Characterization of Patient-Reported Outcomes in Multiple Myeloma Registrational Trials Submitted to the U.S. FDA

Introduction

Multiple myeloma (MM) is a hematologic malignancy characterized by the clonal proliferation of plasma cells. The US Food and Drug Administration (FDA) has approved numerous therapies for the treatment of MM over the past 13 years. During the same time period, there have been advances in characterizing the patient experience in oncology clinical trials, primarily using patient-reported outcomes (PRO). Although there have been advances in MM and collection of patient experience in clinical trials, this review aims to provide details on the heterogeneity of PRO measures and data analyses used in MM registrational trials.

Methods

We reviewed FDA databases for NDA/BLA submissions supporting therapies to treat patients with MM between 2007 and 2020. Pivotal trial protocols, clinical study reports, and FDA clinical/ statistical reviews supporting each application were reviewed for descriptions of included PROs. We focused on PRO measures, collection methods, baseline definition, assessment frequency, compliance, definition of thresholds for clinically meaningfulness, and statistical methods for analysis of PRO.

Results

We examined the results from 25 clinical trials that led to 20 new indications. PRO data were included in 17 of the 25 trials (68%). Of the 17 trials with PRO data, three were single arm trials, and the remainder were randomized trials. All trials were open label except for one blinded randomized trial. Seven trials collected data electronically, five in paper format, and five trials did not specify the collection format. The majority of trials had at least two PRO measures (82%) with two trials (12%) utilizing four measures. Among trials included for analysis, nine unique PRO measures were used, most commonly the EORTC QLQ-30 (87%), EQ-5D (65%), both cancer agnostic and QLQ-MY20 (47%), specifically for MM patients.

We found differences in terms of definition of baseline. One trial defined measurements taken only at screening as the baseline assessment, while the remainder defined baseline as any measurement taken on or before day 1 cycle 1. Assessment frequency was highly variable (e.g. every cycle, every other cycle, every three months) and was largely dependent on cycle length (ranged from 21 to 42 days). We also noted that assessment may have been after a period of drug washout (e.g. the anti-MM regimen was administered on days 1-21 of a 28 day cycle; PRO assessment occurred on day 1 of the next cycle, thus following a 7-day washout). Definition of compliance of the PRO measures varied across the trials ranging from “completing all items” (6%), “completing 50% of the items” (12%), and “completing enough items to calculate the score in any domain” or some variant (82%). There was variability in the definition and justification of clinically meaningful thresholds across trials despite use of the same instrument and similar MM populations. Differences in statistical methods used for analyses of the PRO endpoints were also noted when analyzing data from the same measure across trials. For example, one of the two trials using the BPI-SF analyzed pain severity and intensity as composite endpoints (change from baseline analysis using a mixed model with treatment arm, time point, and baseline score as covariates). The other trial using the BPI-SF used pain response which was defined as a 30% reduction from baseline in worst pain score over the last 24 hours without an increase in analgesic use at 2 consecutive evaluations as an endpoint, and used the stratified Cochran–Mantel–Haenszel test to report treatment differences.

Conclusions

We encountered substantial heterogeneity in terms of PRO collection methods, measures, definitions and analyses in recent MM registrational trials. These differences may hinder the utility of this data for healthcare policy as well as clinician and patient decision making. FDA has proposed a core outcome set to be included in cancer clinical trials which includes assessment of patient-reported adverse events, disease symptoms, physical function, role function, and overall side effect bother. Use of this core outcome set in future MM registration trials is one step towards consistency in PRO collection and analysis. Further work needs to be done to determine the best approach for statistical and analytical methodologies.