UNITED STATES ENVIRONMENTAL PROTECTION AGENCY

WASHINGTON, D.C. 20460      

	OFFICE OF PREVENTION, PESTICIDES

                                                                        
                   AND TOXIC SUBSTANCES

	

  SEQ CHAPTER \h \r 1 MEMORANDUM

Date:  March 15, 2011

SUBJECT:  Bromoxynil.  Benchmark Dose Analysis on Supernumerary Ribs in
Three Developmental Toxicity Studies  

PC Code:  035301, 035302	DP Barcode:   D374609

Decision No.: 425581	Registration No.: 264-1023

Petition No.: 9F7678	Regulatory Action: Section 3 Registration

Risk Assessment Type: Single Chemical/ Aggregate	Case No.: 2070

TXR No.:  0055648	CAS No.: 1689-84-5, 1689-99-2

MRID No.:  00116558, 40466802	40 CFR: 180.324



FROM:	  SEQ CHAPTER \h \r 1 Myron S. Ottley, Ph.D., Senior Biologist 

		Risk Assessment Branch III (RAB3)

		Health Effects Division (HED) (7509P)

THROUGH:	Paula Deschamp, Branch Chief 

		Risk Assessment Branch III (RAB3)

		Health Effects Division (HED) (7509P)

TO:		Bethany Benbow, RM#23

		Registration Division (7505P)

		

I.	BACKGROUND

In the previous risk assessment (D246944.mem, B. Madden 6/24/1998),
dose-response data on supernumerary ribs from three studies were
considered together in hazard characterization and hazard
identification.  Specifically, a NOAEL of 4 mg/kg/day from Rubin and 
Nyska (1982) was combined with a LOAEL of 5 mg/kg/day from Baker &
Cropping (1981).  Based on this combined evaluation, the previous
assessment described concern over the steepness of the dose-response
curve and as a result the FQPA 10X Safety Factor was retained.  In the
last few years, HED has increased its use of benchmark dose (BMD)
analysis in risk assessment.  In an effort to better characterize the
dose response relationship for supernumerary ribs in rat fetuses, a BMD
analysis was conducted on Studies A, B and C at EPA’s National Center
for Environmental Assessment by statistician Jeff Gift, Ph.D., using the
fetal and/or litter data available from each study report using EPA’s
Benchmark Dose Software (BMDS, Version 2.). NCEA provided the full
analysis for each study (including the BMD outputs and plots).  Only the
plots for the best BMDS run for each study presented in this memo. 
However, the complete results of all analyses are presented as
attachments to this memo.  

Briefly, because of the nature of the data available from each study,
different BMD analyses were conducted for each.  These are relatively
old studies and the individual pup data were not available for 2 of 3
studies.  In the case of developmental studies, it is typical for the
responses of pups in the same litter to be more similar to each other
than to the responses of pups in different litters (“intra-litter
correlation”, or “litter-effect”).  When conducting BMD analyses
of developmental studies, nested models are the preferred approach since
these models incorporate two parameters to address this issue:
litter-specific covariate (LSC) and intra-litter correlation (ILC).  The
highest confidence analysis is for the Rubin & Nyska study (study A)
where individual fetus data were available and a nested analysis was
completed.  The lack of individual animal data for the other two studies
(B & C) impacts the degree of confidence on the BMD analysis.    HED
can, however, use the BMD analysis of all three studies to provide an
improved evaluation of the dose response curve compared with that
performed in the previous risk assessment, and thus inform the magnitude
of the FQPA SF.

II.	SUMMARY of RESULTS

Study	Analysis	BMD mg/kg/day

BMDL mg/kg/day	Comment

Study A

(Rubin & Nyska 1982)

	nested dichotomous data analysis, using individual litter and fetus
data	13.66

8.34

	Availability of individual data from both litters and fetuses allows
integration of both into the same analysis

Study B

(Baker & Cropping 1981)

 	dichotomous data analysis, using the litter as the subject and #
litters with anomalies as the response	1.06

0.75

	Lack of individual litter and fetus data disallows any accounting for
litter variation or inter-litter correlation; this is the lowest
confidence BMD run

Study C

(Rogers et al. 1991)	continuous data analysis, using mean percent/litter
9.80

8.86

	Lack of individual litter and fetus data disallows any accounting for
litter variation or inter-litter correlation.  Analysis is based on mean
data only.



Study A:  Rubin & Nyska (1982)

The current DER states that the NOAEL and LOAEL for this study are 4
mg/kg/day and 12.5 mg/kg/day, respectively.  The data-set for this study
included both individual litter and individual fetus data, which allows
for the “nested dichotomous” analysis.  This analysis produced high
confidence results.  The benchmark response of 5% ‘extra risk’
(i.e., incidence above the background) was used here. The use of a BMR5
is normally used for developmental endpoints.  It is notable that the
BMD5 and BMDL5 are both higher than the NOAEL of 4 mg/kg/day.  

Study B: Baker and Cropping (1981)

There is low confidence in the results of the Study B analysis because
the data needed for meaningful and reliable analysis, such as individual
animal and litter variation, were not available from the study,
rendering the data less reliable.  However, visual inspection of the
graph shows large variability of the data, suggesting that there is no
meaningful difference between the control group and the low dose level
(5 mg/kg/day).

  

Study C: Rogers et al (1991) 

Similar to Baker and Cropping (1981), the individual pup data are not
available, thus limiting the utility of the BMD values.  However, visual
inspection of the plots shows a shallow dose-response relationship at
the low and mid dose such that the effects at the mid dose are not
meaningfully different from the control or low dose.

III.	DISCUSSION 

The previous risk assessment used study A and study B as co-critical
studies and established the NOAEL of 4 mg/kg/day based on study A, and
the LOAEL of 5 mg/kg/day based on Study B.  These values suggest a very
steep dose-response curve, which provided the basis for the retaining of
the 10X FQPA factor.

Since that time, use of BMD analyses in HED assessment has accelerated,
and the RAB3 Risk Assessment Team considered this as a good opportunity
to use a more refined dose-response approach.  BMD modeling uses the
entire dose-response curve, considers the variability of the entire
data-set, and takes sample size into account instead of just looking at
a single point (NOAEL or LOAEL) for establishing a point of departure. 
Thus, BMD modeling provides an objective re-consideration of the
available dose-response data.

Study A provided the individual fetal and litter data needed to assess
changes in incidence/litter of supernumerary ribs with dose level,
allowing a nested approach to analysis which integrates litter and fetal
data.  For the two studies without individual animal data (B and C),
visual inspection of the graphs suggest that the dose-response curves
are more shallow than previously interpreted. As such, this analysis
reduces, if not alleviates, the previous concern over the steepness of
the dose-response curve related to the supernumerary ribs and retention
of the FQPA 10X.  At this time, the RAB3 Team has concluded that HED can
have confidence in the NOAEL of 4 mg/kg/day and use it as the point of
departure for supernumerary ribs, without the need to retain a 10X FQPA
factor.  

It is important to note that this analysis is limited to a single
endpoint and was conducted for the purpose of evaluating the
dose-response data only from the supernumerary ribs and thus may not
appropriate for use in setting other points of departure.  Specifically,
the BMDL5 of 8.34 mg/kg/day it is not protective of the decreased body
weight gain observed at 5 mg/kg/day in F1 and F2 parents in a
multi-generational reproduction study,  or the decreased body weight
gain, panting and liquid feces in female dogs at 5 mg/kg/day in a 90-day
oral toxicity study.

IV.	ATTACHMENTS 

 Jeff Gift cover email

BandC 1981  -- BMD analysis of Baker & Cropping (1981) study data

Rand1982-BMR05 – BMD analysis of Rubin & Naska (1982) study data

Rand1982-BMR10 – BMD analysis of Rubin & Naska (1982) study data

Rogers1981-1Std – BMD analysis of Rogers, Francis et al. (1991) study
data

Data set MRID 00116558  Baker & Cropping 1981

Data set MRID 40466802  Rubin & Nyska 1982

Data set Rogers Francis et al 1981

Page   PAGE  3  of   NUMPAGES  5 

