Sex‐by‐formulation interaction in bioequivalence studies: the importance of formulations and experimental conditions

Abstract

In a recently published investigation, the authors argued against the likelihood of sex‐based subject‐by‐formulation interactions in bioequivalence studies, i.e. male and female subjects exhibiting different discriminatory potential to detect bioavailability differences between formulations. The researchers performed a strong methodological study showing the increased probability of false‐positive findings in exploratory subgroup analysis, a well‐known and documented statistical issue. Indeed, the main limitation of assessing a sex‐by‐formulation interaction in average bioequivalence studies lies in the fact that these clinical trials are not designed for this purpose. In this commentary, we further discuss on why the impact of sex differences in gastrointestinal physiology over in vivo drug dissolution and absorption rate cannot remain hidden behind statistical limitations, particularly when average bioequivalence conclusions could be affected. Regulatory agencies should encourage and support these important issues related to biopharmaceutical quality of drug products in both sexes. In addition, a sex‐based analysis of bioequivalence results will enhance the representativeness of conclusions and provide important information regarding formulation performance, thereby promoting the efficacy and safety of generic drugs and reducing consumer risk. The extrapolation of study conclusions from one sex to another is far away from being scientifically proven.

González‐Rojano et al. recently published a study entitled Evaluation of sex‐by‐formulation interaction in bioequivalence studies of efavirenz tablets 1, which we read with great interest. In this article authors argue against the likelihood of sex‐based subject‐by‐formulation interactions in bioequivalence studies, i.e. male and female subjects exhibiting different discriminatory potential to detect bioavailability differences between formulations. The work focuses its criticism on an article we authored, where a sex‐by‐formulation interaction was proposed to explain the results obtained in an efavirenz bioequivalence study between a local product and the innovator Stocrin (Merck Sharp & Dohme) 2. Importantly, in our study the bioequivalence conclusion was not achieved since results showed a dissimilar absorption profile between the evaluated formulations. The difference was also detected by in vitro dissolution testing under biorelevant conditions. All results were published and data were made available 3.

The article by González‐Rojano et al. includes a strong methodology showing the increased probability of false‐positive findings in exploratory subgroup analysis, a well‐known and documented statistical issue. Indeed, the main limitation of assessing a sex‐by‐formulation interaction in average bioequivalence studies lies in the fact that these clinical trials are not designed for this purpose. In accordance with González‐Rojano et al., we consider that observations coming from these analyses should be further confirmed with properly designed experiments. Regulatory agencies should encourage and support discussions on these critical issues related to group‐by‐formulation interactions in average bioequivalence, such as the sex effect, which is probably the most important given its impact on the population. We therefore find this work very important to continue discussing this point.

Considering the previously commented statistical limitations, we supported our hypothesis by correlating biorelevant in vitro dissolution and disintegration testing with in vivo bioequivalence results for the maximum plasma concentration (C_max) bioequivalence ratio obtained in both sexes (107% and 70.5% in female and male subjects, respectively). Results showed that while efavirenz dissolution profiles were similar between assayed formulations under the presence of surfactants, differences were evident in simulated gastric fluid without enzyme. The correlation between these findings and the observed differences in product discrimination between sexes was built in light of a mechanistical background: the vastly reported sex‐related differences in gastrointestinal physiology. The impact of these differences over in vivo drug dissolution and absorption rate cannot be omitted from this discussion. We believe that the statistical analysis and discussion done by González‐Rojano et al. to refute the existence of sex‐by‐formulation interactions should be complemented with a physiological‐based pharmacokinetic analysis. Women present slower gastric emptying rate. Thus, after a meal intake, it is likely for male subjects to achieve the fasting state sooner than women. Given that our bioequivalence study was conducted with nightly dosing separating drug administration 2 h from the meal, gastrointestinal environment differences between sexes could have been present. Pharmacokinetic observations and in vitro–in vivo correlations supported this hypothesis and gave a physiological‐based explanation for the observed sex‐by‐formulation interaction.

Even under the agreement on the sample size limitations of our study, we do not consider that the observed sex‐by‐formulation interaction was explained by extreme values, as suggested by González‐Rojano et al. After removing subjects with extreme individual bioequivalence ratios (female 6 and male 13) sex‐related differences are conserved in the C_max test/reference ratio, with point estimates of 98% for female and 78% for male subjects. Figure 3 shown by González‐Rojano et al. includes an incorrectly estimated individual test and reference difference in C_max for subject 10, as the correct value is 0.022 in log scale. The correct graph is included here (Figure 1).

Figure 1.

Distribution of logarithm (maximum plasma concentration, C_max) differences between test and reference products among the individuals in the efavirenz bioequivalence study previously published 2. Female subjects are represented with green squares and male subjects with blue diamonds

While González‐Rojano et al. presented a very interesting discussion supporting their affirmations with bioequivalence pilot and pivotal studies between efavirenz tablets, it must be emphasized that subject‐by‐formulation interactions are dependent on the assayed formulations and the experimental design. Both aspects differed from the study we reported. Thus, results obtained in their studies should be carefully considered for interpreting and concluding on our results. More importantly, a sex‐by‐formulation interaction could arise only if evaluated formulations are truly different. If we compare the same formulation (i.e. Stocrin vs. Stocrin), no interaction could be observed no matter the sex of the subjects or the administration conditions. The formulations evaluated in our study were in fact different, as proved by the in vitro dissolution and disintegration results. This difference was observed in vivo only in male subjects. In contrast to this, the formulations evaluated by González‐Rojano et al. in the pivotal study (Test A vs. Sustiva, Brystol‐Myers Squibb) were regarded as bioequivalent, with no reported in vitro dissolution differences. Therefore, it is possible that the sex‐by‐formulation interaction observed in their pilot study could have been an artefact. Alternatively, the assumption made by González‐Rojano et al. of unaltered formulations within the 11‐month period that separated the pilot from the pivotal study might be inaccurate. The authors acknowledge that this lapse could have affected the relative bioavailability of formulations, since they were near the expiration date. Differences in relative bioavailability estimation between the pilot and pivotal studies might reflect a significant change in one or both of the formulations. The authors claimed that this eventuality would not necessarily affect the existence or the absence of a sex‐by‐formulation interaction, a key point on which we disagree.

In addition, González‐Rojano et al. claimed that “the hypothetical higher sensitivity of males is inconsequential, even if real, since bioequivalence studies are conducted in fasted state (after fasting for at least 8 h) in the European Union and the USA (not 2 h after a meal), when the product intake is recommended to be taken on an empty stomach.” Efavirenz dosage and administration recommendation made for the brand name drug states that the drug should be taken on an empty stomach, preferably at bedtime 4. Nightly dosing recommendation is done on the basis of an increased efavirenz tolerability. Administration conditions followed in our study took into account the clinical usage of the drug. From a scientific perspective and contemplating the aim of bioequivalence evaluation, we sustain the validity of our study and the conclusions we reached.

Bioequivalence studies, rather than being carried out under the most discriminative conditions, should be conducted to representative conclusions of the targeted population and the prescription conditions at the clinical setting. We dissent from the perspective sustained by González‐Rojano et al.: “current bioequivalence studies are already conducted in the most discriminative condition and in the most sensitive population if the conclusions of Ibarra et al. were real…” Our conclusions regarding men as more discriminative are valid for the assayed efavirenz formulations given 2 h after a meal. We are not implying that men are always the most sensitive population regardless the drug, the formulations and the experimental conditions. Rather, we point out that this phenomenon can happen and should be further investigated in bioequivalence studies disclosing the effect of sex on product discrimination, mainly when the drug is aimed to be given in both sexes.

The importance of sex in bioequivalence evaluation was further assessed in an article including several studies 5. Although we do not encourage a change in bioequivalence regulations requiring powered studies to conclude in both sexes, which would lead to a complex scenario with higher economical and ethical costs, we support the US‐Food and Drug administration recommendation regarding sex: if a product is intended for use in both sexes, the applicant should include a similar proportion of males and females in the study. In addition, a sex‐based analysis of bioequivalence results will enhance the representativeness of conclusions and provide important information regarding formulation performance, thereby promoting the efficacy and safety of generic drugs and reducing consumer risk. The extrapolation of study conclusions from one sex to another is far away from being scientifically proven.

Competing Interests

There are no competing interests to declare.

Ibarra M., Vázquez M., and Fagiolino P. (2019) Sex‐by‐formulation interaction in bioequivalence studies: the importance of formulations and experimental conditions, Br J Clin Pharmacol. 85, 669–671, 10.1111/bcp.13829.