Underrepresentation of blacks, smokers, and obese patients in studies of monoclonal antibodies for asthma

TO THE EDITOR:

The randomized control trial (RCT) is the highest level of scientific evidence, minimizing systematic bias and increasing the reliability of effect estimates. Strict eligibility and extensive exclusion criteria, however, can limit the generalizability of results.¹ If groups underrepresented in trials are prevalent in everyday practice, there may be differences in efficacy observed in trials compared with effectiveness in real-life practice. Prior studies have shown that children, adults ≥65 years, females, Hispanics, and African Americans are often underrepresented in clinical trials.^2,3

RCTs have shown the efficacy of multiple biologics in the treatment of severe asthma. Asthma affects over 25 million individuals in the United States and disproportionately affects African Americans and Hispanics.⁴ These minorities, in addition to postpubertal females, obese individuals, and smokers, are more likely to have severe disease and have decreased response to bronchodilators or corticosteroids.^4–6 In addition, age could modify therapeutic efficacy as childhood-onset and adult-onset asthma are now recognized as distinct phenotypes.⁷ We compared baseline characteristics of participants enrolled in the RCTs for the biologics currently approved by the Food and Drug Administration (FDA) for severe asthma with participants with uncontrolled asthma from a nationally representative sample.

For each of the biologics, omalizumab, mepolizumab, benralizumab, reslizumab, and dupilumab, we identified the seminal placebo-controlled RCTs used in granting FDA approval. We extracted data about trial participants from the medication guides and/or from published results in peer-reviewed journals. For comparison, we used data from the 2005–2012 National Health and Nutrition Examination Survey (NHANES) to generate a nationally representative sample. We identified survey participants aged ≥12 years with asthma. Of these, we identified those who were likely to have uncontrolled asthma in 2 ways. NHANES-A were on inhaled corticosteroid (ICS) therapy singly or in combination with a long-acting β-agonist (LABA) and had ≥2 exacerbations in the prior year requiring a visit to an emergency department or urgent care center. NHANES-B were also on ICS and had ≥2 exacerbations in the prior year, but in addition had been on 1 or more additional controllers (LABA, long-acting muscarinic antagonists, leukotriene antagonists, and/or methylxanthines) or oral steroids for at least 90 days at the time of interview. Covariates of interest included age, gender, race, body mass index (BMI), and smoking status. Detailed methods are discussed in this article’s Online Repository at www.jaci-inpractice.org.

Of 2475 survey participants aged ≥12 years with current asthma, we identified 80 and 48 patients in NHANES-A and NHANES-B, respectively. A total of 16 RCTs were identified (see this article’s Online Repository at www.jaci-inpractice.org). We used the trial names to identify these studies. Where no trial name is available, we name them based on the relevant biologic, for example, XOLAIR-I is “Trial 1” for omalizumab as listed in the medication guide.

Trial participants were significantly older (range of mean age: 48.6–52.0 years) in comparison with NHANES-A and NHANES-B (42.6 and 43.4 years) in all studies for which mean age and standard deviation were reported (Figure 1, A, and Tables E1–E5, available in this article’s Online Repository at www.jaci-inpractice.org). Significantly fewer females were enrolled in the trials (51% to 66%) than in NHANES-A (77% of participants). The direction of effect was similar in comparison with NHANES-B in which 66% of participants were females, but only reached statistical significance in comparison with XOLAIR-II (59%) (Tables E1–E5, available in this article’s Online Repository at www.jaci-inpractice.org).

FIGURE 1.

A, Mean age in years with error bar; B, proportion of whites and blacks; C, body mass index, in kg/m²; D, proportion of never and former smokers. Proportion of current smokers was 26% in both National Health and Nutrition Examination Survey groups but 0% to 1% in all the studies when reported.

Compared with NHANES-A and -B, 11 of the 15 RCTs that reported proportion of Caucasians had significantly higher proportions of white participants. Compared with NHANES-A and -B, 8 of 9 RCTs reporting percentage of blacks had significantly lower proportion of blacks. Only 5 studies reported Hispanic ethnicity. Three of these 5 had a higher proportion of Hispanics than NHANES-A or -B (Figure 1, B, and Tables E2–E4, available in this article’s Online Repository at www.jaci-inpractice.org). In all of the studies where BMI was reported, the mean BMI of the trial participants (27.8–29.6 kg/m²) was significantly lower than in NHANES-A and -B (33.6 and 32.6 kg/m²) (Figure 1, D). The trials included a higher proportion of never smokers (61% to 81%) than NHANES-A or -B (45% and 41%) (Figure 1, D, and Tables E1–E5, available in this article’s Online Repository at www.jaci-inpractice.org). The duration of asthma was similar between trial and survey participants in all the studies where reported.

Overall, we found that participants enrolled in trials for the biologics currently approved for the treatment of asthma are older, more likely to be Caucasian and never smokers, and have lower BMI compared with a nationally representative sample of participants with uncontrolled asthma. These findings suggest that study participants may not be representative of the overall population of those with severe uncontrolled asthma. This lack of representativeness may have implications for the estimated treatment effect size of these biologic medicines.

Blacks, obese patients, and smokers are more likely to have severe asthma and demonstrate lower response to bronchodilators, LABA, and/or corticosteroids.^5,6,8 However, previous work to identify patients likely to respond to biologics have generally focused on asthma severity, IgE levels, and eosinophil counts. Only a few studies have explored the impact of demographic factors on response to these biologics with most of these being retrospective studies or post hoc analyses of RCTs. After initial approval, a randomized trial of omalizumab in inner city children, a group with high allergen sensitivity and over 90% being black or Hispanic, showed that omalizumab was very effective in decreasing exacerbations and dose of inhaled glucocorticoid.⁹ In a postmarketing surveillance study using the Australian Xolair Registry, response to omalizumab was found to be similar between obese and nonobese indviduals,¹⁰ but a post hoc analysis of the DREAM study¹¹ showed greater improvement in exacerbation rates in the cluster of female-predominant obese patients receiving mepolizumab. Similarly, a single-center retrospective study¹² of 42 patients (32 mepolizumab-responders and 10 nonresponders) did not show statistically significant differences between these groups; however, median BMI was 28 kg/m² among the responders, compared with 31 kg/m² among the nonresponders. In the same study, over half of responders were never smokers compared with 40% of nonresponders. Another study using the Australian Xolair Registry¹³ showed that omalizumab improved asthma control and health-related quality of life in patients with asthma-chronic obstructive pulmonary disease overlap syndrome, but there was no improvement in lung function in this group.

It may be reassuring that these studies have, in general, not shown great differences in the efficacy of these biologics based on demographic subgroups; however, the prevailing knowledge of asthma phenotypes suggests that there is likely to be heterogeneity of treatment effect. For instance, multiple studies have shown that obesity, especially in women, might represent a distinct asthma phenotype and might influence response to biologic therapy.^11,14 Furthermore, post hoc and retrospective analyses may lack adequate power and often suffer from reporting bias.¹⁵ Given the low numbers of smokers, obese individuals, and minorities in these trials of biologics in asthma, post hoc analyses are unlikely to have been adequately powered to detect statistically significant differences when multiplicity is appropriately accounted for. These issues could be limited if adequate numbers of these patient subgroups are enrolled in randomized and prospective trials and if prespecified subgroup analyses are appropriately conducted using the test for interaction rather than reporting different effect sizes in the subgroups. Thus, adequate enrollment of racial minorities, obese patients, and smokers in studies of biologics is necessary for obtaining accurate estimates of efficacy of biologics in these subgroups of patients. Clinical researchers and pharmaceutical companies could consider enrollment methods that will help improve the generalizability of trial results. Finally, given the widespread underrepresentation of minorities in clinical trials in general,¹⁶ regulatory bodies should search for these differences in research designs and help ensure that studies are representative of patient population.

This study has limitations. First, asthma diagnosis for NHANES was by self-report, and uncontrolled asthma as we defined it might reflect health care utilization and not true severity. However, baseline characteristics reported are similar to those of prior asthma epidemiology studies. We also attempted to increase our accuracy by using 2 different definitions, and we found estimates very similar regardless of the group used for comparison. Secondly, although these surveys are large, the number of participants with uncontrolled asthma was small. However, this is more likely to decrease rather than increase the probability of identifying differences, and the differences we have found are mostly at the P < .001 level, which is remarkable for such small sample sizes. Furthermore, although we used only 16 studies in these comparisons, there are multiple other studies of these biologics. Some of the studies not considered by the FDA during the approval process, such as the omalizumab in the inner city children study,⁹ may have included a higher proportion of groups underrepresented in these seminal studies. However, these studies were used in granting approval of these biologics. Despite these limitations, we believe that this study highlights the important subject of representativeness of participants in clinical trials of biologics.

Clinical Implications.

• Participants in the randomized control trials foromalizumab, mepolizumab, benralizumab, reslizumab, and dupilumab in asthma are more likely to be Caucasian, never smokers, and have lower body mass index in comparison with patients with uncontrolled asthma in the general population.