Utilisation of parametric methods to improve percentile-based estimates for the carcinogenic potency of nitrosamines.

N-Nitrosamines have recently been the subject of intense regulatory scrutiny, including the setting of low exposure limits (18 ng/day) (European Medicines Agency (EMA), 2020). This paper evaluates different methodologies to determine statistically robust bounds on the carcinogenic potency of chemical classes, using historic TD50 data (Peto et al., 1984; Thresher et al., 2019) and exemplified for N-nitrosamines. Initially, the distribution of TD50 values (TD50s) for N-nitrosamines of known potency was characterised. From this, it is possible to compare parametric and non-parametric methods to obtain percentiles of interest from the distribution of TD50s, which are shown to be robust to uncertainty in the initial TD50 estimates. These methods may then be applied to different chemical subclasses. The values obtained may be of use in refining acceptable intakes for N-nitrosamines and their subclasses.

A comprehensive analysis of the animal carcinogenicity data for glyphosate from chronic exposure rodent carcinogenicity studies.

Since the introduction of glyphosate-tolerant genetically-modified plants, the global use of glyphosate has increased dramatically making it the most widely used pesticide on the planet. There is considerable controversy concerning the carcinogenicity of glyphosate with scientists and regulatory authorities involved in the review of glyphosate having markedly different opinions. One key aspect of these opinions is the degree to which glyphosate causes cancer in laboratory animals after lifetime exposure. In this review, twenty-one chronic exposure animal carcinogenicity studies of glyphosate are identified from regulatory documents and reviews; 13 studies are of sufficient quality and detail to be reanalyzed in this review using trend tests, historical control tests and pooled analyses. The analyses identify 37 significant tumor findings in these studies and demonstrate consistency across studies in the same sex/species/strain for many of these tumors. Considering analyses of the individual studies, the consistency of the data across studies, the pooled analyses, the historical control data, non-neoplastic lesions, mechanistic evidence and the associated scientific literature, the tumor increases seen in this review are categorized as to the strength of the evidence that glyphosate causes these cancers. The strongest evidence shows that glyphosate causes hemangiosarcomas, kidney tumors and malignant lymphomas in male CD-1 mice, hemangiomas and malignant lymphomas in female CD-1 mice, hemangiomas in female Swiss albino mice, kidney adenomas, liver adenomas, skin keratoacanthomas and skin basal cell tumors in male Sprague-Dawley rats, adrenal cortical carcinomas in female Sprague-Dawley rats and hepatocellular adenomas and skin keratocanthomas in male Wistar rats.

Discussion on statistical analysis of carcinogenicity studies with an early terminated treated group.

High rates of mortality on long term carcinogenicity studies can often result in challenges when it comes to the statistical analysis of tumor incidence. The current regulatory advice often results in treated groups being terminated earlier than the control group. However, this advice rarely considers the impact of this action on the statistical analyses. The nature of these analyses means that groups terminated at different times may not be directly comparable due to age differences of the animals. Here we discuss the issues related to this and investigate several approaches of how to incorporate these groups within the statistical analyses. Although no single method appears to resolve these issues consistently, inclusion of the early terminated group is still informative. Depending on the timing of the early termination, either pooling of the groups into a single terminal kill (TK) interval or reassignment of intervals based purely on time of death (ie, no separate TK interval) appear preferable. However, to draw meaningful conclusions the time of onset of a given tumor must also be considered alongside incident rates and any statistical findings.

Comparisons of false negative rates from a trend test alone and from a trend test jointly with a control-high groups pairwise test in the determination of the carcinogenicity of new drugs.

Interest has been expressed in using a joint test procedure that requires that the results of both a trend test and a pairwise comparison test between the control and the high groups be statistically significant simultaneously at the levels of significance recommended in the FDA 2001 draft guidance for industry document for the separate tests in order for the drug effect on the development of an individual tumor type to be considered as statistically significant. Results of our simulation studies show that there is a serious consequence of large inflations of the false negative rate through large decreases of false positive rate in the use of the above joint test procedure in the final interpretation of the carcinogenicity potential of a new drug if the levels of significance recommended for separate tests are used. The inflation can be as high as 204.5% of the false negative rate when the trend test alone is required to test if the effect is statistically significant. To correct the problem, new sets of levels of significance have also been developed for those who want to use the joint test in reviews of carcinogenicity studies.

Penalized estimation for proportional hazards models with current status data.

We provide a simple and practical, yet flexible, penalized estimation method for a Cox proportional hazards model with current status data. We approximate the baseline cumulative hazard function by monotone B-splines and use a hybrid approach based on the Fisher-scoring algorithm and the isotonic regression to compute the penalized estimates. We show that the penalized estimator of the nonparametric component achieves the optimal rate of convergence under some smooth conditions and that the estimators of the regression parameters are asymptotically normal and efficient. Moreover, a simple variance estimation method is considered for inference on the regression parameters. We perform 2 extensive Monte Carlo studies to evaluate the finite-sample performance of the penalized approach and compare it with the 3 competing R packages: C1.coxph, intcox, and ICsurv. A goodness-of-fit test and model diagnostics are also discussed. The methodology is illustrated with 2 real applications.

Relationship between increasing concentrations of two carcinogens and statistical image descriptors of foci morphology in the cell transformation assay.

Cell Transformation Assays (CTAs) have long been proposed for the identification of chemical carcinogenicity potential. The endpoint of these in vitro assays is represented by the phenotypic alterations in cultured cells, which are characterized by the change from the non-transformed to the transformed phenotype. Despite the wide fields of application and the numerous advantages of CTAs, their use in regulatory toxicology has been limited in part due to concerns about the subjective nature of visual scoring, i.e. the step in which transformed colonies or foci are evaluated through morphological features. An objective evaluation of morphological features has been previously obtained through automated digital processing of foci images to extract the value of three statistical image descriptors. In this study a further potential of the CTA using BALB/c 3T3 cells is addressed by analysing the effect of increasing concentrations of two known carcinogens, benzo[a]pyrene and NiCl2 , with different modes of action on foci morphology. The main result of our quantitative evaluation shows that the concentration of the considered carcinogens has an effect on foci morphology that is statistically significant for the mean of two among the three selected descriptors. Statistical significance also corresponds to visual relevance. The statistical analysis of variations in foci morphology due to concentration allowed to quantify morphological changes that can be visually appreciated but not precisely determined. Therefore, it has the potential of providing new quantitative parameters in CTAs, and of exploiting all the information encoded in foci. Copyright © 2016 John Wiley & Sons, Ltd.

Statistical considerations for a chronic bioassay study: Exposure to Decamethylcyclopentasiloxane (D5) and incidence of uterine endometrial adenocarcinomas in a 2-year inhalation study with Fischer rats.

Decamethylcyclopentasiloxane (D5) is a cyclic siloxane used in the production of industrial and consumer products. Four groups of 60 Fischer-344 female rats were analyzed for uterine endometrial adenocarcinoma (inhalation study with exposure levels in ppm/number of observed cases: 0/0, 10/1, 40/0, and 160/5) by exact regression (logistic, Poisson), the max poly-3 trend test, and a random effects probit model. When comparing the 160 ppm group to controls after 24 months, the incidence of adenocarcinomas was elevated (borderline significant); it was significant when all exposure levels were included. Four sets of (historical) control groups were formed, with varying heterogeneity. The effect of D5 was either significant or borderline significant when comparing all control sets to the 160 ppm group. When considering all exposure groups using any of the analysis methods, a significant effect was observed when the high dose group was included in the analysis; the effect was not significant when the high dose group was not included. The evidence tends to support the conclusion that D5 at the highest dose level (160 ppm) results in an increased incidence of adenocarcinomas. However, it is important to verify any potential effect through a biological investigation.

Bayesian model-averaged benchmark dose analysis via reparameterized quantal-response models.

An important objective in biomedical and environmental risk assessment is estimation of minimum exposure levels that induce a pre-specified adverse response in a target population. The exposure points in such settings are typically referred to as benchmark doses (BMDs). Parametric Bayesian estimation for finding BMDs has grown in popularity, and a large variety of candidate dose-response models is available for applying these methods. Each model can possess potentially different parametric interpretation(s), however. We present reparameterized dose-response models that allow for explicit use of prior information on the target parameter of interest, the BMD. We also enhance our Bayesian estimation technique for BMD analysis by applying Bayesian model averaging to produce point estimates and (lower) credible bounds, overcoming associated questions of model adequacy when multimodel uncertainty is present. An example from carcinogenicity testing illustrates the calculations.

Statistical evaluation of mortality in long-term carcinogenicity bioassays using a Williams-type procedure.

Several doses and a control group can be compared under order restriction using the Williams procedure for normally distributed endpoints assuming variance homogeneity. Comparison of the survival functions represents a secondary endpoint in long-term in vivo bioassays of carcinogenicity. Therefore, a Williams-type procedure for the comparison of survival functions is proposed for the assumption of the Cox proportional hazards model or the general frailty Cox model to allow a joint analysis over sex and strains. Interpretation according to both statistical significance and biological relevance is possible with simultaneous confidence intervals for hazard ratios. Related survival data can be analyzed using the R packages survival, coxme, and multcomp. Together with the R packages MCPAN and nparcomp, Dunnett- or Williams-type procedures are now available for the statistical analysis of the following endpoint types in toxicology: (i) normally distributed, (ii) non-normally distributed, (iii) score (ordered categorical) data, (iv) crude proportions, (v) survival functions, and (vi) time-to-tumor data with and without cause-of-death information.

An assessment of the statistical methods used to analyse toxicology studies.

The Statisticians in the Pharmaceutical Industry Toxicology Special Interest Group has collated and compared statistical analysis methods for a number of toxicology study types including general toxicology, genetic toxicology, safety pharmacology and carcinogenicity. In this paper, we present the study design, experimental units and analysis methods.

Soluble suppression of tumorigenicity-2 predicts pneumonia in patients with inhalation injury: Results of a pilot study.

INTRODUCTION: Several mechanisms play a role in the development of pneumonia after inhalation injury. Our aim was to analyze whether higher concentrations of inflammatory markers or of biomarkers of epithelial injury are associated with a higher incidence of pneumonia in patients with inhalation injury. MATERIAL AND METHODS: Secondary analysis of a single-center prospective observational cohort pilot study, performed over a two-year period (2015-2017) at the Burns Unit of the Plastic and Reconstructive Surgery Department of Vall d'Hebron University Hospital. All patients aged 18 with suspected inhalation injury undergoing admission to the Burns Unit were included. Plasma biomarkers of the lung epithelium (RAGE and SP-D), inflammation markers (IL6, IL8), and IL33, as well as soluble suppression of tumorigenicity-2 (sST2) levels, were measured within the first 24 h of admission. RESULTS: Twenty-four patients with inhalation injury were included. Eight (33.3%) developed pneumonia after a median of 7 (4-8) days of hospital stay. Patients with pneumonia presented higher plasma concentrations of sST2 (2853 [2356-3351] ng/mL vs 1352 [865-1839] ng/mL; p < 0.001), IL33 (1.95 [1.31-2.59] pg/mL vs 1.26 [1.07-1.45] pg/mL; p = 0.002) and IL8 (325.7 [221.6-430.0] pg/mL vs 174.1 [95.2-253.0] pg/mL; p = 0.017) on day 1 of inclusion. Plasma sST2 concentration in the first 24 h demonstrated excellent diagnostic accuracy for predicting the occurrence of pneumonia in patients with smoke inhalation (AUROC 0.929 [95%CI 0.818-1.000]). A cutoff point of ≥2825 ng/mL for sST2 had a sensitivity of 75% and a specificity of 100%. The risk ratio of pneumonia in patients with sST2 ≥ 2825 ng/mL was 7.14 ([95% CI 1.56-32.61]; p = 0.016). CONCLUSIONS: Plasma sST2 in the first 24 h of admission predicts the occurrence of pneumonia in patients with inhalation injury.

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.

Evaluation of non-commercial models for genotoxicity and carcinogenicity in the assessment of EFSA's databases.

Over the past years, the European Food Safety Authority (EFSA) released to the public domain several databases, with the main objectives of collecting and storing hazard data on the substances considered in EFSA's risk assessment and secondly to serve as a basis for further development of in silico tools such as quantitative structure-activity relationship (QSAR) models. In this work, we evaluated the ability of freely available QSAR models to estimate genotoxicity and carcinogenicity properties and their possible use for screening purposes on three different EFSA's databases. With an accuracy close to 90%, the results showed good capabilities of QSAR models to predict genotoxicity in terms of bacterial reverse mutation test, while statistics for in vivo micronucleus test are not satisfactory (accuracy in the predictions close to 50%). Interestingly, results on the carcinogenicity assessment showed an accuracy in prediction close to 70% for the best models. In addition, an example of the potential application of in silico models is presented in order to provide a preliminary screening of genotoxicity properties of botanicals intended for use as food supplements.

Development of quantitative structure-activity relationship (QSAR) models to predict the carcinogenic potency of chemicals I. Alternative toxicity measures as an estimator of carcinogenic potency.

Determining the carcinogenicity and carcinogenic potency of new chemicals is both a labor-intensive and time-consuming process. In order to expedite the screening process, there is a need to identify alternative toxicity measures that may be used as surrogates for carcinogenic potency. Alternative toxicity measures for carcinogenic potency currently being used in the literature include lethal dose (dose that kills 50% of a study population [LD(50)]), lowest-observed-adverse-effect-level (LOAEL) and maximum tolerated dose (MTD). The purpose of this study was to investigate the correlation between tumor dose (TD(50)) and three alternative toxicity measures as an estimator of carcinogenic potency. A second aim of this study was to develop a Classification and Regression Tree (CART) between TD(50) and estimated/experimental predictor variables to predict the carcinogenic potency of new chemicals. Rat TD(50)s of 590 structurally diverse chemicals were obtained from the Cancer Potency Database, and the three alternative toxicity measures considered in this study were estimated using TOPKAT, a toxicity estimation software. Though poor correlations were obtained between carcinogenic potency and the three alternative toxicity (both experimental and TOPKAT) measures for the CPDB chemicals, a CART developed using experimental data with no missing values as predictor variables provided reasonable estimates of TD(50) for nine chemicals that were part of an external validation set. However, if experimental values for the three alternative measures, mutagenicity and logP are not available in the literature, then either the CART developed using missing experimental values or estimated values may be used for making a prediction.

The Contribution of In Vivo Mammalian Studies to the Knowledge of Adverse Effects of Radiofrequency Radiation on Human Health.

The proliferation of cellular antennas and other radiofrequency radiation (RFR) generating devices of the last decades has led to more and more concerns about the potential health effects from RFR exposure. Since the 2011 classification as a possible carcinogen by the International Agency for Research on Cancer (IARC), more experimental studies have been published that support a causal association between RFR exposure and health hazards. As regard cancer risk, two long-term experimental studies have been recently published by the US National Toxicology Program (NTP) and the Italian Ramazzini Institute (RI). Despite important experimental differences, both studies found statistically significant increases in the development of the same type of very rare glial malignant tumors. In addition to carcinogenicity, reproductive organs might be particularly exposed, as well as sensitive to RFR. In this work, we reviewed the currently available evidence from in vivo studies on carcinogenicity and reproductive toxicity studies in order to summarize the contribution of experimental research to the prevention of the adverse effects of RFR on human health.

An optimised data analysis for the Balb/c 3T3 cell transformation assay and its application to metal compounds.

The Balb/c3T3 cell transformation assay (CTA) is an available in vitro system to detect the carcinogenic potential of chemicals. Currently, the European Centre for the Validation of Alternative Methods (ECVAM) is validating this test, assessing its reliability and relevance. Its endpoint is the formation of type III foci, which is, when using clone A31-1-1, a very rare event that usually does not occur at all for negative controls. The carcinogenic potential of a compound tested is assessed by comparing the number of foci in treated and untreated cells. The objective of the present work is to optimise the data analysis for this endpoint by applying the most commonly used approach by a t-test and the Fisher's exact test as an alternative approach. For this purpose selected metal compounds classified as carcinogenic (NaAsO2, CdCl2, cisPt), as suspected carcinogenic (C6H5)4AsCl, CH3HgCl), or as compounds without evidence of carcinogenic properties in humans ((NH4)2PtCl6, NaVO3) as well as a non-carcinogenic (AgNO3) were analysed. Our evaluation revealed that the t-test approach, which assumes normality of data, is not appropriate. The results demonstrated that the statistical analysis by Fisher's exact test, better reflecting the data properties, greatly facilitates the interpretation of Balb/c3T3 CTA data regarding carcinogenic potential.

Quantitative structure activity relationship for the computational prediction of nitrocompounds carcinogenicity.

Several nitrocompounds have been screened for carcinogenicity in rodents, but this is a lengthy and expensive process, taking two years and typically costing 2.5 million dollars, and uses large numbers of animals. There is, therefore, much impetus to develop suitable alternative methods. One possible way of predicting carcinogenicity is to use quantitative structure-activity relationships (QSARs). QSARs have been widely utilized for toxicity testing, thereby contributing to a reduction in the need for experimental animals. This paper describes the results of applying a TOPological substructural molecular design (TOPS-MODE) approach for predicting the rodent carcinogenicity of nitrocompounds. The model described 79.10% of the experimental variance, with a standard deviation of 0.424. The predictive power of the model was validated by leave-one-out validation, with a determination coefficient of 0.666. In addition, this approach enabled the contribution of different fragments to carcinogenic potency to be assessed, thereby making the relationships between structure and carcinogenicity to be transparent. It was found that the carcinogenic activity of the chemicals analysed was increased by the presence of a primary amine group bonded to the aromatic ring, a manner that was proportional to the ring aromaticity. The nitro group bonded to an aromatic carbon atom is a more important determinant of carcinogenicity than the nitro group bonded to an aliphatic carbon. Finally, the TOPS-MODE approach was compared with four other predictive models, but none of these could explain more than 66% of the variance in the carcinogenic potency with the same number of variables.

Predicting carcinogenicity by using batteries of dependent short-term tests.

Among the various methods for predicting carcinogenicity from a battery of short-term tests (STTs), the carcinogenicity prediction and battery selection (CPBS) procedure is the most prominent. A major assumption of CPBS is that the STTs used in the prediction are conditionally independent. Results of recent National Toxicology Program studies of four commonly used in vitro STTs contradict this assumption, thereby necessitating modification of CPBS to accommodate dependencies. This is accomplished via log-linear modeling, which then also yields an important dividend: standard errors for the predicted probabilities of carcinogenicity.

Biostatistical issues in the design and analysis of animal carcinogenicity experiments.

Two-year animal carcinogenicity experiments are used to evaluate the potential carcinogenicity from exposure to chemicals. The choice of exposure levels, the allocation of animals to doses, the length of exposure, and the choice of interim sacrifice times all affect the power of statistical tests for carcinogenic effects and the variance of interpolated estimates of carcinogenic risk. In this paper, one aspect of this problems is considered: the ability of tumor incidence data to provide information on carcinogenic mechanism and the optimal choice of design parameters with which to achieve this purpose. The direct application of biochemical data to the estimation of carcinogenic risk is also discussed in detail.