Frailty Subgroup Analysis of Isatuximab with Pomalidomide and Dexamethasone in a UK‐wide Real‐world Cohort of Relapsed Myeloma Patients

Multiple myeloma (MM) is primarily a disease of the elderly with 43% of the UK's newly diagnosed (NDMM) patients aged 75 years and over.1 The highest incidence rates in the UK in both males and females occur in those aged 85–89 years.1 A structured frailty assessment is needed in elderly myeloma patients in order to individualise treatment decisions.2 In the relapsed/refractory (RRMM) setting, frailty-based approach to therapy is equally required to devise patient-centred treatment plans, which can reduce the risk of under-treating fit patients and the risk of over-treating frail patients. The International Myeloma Working Group (IMWG) frailty score and the revised Myeloma Co-morbidity Index (R-MCI), both demonstrated their prognostic value in myeloma patients.3-5 The UK Myeloma Research Alliance (MRA) published a validated frailty-based outcome prediction model named Myeloma Risk Profile (MRP), which was predictive of overall survival (OS) in transplant-ineligible NDMM patients.6 More recently, a simplified frailty score which utilises three components (age, Charlson Co-morbidity Index-CCI, and ECOG performance status-PS) and categorises patients as frail or non-frail, was predictive of outcomes in elderly NDMM patients treated as part of the FIRST trial.7 External validation of this simplified frailty score in a population-based cohort demonstrated that it was also prognostic for OS in real-world transplant-ineligible NDMM patients.8 Novel triplet therapy isatuximab with pomalidomide and dexamethasone (IsaPomDex) has become an important treatment option in the UK for myeloma patients at 3rd relapse, based on efficacy data demonstrated in ICARIA-MM trial.9 Real-world outcomes of IsaPomDex in a UK-wide study showed that efficacy outcomes are comparable to trial data.10 ICARIA-MM trial outcomes by frailty employed the simplified frailty score and demonstrated that overall response rate (ORR) was statistically higher with IsaPomDex compared to PomDex, regardless of frailty status: frail (52.1% vs. 34.2%, p = 0.0476), non-frail (66.3% vs. 35.7%, p < 0.0001), but the difference in the ORR rate between frail and non-frail patients receiving IsaPomDex was not statistically significant (p = 0.3971). Median PFS was higher with IsaPomDex compared to PomDex regardless of frailty status, but the difference was only statistically significant for the non-frail subgroup comparison: frail (9 vs. 4.5 months, p = 0.4928), non-frail patients (12.7 vs. 7.4 months, p = 0.0004).11 However, there are no real-world data to describe IsaPomDex outcomes by frailty in RRMM patients, and no efficacy and toxicity data analysis to describe the utility of the simplified frailty score in RRMM in routine care. We conducted a UK-wide retrospective real-world study of RRMM patients treated with IsaPomDex. Herein, we report patient characteristics, and clinical outcomes according to frailty categories calculated using the simplified frailty score. Frailty scoring utilised 3 components: age (≤75 years: 0, 76–80 years: 1, >80 years: 2), CCI score (≤1:0, >1: 1) and ECOG PS (0: 0, 1:1, ≥2:2). The sum of all three components would define patients as frail (if ≥2) or non-frail (if <2).7, 11 To evaluate the prognostic value of the simplified frailty score in RRMM in this cohort, outcomes by frailty included ORR, progression-free survival (PFS), duration of response (DOR), OS and adverse events (AEs). The method of this study is fully described in File S1. From 107 patients, the total cohort evaluable for frailty comprised 106 patients. Median follow-up (FU) in the total cohort (n = 106) was 12.1 months (IQR 10.1–18.6 months). Period of FU did not differ between the two frailty groups (log-rank test p = 0.7966, inverse Kaplan–Meier method). According to frailty scoring, 72 patients (67.9%) were frail, whilst 34 patients (32.1%) were non-frail. Baseline patient, disease and treatment characteristics of the different frailty subgroups are fully presented in Table 1. Patients in the frail subgroup had a higher median age (frail: 71 vs. non-frail: 57.5 years), a higher proportion of patients with PS score of ≥1 (frail: 95.9% vs. non-frail: 29.4%), a higher co-morbidity burden expressed as a higher median CCI score (frail: 4 vs. non-frail: 1), and a more prevalent renal presentation (e-GFR < 60 ml/min) (frail: 51.4% vs. non-frail: 26.5%). Median number of prior therapies was identical between subgroups (3 vs. 3). Patients in the frail subgroup received a numerically lower median number of IsaPomDex cycles within the period of follow, but without a statistical significance (frail: 6 vs. non-frail: 9.5, p = 0.1664). The rate of pomalidomide dose reductions was numerically higher in the frail subgroup (51.4% vs. 35.3%, p = 0.121). Treatment status as well as reasons for discontinuation in the total cohort and frailty subgroups, are presented in Table 2A. There was no statistical difference in overall discontinuation rates according to frailty (frail: 51.4% vs. non-frail: 50%, p = 0.894). Discontinuation due to treatment toxicity was very low in the total cohort (1.9%) and across frailty subgroups. Discontinuation due to progressive disease was similar between frailty subgroups (frail: 29.2% vs. non-frail:32.4%). Discontinuation rate due to death was numerically higher in the frail subgroup (frail:16.7% vs. non-frail:11.8%, p = 0.234). ORR and response categories in the total cohort and in subgroups by frailty, are presented in Table 2B. ORR was 66% in the total cohort. ORR was comparable across frailty subgroups (frail: 65.3% vs. non-frail: 67.6%, p = 0.779). Median PFS in the total cohort was 10.9 months (95% CI: 3.5–17.6 months). Median PFS was numerically higher in the non-frail group but without statistical difference (frail: 10.1 vs. non-frail: 13.7 months, log-rank p = 0.5259), Figure S1. In 70 patients who achieved an objective response (≥PR) from the total cohort of 106 patients (of whom 40 remain in ongoing remission, 3 died before a relapse event and 27 experienced a relapse), median DOR was 10.1 months (95% CI 7.7–NA). There was no statistical difference in median DOR by frailty (frail: 10.1 vs. non-frail: 10.2 months, Grey's p = 0.685), Figure S1. Median OS in the total cohort was 18.6 months (IQR 7–22.7 months, 95% CI: 14.4-NE). Median OS was numerically lower in the frail group but without statistical difference (frail: 15 months, 95% CI 12-NE vs. non-frail: not reached, 95% CI 16.7-NE, log-rank p = 0.3571), Figure S1. This OS analysis needs to be interpreted with caution due to insufficient length of follow up for OS. Data on AEs was evaluable in all 106 patients and is presented here after a median number of cycles (IQR) of 4 (2–8), and a median follow up (IQR) of 3.7 months (0.5–12.4). The most common any grade AEs (experienced by ≥10% of patients) were neutropenia (65%), thrombocytopenia (23.6%), infections (23.6%), anaemia (16%), and fatigue (10.4%). In 87.7% of patients who experienced ≥1 any grade AEs, the median (range) of any grade AEs per patient was 2 (1–9). Incidence rate of any grade AEs by frailty was (frail: 87.5% vs. non-frail 88.4%, p = 0.763). Incidence rate of any grade haematological AE by frailty was (frail: 76.4% vs. non-frail: 67.6%, p = 0.341). Incidence rate of any grade non-haematological AEs by frailty was (frail: 58.3% vs. non-frail 52.9%, p = 0.601). With 62.3% of patients experiencing ≥1 AEs of ≥G3, the most common ≥G3 AEs (experienced by ≥10% of patients) were neutropenia (45.3%), infections (18.9%) and thrombocytopenia (14.1%). Incidence rate of ≥G3 infections by frailty was (frail: 22.2% vs. non-frail 11.8%, p = 0.199). The nature and number of ≥G3 infections experienced in this cohort (total of 22 episodes) were: COVID-19 pneumonia (6), neutropenic sepsis (5), Escherichia coli infection (2), urinary tract infection (3), lung infection (2), Serratia liquifaciens infection (1), pseudomonas sepsis (1), bacteraemia (1), and skin infection (1). In 51.9% of patients who experienced ≥1 haematological AEs of ≥G3, the median (range) number of episodes per patient was 1 (1-4). Incidence rate of ≥G3 haematological AEs by frailty was (frail: 58.3% vs. non-frail 38.2%, p = 0.053). The total number of all ≥G3 haematological AEs in the total cohort was 81. Of those 81 episodes and taking into account that some patients experienced the same toxicity on more than one occasion, the number of episodes per toxicity in the total cohort and in frailty subgroups were: neutropenia 49 (frail: 36 vs. non-frail: 13), thrombocytopenia 15 (frail: 13 vs. non-frail: 2), anaemia 16 (frail: 16 vs. non-frail: 0), and lymphopenia 1 (frail: 1 vs. non-frail: 0). It is important to bear in mind that comparing the number of episodes between frailty subgroups is limited by the significant difference in cohort size between those subgroups (frail: 72 patients vs. non-frail: 34 patients). Incidence rates of ≥G3 haematological toxicities between frailty subgroups were: neutropenia (frail: 48.6% vs. non-frail: 38.2%, p = 0.316), thrombocytopenia (frail: 18% vs. non-frail: 5.9%, p = 0.1360), and anaemia (frail: 12.5% vs. non-frail: 0%, p = 0.0549).Twenty-four patients (22.6%) in the total cohort experienced one or more AEs leading to ≥1 inpatient admissions. The median (range) duration of hospitalisation per patient was 8 (1–21) days. The cumulative number of inpatient hospitalisation days related to AEs in the total cohort was 207 days, of which 159 days (76.8%) were infection-related hospitalisations. Of the 207 hospitalisation days, the number of days in the different frailty subgroups were (frail: 167 vs. non-frail: 40), but this comparison is limited by the difference in cohort sizes of subgroups and can be more meaningfully described using the incidence rate. Incidence rates of AE-related inpatient hospitalisations were (frail: 26.4% vs. non-frail: 14.7%, p = 0.18). Our study is the first to describe efficacy and toxicity outcomes stratified by the simplified frailty score,7 in RRMM patients in the real-world. However, our work is limited by its retrospective, non-randomised nature with the inherent possibility of unmeasured confounding factors, patient selection bias, the potential for medical chart misinterpretation and under-reporting of toxicities, in addition to lack of sufficient length of FU which limits the meaningfulness of our OS analysis. Despite these limitations, we demonstrated the high prevalence of frailty in relapsed myeloma patients treated with IsaPomDex in the real-world (67.9%), which is significantly higher than 31.2% in the IsaPomDex arm of ICARIA-MM trial.11 We showed comparable real-world efficacy outcomes (ORR/PFS/DOR) between frailty subgroups. Our ORR is comparable to ICARIA-MM frailty subgroup analysis for non-frail patients (67.6% vs. 66.3%) but less comparable for frail patients (65.3% vs. 52.1%).11 Our median PFS data (frail: 10.1 vs. non-frail: 13.7 months) is consistent with ICARIA-MM (frail: 9 vs. non-frail: 12.7 months).11 ORR and PFS comparisons between our real-world data and ICARIA-MM frailty subgroup analysis are limited by the previously mentioned difference in cohort frailty composition (67.9% vs. 31.2%). We demonstrated in our results a trend for worse toxicity outcomes (≥G3 haematological toxicities, ≥G3 infections, and AE-related inpatient hospitalisations) in frail patients compared to non-frail patients, but treatment discontinuation rate due to toxicity remained very low in the total cohort (1.9%) and across frailty subgroups. Neutropenia and thrombocytopenia can be attributed to pomalidomide. It is important to add that a recently published phase 2 trial of 109 RRMM patients demonstrated that isatuximab monotherapy led to ≥G3 neutropenia and ≥ G3 thrombocytopenia in 18.3% of patients (equal incidence rate of both toxicities); however isatuximab dose used in this phase 2 trial was 20 mg/kg which is higher than 10 mg/kg used in the IsaPomDex triplet combination.12 Pomalidomide dosing should be adjusted for neutropenia and thrombocytopenia according to manufacturer's recommendation. There is no data to suggest dose-adjustment of isatuximab during IsaPomDex therapy but it is recommended to delay isatuximab administration in the event of G4 neutropenia until neutrophil count improves to at least 1.0 × 109/L. It is reasonable to consider the use of granulocyte colony-stimulating factors (G-CSF) to mitigate the risk of profound or recurrent neutropenia. Infections observed in this IsaPomDex cohort can be driven by all three drugs in the triplet combination; isatuximab (being an anti-CD38 monoclonal antibody) adds to infection risk. We demonstrated that infections contributed to 76.8% of the cumulative inpatient hospitalisation days in the total cohort. Therefore, optimal infection prevention and management strategies can help reduce infection-related hospitalisations whilst on this therapy. We recently published infection morbidity analysis from this IsaPomDex dataset. We found an independent association between a high co-morbidity burden (CCI ≥4) and the incidence of ≥G3 infections, and a protective effect of an objective myeloma response (≥PR) from ≥G3 infections.13 In order to reduce the incidence of infections, particularly in frail patients and those with co-morbidities, we recommend close monitoring, and the optimal use of anti-infective prophylaxis strategies. Anti-infective prophylaxis should include the preventative use of antivirals in addition to antifungals and pneumocystis pneumonia (PCP) prophylaxis. Optimal vaccination strategies are also recommended to limit infections including the use of COVID-19 boosters. Patients should be advised to seek urgent medical advice if they test positive for SARS-CoV-2, in order to receive early treatment such as with antivirals or monoclonal antibody therapies for COVID-19. Intravenous immunoglobulin should be considered on a case-by-case basis, in consultation with immunology, for myeloma patients presenting with drug-associated hypogammaglobinaemia and recurrent infections. FD designed and led the study, contributed to the setup of the OpenClinica online data collection platform, co-ordinated the study across the UK, collected, analysed and curated data. SB and KR are senior authors. GV executed the set up the OpenClinica platform. FP conducted statistical analysis. AR collected and analysed data. IT collected and analysed data. All other authors collected data for each of their respective hospital sites. FD wrote the manuscript, which all authors critically reviewed and approved. FD (Sanofi: honoraria for education evening for haematologists). SB (Sanofi: honoraria and advisory board). KR (Sanofi: honoraria and advisory board; BMS: research support, honoraria, advisory board, travel support). All other authors have no conflicts of interest to declare. Figure S1. Appendix S1. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.