Exploring Standard Endpoints for Clinical Trials of Pneumonia Therapy

The Journal of Infectious Diseases E D I T O R I A L C O M M E N TA R Y Exploring Standard Endpoints for Clinical Trials of Pneumonia Therapy Joshua C. Eby Division of Infectious Diseases and International Health, University of Virginia, Charlottesville (See the Major Article by Talbot et al on pages 1536–44.) Bringing new antimicrobials through regulatory approval and into clinical use is an expensive and time-consuming process. Nonetheless, in an era of increasing antimicrobial resistance and continued antimicrobial adverse effects, the need for new antibiotics has increased. While regulatory guidance for trial design must address study feasibility, any changes must be balanced by the need for scientific validity. This balance must be achieved with the ultimate goal in mind of benefiting patient care in a meaningful way. The Foundation for the National Institutes of Health Biomarkers Consortium Project Team, at the request of the US Food and Drug Administration (FDA), has been working to find this delicate balance by addressing the need for standard endpoints for noninferiority (NI) clinical trials. The group has used historical placebo-controlled trial data to identify early endpoints upon which to base NI trial design for acute bacterial skin and skin structure infection (ABSSSI) and community-acquired bacterial pneumonia (CABP) [1]. Their Received 5 December 2018; editorial decision 5 December 2018; accepted 5 December 2018; published online December 10, 2018. Correspondence: J. C. Eby, MD, University of Virginia, PO Box 800419, Charlottesville, VA 22901 (). The Journal of Infectious Diseases®2019;219:1515–7 © The Author(s) 2018. Published by Oxford University Press for the Infectious Diseases Society of America. All rights reserved. For permissions, e-mail: . DOI: 10.1093/infdis/jiy708 resulting recommendations were incorporated into FDA guidance documents and used in NI clinical trials that resulted in approval of new antibiotics [2–4]. This standardization has been a great step forward for regulatory NI trial design. Building upon their progress with ABSSSI and CABP, in this issue of The Journal of Infectious Diseases, Talbot et al present a summary of their most recent work examining endpoints for use in NI trials of antibiotics for treating hospital-acquired bacterial pneumonia (HABP) and ventilator-associated bacterial pneumonia (VABP) [5]. The project team has incorporated the input of a range of stakeholders into their recommendations, and an important part of this work is publishing the analysis for evaluation by an audience who might otherwise not see it. The authors reanalyze data from modern trials of antibiotic therapy of HABP and VABP to characterize all-cause mortality (ACM), an accepted endpoint for HABP and VABP used in these trials [6]. In the reanalyzed trials, the 28-day ACM is low in some groups, with a rate of 9.8% among patients with nonventilated HABP (nv-HABP) in one study. The ACM rates for ventilated HABP (v-HABP) and VABP are higher than for nv-HABP within each study. The reasons for the differences in mortality among groups are not characterized in the present study, though others have characterized HABP populations with variable findings [7–9]. If the ACM event rate is <15% in the active control group in an NI study, a fixed NI margin of 10%–15% would not be appropriate, and a study would need to be designed with a smaller fixed NI margin. Alternatively, an odds ratio analysis approach could be used. With either adjustment in design, the sample size would have to be increased to confidently draw a conclusion of NI. The authors recommend, appropriately, to consider nv-HABP separately from v-HABP or VABP when planning enrollment for clinical trials. Because trials are already difficult to conduct, a larger sample size required in nv-HABP populations would further decrease trial feasibility, and the authors explore trial endpoints in this context. The authors suggest that endpoint event frequency could be increased by making a composite endpoint, which they term “ACM plus” (ACM+), which combines ACM with nonfatal adverse events. Based on adverse events identified using MedDRA Toxic/Septic Shock Standardized MedDRA Queries, the 28-day ACM+ rate was at least modestly greater in magnitude relative to ACM within most patient groups in the trials analyzed. In general, a composite endpoint can increase the event rate in comparison to the individual endpoint components, but will also add more complexity to the interpretation of results. The events added EDITORIAL COMMENTARY • jid 2019:219 (15 May) • 1515 Keywords. hospital-acquired pneumonia; ventilator-associated pneumonia; clinical outcome; patient-reported outcome; antimicrobial drug development. Information conveyed by ACM frequency is limited, excluding potentially important information about outcomes of surviving patients; ACM+ captures some of this missed information, and the authors examine the potential for using another measure, symptom response. The same studies used to assess ACM were also reanalyzed for patient symptoms during the study. Assessment of v-HABP and VABP patient symptoms was understandably limited by the low rate of recorded symptoms; however, 2 symptoms were reported at the time of enrollment for 71%–96% of patients with nv-HABP. The authors show evidence for an early improvement in symptoms for nv-HABP patients. Rather than pursue a symptom-based endpoint, the authors use these data as support for developing a patient-reported outcome (PRO) instrument for nv-HABP. The group has already developed PRO instruments for CABP and ABSSSI. In fact, the authors have also completed a draft PRO instrument for nv-HABP. PRO instruments are tools for measuring patient symptoms, function, and quality of life without interpretation by another entity. Despite the seeming subjectivity of symptoms, PROs are designed carefully to provide objective, standardized, low-variability measurements. It is not clear yet whether a PRO could be used as a primary endpoint for a drug trial for treatment of pneumonia because noninferiority studies require a base in historical data, and the performance of the drafted PRO has not yet been adequately determined. If not used as an efficacy endpoint, it nonetheless remains an important tool for assessing patient status. Use of PRO instruments to measure treatment response in clinical trials is supported by the FDA. PRO instruments have been used in trials of antimicrobial therapy for infections, including hepatitis C, pseudomonal infection in cystic fibrosis, and exacerbations of chronic bronchitis [10–12]. There is increasing evidence in other specialties, such as oncology and rheumatology, that PROs 1516 • jid 2019:219 (15 May) • EDITORIAL COMMENTARY benefit patient care when incorporated into clinical assessment. Evidence of the utility of PRO instruments in regulatory drug trials or other clinical studies should be a strong motivator f (...truncated)