现在的位置: 首页研究点评, 进展交流>正文
[JAMA述评]:干扰素β-1a对重度ARDS患者缺乏临床疗效
2020年02月29日 研究点评, 进展交流 暂无评论

Editorial February 17, 2020

Lack of Clinical Benefit of Interferon β-1a Among Patients With Severe Acute Respiratory Distress Syndrome: Time to Overhaul Drug Trials in ARDS?

Manu Shankar-Hari, Carolyn S. Calfee

JAMA. Published online February 17, 2020. doi:10.1001/jama.2019.2252

In acute respiratory distress syndrome (ARDS), the alveolar-capillary units are disrupted, with lung endothelial and epithelial injury resulting in exudative pulmonary edema containing inflammatory mediators, solutes, proteins, and leukocytes.1 Treating pulmonary endothelial injury in ARDS may improve patient outcomes. Interferon β-1a (IFN-β-1a), a type 1 IFN, is one such treatment that may improve pulmonary endothelial barrier function. In addition to its myriad immunologic effects, IFN-β-1a upregulates cluster of differentiation 73 (CD73) on pulmonary endothelial cells, thereby increasing extracellular adenosine concentrations, which acting via adenosine receptors improves pulmonary endothelial barrier function through junctional reorganization, cytoskeleton rearrangement, and further transcriptional upregulation of CD73.2-4

Based on these preclinical experimental observations, the clinical utility of IFN-β-1a for ARDS was previously tested by Bellingan and colleagues5 in an open-label, phase 1/2 dose-finding study in adult patients with ARDS, with 28-day mortality as the primary end point. In this nonrandomized study, IFN-β-1a was administered in escalating doses, with an optimum tolerated dose of 10 μg. Among the 37 patients who received IFN-β-1a, mortality was 8% compared with a higher mortality rate of 32.2% among 59 control patients (recruited contemporaneously but not randomized). This mortality difference, albeit in a nonrandomized sample, prompted the authors to design an efficacy trial.

In this issue of JAMA, Ranieri and colleagues6 report the results of a clinical trial in which 301 adults admitted with moderate or severe ARDS based on the Berlin definition7 were randomized to receive intravenous IFN-β-1a (10 μg) or placebo once daily for 6 days. The trial enrolled patients from 74 intensive care units in 8 European countries between December 2015 and December 2017. A novel approach was used to confirm ARDS, with eligibility verified by 1 of the 2 clinicians (ie, medical monitors) and the requirement that the enrolling clinician match the patient’s chest radiograph against a panel of unlabeled chest radiographs to ensure that imaging was truly consistent with ARDS. The primary outcome, a composite end point of days alive and free of mechanical ventilation at day 28 (ventilator-free days), was analyzed using a modified intention-to-treat analysis consisting of all randomized patients who received at least 1 dose of the study drug (IFN-β-1a [n = 144] vs placebo [n = 152]). At baseline, the cohort was well-balanced across the 2 groups, with features typical of moderate to severe ARDS: pneumonia was the most common predisposing diagnosis (in approximately 67% overall) and use of ARDS-specific interventions such as prone position (in about 20%), neuromuscular blockade (in about 30%), and corticosteroids (in about 35%).

The primary outcome of ventilator-free days at day 28 did not differ between the 2 groups (median, 10.0 days [interquartile range, −1.0 to 20.0] with IFN-β-1a and 8.5 days [interquartile range, −0 to 20.0] with placebo; P = .82). The 28-day mortality was 26.4% in the IFN-β-1a group and 23.0% in the placebo group. Concomitant treatments for ARDS during the first 28 days, such as neuromuscular blockade, corticosteroids, prone position, and use of extracorporeal membrane oxygenation therapy, were comparable between the groups. Similarly, the days free of organ failure, days free of organ support, and days alive without critical care were comparable between the groups. IFN-β-1a appeared to have a biological effect, as a specific biomarker of IFN-β-1a activity, myxovirus resistance protein A, was significantly higher in the IFN-β-1a group over the first 2 weeks.

Several reasons may explain the failure to reject the null hypothesis that IFN-β-1a is similar to placebo in the INTEREST Trial (P = .82). First, the sample size calculations may have been overly optimistic, despite being informed by the previous nonrandomized dose-finding study.5 Sample size calculations were based on expected ventilator-free days in the IFN-β-1a group of 3 days or more than placebo, and a 15% absolute reduction in 28-day mortality in the IFN-β-1a group compared with the placebo group. No intervention tested to date in critical care has shown mortality effects of such magnitude. Furthermore, although mortality data are collected in early-phase trials, these trials are seldom powered for these outcomes and sampling errors may generate statistically significant differences. In addition, in their analysis of ventilator-free days, the investigators assigned an arbitrary score of −1 for death within 28 days, instead of the conventional score of 0.8 This approach of assigning a higher penalty for death than ventilation duration should have increased the chance of detecting treatment effect.

Second, the authors highlight a potential drug-drug interaction between IFN-β-1a and corticosteroids, identified in post hoc analysis, that may have contributed to the lack of benefit. In the IFN-β-1a group, 50% of patients (22/44) who received corticosteroids died compared with 28.3% of placebo patients (17/60) treated with corticosteroids, a risk difference of 21.7% (95% CI, 3.0%-40.3%; P = .04) favoring placebo. Among patients without concomitant corticosteroid treatment, 16% of patients (16/100) who received IFN-β-1a died compared with 19.6% of patients (18/92) who received placebo, a risk difference of −3.6% (95% CI, −14.4% to 7.3%; P = .32). However, undue focus on this one speculative mechanism that corticosteroid administration could have inhibited the IFN-β-1a–mediated upregulation of CD739 ignores the numerous potential immunological interactions between glucocorticoids and IFN-β-1a that may also have contributed. Infections, both bacterial and viral, induce type 1 IFN responses, which may or may not improve pathogen clearance and often suppress the adaptive immune response.10,11 In patients with ARDS, extremes of IFN-stimulated gene expression have been associated with poor clinical outcomes.12Similarly, the adenosine (purinergic) signaling in ARDS can have beneficial and harmful effects.13 One or more of these wide-ranging immunologic effects could counteract the putative beneficial effects of IFN-β-1a on the endothelium. Collectively, these results imply that IFN-β-1a may negatively interact with corticosteroids, has limited clinical effect in the absence of steroids, and that IFN-β-1a may not have any benefit in ARDS. If further trials were to be considered, notwithstanding these overall INTEREST Study results, at a minimum, corticosteroids should be accounted for in the exclusion criteria and in adverse events monitoring.

Clinical trials of novel pharmacologic agents for ARDS have been consistently negative, despite decades of progress toward understanding ARDS biology in experimental models and observational human studies.1,4 So, the IFN-β-1a trial results are not unique. The current model for ARDS drug trials is typically focused around modifying one or more pathways related to alveolar epithelial injury, pulmonary endothelial injury, or the systemic immune response. Although endothelial injury may be central to the pathogenesis of ARDS, the dominant causal pathway to outcome may not be endothelial barrier function in all patients with ARDS.14 Similar considerations apply for epithelial injury. Although the dominant mechanism for death in a fraction of patients with ARDS might be the lung injury, the dominant mechanisms in the causal pathway to death could be the nonpulmonary systemic insults in others. Therefore, ARDS should be considered as a multipathway illness, with the ARDS to outcome relationship a multicausality relationship and every causal mechanism representing the joint action of a multitude of component causes.15 The pathways that determine outcome may differ between patients. This heterogeneity suggests the importance of highlighting 2 essential concepts in early-phase trials, prognostic and predictive enrichment and surrogate outcomes, both of which need additional development in ARDS.

The Berlin ARDS criteria provide prognostic enrichment14 (enriching for outcome risk), which was applied in this trial as in other prominent recent trials in ARDS. Focusing on predictive enrichment14 (enriching for treatment response based on a biological or physiological mechanism) may hold the key to success in future drug trials in ARDS and in other heterogeneous conditions in critical care. When new drugs are being developed for ARDS, the drug target should be in the causal pathway to a clinical outcome such as death. Subsequent phase 2 trials should focus on identifying the likelihood of a positive phase 3 trial, alongside key clinical factors, biological markers, or both to enable subsequent predictive and prognostic enrichment. When discussing enrichment approaches, examples to highlight are successful oncology trials, in which the goal is to eradicate neoplastic cells, often with either targetable cell surface markers or intracellular mechanisms. In contrast, in critically ill patients, drugs with multiple mechanisms of action are systemically administered, with the overall goal of achieving homeostasis within complex systems.

However, the ability to predict treatment responses in the complex systems scenario is limited. The outcomes of interest in early-phase trials are often surrogate outcomes (such as biomarkers, physiologic changes, imaging data). The ideal surrogate outcome would represent the single causal pathway of ARDS to patient-centered outcomes (such as mortality), the effect of treatment on the surrogate outcome mirrors the likely effect of treatment on patient-centered outcome, and the entire effect of the treatment on the patient-centered outcomes is mediated through its effect on the surrogate outcome. ARDS research currently lacks such surrogate outcomes, as causal pathways are extrapolated between nonideal surrogate outcomes and patient-centered outcomes in the context of incomplete knowledge. In the future, early-phase trials of novel pharmacotherapies in ARDS should develop treatment-specific surrogate outcomes, include enrichment strategies, and test patient-centered outcomes as primary outcomes to potentially break this pattern of relentlessly “negative” clinical trials.

In conclusion, Ranieri and colleagues6 report a well-conducted study of a novel immunological intervention, IFN-β-1a, for an important clinical problem, ie, ARDS. The authors found that IFN-β-1a administered for 6 days, compared with placebo, resulted in no significant difference in a composite score that included death and ventilator-free days with the current sample size and trial design. Until further prospective evaluations are performed, IFN-β-1a remains an experimental intervention and should not be included in routine clinical management of patients with ARDS.

给我留言

您必须 [ 登录 ] 才能发表留言!

×
腾讯微博