Based on the deliberations of the 1995 Gender Workshop, two central questions for assessing gender effects during drug development emerge: 1) How can we best determine whether the dose regimen should be adjusted based on gender? 2) When during drug development, should this determination be made? The answers to these questions depend on the specific drug and its intended use, the degree of certainty that must be achieved, and the information available about the drug itself or related drugs. The FDA is involved in probing these questions further in order to develop strategies that will assist sponsors and other researchers to analyze gender-related drug effects which provided the impetus for this workshop.
As noted by the Workshop participants and delineated in the 1993 Guideline, information about potentially important gender effects can be provided by a number of clinical and non-clinical approaches, including:
in vitro studies in human or non-human tissue;
in vivo studies in non-human species;
early phase exploratory clinical studies using pharmacokinetic and pharmacodynamic approaches, coupled with a mechanistic understanding of drug action, if available;
late phase confirmatory studies in humans;
A combination of these approaches is likely to prove useful in defining gender and other population subset effects. To build on the data already being collected in clinical trials, the need is also apparent for: 1) more complete tabulations and graphics of safety and efficacy data grouped by gender; 2) more by-gender analyses, generally by greater use of meta-analysis to maximize the amount of information gained for both genders; and 3) new ways to think about and perform meta-analyses of information from studies of varying design.
A ‘decision tree’ approach to thinking about ways to assess gender effects during the development of medical products may be useful (Figure 1). At the start of clinical development, prior information - such as the intended use of the drug, the likelihood of the drug having a narrow therapeutic range, the drug’s possible routes of elimination, its use for critical and lifesaving illnesses, animal PK/PD data, and other pertinent factors - should be coupled with initial PK/PD studies to determine whether signals of a gender effect exist. Included in this initial screen might be data from non-clinical studies, such as in vitro studies of human tissue, which are routinely used to identify possible important routes of metabolism and potential drug-drug interactions and which might signal the presence or absence of a gender effect. Workshop discussions suggest, however, that further work may be necessary before this approach can be used to predict gender effects reliably. Some studies do suggest, however, that if significant metabolism occurs via CYP 450 1A and 3A, a gender and/or hormonal effect may be more likely.
In early phase clinical trials, entry of sufficient numbers of both genders will allow initial pharmacokinetics and perhaps pharmacodynamic evaluation for a gender effect. In general, important pharmacokinetic differences by gender is readily obtained even when the numbers of subjects in these studies is relatively small. Where good pharmacodynamic measures exist (as they do, for example, with beta-blockers, ACE inhibitors and others), it is relatively easy to obtain a good pharmacokinetic/pharmacodynamic relationship that allows important clinical differences to be anticipated. The availability of a suitable pharmacologic endpoint to allow a pharmacodynamic or other clinical assessment of a gender difference may be problematic, although recent surveys per- formed by FDA suggest that some useful pharmacodynamic endpoints may be available in up to 50 percent of investigational drugs, and possibly more. If a signal of gender effect emerges, more elaborate pharmacokinetics and pharmacodynamic studies may be indicated. An early indication of by-gender differences, subsequently confirmed by targeted pharmacokinetics and/or pharmacodynamic studies, suggest that later phase studies should be designed and powered to assess gender effects. If pharmacokinetics, and, if possible, pharmacodynamic results from early studies are negative, subsequent reliance on population pharmacokinetic/pharmacodynamic approaches (pharmacokinetics screens) and post-hoc gender subgroup analyses should generally be sufficient to investigate the impact of gender on the safety and efficacy of a drug.
The post-hoc subset analyses now relied on by the FDA to assess gender and other subset group effects are likely to detect large differences in safety and efficacy, but smaller differences may go undetected. A remaining question, then, is whether smaller differences - which might be discovered from more precise pharmacokinetic and pharmacodynamic studies, should be translated into labeling instructions. The answer will be made on a case-by-case basis, recognizing that the population dose response relationship and the steepness of the dose/response curve will determine whether different dosing recommendations should be provided to men and women.