Patient-reported frailty phenotype (PRFP) vs. International Myeloma Working Group frailty index (IMWG FI) proxy: A comparison between two approaches to measuring frailty

Introduction: Frailty assessments may help to identify patients at highest risk for treatment-related toxicity, early treatment discontinuation due to toxicity, and death in Multiple Myeloma. We aimed to compare the patientreported frailty phenotype (PRFP) and a modified version of the International Myeloma Working Group frailty index (IMWG FI) in terms of their strengths, limitations, and classification of frailty in a cohort of patients with relapsed/refractory multiple myeloma (RRMM).Materials and Methods: Data were pooled from six RRMM Phase 3 randomized clinical trials submitted to the Food and Drug Administration for regulatory review between 2010 and 2021. Patients were classified as fit, intermediate fit/pre-frail, or frail using both PRFP and the IMWG FI proxy. Agreement between the two approaches in classification of patient frailty was assessed using weighted Cohen's kappa. A contingency table and Venn diagram were generated to analyze overlap in categorization of patient frailty across the different severity groups. Descriptive statistics were used to summarize and compare the clinical and demographic characteristics of patients categorized as frail by PRFP vs. IMWG FI proxy.Results: Of the 2,750 patients included in this analysis, IMWG FI proxy classified 16.4% (452) patients as frail, 28.1% (772) as intermediate fit/pre-frail, and 55.5% (1,526) as fit. Meanwhile, PRFP classified 21.7% (597) of patients as frail, 24.5% (675) as intermediate fit/pre-frail, and 53.8% (1478) as fit. Fair agreement was observed between PRFP and IMWG FI proxy (weighted Cohen's Kappa = 0.34 [0.31-0.37]). On average, patients who were categorized as frail by IMWG FI proxy were older and had higher Charlson Comorbidity Index scores than patients classified as frail by PRFP. In contrast, patients who were classified as frail by PRFP had worse EORTC QLQ-C30 Physical Functioning subscale summary scores as compared to patients in the IMWG FI proxy frail group (median score of 40 vs. 47 out of 100). Discussion: Our analysis found fair concordance between IMWG FI proxy and PRFP. This demonstrates that while both frailty models measure the same underlying construct, the variables that constitute each approach may result in differing frailty categorizations for the same patient. Further prospective studies are needed to establish and compare the predictive and prognostic abilities of the different frailty indices in MM.