The prognostic value for survival of frailty and patient-reported outcomes in relapsed/refractory multiple myeloma: Results from a prospective multicenter observational study

Abstract

Abstract Background Clinical decisions for patients with relapsed/refractory multiple myeloma (RRMM) are challenging and assessment of frailty is critical to inform treatment strategies. The International Myeloma Working Group Frailty Index (IMWG-FI) is widely used in patients with newly diagnosed multiple myeloma (NDMM), given its ability to identify patients with distinct survival outcomes. However, very limited evidence exists on its prognostic value in the setting of RRMM and on the potential role of patient-reported outcomes (PROs) in the definition of frailty this setting. Objective The primary objective of this analysis was to investigate the prognostic value for overall survival (OS) of the IMWG-FI in patients with RRMM. A secondary objective was to examine whether PROs data could provide prognostic information for OS beyond the IMWG-FI and other key sociodemographic and clinical factors. Methods Adult patients with RRMM were enrolled in an international prospective cohort observational study by the GIMEMA. Eligibility criteria also included the availability of all the individual components of the geriatric assessment needed to calculate the IMWG-FI and a PRO assessment including the EORTC QLQ-C30 and the QLQ-MY20. OS was defined as the time from study entry to death from any cause. The prognostic value of the IMWG-FI was evaluated using Kaplan-Meier estimates and log-rank tests to compare OS across IMWG-FI groups. Additionally, we investigated the prognostic value of EORTC QLQ-C30 and QLQ-MY20 scales using univariate and multivariate Cox regression models. The initial univariate model included the following key variables: sex, time since initial MM diagnosis, living arrangements, education level, number of previous therapy lines, transplantation, best response to previous therapies, and IMWG frailty group. Likelihood ratio tests and C-indexes were used to evaluate potential improvement in prognostic information when adding PROs to the IMWG-FI. A bootstrap resampling procedure (1000 iterations) was applied to evaluate prognostic importance of each variable through inclusion frequencies across bootstrap generated samples. Results Between November 2017 and September 2022, 511 patients were enrolled from 31 centers. Patients had a median age at study entry of 69.8 years, and 287 (56%) were male. Median time since initial diagnosis of MM was 5 years (IQR 3- 7), 428 (84%) patients had received ≥2 lines of therapy before study entry, and 201 (39%) had at least one comorbidity. According to the IMWG-FI, 270 (52.8%) patients were classified as fit, 116 (22.7%) as intermediate, and 125 (24.5%) as frail. The 3 original IMWG-FI groups had distinct OS (p=0.001). However, similar OS were observed for the fit and the intermediate group, leading to the development of a two-tier classification by grouping fit/intermediate vs. frail. Indeed, well distinct median OS were observed for these two newly derived groups, being 38 and 33 months for fit/intermediate and frail patients (p<0.001), respectively. OS probabilities at 1, 2, and 3 years were clearly differentiated between the two IMWG-FI groups. For example, the two-year OS probabilities were 75.2%. and 62.1%, for fit/intermediate and frail patients, respectively. Prognostic analysis of PROs data indicated that physical functioning (PF) had the highest inclusion frequency (90%) across the 1000 bootstrap generated samples, thereby pointing to the prognostic importance of this variable. Multivariate analysis adjusting for key potential confounders, including IMWG-FI, showed that PF remained independently associated with OS (HR 0.933, 95% CI 0.875–0.995; p=0.034), which translates into a 7% increase in the hazard of death for every 10-point decrease (worsening) in the EORTC QLQ-C30 PF scale. Given these results, we investigated whether adding PF to a survival model based only on IMWG-FI (fit/intermediate vs. frail) could enhance its predictive accuracy. This resulted in a statistically significant increase in the likelihood ratio test (p=0.006) with the addition of PF, along with an improvement in the C-index (from 0.56 to 0.61). Conclusions In the setting of RRMM, the IMWG-FI is able to well discriminate two patients' groups with distinct OS. Our findings also indicate that patient-reported PF provides prognostic information for OS beyond the IMWG-FI, thereby laying the groundwork for the future development of a simplified patient-centric frailty index which would include PROs.