Deriving an RfD from a BMDL10 and Comparing It with an Occupational Aerosol Exposure
A two-year rat drinking-water study of a non-genotoxic chlorinated by-product reports dose-related centrilobular hepatocellular hypertrophy and vacuolation. Benchmark dose modeling of liver hypertrophy gives a BMDL10 of 20 mg/kg-day; the NOAEL by pairwise testing is 30 mg/kg-day and the LOAEL 60 mg/kg-day. Only rats were tested, and there is no repeat-dose reproductive or developmental study. A mouse carcinogenicity study produced liver tumors only at the top dose (400 mg/kg-day), with a negative genotoxicity battery. Industrial hygiene sampling where the chemical is made gives an 8-hour TWA of 0.6 mg/m3 as respirable aerosol; the material is also a skin permeant and workers handle it without gloves.
1. Derive a health-protective oral RfD. State your choice of point of departure, the dosimetric adjustment, and each uncertainty factor with its justification.
Read the worked answer
Choice of POD. Use the BMDL10 (20 mg/kg-day) rather than the NOAEL (30). The BMDL is defined by the dose-response curve rather than by the arbitrary dose spacing of the study, is the lower confidence bound (therefore conservative relative to the central estimate), and is not inflated by a fortunate choice of tested doses. Had the BMDL exceeded the NOAEL, that discrepancy would itself signal model or endpoint problems worth investigating.
Dosimetric adjustment (rat to human). Default oral scaling uses body weight to the 3/4 power:
HED = 20 mg/kg-day x (0.25 kg / 70 kg)^0.25 = 20 x 0.244 ≈ 4.9 mg/kg-day
Uncertainty factors:
- Interspecies: 3 (residual, toxicodynamic only). The allometric scaling has already accounted for kinetic differences; applying a full 10 would double-count.
- Intraspecies: 10 to protect sensitive humans, including children, given drinking-water exposure.
- Database deficiency: 3 - a single species, a single sex-relevant design, and no reproductive/developmental data. A factor of 10 would be justified if the endpoint suggested a developmental concern; 3 is proportionate here.
RfD = 4.9 / (3 x 10 x 3) = 4.9 / 90 ≈ 0.05 mg/kg-day.
A defensible alternative of 4.9/(10 x 10 x 3) = 0.016 mg/kg-day exists if a full interspecies factor is retained; the reasoning, not the single number, is what is being tested.
2. Do the mouse liver tumors force a linear (cancer) extrapolation instead of the RfD approach? What would you need to decide?
Read the worked answer
Not automatically. A negative genotoxicity battery plus tumors confined to the top dose, in an organ showing hypertrophy and vacuolation at lower doses, is a pattern consistent with a non-genotoxic, threshold-mediated mode of action (enzyme induction, hepatocellular cytotoxicity, and compensatory proliferation). Under the 2005 EPA cancer guidelines, a nonlinear approach is permissible for such an agent, and the RfD-style evaluation can serve as the cancer assessment.
What must be established before relying on that:
- Is the tumor response above the concurrent and historical control range? A top-dose-only excess can be an artifact of an inadequate maximum tolerated dose or of survival differences.
- Is there a documented MOA with supportive key events (e.g., CAR/PXR activation, Cyp induction, increased replication) in the mouse, and does the human plausibility argument hold?
- Is the cancer-based value actually protective? If the linear default were applied, the resulting risk-specific dose must be compared with 0.05 mg/kg-day. If the linear value is lower, the RfD does not cover the cancer endpoint and the more conservative value governs.
Judgment: the worst answer is to ignore the tumors because the chemical is 'non-genotoxic,' or to accept the MOA narrative from the sponsor without dose-response and replication data. Request the full pathology tables, the historical control database, and any MOA studies; state the conclusion conditionally on them.
3. Convert the 0.6 mg/m3 occupational exposure to an intake and evaluate it against your RfD. What weight should the RfD carry for a worker population?
Read the worked answer
Conversion. Assuming 10 m3 inhaled per 8-hour shift and 70 kg body weight, with complete absorption of respirable aerosol:
Intake = (0.6 mg/m3 x 10 m3) / 70 kg = 0.086 mg/kg-day
This is roughly 1.7x the RfD of 0.05 mg/kg-day, before any dermal contribution - and the chemical is a skin permeant handled without gloves, so the true internal dose may be substantially higher. Normalising to the HED BMDL gives a margin of exposure of 4.9/0.086 ≈ 57, well short of the ~100-300 that a BMDL-based margin would typically require.
How much weight the RfD deserves here. The RfD applies to lifetime exposure of the general population, including children, and already contains a 10-fold intraspecies factor. Comparing a healthy working-age population with it is therefore formally conservative - but the conservatism is more than offset by (a) the omitted dermal route, (b) the 8-hour versus 24-hour averaging, and (c) the absence of reproductive/developmental data. So the correct conclusion is not 'the RfD is exceeded, therefore unsafe,' nor 'the RfD doesn't apply to workers, therefore fine.' It is that the margin is inadequate, that air concentration alone understates dose, and that the exposure must be controlled by enclosure/ventilation and glove selection, with biological monitoring if a suitable biomarker exists.