groups 
  of 
  female 
  mice 
  compared 
  to 
  controls 
  
(controls: 
  1.5% 
  vs 
  15.6%, 
  and 
  12.9% 
  in 
  the 
  higher 
  dose 
  groups) 
  after 
  the 
  reanalysis. 
  In 
  

 male 
  rats, 
  there 
  was 
  a 
  significant 
  positive 
  trend 
  for 
  hepatocellular 
  carcinomas, 
  and 
  the 
  
incidence 
  of 
  liver 
  carcinomas 
  in 
  the 
  two 
  highest 
  RDX 
  treatment 
  groups 
  (3.6% 
  and 
  6.5%) 
  
exceeded 
  the 
  National 
  Toxicology 
  Program’s 
  (NTP) 
  historical 
  control 
  range 
  for 
  this 
  rare 
  
tumor 
  in 
  F344 
  rats 
  (0-­‐2%) 
  from 
  studies 
  conducted 
  during 
  that 
  same 
  time 
  period. 
  To 
  
evaluate 
  the 
  occurrence 
  of 
  rare 
  tumors 
  in 
  a 
  carcinogenicity 
  study 
  for 
  a 
  possible 
  chemical-­‐
related 
  effect, 
  it 
  is 
  important 
  to 
  compare 
  incidence 
  values 
  with 
  historical 
  control 
  rates. 
  
Using 
  NTP 
  historical 
  control 
  data, 
  the 
  trend 
  for 
  liver 
  carcinomas 
  in 
  RDX-­‐treated 
  F344 
  rats 
  
is 
  highly 
  significant 
  (p 
  = 
  0.003). 
   
  
 
  
Regarding 
  lung 
  tumors, 
  the 
  incidence 
  of 
  alveolar/bronchiolar 
  (A/B) 
  carcinomas 
  and 
  the 
  
combined 
  incidence 
  of 
  A/B 
  adenomas 
  and 
  carcinomas 
  in 
  the 
  Lish 
  et 
  al. 
  (1984) 
  study 
  
showed 
  a 
  significant 
  positive 
  trend 
  in 
  female 
  mice. 
   
  The 
  incidence 
  of 
  A/B 
  carcinomas 
  in 
  
male 
  mice 
  also 
  showed 
  a 
  significant 
  positive 
  trend. 
   
  The 
  draft 
  RDX 
  document 
  should 
  also 
  
have 
  noted 
  that 
  the 
  increased 
  incidence 
  of 
  A/B 
  carcinomas 
  in 
  the 
  high 
  dose 
  group 
  of 
  male 
  
mice 
  appears 
  to 
  be 
  significant 
  (18.5% 
  vs 
  4.8% 
  in 
  controls) 
  and 
  should 
  list 
  the 
  p 
  value 
  for 
  
this 
  increase. 
  The 
  above 
  findings 
  provide 
  consistency 
  for 
  the 
  lung 
  tumor 
  response 
  in 
  male 
  
and 
  female 
  mice. 
   
  
 
  
Based 
  on 
  the 
  analysis 
  of 
  the 
  liver 
  and 
  lung 
  tumor 
  responses 
  observed 
  in 
  mice 
  or 
  rats 
  
exposed 
  to 
  RDX, 
  it 
  is 
  not 
  clear 
  why 
  the 
  cancer 
  WOE 
  categorization 
  for 
  RDX 
  was 
  reported 
  in 
  
the 
  draft 
  document 
  as 
  suggestive 
  evidence 
  of 
  carcinogenic 
  potential 
  when 
  the 
  tumor 
  data 
  for 
  
RDX 
  clearly 
  meet 
  EPA’s 
  criteria 
  for 
  the 
  descriptor 
  likely 
  to 
  be 
  carcinogenic 
  to 
  humans: 
  
According 
  to 
  EPA’s 
  2005 
  Guidelines 
  for 
  Carcinogen 
  Risk 
  Assessment, 
  supporting 
  data 
  for 
  
the 
  descriptor 
  “likely 
  to 
  be 
  carcinogenic 
  to 
  humans” 
  may 
  include 
  “an 
  agent 
  that 
  has 
  tested 
  
positive 
  in 
  animal 
  experiments 
  in 
  more 
  than 
  one 
  species, 
  sex, 
  strain, 
  site, 
  OR 
  exposure 
  
route, 
  with 
  or 
  without 
  evidence 
  of 
  carcinogenicity 
  in 
  humans.” 
  The 
  tumor 
  data 
  on 
  RDX 
  
certainly 
  meets 
  the 
  criteria 
  covered 
  by 
  this 
  descriptor 
  and 
  is 
  much 
  stronger 
  than 
  all 
  of 
  the 
  
criteria 
  listed 
  in 
  EPA’s 
  cancer 
  guidelines 
  for 
  suggestive 
  evidence 
  of 
  carcinogenicity. 
   
  
 
  
In 
  attempting 
  to 
  justify 
  the 
  lower 
  cancer 
  WOE 
  category, 
  the 
  report 
  gives 
  a 
  misdirected 
  
emphasizes 
  the 
  number 
  of 
  hepatocellular 
  carcinomas 
  in 
  male 
  F344 
  rats 
  (two 
  in 
  the 
  8 
  and 
  
40 
  mg/kg 
  groups 
  and 
  one 
  in 
  the 
  controls). 
   
   
  However, 
  in 
  doing 
  so, 
  the 
  statement 
  neglects 
  to 
  
note 
  that 
  the 
  denominator 
  values 
  were 
  different 
  between 
  the 
  control 
  group 
  (55) 
  and 
  the 
  40 
  
mg/kg 
  group 
  (31). 
  With 
  proper 
  adjustment 
  for 
  differences 
  in 
  the 
  denominators, 
  incidence 
  
values 
  are 
  1.8% 
  and 
  6.5%, 
  respectively. 
  As 
  noted 
  above, 
  the 
  latter 
  value 
  exceeds 
  NTP’s 
  
historical 
  control 
  rate 
  for 
  this 
  tumor. 
   
  Surely 
  the 
  EPA 
  staff 
  and 
  managers 
  know 
  that 
  it 
  is 
  
inappropriate 
  to 
  highlight 
  tumor 
  numbers 
  that 
  are 
  not 
  adjusted 
  for 
  differences 
  in 
  the 
  
number 
  of 
  animals 
  at 
  risk. 
  Misleading 
  information 
  should 
  not 
  be 
  used 
  to 
  justify 
  the 
  lower 
  
cancer 
  WOE 
  descriptor. 
  
 
  
The 
  suggestive 
  evidence 
  of 
  carcinogenic 
  potential 
  descriptor 
  is 
  also 
  promoted 
  in 
  the 
  draft 
  
report 
  by 
  the 
  claim 
  that 
  “RDX 
  did 
  not 
  increase 
  the 
  incidence 
  of 
  carcinomas 
  at 
  any 
  other 
  site 
  
in 
  F344 
  or 
  Sprague-­‐Dawley 
  rats.” 
  However, 
  in 
  earlier 
  text 
  (pages 
  1-­‐60 
  to 
  1-­‐63), 
  it 
  was 
  
noted 
  that 
  “interpretation 
  of 
  results 
  from 
  this 
  study 
  [Hart, 
  1976 
  in 
  Sprague-­‐Dawley 
  rats] 
  is 
  
limited 
  by 
  the 
  comparatively 
  lower 
  doses 
  employed 
  in 
  the 
  study, 
  and 
  the 
  recording 
  of 
  
effects 
  only 
  at 
  the 
  control 
  and 
  high 
  dose 
  groups” 
  and 
  “the 
  lack 
  of 
  characterization 
  of 
  the 
  
test 
  material.” 
  These 
  omissions 
  in 
  the 
  draft 
  document 
  on 
  RDX 
  result 
  in 
  a 
  misleading 
  

 justification 
  for 
  the 
  selection 
  of 
  the 
  descriptor 
  suggestive 
  evidence 
  of 
  carcinogenic 
  potential. 
  
 
  
In 
  its 
  role 
  of 
  protecting 
  people 
  from 
  toxic 
  agents 
  in 
  the 
  environment, 
  it 
  is 
  EPA’s 
  obligation 
  
to 
  promote 
  policies 
  that 
  reduce 
  human 
  exposures 
  to 
  adverse 
  environmental 
  agents. 
  The 
  
faulty 
  cancer 
  evaluation 
  of 
  RDX 
  will 
  result 
  in 
  risk 
  management 
  decisions 
  that 
  protect 
  
polluters 
  rather 
  than 
  protecting 
  US 
  citizens 
  who 
  are 
  exposed 
  to 
  this 
  cancer-­‐causing 
  
chemical. 
   
  
 
  
The 
  RDX 
  document 
  does 
  a 
  poor 
  job 
  of 
  explaining 
  how 
  or 
  why 
  EPA 
  opted 
  for 
  the 
  suggestive 
  
evidence 
  WOE 
  descriptor. 
  I 
  urge 
  the 
  EPA 
  to 
  correct 
  the 
  serious 
  defects 
  in 
  this 
  IRIS 
  
assessment 
  in 
  a 
  timely 
  and 
  transparent 
  fashion 
  because 
  the 
  inappropriate 
  application 
  of 
  
the 
  cancer 
  descriptor 
  differs 
  markedly 
  from 
  evaluations 
  of 
  animal 
  cancer 
  data 
  performed 
  
by 
  the 
  NTP, 
  by 
  the 
  International 
  Agency 
  for 
  Research 
  on 
  Cancer, 
  and 
  previously 
  by 
  EPA; 
  
and 
  most 
  importantly, 
  it 
  sets 
  a 
  bad 
  precedent 
  for 
  future 
  assessments 
  of 
  environmental 
  
carcinogens. 
   
  
 
   
  
 
  
Sincerely, 
  
 
  
 
  
Ronald 
  L. 
  Melnick, 
  PhD 
  
Senior 
  Toxicologist, 
  retired 
  
National 
  Toxicology 
  Program 
  
National 
  Institute 
  of 
  Environmental 
  Health 
  Sciences