Low dose risk estimation via simultaneous statistical inferences

Summary.  The paper develops and studies simultaneous confidence bounds that are useful for making low dose inferences in quantitative risk analysis. Application is intended for risk assessment studies where human, animal or ecological data are used to set safe low dose levels of a toxic agent, but where study information is limited to high dose levels of the agent. Methods are derived for estimating simultaneous, one-sided, upper confidence limits on risk for end points measured on a continuous scale. From the simultaneous confidence bounds, lower confidence limits on the dose that is associated with a particular risk (often referred to as a bench-mark dose ) are calculated. An important feature of the simultaneous construction is that any inferences that are based on inverting the simultaneous confidence bounds apply automatically to inverse bounds on the bench-mark dose.     

A Diversity Index for Model Space Selection in the Estimation of Benchmark and Infectious Doses via Model Averaging

In chemical and microbial risk assessments, risk assessors fit dose‐response models to high‐dose data and extrapolate downward to risk levels in the range of 1–10%. Although multiple dose‐response models may be able to fit the data adequately in the experimental range, the estimated effective dose (ED) corresponding to an extremely small risk can be substantially different from model to model. In this respect, model averaging (MA) provides more robustness than a single dose‐response model in the point and interval estimation of an ED. In MA, accounting for both data uncertainty and model uncertainty is crucial, but addressing model uncertainty is not achieved simply by increasing the number of models in a model space. A plausible set of models for MA can be characterized by goodness of fit and diversity surrounding the truth. We propose a diversity index (DI) to balance between these two characteristics in model space selection. It addresses a collective property of a model space rather than individual performance of each model. Tuning parameters in the DI control the size of the model space for MA.     

Using the Biological Two-Stage Model to Assess Risk from Short-Term Exposures

Two assumptions used in risk assessment are investigated: (1) the assumption of fraction of lifetime dose rate assumes that the risk from a fractional lifetime exposure at a given dose rate is equal to the risk from full lifetime exposure at that same fraction of the given dose rate; (2) the assumption of fraction of lifetime risk assumes that the risk from a fractional lifetime exposure at a given dose rate is equal to that same fraction of the risk from full lifetime exposure at the same dose rate. These two assumptions are equivalent when risk is a linear function of dose. Thus both can be thought of as generalizations of the assumption that cancer risk is proportional to the total accumulated lifetime dose (or average daily dose), which is often made to assess the risk from short-term exposures. In this paper, the age-specific cumulative hazard functions are derived using the two-stage model developed by Moolgavkar, Venzon, and Knudson for situations when the exposure occurs during a single period or a single instant. The two assumptions described above are examined for three types of carcinogens, initiator, completer, and promoter, in the context of the model. For initiator and completer, these two assumptions are equivalent in the low-dose region; for a promoter, using the fraction of lifetime risk assumption is generally more conservative than that of the fraction of lifetime dose rate assumption. Tables are constructed to show that the use of either the fraction of lifetime dose rate assumption or the fraction lifetime risk assumption can both underestimate and overestimate the true risk for the three types of carcinogens.     

Estimation and Extrapolation of Tumor Probabilities from a Mouse Bioassay with Survival/Sacrifice Components

A nonparametric estimator of the probability distribution of time-to-tumor is incorporated into an algorithm for calculating linearly extrapolated dosage limits from an animal carcino-genesis bioassay. The procedure is illustrated with tumor data from a mouse bioassay with 2-acetylaminofluorene. Extrapolated dosage limits for an excess risk of 10^-6 differ by only a factor of 2 across the six replicates of the experiment.     

Handling cause of death in equivocal cases using the em algorithm

The statistical analysis of animal bioassays fore carcinogenicity often involves utilizing the cause of death of each animal. There is considerable disagreement among veterinary pathologists as to the reliability of cause of death information. Recent recommendations for assigning cause of death in animal studies have allowed for uncertainty on the part of the pathologist. This has given rise to data that contain acknowledged equivocal cases with respect to cause of death. The present paper proposes a method for incorporating these equiYocal cases into an existing estimation procedure that requires distinguishing between tumors that caused death and those that did not.     

On the Use of Hierarchical Probabilistic Models for Characterizing and Managing Uncertainty in Risk/Safety Assessment

A general probabilistically-based approach is proposed for both cancer and noncancer risk/safety assessments. The familiar framework of the original ADI/RfD formulation is used, substituting in the numerator a benchmark dose derived from a hierarchical pharmacokinetic/pharmacodynamic model and in the denominator a unitary uncertainty factor derived from a hierarchical animal/average human/sensitive human model. The empirical probability distributions of the numerator and denominator can be combined to produce an empirical human-equivalent distribution for an animal-derived benchmark dose in external-exposure units.     

Upper Confidence Limits on Excess Risk for Quantitative Responses

The definition and observation of clear-cut adverse health effects for continuous (quantitative) responses, such as altered body weights or organ weights, are difficult propositions. Thus, methods of risk assessment commonly used for binary (quantal) toxic responses such as cancer are not directly applicable. In this paper, two methods for calculating upper confidence limits on excess risk for quantitative toxic effects are proposed, based on a particular definition of an adverse quantitative response. The methods are illustrated with data from a dose-response study, and their performance is evaluated with a Monte Carlo simulation study.     

Experimental Design of Bioassays for Screening and Low Dose Extrapolation

Relatively high doses of chemicals generally are employed in animal bioassays to detect potential carcinogens with relatively small numbers of animals. The problem investigated here is the development of experimental designs which are effective for high to low dose extrapolation for tumor incidence as well as for screening (detecting) carcinogens. Several experimental designs are compared over a wide range of different dose response curves. Linear extrapolation is used below the experimental data range to establish an upper bound on carcinogenic risk at low doses. The goal is to find experimental designs which minimize the upper bound on low dose risk estimates (i.e., maximize the allowable dose for a given level of risk). The maximum tolerated dose (MTD) is employed for screening purposes. Among the designs investigated, experiments with doses at the MTD, 1/2 MTD, 1/4 MTD, and controls generally provide relatively good data for low dose extrapolation with relatively good power for detecting carcinogens. For this design, equal numbers of animals per dose level perform as well as unequal allocations.     

Estimation of the Joint Risk from Multiple-Compound Exposure Based on Single-Compound Experiments

In the absence of data from multiple-compound exposure experiments, the health risk from exposure to a mixture of chemical carcinogens is generally based on the results of the individual single-compound experiments. A procedure to obtain an upper confidence limit on the total risk is proposed under the assumption that total risk for the mixture is additive. It is shown that the current practice of simply summing the individual upper-confidence-limit risk estimates as the upper-confidence-limit estimate on the total excess risk of the mixture may overestimate the true upper bound. In general, if the individual upper-confidence-limit risk estimates are on the same order of magnitude, the proposed method gives a smaller upper-confidence-limit risk estimate than the estimate based on summing the individual upper-confidence-limit estimates; the difference increases as the number of carcinogenic components increases.