Review of SAB Comments on EPA’s Toxicological Review of Libby Amphibole Asbestos (August 2011)
By Tony Ward, The University of Montana, and
Steve Ackerlund, Ackerlund, Inc.
Draft Toxicity Factors Proposed by EPA
The EPA has proposed cancer inhalation unit risk (IUR) factor for Libby Amphibole (LA) that is lower (i.e. less toxic) than the current asbestos standard and a first of its kind non-cancer reference concentration (RfC). The implications of these values to the cleanup is that the IUR is the more stringent value that will drive cleanup decisions at a 1 x 10-6 risk level, but the RfC would drive cleanup decisions if a 1 x 10-5 risk level is applied. Note that EPA generally seeks to develop remedies that achieve a cancer risk level of between 1 x 10-4 to 1 x 10-6 (one-in-ten thousand to one-in-one million), and a non-cancer hazard quotient of 1.
Toxicity Factor Type
Cancer (IUR), risk per fibers/cc
Non-cancer (RfC), fibers/cc
The current cancer IUR for asbestos was posted online in IRIS in 1988 (as measured by PCM). It is based on the central tendency—not the upper bound—of the risk estimates. Although other cancers have been associated with asbestos (e.g., laryngeal, stomach, ovarian), the current IUR for asbestos accounts for only lung cancer and mesothelioma.
The Libby, MT worker cohort was used to derive the new IUR. No data were available pertaining to cancer incidence or mortality in Marysville, OH, and mortality and exposure data for other populations exposed to Libby Amphibole asbestos are very limited. . Derivation of the new IUR is based on the upper bound estimate of risk as a health protective measure to account for certain unknowns. It uses much more detailed information on exposure to individuals than were available to support the development of the prior cancer IUR. EPA states that the Libby data set is a well-documented and well-studied cohort of workers with adequate years of follow-up to evaluate mesothelioma and lung-cancer mortality risks. Like all studies, some limitations were acknowledged, such as:
- Use of historical phase contrast microscopy (PCM) exposure measurements. Ambient air may have contained material other than asbestos that could have contributed to fibers counted by PCM. Therefore, it is possible that exposure estimates for some or possibly a large portion of the cohort are overestimated, and, therefore, the resulting IUR may be underestimated.
- Measurement error in exposure assessment and assignment. There was insufficient work history information to estimate exposures for many workers that required manipulation of the exposure data. Also, exposure measurements are imperfect. EPA estimates this led to a possible underestimated risk.
- Limited length of follow-up. The IUR for mesothelioma mortality could be larger than was estimated from existing data, since latency of mesothelioma can be as long as 60 years. The maximum length of follow-up was 46 years in this cohort.
- Small number of women and ovarian cancer. While asbestos is causally associated with increased risks of ovarian cancer, there were only 84 women in the whole cohort, and there were no deaths from ovarian cancer among 24 total deaths. The lack of observed ovarian cancer in this cohort may be a function of the limited number of female deaths in the cohort allowing for the possibility that exposure to Libby Amphibole asbestos could result in increased risk of ovarian cancer. However, it was not possible to estimate the magnitude of this underestimation on the total cancer risk.
- Potential residual confounding and effect modification. The unit risk of lung-cancer mortality estimated herein, and the combined mesothelioma and lung-cancer mortality IUR, would over-estimate the risk in any population that had a lower prevalence of smoking than that of the Libby worker cohort. Because the Libby worker cohort had a large prevalence of smokers and ex-smokers and no known nonsmokers developed lung cancer, it is also possible that estimated risk for lung cancer is actually risk for an interaction of lung cancer and smoking, and effects of smoking and asbestos are known to be between additive and multiplicative (see Section 4). However, the company imposed smoking ban, and the observation that there were many ex-smokers in the cohort, would tend to lessen risks that would have occurred if these individuals continued smoking.
Confidence in the principal study supporting the non-cancer RfC is considered medium. It is based on human, epidemiological data from Marysville, OH occupational exposure. Human exposure is generally preferred over laboratory studies on animals. However, as no studies are perfect, principal shortcomings noted in the study are:
- Small data set which reduces statistical power.
- Self reported disease which may lead to underreporting and a lower RfC.
- Older radiographic technology for identifying disease which may lead to underreporting and a lower RfC.
- Weak exposure data supporting an exposure-response relationship.
The SAB cover letter provides a polite presentation of what amounts to a potentially large request for additional assessment:
“The SAB finds the EPA’s draft assessment to be comprehensive and generally clear, logical and well-written. There are several areas that need more consideration, and we provide recommendations to further enhance the clarity and strengthen the scientific basis for the conclusions presented.”
In brief, additional assessments for the non-cancer RfC include:
- Consider additional literature. While recommended, the SAB stops short of suggesting these additional studies would be preferred studies for establishing a standard, so it seems this is unlikely to result in changes.
- More assessment and justification of disease indicators. SAB recommends that EPA consider any X-ray abnormalities as the outcome: LPT, diffuse pleural thickening (DPT), or asbestosis. SAB also suggests that the EPA compare and contrast pleural abnormalities among the Libby workers and the Minneapolis Community to further support the selection of the disease onset indicator.
- More justification on the exposure-response model. Again, they stop short of advocating for a different model; however, the outcome of this additional assessment once additional literature is considered is uncertain.
- Reassessment of uncertainty factors. Uncertainty factors are used to lower the RfC in response to uncertainty to ensure health protection. SAB argues EPA has been too cautious in some cases and not cautious enough in other cases. On balance, and if accepted as argued, this might lead to a slightly higher standard (i.e. considering the asbestos to be less toxic).
Additional assessments for the cancer IUR include:
- More detailed assessment of the mortality bias. While EPA recognized the issues associate with using mortality rather than incidence data, the SAB recommends more detailed discussion on how the use of mortality data impacts the IUR. More data on other major categories of mortality are requested to better judge the magnitude of any bias.
- Reconsideration of the exposure estimates. SAB has recommended alternative statistical measures for determining the exposure levels received by Libby workers. SAB hints that the arithmetic mean might be more appropriate than the geometric mean, which would lead to a higher (more toxic) IUR.
- Formal mode of action analyses. A number of different models have been developed to apply experimental data to define the dose-response relationship. The choice of model depends on how the cancer disease occurs, i.e. the model of action. The SAB has recommended more detailed assessment of the mode of action and consideration of several models in addition to the Poisson and Cox models used in the draft assessment. This request might lead to using a different model for low-dose extrapolation and a different IUR.
- Additional sensitivity analyses. The SAB recommends a more detailed discussion and justification on the use of the subcohort post-1959 and on the use of mortality data rather than incidence data. These assumptions by EPA may have resulted in an undercount of cases of lung cancer and mesothelioma and therefore underestimate of the IUR.
- More model fit analyses. The SAB contends that the agency has been overly constrained by reliance on model fit statistics as the primary criterion for model selection. The SAB recommends graphical display of the fit to the data for both the main models. They also recommend EPA present the fit for a broader range of models to provide a more complete and transparent justification of the current choice. The SAB also recommends that the EPA consider literature on epidemiological studies of other amphiboles for model selection for dose-response assessment, since the size of the Libby subcohort used in the exposure-response modeling is small.
- Definition of lifetime exposure. SAB recommends that the standard 70 year lifetime be used, rather than the 60 year period proposed by EPA based on the Libby data set. This change would increase the IUR (make it more toxic).
A large number of qualified experts have participated in EPA’s assessment and in the SAB review. The SAB recommendations reflect a detailed and through critique of EPA’s assessment. Importantly, there is expressed agreement on many of the big questions (e.g. choice of study for deriving toxicity factors, choice of the most sensitive toxic endpoint). Regarding the details, LATAG should recognize that the methodology for establishing toxicity factors has undergone considerable expansion and improvement that culminated in Guidelines for Carcinogen Risk Assessment(USEPA, 1995) and Benchmark Dose Technical Guidance (USEPA, 2012).. This work on LA seems to reflect the ongoing development of expertise and standards of care in the practical application of these new methodologies. While it is difficult to anticipate how the EPA will response to each particular comment, it does seem likely that the process will take considerable time and is likely to lead to some adjustment to the currently proposed toxicity factors. However, it seems unlikely that the new toxicity factors for LA will be significantly different from those used to craft the current remedies. Cleanup is likely to continue to be driven by the concern for cancer upon inhalation at roughly the same level that it has been.