Women historically were excluded from clinical pharmaceutical trials because of potential risks to individuals of childbearing potential. The now discredited belief that studies of men apply without modification to women also contributed to this oversight. Remediation by the US National Institute of Health (NIH) Revitalization Act of 1993 mandated enrollment of women in federally supported phase III clinical trials; a European Union initiative provided practical tools regarding sex differences in drug study design [1]. Although the inclusion of women in clinical research subsequently increased in the USA, the EU, and Australia [1,2,3,4], most studies did not provide sex-specific data analyses [5, 6]. A 2018 review of 107 NIH funded randomized control trial studies that enrolled both men and women found that only 26% reported even one outcome by sex or included both sexes as a covariate [7]; 72% simply did not include sex in their analyses. NIH policies mandated over a quarter century ago have yet to yield the intended increases in reporting by sex.
A consequence of this sex inequality hides in plain sight today: most drugs are prescribed to women and men at the same dose. Many currently prescribed drugs were approved by the US Food and Drug Administration (FDA) prior to 1993, with inadequate enrollment of female animals in preclinical research and of women in clinical trials [8]. An illustrative example is found in the sedative-hypnotic drug zolpidem, which has been marketed under several names (e.g., Ambien, Edluar, Zolpimist) since it was first approved by the FDA, 1 year before the NIH Revitalization Act. Only after decades of post-marketing reports of cognitive deficits in women given the standard male dose were sex-based dose adjustments developed (NDA 021774). Many other drugs administered in equal doses to women and men likely require re-evaluation for sex-specific dose adjustment. Even within sex, dosing is not usually weight-adjusted in adults [9]; it remains uncertain whether weight-adjusted doses will suffice to offset the majority of sex-specific ADRs. One survey of clinical pharmacology data for 300 new drug applications (NDAs) evaluated by the FDA between 1994 and 2000 [10] indicated that 31% of studies showed a possible sex effect based on pharmacokinetic (PK) sex differences greater than 20%. In the same report, 11 drugs showed a > 40% difference between males and females in PK measures, yet no dosing recommendations to consider sex were issued, implicitly based on the unsubstantiated grounds that these differences were not clinically relevant. The existing knowledge base for sex-aware prescribing lacks information on sex differences for one-third of all drugs [9, 10]. Pharmaceutical companies responsible for generating pre-approval data often fail to include information on sex differences in NDA documents, and the FDA has previously failed to enforce its own requirements before approving new drugs [11]. Consequently, potential sex differences in PK measures and their relation to unwanted side effects often remain unknown. Most of the data submitted to the FDA by drug companies are not publicly available and not subject to peer-review by the broader scientific community [9]. Regulatory agencies have historically paid insufficient attention to differences between women and men in terms of both sex and gender, which perpetuates inequalities by neglecting drug safety problems that are sex-specific. In addition, this disparity allows for misleading drug marketing [11].
The present review study inventories drugs that elicit different responses in women and men and considers sex differences in adverse drug reactions (ADRs) that occur in individuals treated with therapeutic doses of medications. The range of ADRs included, but was not limited to, nausea, headache, drowsiness, depression, excessive weight gain, cognitive deficits, seizures, hallucinations, agitation, and cardiac anomalies. Sex differentiated ADRs were operationally defined as statistically significant differences in unintended drug effects in one sex, as reported in peer-reviewed literature or in NDAs. Supplementary evidence of ADRs was obtained from VigiBase [12], the World Health Organization (WHO) global database of individual case safety reports (ICSRs), which since 1968 has aggregated ICSRs from member countries of the WHO Programme for International Drug Monitoring. It is important to note that pharmacovigilance datasets, including but not limited to VigiBase, have a number of shortcomings: the information in VigiBase comes from a variety of sources, and the probability that the suspected ADR is, in fact, drug-related is not the same in all cases. Thus, it cannot be proven that a specific drug caused an ADR, rather than an underlying illness or other concomitant medication. Reports submitted to VigiBase come from both regulated and voluntary sources, and the volume of reports about a given substance may be influenced by extent of use of the product, the nature of the ADR, and/or publicity. Lastly, many drugs are disproportionately prescribed to one sex, and VigiBase data do not account for the number of patients of each sex exposed to the drug [13].
In contrast, much stronger and less confounded evidence comes from clinical trials and experimentally controlled empirical studies of drugs, in which ADRs can be quantified in the context of a known number of subjects administered the drug. Because the present report is, to our knowledge, the most comprehensive attempt yet to identify sex-biased ADRs using clinically identified ADR data, it also provided an opportunity to evaluate the accuracy of VigiBase data in estimating sex differences in ADRs compared to estimates obtained under more controlled experimental conditions.
Reasons why men and women respond differently to drugs
Women have a nearly 2-fold greater risk than men for exhibiting ADRs across all drug classes and are significantly more likely to be hospitalized secondary to an ADR [14,15,16]. This disparity is pervasive: among 668 drugs of the 20 most frequent treatment regimens in the USA, 307 (46%) report significant sex differences in ADRs [17]. In an analysis of cohort studies on 48 drugs completed between 1982 and 1997, compiling data from > 500,000 patients, Martin et al. [18] concluded that women over the age of 19 were 43 to 69% more likely to have an ADR recorded by their general practitioner. ADRs also peaked earlier in development among women (30–39 years of age) than men (50–59 years of age [18];). Women are also more likely than men to use two or more medications concurrently (polypharmacy), and women use more unique medications per year (5.0 vs. 3.7, respectively), which may contribute to increased female ADRs [19] but also renders the issue of sex-aware dosing all the more critical.
A comprehensive survey of the chemical and biological processes that underlie PKs and pharmacodynamics (PDs), and sex differences therein, is beyond the scope of the present work. We refer the reader to several thorough reviews of mechanisms relevant to PKs and PDs [20, 21] and sex differences in drug disposition [22, 23]. Here we discuss biophysical and molecular mechanisms only as required for illustration. Future analyses of drug elimination mechanisms (e.g., via CYP3A4) may facilitate extrapolation of the present results to a larger sample of drugs.
In general, drug disposition occurs through separate phases: absorption, distribution, bioavailability, metabolism, and excretion, and sex differences have been documented for each phase [22, 23]. Women generally have a lower body weight and organ size and a higher percentage of body fat, which affects the absorption and distribution of drugs. The larger the volume of distribution (Vd), the more likely the drug will be found in body tissues.
A number of biological, psychological, and cultural factors may contribute to why sex is such a strong risk factors for ADRs, including the following: sex differences in PKs and PDs, sex-specific organizational (early life) and activational (peripubertal through adulthood) endogenous steroid hormone exposure, and sex differences in exogenously administered steroids, higher rates of polypharmacy in women, sex differences in the expression of somatoform disorders, and sex differences in reporting rates [24].
Drug clearance is strongly linked to sex-specific expression of metabolic enzyme systems [25,26,27]; renal clearance of drugs is decreased in women because of a relatively lower glomerular filtration rate compared to men [28]. Women have a slower gastric emptying time and lower gastric pH, lower plasma volume, body mass index, average organ blood flow, and total body water differences, all of which affect drug distribution and PKs [29]. Responses to drugs are also affected by physiological changes during the menstrual cycle. The striking hormonal variations across days over the course of the human menstrual have no parallel in men in which hormonal variations largely occur within rather than across days [30].
The anticoagulant drug lepirudin is excreted by the kidney with systemic clearance in women about 25% lower than in men (NDA 020807 [31]). But PK variables do not translate linearly into phenotypes; thus, in women lepirudin is detectible in the circulation for up to 48 h, compared to just 2 h in men, which greatly increases the potential for undesired bleeding [32]; indeed, in this example, low molecular weight heparin-induced thrombocytopenia is a clinically important ADR which occurs more frequently in women than men [33], a difference that corresponded to much higher drug exposure as indicated by female-bias in multiple PK measures.
Arpon et al. [34] maintain that dosing adjusted according to a range of patient-specific factors is of large and increasing clinical and financial concern but lament that the amount of body weight adjusted dosing change that occurs in clinical practice is presently unknown. Adjustments for body mass in most cases do not ameliorate the high incidence of female ADRs. Thus, a multivariate regression analysis controlling for age, body mass index, and number of prescribed drugs identified a strong and significant effect of female sex on the increased risk of encountering ADRs, indicating that the sex disparity in ADRs does not merely reflect body mass masquerading as an effect of sex [35].
Most drugs are not administered on a milligram/kilogram basis but as a “one size fits all” dose, leading to higher exposures in women [10]. Under optimal circumstances, the dose should be based on milligram/kilogram body weight, or titrated to the desired clinical effect [36]. Correction for height, weight, surface area, or body composition eliminates a minority of sex-dependent PK differences [10]. The inference that weight-corrected PKs are comparable between men and women is not generally warranted but must be examined on a case-by-case basis, if data on both sexes exist. If for a given drug with a sexually differentiated pattern of ADR expression, correction for body weight eliminates the sex differences in PKs, this may or may not have bearing on PK-driven exposure leading to sex-differentiated ADRs. In short, sex differences in PK may be sufficient but not necessary for the manifestation of sex differences in ADRs.
PKs differ in men and women for many drugs [4, 21,22,23, 37,38,39,40]; this impacts drug efficacy and toxicity [41, 42]. Data from bioequivalence trials identified significant sex differences in PKs in ~ 28% of data sets [43]. Despite these differences, sex-specific dosing recommendations are absent for most drugs [37]. When women consistently experience less therapeutic effect or more adverse effects, a change in dosing regimen may be necessary.
Sex-related differences in PKs present a more significant challenge for medications with a low therapeutic index, i.e., those in which the lowest effective dose approaches toxic concentrations [44, 45]. If the therapeutic index is narrow, these differences are more likely to become clinically significant. A detailed study of sex differences for drugs with either steep dose response curves or narrow therapeutic indices is warranted [46]. For drugs with a relatively wide therapeutic index, a fixed dosing regimen is less of an impediment, but even in these instances, selecting the lowest effective dose would be prudent in women, at a minimum to decrease the potential for ADRs, as women are more likely than men to be prescribed more than one drug at a time.
Population PK studies often rely on sparse blood sampling collected from many subjects in phase II and III clinical trials. PK information is included in only a small minority of approved drug labels [10]; among > 2500 compounds listed in the Physicians Desk Reference (PDR), only 88 (fewer than 4%) presented population PK information in labelling [47]. PK sex difference data routinely are derived from small clinical pharmacology studies, typically enrolling 12–24 healthy subjects. Studies with such relatively small sample sizes have lower statistical power: they may be adequate to detect only very large sex differences in PK attributes, but as effect sizes decrease, these under-powered experimental designs generate widespread type II statistical errors—they become less and less capable of identifying real sex differences.
Aims
Here we examine relations between sex differences in drug PKs and ADRs to critically evaluate the hypothesis that drug exposure (PKs) and bodily responses to drugs (PDs; more specifically, clinically unintended effects, or ADRs) should be considered in the development of rational, feasible sex-based dosing adjustments. We conclude that such considerations are presently missing and recommend that sex-based dosing recommendations be disseminated to physicians and appear on drug labels. In many cases, these changes can be implemented at little cost. As demonstrated below, the data required to implement these procedures already exist for a number of drugs but have been ignored.
To determine the extent to which sex as a biological variable has been incorporated in the development of therapeutic pharmaceuticals, we reviewed whether sex differences in PK data were identified for any given drug and, where present, if such differences were incorporated into recommendations for sex-aware dosing/prescribing. Additionally, we reviewed an extensive literature on sex-specific ADRs, to evaluate whether sex differences in PK data predict sex disparities in ADRs. The hypothesis tested was that sex differences in PKs, specifically higher drug exposure in women than men, would be associated with clinically significant sex differences in ADRs. Support for this hypothesis, based on the broad net cast in the present investigation, would support dose reductions in women for the drugs under investigation here, and perhaps warrant extrapolation to any drugs for which sex differences in PKs exist. Several recent comprehensive reviews have addressed sex differences in drug treatment targeting specific diseases [48,49,50,51,52,53]. Here we report, for the first time, substantial sex differences in PKs and ADRs for 86 drugs, spanning multiple (> 10) therapeutic categories that support dose adjustments for women.