Chan et al. (19) evaluated CDER's approach to selection of the MRSD, using data from 35-36 non-cytotoxic drugs tested in humans during the years 1996-2002. Although not stated, it is assumed that the SF used in each case was the default factor of 10, since there was no discussion of a range of SFs. For all 35-36 drugs, the cardiovascular, gastrointestinal, and nervous systems were the most frequent sites of adverse effects. For 20-21 drugs, a no-effect or minimal-effect dose was identified. The calculated MRSD was lower than the no- or minimal-effect clinical dose in 11 of the 21 cases (52%). In the worst case, the calculated MRSD was 20 times higher than the no- or minimal-effect clinical dose. For 15 drugs, the lowest dose associated with what were considered significant adverse effects was identified. The calculated MRSD was lower than the clinical dose associated with significant adverse effects in 12 of the 15 cases (80%). In the worst case, the calculated MRSD was approximately six times higher than the clinical dose associated with significant adverse effects.
An internal effort has been initiated in order to validate CDER's approach described in the guidance for industry. The first step in this validation was to identify a number of drugs for which sufficient data were available to determine whether or not selection of the MRSD according to CDER's algorithm would have been successful, that is, starting dose associated with no (or minimal) toxicity. The analysis was based on data for 69 drugs, representing all review divisions. Reviewers were asked to provide the following information for each case: (i) therapeutic indication, (ii) the route of administration used and the duration of the toxicity studies used to determine NOAELs in animals, (iii) the NOAEL in the most sensitive species, (iv) the start dose and the route of administration used in the FIH trial, (v) the findings in humans at the start dose, (vi) the highest dose tested in the FIH trial, if known, and (vii) the human dose associated with significant toxicity, if identified.
For six drugs, there were insufficient data in humans to be included in the analysis. Therefore, the overall analysis included data from 63 drugs. For 20 of the 63 drugs, a toxic dose was identified in humans. For 35 drugs, data at doses above the starting dose were available; however, no toxic dose was identified. For eight drugs, human data were only available at the starting dose. For one drug, the route of administration differed between animals (subcutaneous) and humans (sublingual). In the majority of the cases, NOAELs in animals were based on toxicity studies of up to 3 months in duration; for two drugs, NOAELs were based on 6-month studies; for one drug, NOAELs were based on data from 13- to 26-week studies. It is not known how the actual starting dose used was selected in each case. Generally, the reviewer calculates a MRSD, using the method described, and compares that with the start dose proposed by the sponsor. However, reviewers do not routinely recommend adjusting the starting dose proposed by the sponsor unless a safety concern is raised.
In all but two of the 63 cases, the starting dose was successful, in that it was not associated with adverse effects. However, an important factor in some cases was selection of the safety margin. For the cases in which a toxic dose had been identified, the safety margins selected (i.e., HED/HSD) ranged from 0.7 to 300. Overall, safety margins used ranged from 0.4 to 2700. By category, 25/63 cases used a safety margin of < 10, 3/63 cases used a safety margin of 10, 24/63 used a safety margin between 10 and 100, and 11/63 used a safety margin of > 100. For the cases in which a toxic dose in humans had been identified, use of a 10-fold safety margin would have resulted in a HSD greater than the toxic dose in two cases. In 12 of the 20 cases, the toxic dose:HSD ratio would have been <10. Therefore, the selection of a sufficient safety margin would appear to be critical.
The conclusions of CDER's preliminary analysis are limited by the fact that the cases examined were not random samples of relevant cases, that in some instances selection of the starting dose was based on human data, and by the lack of sufficient data from the initial clinical trial. Cases are continuing to be collected in order to complete the evaluation of CDER's approach to selection of the MRSD. However, the results of Chan et al. (19) and the preliminary internal analysis both indicate that serious consideration should be given to selection of an appropriate safety margin, and that a 10-fold SF may not be sufficient in some cases.
Was this article helpful?