Outbred level in safety studies to achieve reproducibility and

Outbred rats are the most common animals used in drug safety testing at this time. Charles River is the leading global supplier of standard rat models for biomedical research. These ratsare often used in bioassays such as carcinogenicity studies. According to ICH14, future carcinogenicity testing applications should consist of one set of animal studies for joint new drugapplications, which must be submitted concurrently in each country. In such study the animals should be carefully selected standardized outbred rats of the same strain,” which can be used on an international level in safety studies to achieve reproducibility and comparability of results.13 Food safety risk assessment is the scienti c evaluation of known or potential adverse health effects resulting from human exposure to food borne hazards.2 The most important aspect of risk assessment in relation to food safety is that it should be rooted in scienti c data. The sources of the data should be assembled in a systematic manner and should stem from valid scienti c studies and communities across the world.1 A joint WHO/FAO consultation provides an advice on practical approaches for the application of risk analysis to food standard issues.During this consultation di erent issues were discussed to agree on risk assessment model application. Finally, the consultation limited its consideration to biological and chemical agents in oron food. Considering this model, the consultation emphasized the need for better information to enhance the risk assessment process and it also considered the way in which uncertainty was associated with risk assessment. 15A proper risk assessment can be described as being objective and unbiased, with absolute transparency. A Proper risk assessment is a constantly revolving process consisting of hazard identi cation, hazard characterization, exposure assessment, and risk characterization. 1 Identi cation of biological, chemical and physical hazards in the early stage of the production process of food is an important step in risk assessment since it can prevent from outbreak of devastating effects in later stages. The nature and extent of the adverse health effects known to be associated with the speci fic hazard. Using toxicity studies and epidemiological data, a dose-response relationship should be established between different levels of exposure to the hazard and the likelihood of di erent adverse health effects.An exposure assessment examines the exposure to the hazard over a particular period of time in foods that are actually consumed,taking into account the food consumption patterns of the target population and levels of hazard in all steps of the production process. The assessment should also account for varying levels of hazard throughout production to estimate the likely hazard level at point of consumption.1The application of transcriptomic technology to chemical risk assessment has been proposed due to its sensitivity and the ability to examine more comprehensively the molecular changes resulting from chemical exposure 3. In most cases, toxicity is not expected to occur without alterations at the transcriptional level.4 These transcriptional alterations include both the direct and indirect e ects on the cell or tissue. The direct effects include potential key events in the mode of action for a chemical while the indirect effects include secondary processes that are activated following the initial damage. When performed in dose{response format, the transcriptionalchanges can provide both quantitative and qualitative information on the dose atwhich cellular processes are a ected. This information can then be used to identify a transcriptional point-of-departure for chemical risk assessment 5. “Risk” is being used to mean the probability of the detrimental effect of interest. Risk is typically zero when dose is zero,for acute or short-term exposures, and it is commonly assumed that risk is determined from cumulative exposure, de ned as dose duration. Habers Rule usually stated as Ct 14 k, where C is an airborne concentration, t is time (duration) of exposure, and k is a constant.11 Toxicological response measurements generally can be classi ed into two distinct types: nominal(quantal) scale and continuous (quantitative) scale. Quantal responses characterize the effect by the presence or absence of a condition on a toxicity endpoint while Quantitative responses measure the effect in quantitative terms such as a change in weight, survival time, or some hematology or clinical chemistry measurements. Although the use of mathematical models in risk assessment for quantal response data has been widely developed for carcinogenic effects and to a lesser extent for reproductive/developmental effects, the use of dose response models in risk assessment for quantitative response data has been slow to develop.16 An essential step in risk assessment analysis is that selection of end points. Multiple end-points are collected and the most sensitive endpoint associated with an adverse e ect is used as the basis of evaluation for the risk assessment, although current guidance suggests a more nuanced approach.6;7 Either a no observed adverse effect level (NOAEL) or the statistical lower con dence limit on the benchmark dose (BMDL) is used as the point-of-departure (POD), which is then adjusted by uncertainty factors to take into account potential differences in inter-species extrapolation, pharmacodynamic and pharmacokinetic variability (i.e., intra-species variability), database de ciency, and exposure duration. The nal result is a reference dose (RfD) or reference concentration (RfC) which is de ned by the EPA as an estimate (with uncertainty spanning perhaps an order of magnitude) of a continuous exposure to the human population(including sensitive subgroups) that is likely to be without an appreciable risk of deleterious effects during a lifetime. Determination of NOEL is arbitrary and varies from investigator to investigator; the precision of determination varies from experiment to experiment. In addition, the NOEL may not represent a safe” dose for the laboratory animal tested because the power of the experimentmay be inadequate to detect subtle toxic effects. This is an indication that the NOEL may have been inadequately established. Also the safety factors are somewhat arbitrary and may be adequate on the average, but may be inadequate for any particular case.16 Benchmark dose (BMD) is introduced as better alternative to detect toxic effect. First it introduced by Crump8, and its statistical lower limit, the BMDL, have become essential in the effort to identify exposure limits .The idea is to use the data at all doses groups in a study and to use a lower bound to take sample size into account: the smaller the sample size, the less information is availableand the more conservative (lower) the lower bound.10 Now a days, the BMD is being used increasingly in place of the NOAEL and EPA has produced software for the BMD.9 The BMD approach is applicable to all toxicological effects. It makes use of all of the dose{response data to estimate the shape of the overall dose{response relationship for a particular endpoint.22 Due to this, the BMD approach is suitable for comparison of the potencies of different substance, or same substance under different exposure conditions 17 and analyzing the effects of covariates on the dose-response.20 It is also suitable for the derivation of Toxic EquivalencyFactors (TEF) for individual substances in a mixture that share a common mode of toxicological action (the BMD approach has been used to provide relative potency estimates for different organophosphates21). In mode of action analyses, relative potency estimates of BMD approach are more appropriate than NOAELs approach.18 Probabilistic approaches in risk assessment are receiving increasing attention, regarding both exposure assessment and hazard.17 The BMD approach is also compatible with probabilistic hazard characterization, as the uncertainty in the BMD can be quanti ed in the form of a distribution.19 Further, the dose-response modeling behind the BMD approach provides a means of estimating the magnitude of a potential health effect in the human population, given a particular exposure level.17 As it is mentioned earlier, assessing a benchmark dose based on the NOAEL approach may not be optimal because of its limited number of dose groups. The benchmark dose is derived from the complete dose- response curve, which is in uenced by both the number of dose groups and the number of animals per dose group.The use of real data appears favorable for obvious reasons. However, the problem is that the conclusions drawn from the results only hold for the particular situation in that particular study, while the number of possible situations (i.e., combinations of designs, shapes of dose-response relationships, and residual variance) is without bounds. Computer simulations o er the advantage that many more situations can be explored compared to the use of real toxicity data. Besides that, in computer simulations the `true’ dose-response relationship is known by de nition, providing a reference for the results based on the generated data.12 Currently, the term dose{response model is used for a mathematical expression (function) that describes the relationship between (mean) response (biological e ect ) and dose (chemical agent).Ideally, the relationship between dose and response would be described by a biologically based model that describes (models) the essential toxicokinetic and {dynamic processes related to the speci c compound. For most compounds, such models are not available, and therefore, the BMD approach uses fairly simple models that do not describe the underlying biology in any detail, and should be treated as purely statistical models. As the purpose of a BMD analysis is not to nd the best estimate of the (true) BMD but rather to nd all plausible values of the (true) BMD, given the data available, not only the best- tting model but also the modelsresulting in a slightly poorer t need to be taken into account.22The Scienti c Committee (SC) reviewed EFSA’s, 2009 document how they implement the BMD approach; the experience gained with its application and the latest methodological developments in regulatory risk assessment, and concluded that an update of its guidance from 2009 was necessary. Most of the modi cations made to the SC guidance of 2009 concern the section providing guidance on how to apply the BMD approach in practice. The updated EFSA document suggest that model averaging preferred method for calculating the BMD con dence interval, while acknowledging that the respective tools are still under development. As these tools may currently not be easily accessible to every risk assessor, the simpler approach of selecting or rejecting models is still considered as a suboptimal alternative. The set of default models to be used for the BMD analysis has been reviewed, and the Akaike information criterion (AIC) has been introduced instead of the log-likelihood to characterize the relative goodness of t of different mathematical models to a dose{response data set. All this method was implemented for a single endpoint of dose response data (quantitative). Current EFSA guidance providesthe implementation of benchmark dose on multiple endpoints and multiple studies.22 The aim of this study was to t the existence dose response models in EFSA, 2016 and extend them to multiple studies in the case of quantitative approach.