She Loves Me, She Loves Me Not: Do Biologics Decrease Glucocorticoid Morbidity in Adults with Severe T-Helper Cell Type 2 Asthma?

The development of biologics that target in particular T-helper cell type 2 (Th2) inflammation has revolutionized the treatment of severe asthma. Although daily or alternate-day steroids, with all their attendant side effects, were the court of last appeal to control severe asthma in the past, now the advances in biologics mean that “steroid-dependent asthma” is becoming a thing of the past for patients with type 2 asthma.

Biologics have undoubtedly improved asthma outcomes, especially reducing asthma attacks while allowing steroid treatment to be tapered. In this issue of the Journal (pp. 1165–1174) Sadatsafavi and colleagues (1) present the results of a substantial research effort. They analyzed data from two large cohorts, the International Severe Asthma Registry (ISAR) cohort and a United Kingdom– based Optimum Patient Care Research Database (OPCRD). ISAR contained 11,883 patients, of whom 8,432 (n = 4,280 biologic initiators; n = 4,152 noninitiators) were included; thus, 29% were excluded. OPCRD contained 41,154 patients, and 34,476 (n = 1,410 initiators, 33,066 noninitiators) were included; thus, 16% were excluded. This large dataset was used to address an important question. Specifically, they asked whether those patients with asthma requiring chronic oral corticosteroid (OCS) treatment had a lower probability of developing long-term OCS-related complications if therapy with an asthma biologic was initiated. Because several major publications have demonstrated an OCS-sparing effect of different biologic therapies for severe asthma (for example, see Reference 2) it would seem obvious that biologic therapies should decrease OCS-related side effects. However, their results could be interpreted in more than one way.

The strengths of the study were the use of real-world data and a huge number of patients from diverse settings (17 different countries). However, as with all studies, when one digs deep there are inevitable problems. First, only 34 patients were prescribed tezepelumab (3), the only monoclonal with activity against Th2_lo asthma, so we have to assume that all other patients met criteria for Th2_hi asthma, and thus this is the only group for which the findings are relevant. Second, the statistical analyses are complex. Because of the difficulty of doing a large longitudinal study in a prospective fashion, the authors used inverse probability weighting to compare propensity scores from what were two quite different longitudinal asthma cohorts. The authors used a logistic regression model to account for variables and compared hazard ratios for developing OCS-related side effects among those who were and were not commenced on any asthma biologic within specified time frames. They studied patients >18 years old meeting similar, but slightly different, global initiative for asthma (GINA) criteria for severe asthma in the two cohorts. In the ISAR, approximately half the patients with severe asthma were started on biologics. By contrast, in the OPCRD, this proportion was around 4%, which raises questions about the comparability of the cohorts and the validity of combining them. Next, could there have been an inadvertent selection bias as a result of excluding around 10,000 patients? Finally, as in any huge study, the diagnoses have to be taken on trust; some of the patients could in fact have had chronic obstructive pulmonary disease (COPD) that was misdiagnosed as asthma. The consequences of this are discussed below.

The authors evaluated 11 established complications of chronic OCS use. Surprisingly, the hazard ratio for five of these exhibited a trend for being lower in the group not started on biologics. Only hazard ratios for developing diabetes, major cardiovascular events, and anxiety or depression reached significance. These three drove a slight advantage to biologic initiation. There are thus problems of interpretation, not least driven by the fact that biologic initiation was not randomized. Those not initiated were more than fourfold more likely to smoke (5% vs. 23%) and were more obese. Indeed, smoking and obesity-related asthma may have made the respiratory health provider less likely to initiate biologics, given the sometimes different phenotypes of smoking- and obesity-related asthma. These considerations were meant to be addressed by the multiple regression model. Smoking and obesity are uniquely associated with cardiovascular events and diabetes, unlike most other OCS complications. Could the differences between the two groups merely reflect a different baseline risk between the two groups, driven by smoking and obesity?

There is also a confounder that was not comprehensively studied because data were not available: COPD. Most of the analysis was done without considering whether or not subjects had COPD. Because of the discrepancy in smoking prevalence, it would have been helpful to know that the logistic regression analysis accounted for COPD as a confounder. It is only recently that some biologics are indicated for COPD (4). Concomitant asthma and COPD might have created hesitancy in the provider’s mind with regard to starting a biologic, while at the same time increasing the subject’s risk for cardiovascular events, a sedentary lifestyle, obesity, and depression. So, it is arguable that the hazard ratios do not represent an accurate estimate of the benefit of biologics. A priori, we would have assumed that the hypothesis was correct. But we are concerned that the Reverend Bayes has been stretched here to the very limit of his capabilities (3, 5).

Last, the healthcare economics deserve mention. In the introduction, there is a discussion of the potential financial implications of OCS-related complications. However, quite correctly, the authors do not make any claim about a benefit to healthcare systems that could be realized by initiating biologic therapy: there are no data presented to any such claims. This is particularly important because of the exceptionally long list of conflicts of interest. It would be tempting to use this manuscript to argue that lowering OCS-related morbidity and cost offsets the cost of initiating biologic therapy for asthma, but this is not valid.

In conclusion, there are major benefits on asthma outcomes with biologics for Th2_hi disease. We congratulate Sadatsafavi and coworkers on a magnificent effort, but whether their data definitively demonstrate a reduction in systemic complications of OCS is debatable. We would be surprised if there was not a beneficial effect, but life is full of surprises. We need a longer-term, prospective study to determine the answer to the questions posed in this manuscript.