The Potential Effects of Recall Bias and Selection Bias on the Epidemiological Evidence for the Carcinogenicity of Glyphosate.

Glyphosate is a widely used herbicide worldwide. The International Agency for Research on Cancer in 2015 declared that glyphosate is probably carcinogenic to humans, noting a positive association for non-Hodgkin lymphoma (NHL). The principal human data on glyphosate and NHL come from five case-control studies and two cohort studies. The case-control studies are at risk of recall bias resulting from information on exposure to pesticides being collected from cases and controls based on their memories. In addition, two of the case-control studies are additionally at risk of a form of selection bias that can exacerbate the effect of recall bias. Both biases are in the direction of making glyphosate appear carcinogenic. If odds ratios (ORs) are not biased and a pesticide plays no role in causing NHL, the probability that an OR for that pesticide is greater than 1.0 is approximately 0.5. The fractions of ORs for pesticides other than glyphosate that are greater than 1.0 in the case-control studies are 0.90 (n = 92), 0.90 (n = 152), 0.93 (n = 59), 0.76 (n = 140), and 0.53 (n = 54), the first two from studies that are at risk for both types of bias. In the two cohort studies, which are not subject to these biases, the comparable fractions for relative risks for all cancers are 0.51 (n = 70) and 0.48 (n = 158). These results comply closely with what would be expected if evidence for carcinogenicity of glyphosate in these studies results from statistical bias in the case-control studies.

Reanalysis of Diesel Engine Exhaust and Lung Cancer Mortality in the Diesel Exhaust in Miners Study Cohort Using Alternative Exposure Estimates and Radon Adjustment.

The Diesel Exhaust in Miners Study (DEMS) (United States, 1947-1997) reported positive associations between diesel engine exhaust exposure, estimated as respirable elemental carbon (REC), and lung cancer mortality. This reanalysis of the DEMS cohort used an alternative estimate of REC exposure incorporating historical data on diesel equipment, engine horsepower, ventilation rates, and declines in particulate matter emissions per horsepower. Associations with cumulative REC and average REC intensity using the alternative REC estimate and other exposure estimates were generally attenuated compared with original DEMS REC estimates. Most findings were statistically nonsignificant; control for radon exposure substantially weakened associations with the original and alternative REC estimates. No association with original or alternative REC estimates was detected among miners who worked exclusively underground. Positive associations were detected among limestone workers, whereas no association with REC or radon was found among workers in the other 7 mines. The differences in results based on alternative exposure estimates, control for radon, and stratification by worker location or mine type highlight areas of uncertainty in the DEMS data.

Bogen's Critique of Linear-No-Threshold Default Assumptions.

In an article recently published in this journal, Bogen(1) concluded that an NRC committee's recommendations that default linear, nonthreshold (LNT) assumptions be applied to dose- response assessment for noncarcinogens and nonlinear mode of action carcinogens are not justified. Bogen criticized two arguments used by the committee for LNT: when any new dose adds to a background dose that explains background levels of risk (additivity to background or AB), or when there is substantial interindividual heterogeneity in susceptibility (SIH) in the exposed human population. Bogen showed by examples that SIH can be false. Herein is outlined a general proof that confirms Bogen's claim. However, it is also noted that SIH leads to a nonthreshold population distribution even if individual distributions all have thresholds, and that small changes to SIH assumptions can result in LNT. Bogen criticizes AB because it only applies when there is additivity to background, but offers no help in deciding when or how often AB holds. Bogen does not contradict the fact that AB can lead to LNT but notes that, even if low-dose linearity results, the response at higher doses may not be useful in predicting the amount of low-dose linearity. Although this is theoretically true, it seems reasonable to assume that generally there is some quantitative relationship between the low-dose slope and the slope suggested at higher doses. Several incorrect or misleading statements by Bogen are noted.

Influence of Alternative Exposure Estimates in the Diesel Exhaust Miners Study: Diesel Exhaust and Lung Cancer.

The landmark Diesel Exhaust in Miners Study (DEMS) studied the relationship between diesel exhaust exposure (DEE) and lung cancer mortality of workers at eight nonmetal mines who were followed from beginning of dieselization of the mines (1947-1967) through December 31, 1997. The original analyses quantified DEE exposures using exposure to respirable elemental carbon (REC) to represent DEE, and CO as a surrogate for REC. However, this use of CO data, and the CO data themselves, have numerous shortcomings. We developed new estimates of REC exposures using historical data on use of diesel equipment, diesel engine horsepower (HP), mine ventilation rates, and the documented reduction in particulate matter emissions per HP in diesel engines from 1975 through 1995. These new REC estimates were applied in a conditional logistic regression of the DEMS nested case-control data very similar to the one applied in the original DEMS analyses. None of the trend slopes calculated using the new REC estimates were statistically significant (p > 0.05). Moreover, these trend slopes were smaller by roughly factors of five without control for radon exposure and factors of 12 with control for radon exposure compared to those estimated in the original DEMS analyses. Also, the 95% confidence intervals for these trend slopes had only minimal overlap with those for the slopes in the original DEMS analyses. These results underscore the uncertainty in estimates of the potency of diesel exhaust in causing lung cancer based on analysis of the DEMS data due to uncertainty in estimates of exposures to diesel exhaust.

Reanalysis of the DEMS nested case-control study of lung cancer and diesel exhaust: suitability for quantitative risk assessment.

The International Agency for Research on Cancer (IARC) in 2012 upgraded its hazard characterization of diesel engine exhaust (DEE) to "carcinogenic to humans." The Diesel Exhaust in Miners Study (DEMS) cohort and nested case-control studies of lung cancer mortality in eight U.S. nonmetal mines were influential in IARC's determination. We conducted a reanalysis of the DEMS case-control data to evaluate its suitability for quantitative risk assessment (QRA). Our reanalysis used conditional logistic regression and adjusted for cigarette smoking in a manner similar to the original DEMS analysis. However, we included additional estimates of DEE exposure and adjustment for radon exposure. In addition to applying three DEE exposure estimates developed by DEMS, we applied six alternative estimates. Without adjusting for radon, our results were similar to those in the original DEMS analysis: all but one of the nine DEE exposure estimates showed evidence of an association between DEE exposure and lung cancer mortality, with trend slopes differing only by about a factor of two. When exposure to radon was adjusted, the evidence for a DEE effect was greatly diminished, but was still present in some analyses that utilized the three original DEMS DEE exposure estimates. A DEE effect was not observed when the six alternative DEE exposure estimates were utilized and radon was adjusted. No consistent evidence of a DEE effect was found among miners who worked only underground. This article highlights some issues that should be addressed in any use of the DEMS data in developing a QRA for DEE.

Diesel engine exhaust and lung cancer mortality: time-related factors in exposure and risk.

To develop a quantitative exposure-response relationship between concentrations and durations of inhaled diesel engine exhaust (DEE) and increases in lung cancer risks, we examined the role of temporal factors in modifying the estimated effects of exposure to DEE on lung cancer mortality and characterized risk by mine type in the Diesel Exhaust in Miners Study (DEMS) cohort, which followed 12,315 workers through December 1997. We analyzed the data using parametric functions based on concepts of multistage carcinogenesis to directly estimate the hazard functions associated with estimated exposure to a surrogate marker of DEE, respirable elemental carbon (REC). The REC-associated risk of lung cancer mortality in DEMS is driven by increased risk in only one of four mine types (limestone), with statistically significant heterogeneity by mine type and no significant exposure-response relationship after removal of the limestone mine workers. Temporal factors, such as duration of exposure, play an important role in determining the risk of lung cancer mortality following exposure to REC, and the relative risk declines after exposure to REC stops. There is evidence of effect modification of risk by attained age. The modifying impact of temporal factors and effect modification by age should be addressed in any quantitative risk assessment (QRA) of DEE. Until there is a better understanding of why the risk appears to be confined to a single mine type, data from DEMS cannot reliably be used for QRA.

A statistical reevaluation of the data used in the Lanphear et al. ( 2005 ) pooled-analysis that related low levels of blood lead to intellectual deficits in children.

A pooled-analysis by Lanphear et al. (2005) of seven cohort studies of the association between blood lead (BPb) concentrations in children and measures of their intelligence concluded that "environmental lead exposure in children who have maximal blood lead levels <7.5 µg/dL is associated with intellectual deficits." This study has played a prominent role in shaping the public understanding of the effects upon children's IQ of low BPb exposures (e.g., BPb ≤ 10 µg/dL). Here we present a reanalysis of the data used by Lanphear et al. to evaluate the robustness of their conclusions. Our analysis differed from that of Lanphear et al. primarily in how we controlled for non-lead variables (allowing a number of them to be site-specific), how we defined summary measures of BPb exposure, and in how we decided which BPb measures and transformations best modeled the data. We also reproduced the Lanphear et al. analysis. Although we found some small errors and questionable decisions by Lanphear et al. that, taken alone, could cause doubt in their conclusions, our reanalysis tended to support their conclusions. We concluded that there was statistical evidence that the exposure-response is non-linear over the full range of BPb evaluated in these studies, which implies that, for a given increase in blood lead, the associated IQ decrement is greater at lower BPb levels. However at BPb below 10 µg/dL, the exposure-response is adequately modeled as linear. We also found statistical evidence for an association with IQ among children who had maximal measured BPb levels ≤7 µg/dL, and concurrent BPb levels as low as ≤5 µg/dL.

Evaluation of an exposure assessment used in epidemiological studies of diesel exhaust and lung cancer in underground mines.

NIOSH/NCI (National Institute of Occupational Safety and Health and National Cancer Institute) developed exposure estimates for respirable elemental carbon (REC) as a surrogate for exposure to diesel exhaust (DE) for different jobs in eight underground mines by year beginning in the 1940s-1960s when diesel equipment was first introduced into these mines. These estimates played a key role in subsequent epidemiological analyses of the potential relationship between exposure to DE and lung cancer conducted in these mines. We report here on a reanalysis of some of the data from this exposure assessment. Because samples of REC were limited primarily to 1998-2001, NIOSH/NCI used carbon monoxide (CO) as a surrogate for REC. In addition, because CO samples were limited, particularly in the earlier years, they used the ratio of diesel horsepower (HP) to the mine air exhaust rate as a surrogate for CO. There are considerable uncertainties connected with each of these surrogate-based steps. The estimates of HP appear to involve considerable uncertainty, although we had no data upon which to evaluate the magnitude of this uncertainty. A sizable percentage (45%) of the CO samples used in the HP to CO model was below the detection limit which required NIOSH/NCI to assign CO values to these samples. In their preferred REC estimates, NIOSH/NCI assumed a linear relation between C0 and REC, although they provided no credible support for that assumption. Their assumption of a stable relationship between HP and CO also is questionable, and our reanalysis found a statistically significant relationship in only one-half of the mines. We re-estimated yearly REC exposures mainly using NIOSH/NCI methods but with some important differences: (i) rather than simply assuming a linear relationship, we used data from the mines to estimate the CO-REC relationship; (ii) we used a different method for assigning values to nondetect CO measurements; and (iii) we took account of statistical uncertainty to estimate bounds for REC exposures. This exercise yielded significantly different exposure estimates than estimated by NIOSH/NCI. However, this analysis did not incorporate the full range of uncertainty in REC exposures because of additional uncertainties in the assumptions underlying the modeling and in the underlying data (e.g. HP and mine exhaust rates). Estimating historical exposures in a cohort is generally a very difficult undertaking. However, this should not prevent one from recognizing the uncertainty in the resulting estimates in any use made of them.

A meta-analysis of evidence for hormesis in animal radiation carcinogenesis, including a discussion of potential pitfalls in statistical analyses to detect hormesis.

A database containing 800 datasets on the incidence of specific tumor types from 262 radiation carcinogenicity experiments identified in a comprehensive literature search through September 2000 was analyzed for evidence of hormesis. This database includes lifetime studies of tumorigenic responses in mice, rats, and dogs to exposures to alpha, beta, gamma, neutron, or x-ray radiation. A J-shaped dose response, in the form of a significant decreased response at some low dose followed by a significant increased response at a higher dose, was found in only four datasets from three experiments. Three of these datasets involved the same control animals and two also shared dosed animals; the J shape in the fourth dataset appeared to be the result of an outlier within an otherwise monotonic dose response. A meta-analysis was conducted to determine whether there was an excess of dose groups with decreases in tumor response below that in controls at doses below no-observed-effect levels (NOELs) in individual datasets. Because the probability of a decreased response is generally not equal to the probability of an increased response even in the null case, the meta-analysis focused on comparing the number of statistically significant diminished responses to the number expected, assuming no dose effect below the NOEL. Only 54 dose groups out of the total of 2579 in the database had doses below the dataset-specific NOEL and that satisfied an a priori criterion for sufficient power to detect a reduced response. Among these 54, a liberal criterion for defining a significant decreases identified 15 such decreases, versus 54 × 0.2 = 10.8 expected. The excess in significant reductions was accounted for almost entirely by the excess from neutron experiments (10 observed, 6.2 expected). Nine of these 10 dose groups involved only 2 distinct control groups, and 2 pairs from the 10 even shared dosed animals. Given this high degree of overlap, this small excess did not appear remarkable, although the overlap prevented a formal statistical analysis. A comprehensive post hoc evaluation using a range of NOEL definitions and alternative ways of restricting the data entering the analysis did not produce materially different results. A second meta-analysis found that, in every possible low dose range ([0, d] for every dose, d) of each of the radiation types, the number of dose groups with significantly increased tumorigenic responses was either close to or exceeded the number showing significantly reduced responses. This meta-analysis was considered to be the more definitive one. Not only did it take dose into account by looking for consistent evidence of hormesis throughout defined low-dose ranges, it was also potentially less susceptible to limitations in experimental protocols that would cause individual animals to respond in a non-independent fashion. Overall, this study found little evidence in a comprehensive animal radiation database to support the hormesis hypothesis. However, the ability of the database to detect a hormetic effect was limited both by the small number of dose groups with doses below the range where positive effects have been found in epidemiological studies (≤ 0.1 Gy) and by the limited power of many of these dose groups for detecting a decrease in response.

Use of threshold and mode of action in risk assessment.

Under current guidelines, exposure guidelines for toxicants are determined by following one of two different tracks depending on whether the toxicant's mode of action (MOA) is believed to involve an exposure threshold. Although not denying the existence of thresholds, this paper points out problems with how the threshold concept and MOA is used in risk assessment. Thresholds are frequently described using imprecise terms that imply some unspecified increase in risk, which robs them of any meaning (any reasonable dose response will satisfy such a definition) and tacitly implies a value judgment about how large a risk is acceptable. MOA is generally used only to inform a threshold's existence and not its value. Often MOA is used only to conclude that the adverse effect requires an upstream cellular or biochemical response for which a threshold is simply assumed. Data to inform MOA often come from animals, which complicates evaluation of the role of human variation in genetic and environmental conditions, and the possible interaction of the toxicant with processes already producing background toxicity in humans. In response to these and other problems with the current two-track approach, this paper proposes a modified point of departure/safety factor approach to setting exposure guidelines for all toxicants. MOA and the severity of the toxic effect would be addressed using safety factors calculated from guidelines established by consensus and based on scientific judgment. The method normally would not involve quantifying low-dose risk, and would not require a threshold determination, although MOA information regarding the likelihood of a threshold could be used in setting safety factors.

Counting rules for estimating concentrations of long asbestos fibers.

Mounting evidence that long asbestos fibers (e.g. >20 or even 40 µm) pose the greatest cancer risk underscores the need for accurate measurement of concentrations of such fibers. These fiber lengths are of the same order of magnitude as the size of openings in the grids (typically ≈90 µm per side) used to analyze asbestos samples by transmission electron microscopy. This means that a substantial proportion of long fibers will cross the edge of a grid opening (GO) and therefore not be completely visible. Counting rules generally deal with such fibers by assigning a length equal to twice the visible length. Using both theoretical and simulation methods, we show that this doubling rule introduces bias into estimates of fiber concentrations and the amount of bias increases with fiber length. We investigate an alternative counting rule that counts only fibers that lie completely within a GO and weights those fibers by the reciprocal of the probability that a fiber of that length lies totally within a GO. This approach does not have the bias inherent in the doubling rule and is essentially unbiased if the stopping rule specifies a fixed number of GOs to be scanned. However, a stopping rule based on successively scanning GOs until a fixed number of fibers have been counted will introduce bias into any counting method, although this bias may typically not be large enough to be of practical concern. We recommend use of the weighted approach as a supplement to use of the doubling rule when estimating concentrations of long fibers, irrespective of the stopping rule employed.

Accounting for Multiple Comparisons in Statistical Analysis of the Extensive Bioassay Data on Glyphosate.

Glyphosate is a widely used herbicide worldwide. In 2015, the International Agency for Research on Cancer (IARC) reviewed glyphosate cancer bioassays and human studies and declared that the evidence for carcinogenicity of glyphosate is sufficient in experimental animals. We analyzed 10 glyphosate rodent bioassays, including those in which IARC found evidence of carcinogenicity, using a multiresponse permutation procedure that adjusts for the large number of tumors eligible for statistical testing and provides valid false-positive probabilities. The test statistics for these permutation tests are functions of p values from a standard test for dose-response trend applied to each specific type of tumor. We evaluated 3 permutation tests, using as test statistics the smallest p value from a standard statistical test for dose-response trend and the number of such tests for which the p value is less than or equal to .05 or .01. The false-positive probabilities obtained from 2 implementations of these 3 permutation tests are: smallest p value: .26, .17; p values ≤ .05: .08, .12; and p values ≤ .01: .06, .08. In addition, we found more evidence for negative dose-response trends than positive. Thus, we found no strong evidence that glyphosate is an animal carcinogen. The main cause for the discrepancy between IARC's finding and ours appears to be that IARC did not account for the large number of tumor responses analyzed and the increased likelihood that several of these would show statistical significance simply by chance. This work provides a more comprehensive analysis of the animal carcinogenicity data for this important herbicide than previously available.

Update of potency factors for asbestos-related lung cancer and mesothelioma.

The most recent update of the U.S. Environmental Protection Agency (EPA) health assessment document for asbestos (Nicholson, 1986, referred to as "the EPA 1986 update") is now 20 years old. That document contains estimates of "potency factors" for asbestos in causing lung cancer (K(L)'s) and mesothelioma (K(M)'s) derived by fitting mathematical models to data from studies of occupational cohorts. The present paper provides a parallel analysis that incorporates data from studies published since the EPA 1986 update. The EPA lung cancer model assumes that the relative risk varies linearly with cumulative exposure lagged 10 years. This implies that the relative risk remains constant after 10 years from last exposure. The EPA mesothelioma model predicts that the mortality rate from mesothelioma increases linearly with the intensity of exposure and, for a given intensity, increases indefinitely after exposure ceases, approximately as the square of time since first exposure lagged 10 years. These assumptions were evaluated using raw data from cohorts where exposures were principally to chrysotile (South Carolina textile workers, Hein et al., 2007; mesothelioma only data from Quebec miners and millers, Liddell et al., 1997) and crocidolite (Wittenoom Gorge, Australia miners and millers, Berry et al., 2004) and using published data from a cohort exposed to amosite (Paterson, NJ, insulation manufacturers, Seidman et al., 1986). Although the linear EPA model generally provided a good description of exposure response for lung cancer, in some cases it did so only by estimating a large background risk relative to the comparison population. Some of these relative risks seem too large to be due to differences in smoking rates and are probably due at least in part to errors in exposure estimates. There was some equivocal evidence that the relative risk decreased with increasing time since last exposure in the Wittenoom cohort, but none either in the South Carolina cohort up to 50 years from last exposure or in the New Jersey cohort up to 35 years from last exposure. The mesothelioma model provided good descriptions of the observed patterns of mortality after exposure ends, with no evidence that risk increases with long times since last exposure at rates that vary from that predicted by the model (i.e., with the square of time). In particular, the model adequately described the mortality rate in Quebec chrysotile miners and millers up through >50 years from last exposure. There was statistically significant evidence in both the Wittenoom and Quebec cohorts that the exposure intensity-response is supralinear(1) rather than linear. The best-fitting models predicted that the mortality rate varies as [intensity](0.47) for Wittenoom and as [intensity](0.19) for Quebec and, in both cases, the exponent was significantly less than 1 (p< .0001). Using the EPA models, K(L)'s and K(M)'s were estimated from the three sets of raw data and also from published data covering a broader range of environments than those originally addressed in the EPA 1986 update. Uncertainty in these estimates was quantified using "uncertainty bounds" that reflect both statistical and nonstatistical uncertainties. Lung cancer potency factors (K(L)'s) were developed from 20 studies from 18 locations, compared to 13 locations covered in the EPA 1986 update. Mesothelioma potency factors (K(M)'s) were developed for 12 locations compared to four locations in the EPA 1986 update. Although the 4 locations used to calculate K(M) in the EPA 1986 update include one location with exposures to amosite and three with exposures to mixed fiber types, the 14 K(M)'s derived in the present analysis also include 6 locations in which exposures were predominantly to chrysotile and 1 where exposures were only to crocidolite. The K(M)'s showed evidence of a trend, with lowest K(M)'s obtained from cohorts exposed predominantly to chrysotile and highest K(M)'s from cohorts exposed only to amphibole asbestos, with K(M)'s from cohorts exposed to mixed fiber types being intermediate between the K(M)'s obtained from chrysotile and amphibole environments. Despite the considerable uncertainty in the K(M) estimates, the K(M) from the Quebec mines and mills was clearly smaller than those from several cohorts exposed to amphibole asbestos or a mixture of amphibole asbestos and chrysotile. For lung cancer, although there is some evidence of larger K(L)'s from amphibole asbestos exposure, there is a good deal of dispersion in the data, and one of the largest K(L)'s is from the South Carolina textile mill where exposures were almost exclusively to chrysotile. This K(L) is clearly inconsistent with the K(L) obtained from the cohort of Quebec chrysotile miners and millers. The K(L)'s and K(M)'s derived herein are defined in terms of concentrations of airborne fibers measured by phase-contrast microscopy (PCM), which only counts all structures longer than 5 microm, thicker than about 0.25 microm, and with an aspect ratio > or =3:1. Moreover, PCM does not distinguish between asbestos and nonasbestos particles. One possible reason for the discrepancies between the K(L)'s and K(M)'s from different studies is that the category of structures included in PCM counts does not correspond closely to biological activity. In the accompanying article (Berman and Crump, 2008) the K(L)'s and K(M)'s and related uncertainty bounds obtained in this article are paired with fiber size distributions from the literature obtained using transmission electron microscopy (TEM). The resulting database is used to define K(L)'s and K(M)'s that depend on both the size (e.g., length and width) and mineralogical type (e.g., chrysotile or crocidolite) of an asbestos structure. An analysis is conducted to determine how well different K(L) and K(M) definitions are able to reconcile the discrepancies observed herein among values obtained from different environments.

A meta-analysis of asbestos-related cancer risk that addresses fiber size and mineral type.

Quantitative estimates of the risk of lung cancer or mesothelioma in humans from asbestos exposure made by the U.S. Environmental Protection Agency (EPA) make use of estimates of potency factors based on phase-contrast microscopy (PCM) and obtained from cohorts exposed to asbestos in different occupational environments. These potency factors exhibit substantial variability. The most likely reasons for this variability appear to be differences among environments in fiber size and mineralogy not accounted for by PCM. In this article, the U.S. Environmental Protection Agency (EPA) models for asbestos-related lung cancer and mesothelioma are expanded to allow the potency of fibers to depend upon their mineralogical types and sizes. This is accomplished by positing exposure metrics composed of nonoverlapping fiber categories and assigning each category its own unique potency. These category-specific potencies are estimated in a meta-analysis that fits the expanded models to potencies for lung cancer (KL's) or mesothelioma (KM's) based on PCM that were calculated for multiple epidemiological studies in our previous paper (Berman and Crump, 2008). Epidemiological study-specific estimates of exposures to fibers in the different fiber size categories of an exposure metric are estimated using distributions for fiber size based on transmission electron microscopy (TEM) obtained from the literature and matched to the individual epidemiological studies. The fraction of total asbestos exposure in a given environment respectively represented by chrysotile and amphibole asbestos is also estimated from information in the literature for that environment. Adequate information was found to allow KL's from 15 epidemiological studies and KM's from 11 studies to be included in the meta-analysis. Since the range of exposure metrics that could be considered was severely restricted by limitations in the published TEM fiber size distributions, it was decided to focus attention on four exposure metrics distinguished by fiber width: "all widths," widths > 0.2 micro m, widths < 0.4 microm, and widths < 0.2 microm, each of which has historical relevance. Each such metric defined by width was composed of four categories of fibers: chrysotile or amphibole asbestos with lengths between 5 microm and 10 microm or longer than 10 microm. Using these metrics three parameters were estimated for lung cancer and, separately, for mesothelioma: KLA, the potency of longer (length > 10 microm) amphibole fibers; rpc, the potency of pure chrysotile (uncontaminated by amphibole) relative to amphibole asbestos; and rps, the potency of shorter fibers (5 microm < length < 10 microm) relative to longer fibers. For mesothelioma, the hypothesis that chrysotile and amphibole asbestos are equally potent (rpc = 1) was strongly rejected by every metric and the hypothesis that (pure) chrysotile is nonpotent for mesothelioma was not rejected by any metric. Best estimates for the relative potency of chrysotile ranged from zero to about 1/200th that of amphibole asbestos (depending on metric). For lung cancer, the hypothesis that chrysotile and amphibole asbestos are equally potent (rpc = 1) was rejected (p < or = .05) by the two metrics based on thin fibers (length < 0.4 microm and < 0.2 microm) but not by the metrics based on thicker fibers. The "all widths" and widths < 0.4 microm metrics provide the best fits to both the lung cancer and mesothelioma data over the other metrics evaluated, although the improvements are only marginal for lung cancer. That these two metrics provide equivalent (for mesothelioma) and nearly equivalent (for lung cancer) fits to the data suggests that the available data sets may not be sufficiently rich (in variation of exposure characteristics) to fully evaluate the effects of fiber width on potency. Compared to the metric with widths > 0.2 microm with both rps and rpc fixed at 1 (which is nominally equivalent to the traditional PCM metric), the "all widths" and widths < 0.4 microm metrics provide substantially better fits for both lung cancer and, especially, mesothelioma. Although the best estimates of the potency of shorter fibers (5 < length < 10 microm) is zero for the "all widths" and widths < 0.4 microm metrics (or a small fraction of that of longer fibers for the widths > 0.2 microm metric for mesothelioma), the hypothesis that these shorter fibers were nonpotent could not be rejected for any of these metrics. Expansion of these metrics to include a category for fibers with lengths < 5 microm did not find any consistent evidence for any potency of these shortest fibers for either lung cancer or mesothelioma. Despite the substantial improvements in fit over that provided by the traditional use of PCM, neither the "all widths" nor the widths < 0.4 microm metrics (or any of the other metrics evaluated) completely resolve the differences in potency factors estimated in different occupational studies. Unresolved in particular is the discrepancy in potency factors for lung cancer from Quebec chrysotile miners and workers at the Charleston, SC, textile mill, which mainly processed chrysotile from Quebec. A leading hypothesis for this discrepancy is limitations in the fiber size distributions available for this analysis. Dement et al. (2007) recently analyzed by TEM archived air samples from the South Carolina plant to determine a detailed distribution of fiber lengths up to lengths of 40 microm and greater. If similar data become available for Quebec, perhaps these two size distributions can be used to eliminate the discrepancy between these two studies.

Sensitivity analysis of biologically motivated model for formaldehyde-induced respiratory cancer in humans.

Conolly et al. (2003, 2004) developed biologically motivated models of formaldehyde carcinogenicity in F344 rats and humans based on a two-stage clonal expansion model of cancer. Based on the human model, Conolly et al. (2004) claimed that cancer risks associated with inhaled formaldehyde are deminimis at relevant human exposure levels. However, they did not conduct a sensitivity analysis to evaluate the robustness of this conclusion. Here, we present a limited sensitivity analysis of the formaldehyde human model. We show that when the control animals from the National Toxicology Program (NTP) studies are replaced with control animals only from NTP inhalation studies, estimates of human risk are increased by 50-fold. When only concurrent control rats are used, the model does not provide any upper bound (UB) to human risk. No data went into the model on the effect of formaldehyde on the division rates and death rates of initiated cells. We show that slight numerical perturbations to the Conolly et al. assumptions regarding these rates can be made that are equally consistent with the underlying data used to construct the model, but produce estimates of human risk ranging anywhere from negative up to 10,000 times higher than those deemed by Conolly et al. to be 'conservative'. Thus, we conclude that estimates of human risk by Conolly et al. (2004) are extremely sensitive to modeling assumptions. This calls into question the basis for the Conolly et al. claim of de minimis human risk and suggests caution in using the model to derive human exposure standards for formaldehyde.

Uncertainties in biologically-based modeling of formaldehyde-induced respiratory cancer risk: identification of key issues.

In a series of articles and a health-risk assessment report, scientists at the CIIT Hamner Institutes developed a model (CIIT model) for estimating respiratory cancer risk due to inhaled formaldehyde within a conceptual framework incorporating extensive mechanistic information and advanced computational methods at the toxicokinetic and toxicodynamic levels. Several regulatory bodies have utilized predictions from this model; on the other hand, upon detailed evaluation the California EPA has decided against doing so. In this article, we study the CIIT model to identify key biological and statistical uncertainties that need careful evaluation if such two-stage clonal expansion models are to be used for extrapolation of cancer risk from animal bioassays to human exposure. Broadly, these issues pertain to the use and interpretation of experimental labeling index and tumor data, the evaluation and biological interpretation of estimated parameters, and uncertainties in model specification, in particular that of initiated cells. We also identify key uncertainties in the scale-up of the CIIT model to humans, focusing on assumptions underlying model parameters for cell replication rates and formaldehyde-induced mutation. We discuss uncertainties in identifying parameter values in the model used to estimate and extrapolate DNA protein cross-link levels. The authors of the CIIT modeling endeavor characterized their human risk estimates as "conservative in the face of modeling uncertainties." The uncertainties discussed in this article indicate that such a claim is premature.