This scientific output, published on the 10th of February 2011 replaces the the earlier version published on the 8th of February 2011
The CONTAM Unit asked support from the Assessment Methodology Unit to re-assess the elements linked to the establishment of the HBGV in the opinion on cadmium in food of the EFSA’s CONTAM panel. The EFSA opinion on Cadmium (EFSA, 2009a) needed to be reviewed based on a report established by the Joint FAO/WHO Expert Committee on Food Additives (JECFA, 2010) where a different provisional tolerable monthly intake was reported.
The EFSA opinion (EFSA, 2009a) was based on a meta-analysis of epidemiological studies assessing the concentration-effect relationship between Beta-2-microglobulin (B2M), a biomarker of tubular toxicity, and cadmium in urine which was considered as the most reliable basis on which to determine a critical concentration of cadmium. This meta-analysis was published in a separate report (EFSA, 2009b).
Based on this analysis the CONTAM Panel derived a group-based BMDL5 of 4 µg cadmium/g creatinine that was further adjusted based on the fact that modelling was done using summary measures (geometric mean and standard deviations) resulting in a reference point of 1 µg cadmium/g creatinine. In the JECFA report a breakpoint of 5.24 (CI: 4.94 - 5.57) µg cadmium/g creatinine was used as a point of departure.
In this scientific report the methodology used in the meta-analysis (EFSA, 2009b) is described in detail, focussing on how to deal with summary values from the selected studies as no individual data was available to build the concentration-effect model. The rationale behind the approach undertaken is explained and simulation based evidence is provided for the need of an adjustment factor to establish a health-based guidance value. The importance on the selection of the adjustment factor on the final establishment of a health-based guidance value is studied by providing examples of the impact of selecting different percentiles in the calculation of the adjustment factor, recognising that this is in general a policy decision and could be influenced by aspects such as the severity of the effect, the robustness of the data, the nature of the distribution and risk management considerations. A comparison between JECFA and EFSA approaches is presented, illustrating differences between the methods, as well as assumptions made by both approaches, which induces in both methods uncertainties when applying them. The JECFA approach plug-in distributions for the parameters instead of posterior mean values obtained from the posterior distribution when Bayesian inferences were used in the TK modelling (Amzal et al., 2009) in order to account for uncertainty. It is shown that by doing that, uncertainties about the right distribution to use are introduced. A sensitivity analysis in relation to the choice of the toxicodynamic variability function used by JECFA in their approach is also undertaken, in order to illustrate the impact of the choice made when establishing health-based guidance value.