Evaluating the evidence for non-monotonic dose-response relationships: A systematic literature review and (re-)analysis of in vivo toxicity data in the area of food safety.

This study aims to evaluate the evidence for the existence of non-monotonic dose-responses (NMDRs) of substances in the area of food safety. This review was performed following the systematic review methodology with the aim to identify in vivo studies published between January 2002 and February 2015 containing evidence for potential NMDRs. Inclusion and reliability criteria were defined and used to select relevant and reliable studies. A set of six checkpoints was developed to establish the likelihood that the data retrieved contained evidence for NMDR. In this review, 49 in vivo studies were identified as relevant and reliable, of which 42 were used for dose-response analysis. These studies contained 179 in vivo dose-response datasets with at least five dose groups (and a control group) as fewer doses cannot provide evidence for NMDR. These datasets were extracted and analyzed using the PROAST software package. The resulting dose-response relationships were evaluated for possible evidence of NMDRs by applying the six checkpoints. In total, 10 out of the 179 in vivo datasets fulfilled all six checkpoints. While these datasets could be considered as providing evidence for NMDR, replicated studies would still be needed to check if the results can be reproduced to rule out that the non-monotonicity was caused by incidental anomalies in that specific study. This approach, combining a systematic review with a set of checkpoints, is new and appears useful for future evaluations of the dose response datasets regarding evidence of non-monotonicity.

Does EU legislation allow the use of the Benchmark Dose (BMD) approach for risk assessment?

Hazard characterisation is largely based on an approach of (statistically) comparing dose groups with the controls in order to derive points of departure such as no-observed-adverse-effect levels (NOAELs) or lowest-observed-adverse-effect levels (LOAELs). This approach suggests the absence of any relevant effect at the NOAEL. The NOAEL approach has been debated for decades. A recent Scientific Opinion by the European Food Safety Authority (EFSA) concluded that the Benchmark Dose (BMD) approach should be preferred over the NOAEL approach for deriving human (health-based) limit or guidance values. Nonetheless, the BMD approach is used infrequently within European regulatory frameworks. The reason for this may lie in legislation or guidelines requiring the use of the NOAEL approach. In this context, various EU regulatory frameworks were examined on such demands. Interestingly, no single legislation was identified containing statutory requirements in conflict with the use of the BMD approach.

Approaches to the risk assessment of genotoxic carcinogens in food: a critical appraisal.

The present paper examines the particular difficulties presented by low levels of food-borne DNA-reactive genotoxic carcinogens, some of which may be difficult to eliminate completely from the diet, and proposes a structured approach for the evaluation of such compounds. While the ALARA approach is widely applicable to all substances in food that are both carcinogenic and genotoxic, it does not take carcinogenic potency into account and, therefore, does not permit prioritisation based on potential risk or concern. In the absence of carcinogenicity dose-response data, an assessment based on comparison with an appropriate threshold of toxicological concern may be possible. When carcinogenicity data from animal bioassays are available, a useful analysis is achieved by the calculation of margins of exposure (MOEs), which can be used to compare animal potency data with human exposure scenarios. Two reference points on the dose-response relationship that can be used for MOE calculation were examined; the T25 value, which is derived from linear extrapolation, and the BMDL10, which is derived from mathematical modelling of the dose-response data. The above approaches were applied to selected food-borne genotoxic carcinogens. The proposed approach is applicable to all substances in food that are DNA-reactive genotoxic carcinogens and enables the formulation of appropriate semi-quantitative advice to risk managers.

A quantitative method to evaluate the quality of interrupted animal cultures in aging studies.

A method is described for a quantitative estimation of the quality of interrupted animal cultures. It is applicable when survival can be described by a Weibull or by a Gompertz function. Application to four healthy and three infected cultures of Lymnaea stagnalis shows that survival curves can be successfully predicted from cultures censored at 25%, 50%, and 75% survival. The parameters of these predicted survival curves can be used to evaluate experimental results from single (longitudinal or cross-sectional) aging studies, and to evaluate comparisons among different aging studies.

Survival characteristics of the mollusc Lymnaea stagnalis under constant culture conditions: effects of aging and disease.

Survival characteristics of seven different populations of the mollusc Lymnaea stagnalis were studied under constant culture conditions. On the basis of these characteristics four populations were considered as healthy and three as infected. In the healthy populations senescence started at an age of about 200 days, 50% survival age varied from 282 to 372 days, 10% survival age from 417 to 508 days and the maximum age from 528 to 673 days. Infected populations differed from healthy ones: (1) in behaviour of the animals; and (2) in shape of survival curve and age-specific death rate. It is concluded that in aging studies in addition to the absolute age of the animals, the following information should be given: the percentage of survival of the population at the time of sampling and a quantitative estimate of the quality of the cultures from which the animals were sampled. The parameters of the Weibull function seem to be suitable for such estimations.

Risk assessment for the harmful effects of UVB radiation on the immunological resistance to infectious diseases.

Risk assessment comprises four steps: hazard identification, dose-response assessment, exposure assessment, and risk characterization. In this study, the effects of increased ultraviolet B(UVB, 280-315 nm) radiation on immune functions and the immunological resistance to infectious diseases in rats were analyzed according to this strategy. In a parallelogram approach, nonthreshold mathematical methods were used to estimate the risk for the human population after increased exposure to UVB radiation. These data demonstrate, using a worst-case strategy (sensitive individuals, no adaptation), that exposure for approximately 90 min (local noon) at 40 degrees N in July might lead to 50% suppression of specific T-cell mediated responses to Listeria monocytogenes in humans who were not preexposed to UVB (i.e., not adapted). Additionally, a 5% decrease in the thickness of the ozone layer might shorten this exposure time by approximately 2.5%. These data demonstrate that UVB radiation, at doses relevant to outdoor exposure, may affect the specific cellular immune response to Listeria bacteria in humans. Whether this will also lead to a lowered resistance (i.e.,increased pathogenic load) in humans is not known, although it was demonstrated that UVB-induced immunosuppression in rats was sufficient to increase the pathogenic load. Epidemiology studies are needed to validate and improve estimates for the potential effects of increased UVB exposure on infectious diseases in humans.

Balancing the risks and benefits of drinking water disinfection: disability adjusted life-years on the scale.

To evaluate the applicability of disability adjusted life-years (DALYs) as a measure to compare positive and negative health effects of drinking water disinfection, we conducted a case study involving a hypothetical drinking water supply from surface water. This drinking water supply is typical in The Netherlands. We compared the reduction of the risk of infection with Cryptosporidium parvum by ozonation of water to the concomitant increase in risk of renal cell cancer arising from the production of bromate. We applied clinical, epidemiologic, and toxicologic data on morbidity and mortality to calculate the net health benefit in DALYs. We estimated the median risk of infection with C. parvum as 10(-3)/person-year. Ozonation reduces the median risk in the baseline approximately 7-fold, but bromate is produced in a concentration above current guideline levels. However, the health benefits of preventing gastroenteritis in the general population and premature death in patients with acquired immunodeficiency syndrome outweigh health losses by premature death from renal cell cancer by a factor of > 10. The net benefit is approximately 1 DALY/million person-years. The application of DALYs in principle allows us to more explicitly compare the public health risks and benefits of different management options. In practice, the application of DALYs may be hampered by the substantial degree of uncertainty, as is typical for risk assessment.

Ranking of allergenic potency of rubber chemicals in a modified local lymph node assay.

A modified local lymph node assay (LLNA) with ex vivo tritium thymidine (3H-TdR) labeling of the proliferating lymph node cells was used for determination of the allergenic potency of chemicals used in the production of rubber for latex medical gloves. Fifteen chemicals known to induce contact hypersensitivity reactions in man, including various thiuram, carbamate, and benzothiazole compounds, and one amine were tested. The EC3 (effective concentration inducing a 3-fold increase in proliferation of lymph node cells [Stimulation Index, SI = 3]) was calculated with nonlinear regression analysis, including a bootstrap method for determination of the 5-95% confidence interval of the EC3 value. This procedure identified 14 out of the 15 chemicals tested as sensitizers, while for one chemical, ZDBC, no EC3 could be calculated due to low responses and a lack of a dose-response relationship in the data obtained. The ranking order of the chemicals with increasing EC3 values (and thus decreasing allergenic potency) was found to be in the following order: ZDEC < TMTD < TETD < ZPC < ZDMC < MBTS < PTD < TMTM < MBT < MBI < PTT < ZMBT < TBTD < DEA < ZDBC. Our results indicate that the chemicals of choice for use in the production of natural rubber latex products would be for the thiuram compounds, TBTD; for the carbamates, ZDBC; and for the benzothiazoles, ZMBT. However, one has to be aware that besides potency, the total amount of residual chemical present in the final product is also important for allergy induction.

Age-related decrease in electrical coupling of two identified neurons in the mollusc Lymnaea stagnalis.

Electrophysiological characteristics of two identified giant electrotonically coupled neurosecretory cells in the central nervous system of the mollusc Lymnaea stagnalis were studied in mature animals of different age. The coupling coefficient of the neurons decreased considerably with age. The possibility that the decrease is due to an increase in the junctional resistance between the cells is discussed.

Risk assessment of UVB effects on resistance to infectious diseases.

The aim of this study was to develop a quantitative risk assessment of lowered resistance to infections in humans due to (solar) ultraviolet B (UVB) exposure. We followed the steps for risk assessment as defined by the U.S. National Academy of Sciences: (1) hazard identification, (2) dose-response assessment, (3) exposure assessment, and (4) risk characterization. For step 1, the suppressory effects of UVB radiation on the immune system have been reviewed, supplemented with new data, and analyzed. Experiments on UV-induced immunosuppression cannot be performed with humans for ethical reasons, but herpes simplex virus infection appears to be the human paradigm. Thus, UVB radiation appears to be a potential hazard to immunologic functions. Step 2 is crucial, but dose-response relationships for infections have never been measured in humans. We used our earlier dose-response rat data for suppression of lymphocyte stimulation and computed that the UVB dose resulting in a 50% reduction of lymphocyte stimulation by Listeria monocytogenes is 6.800 J/m2. Using mixed skin lymphocyte response assays we found that humans are 3.8 times less sensitive than rats (interspecies variation [IEV]). To account for the 2.5 percentile of most susceptible individuals in a population, an additional factor (intraspecies variation [IAV]) was introduced (0.5 for humans). Using these data, we computed that 13.100 J/m2 of UVB radiation emitted by FS40 lamps would suppress 50% of the proliferative response of lymphocytes to L. monocytogenes in most sensitive skin type 2 humans. In step 3, we assumed the action spectrum for the responses analyzed by us as identical to an action spectrum for suppression of contact hypersensitivity that is available in the literature. This led us to step 4, where we calculated that approximately 100 min of solar exposure at around noon in Italy or Spain would suppress the resistance to infections by L. monocytogenes in the most sensitive humans.

Risk assessment for the harmful effects of immunotoxic agents on the immunological resistance to infectious diseases: ultraviolet light as an example.

Risk assessment comprises four steps: hazard identification, dose-response assessment, exposure assessment and risk characterization. As an example we have assessed the risk of decreased host resistance to infections due to effects of the immunotoxic agent - ultraviolet light. We have analysed effects of UVB radiation on basal immune functions in rats and man, and the immunological resistance to infectious diseases in rats. Non-threshold mathematical methods were used in order to estimate the risk for the human population after increased exposure to UVB radiation. These data demonstrate that UVB radiation, at doses relevant to outdoors exposure, may affect the immunological resistance to infectious diseases in human individuals. This study may also provide a basis for a strategy to assess the risk of adverse effects of exposure to immunotoxic agents.

A quantitative method for assessing the sensitizing potency of low molecular weight chemicals using a local lymph node assay: employment of a regression method that includes determination of the uncertainty margins.

Risk assessment of sensitizing chemicals requires, besides hazard identification, the assessment of potency. To examine the sensitizing capacity of low molecular weight chemicals, a murine local lymph node assay (LLNA) was used. The sensitizing capacity of known allergens was quantified by dose-response modeling. At a stimulatory index (SI) of 3, the corresponding estimated concentration was calculated (EC(3)), together with a confidence interval to take account of the quality of the particular data set. We tested ten allergens (ethyl-p-aminobenzoate (benzocaine), diethylamine (DEA), 2,4-dinitrochlorobenzene (DNCB), 2-mercaptobenzothiazole (MBT), 4-ethoxymethylene 2-phenyl oxazol-5-one (oxazolone), phthalic anhydride (PA), toluene diisocyanate (TDI), trimellitic anhydride (TMA), tetramethylthiuramdisulfide (TMTD) and zincdimethyldithiocarbamate (ZDMC)). Oxazolone showed the strongest sensitizing potency followed in this order by DNCB, TDI, TMA, PA, TMTD, ZDMC, MBT, benzocaine and DEA. The approach performed in this study is a way to accurately assess the potency of sensitizing chemicals and thus a possibility for classification.

Comparison of dose-responses of contact allergens using the guinea pig maximization test and the local lymph node assay.

The guinea pig maximization test (GPMT) has been used as a method for the prediction of skin sensitizing potential for over 30 years. Besides hazard identification, risk assessment of sensitizing chemicals requires the assessment of potency. For the determination of potency based on lowest effective dose levels, dose-response studies are required. In the standard GPMT a single concentration is used for intracutaneous and topical induction and the assay provides a qualitative assessment of allergenicity. This paper presents data derived from quantitative evaluation of the sensitizing potency of chemicals in the GPMT, based on multiple concentrations. We performed the GPMT in accordance with the original procedure of Magnusson and Kligman; and included in this procedure a range of intradermal and topical concentrations for induction. Three allergens with different sensitizing potencies, diethylamine (DEA), tetramethyl thiuram disulfide (TMTD) and zinc dimethyl dithiocarbamate (ZDMC) were tested. The data obtained with this test procedure were compared to data we previously obtained using the local lymph node assay (LLNA). Both the GPMT and the LLNA showed dose response relationships for the three chemicals tested. For the chemicals tested, both tests differed in the relative potencies based on benchmark concentrations. While both tests ranked DEA as the least potent allergen, the GPMT ranked ZDMC more potent than TMTD, the reverse being found in the LLNA. The nature of the data provided in the LLNA makes it likely that benchmarks as defined with this test are more reliable than that defined in the GPMT. However, further validation with human data is necessary.

Developmental toxicity of butyl benzyl phthalate in the rat using a multiple dose study design.

The developmental toxicity of butyl benzyl phthalate (BBP) was investigated in the rat using ten dose groups between 270 and 2100 mg/kg/day. Exposure was by daily gavage from gestation day 5 through 16 or gestation day 5 through 20. Dose-response data were analyzed using the benchmark approach by fitting dose-response models to the various endpoints. BBP induced increased liver and kidney weights in dams, accompanied by liver enzyme increases in maternal serum. Extramedullary hematopoiesis, which was already substantial in control pregnant animals, was increased after BBP treatment. Fetotoxicity included increased resorptions, reduced fetal weights, increased incidence of skeletal anomalies, and reduced fetal testis weights in the presence of an increased incidence of retarded testicular descent. As embryotoxicity was found at lower dosages compared to observed maternal toxicity, BBP appeared to be a specifically embryotoxic compound. The extended exposure protocol (gestation day 5 through 20) appeared more sensitive for measuring fetotoxic effects. We recommend the use of more doses in toxicity tests, together with the benchmark approach as an appropriate and more accurate method for analyzing dose-response data compared to the NOAEL approach.

Estimation of dioxin and furan elimination rates with a pharmacokinetic model.

Quantitative description of the pharmacokinetics of dioxins and furans in humans can be of great help for the assessment of health risks posed by these compounds. To that the elimination rates of sixteen 2,3,7,8-chlorinated dibenzodioxins and dibenzofurans are estimated from both a longitudinal and a cross-sectional data set using the model of Van der Molen et al. [Van der Molen G.W., Kooijman S.A.L.M., and Slob W. A generic toxicokinetic model for persistent lipophilic compounds in humans: an application to TCDD. Fundam Appl Toxicol 1996: 31: 83-94]. In this model the elimination rate is given by the (constant) specific elimination rate multiplied with the ratio between the lipid weight of the liver and total body lipid weight. Body composition, body weight and intake are assumed to depend on age. The elimination rate is, therefore, not constant. For 49-year-old males, the elimination rate estimates range between 0.03 per year for 1,2,3,6,7,8-hexaCDF to 1.0 per year for octaCDF. The elimination rates of the most toxic congeners, 2,3,7,8-tetraCDD, 1,2,3,7,8-pentaCDD, and 2,3,4,7,8-pentaCDF, were estimated at 0.09, 0.06, and 0.07, respectively, based on the cross-sectional data, and 0.11, 0.09, and 0.09 based on the longitudinal data. The elimination rates of dioxins decrease with age between 0.0011 per year for 1,2,3,6,7,8-hexaCDD and 0.0035 per year for 1,2,3,4,6,7,8-heptaCDD. For furans the average decrease is 0.0033 per year. The elimination rates were estimated both from a longitudinal and a cross-sectional data set, and agreed quite well with each other, after taking account of historical changes in average intake levels.

Human variability in polymorphic CYP2D6 metabolism: is the kinetic default uncertainty factor adequate?

Human variability in the kinetics of CYP2D6 substrates has been quantified using a database of compounds metabolised extensively (>60%) by this polymorphic enzyme. Published pharmacokinetic studies (after oral and intravenous dosing) in non-phenotyped healthy adults, and phenotyped extensive (EMs), intermediate or slow-extensive (SEMs) and poor metabolisers (PMs) have been analysed using data for parameters that relate primarily to chronic exposure (metabolic and total clearances, area under the plasma concentration time-curve) and primarily to acute exposure (peak concentration). Similar analyses were performed with the available data for subgroups of the population (age, ethnicity and disease). Interindividual differences in kinetics for markers of oral exposure were large for non-phenotyped individuals and for EMs (coefficients of variation were 67-71% for clearances and 54-63% for C(max)), whereas the intravenous data indicated a lower variability (34-38%). Comparisons between EMs, SEMs and PMs revealed an increase in oral internal dose for SEMs and PMs (ratio compared to EMs=3 and 9-12, respectively) associated with lower variability than that for non-phenotyped individuals (coefficients of variation were 32-38% and 30% for SEMs and PMs, respectively). In relation to the uncertainty factors used for risk assessment, most subgroups would not be covered by the kinetic default of 3.16. CYP2D6-related factors necessary to cover 95-99% of each subpopulation ranged from 2.7 to 4.1 in non-phenotyped healthy adults and EMs to 15-18 in PMs and 22-45 in children. An exponential relationship (R(2)=0.8) was found between the extent of CYP2D6 metabolism and the uncertainty factors. The extent of CYP2D6 involvement in the metabolism of a substrate is critical in the estimation of the CYP2D6-related factor. The 3.16 kinetic default factor would cover PMs for substrates for which CYP2D6 was responsible for up to 25% of the metabolism in EMs.

The benchmark approach applied to a 28-day toxicity study with Rhodorsil Silane in rats. the impact of increasing the number of dose groups.

The OECD study design, aimed at obtaining a no-observed-adverse-effect level (NOAEL), may be suboptimal for deriving a benchmark dose. Therefore the present subacute (28-day) study was carried out to evaluate a multiple dose study design and to compare the results with the common OECD design. Seven groups of 10 female rats each were intragastrically administered corn oil without (controls) or with 50, 150, 300, 450, 600 or 750 mg Rhodorsil Silane/kg body weight/day, once daily (7 days/week) for 4 weeks. From the complete dataset, two subsets were selected, one representing a study design with seven dose groups of five animals (7 x 5 design), the other representing a study design with four dose groups of 10 animals (4 x 10 design). Under the conditions of the present study, the NOAEL for Rhodorsil Silane 198 was assessed at 50 mg/kg body weight/day, based on the data of the 4 x 10 design. The benchmark approach resulted in a benchmark dose of 19 mg/kg body weight/day, based on the data of the 7 x 5 design. Comparison of the results demonstrated that the multiple dose (7 x 5) design led to a more reliable result than the OECD (4 x 10) design, despite the smaller total number of animals. The dose-response analysis showed that at "the NOAEL" the effect on relative spleen weight was larger than 10%, illustrating that at the NOAEL, adverse effects may occur.

Mathematical modelling and quantitative methods.

The present review reports on the mathematical methods and statistical techniques presently available for hazard characterisation. The state of the art of mathematical modelling and quantitative methods used currently for regulatory decision-making in Europe and additional potential methods for risk assessment of chemicals in food and diet are described. Existing practices of JECFA, FDA, EPA, etc., are examined for their similarities and differences. A framework is established for the development of new and improved quantitative methodologies. Areas for refinement, improvement and increase of efficiency of each method are identified in a gap analysis. Based on this critical evaluation, needs for future research are defined. It is concluded from our work that mathematical modelling of the dose-response relationship would improve the risk assessment process. An adequate characterisation of the dose-response relationship by mathematical modelling clearly requires the use of a sufficient number of dose groups to achieve a range of different response levels. This need not necessarily lead to an increase in the total number of animals in the study if an appropriate design is used. Chemical-specific data relating to the mode or mechanism of action and/or the toxicokinetics of the chemical should be used for dose-response characterisation whenever possible. It is concluded that a single method of hazard characterisation would not be suitable for all kinds of risk assessments, and that a range of different approaches is necessary so that the method used is the most appropriate for the data available and for the risk characterisation issue. Future refinements to dose-response characterisation should incorporate more clearly the extent of uncertainty and variability in the resulting output.

Congener-specific bioavailability of PCDD/Fs and coplanar PCBs in cows: laboratory and field measurements.

To estimate congener-specific bioavailabilities for 17 PCDD/Fs in cows grazing near a MSW incinerator both a controlled lab study and a field study were performed. In the lab study the estimates were derived from the elevated concentrations in milk from two cows after administration of a single dose of contaminated fly ash. In the field study, located near a large MSW incinerator, daily samples of grass and milk collected over a period of 60 days were pooled to two monthly bulk samples. The concentrations in these bulk samples of grass and milk were used to estimate the bioavailabilities of the 17 PCDD/Fs as well as of three coplanar PCBs. With the concentrations of PCDD/Fs expressed in I-TEQs the bioavailability in cows was estimated at +/- 7.5% in the field study. The congener-specific bioavailabilities correlated well between the two studies, as well as with previously reported values in the literature. However, the absolute levels differed considerably between studies, indicating a strong matrix effect.

The estimation of elimination rates of persistent compounds: a re-analysis of 2,3,7,8-tetrachlorodibenzo-p-dioxin levels in Vietnam veterans.

The elimination rate of TCDD was re-estimated from measurements of internal concentrations in Vietnam veterans with a model that can account for age dependent body composition, and both age and (calendar) time dependent background intake. Estimates of the specific elimination rate, which is independent of body composition, did not differ much between fits with different simplifications of this model for this particular data set. However, the assumption that the associated half-life is constant over the human lifespan is improper. The overall elimination rate and half-life are only constant when background intake and body composition are approximately constant. For example, our model predicts that half-life ranges between 5.5 in young adults and 11 years in elderly men. The change of half-life with age should be accounted for in toxicokinetic calculations.