The impact of blinding on patient-reported outcomes (PROs) in randomized controlled trials of immune checkpoint inhibitors versus traditional chemotherapies

e23160 Background: Immune checkpoint inhibitors (ICIs) have been met with a wave of excitement due to their novel mechanism. We hypothesized that this may impact how patients (via PROs) report treatment tolerability in comparison to traditional therapies. We sought to examine if there was a notable difference in PROs in blinded vs unblinded trials of ICIs. Methods: We systematically searched PubMed, CINAHL, Embase, Web of Science, and Scopus in August 2018 for publications with quantitative PRO data comparing ICIs to cytotoxic chemotherapy. Case series, narrative reviews, and publications lacking original data were excluded. Eligible publications were reviewed to assess if patients were blinded to the agent received, and a comparison for common PRO metrics was performed. Results: Of the 1,450 unique references identified, eight met inclusion criteria: 1 double blinded placebo-controlled trial and 7 trials where patients were aware of the assigned arm. The blinded trial had quantitative PRO data in the form of the European Organisation for Research and Treatment of Cancer (EORTC) global health status (GHS) score and patient reported symptom burden at week 12. Most (6 of 7; 86%) unblinded trials reported the GHS at either week 12 or 15, and patient symptom burden at these time points as well (5 of 7; 71%). For the EORTC GHS, the blinded trial showed no inter-arm difference at week 12. 4 of 6 (67%) open label trials noted statistically significant differences in GHS favoring the ICI arm. For symptom burden at week 12 or 15, there was no difference found in the blinded study. In unblinded trials, there were domains where patients receiving ICIs reported a statistically significant lower symptom burden than those receiving chemotherapy: fatigue (4 of 5 trials favoring ICIs; 80%), dyspnea (2 of 5; 40%), insomnia (1 of 4; 25%), appetite loss (1 of 4; 25%), and diarrhea (1 of 5; 20%). There were no differences in pain (n = 5), nausea/vomit (n = 5), and constipation (n = 5). Conclusions: We found a trend towards more favorable reporting on common symptoms in unblinded studies of patients receiving ICIs. Our analysis is limited by the lack of available comparisons in the published literature.