We conducted a real-world study with 389 patients with NDMM prospectively treated with VRd or KRd from a single institution during a 5-year period. Our primary aims were to investigate measures of efficacy, including depth of response, PFS, EFS, and OS, and characterize the safety profile of these two regimens in standard and high-risk patients. Overall, we found better outcomes associated with KRd compared to VRd, including depth of response with patients achieving ≥ CR rate (25% vs. 41%, P < 0.01), 5-year PFS rates (56% vs. 67%, P = 0.043), and 5-year EFS rates (34% vs. 52%, P < 0.001) with the VRd and KRd groups, respectively. There was a trend toward improved 5-year OS rates associated with KRd (80% vs. 90%, P = 0.053). Of note, our KRd cohort was enriched for high-risk MM patients, but propensity weight scoring analysis still confirmed KRd benefit in the PFS evaluation. Additionally, KRd demonstrated PFS improvement over VRd, regardless of early ASCT status. In the ENDURANCE trial, the median PFS was 34.4 and 34.6 months for VRd and KRd-treated standard-risk patients, respectively (17). Other studies have reported findings suggesting that VRd and KRd induction can achieve better results than findings from the ENDURANCE trial. Our results, demonstrating median PFS for VRd and KRd were not reached after a median follow-up of 58.8 months, differ from ENDURANCE, but are consistent with other VRd (DETERMINATION, RVD1000) and KRd (MMRC, FORTE) datasets (12, 13, 25, 26).
In this real-world study, we also found a high proportion of patients changing therapy in the absence of progressive disease per IMWG criteria. To overcome issues of statistical bias, we performed EFS sensitivity analysis and found that KRd was associated with significantly improved EFS compared to VRd after adjusting for age, cytogenetic risk, R-ISS stage, and early ASCT. Importantly, clinical trials allowing patients to change therapy without meeting IMWG criteria for progression should report EFS and time to treatment failure in the main analysis in order to properly overcome the influence of censoring bias, which inherently will occur in PFS analysis (27). It seems reasonable to conjecture that increased access to more sensitive testing (for example, blood-based MRD testing) will increase the proportion of patients changing therapy in the absence of progressive disease by IMWG criteria in the future.
On subgroup analysis, we did not detect a significant improvement in PFS associated with KRd induction compared to VRd in patients with standard-risk cytogenetics. Median PFS was not reached in both VRd and KRd groups (P = 0.20). The ENDURANCE trial, which was conducted only in standard-risk patients also did not detect PFS differences between VRd and KRd regimens. Conversely, our real-world study shows a clear PFS benefit associated with KRd compared to VRd in patients with HRCA demonstrating median PFS of 41 months vs. 70.9 months in VRd and KRd groups, respectively (P = 0.016). Compared to other studies, our study is consistent with PFS rates reported in high-risk multiple myeloma patients receiving either VRd or KRd, including RVD1000, SWOGS1211, and FORTE (12, 18, 26). Importantly, these trends for KRd benefit over VRd were seen in both PFS and OS (truncated at 5.5 years of follow-up) multivariate analysis for high-risk and overall group but not seen in standard-risk multivariate analysis.
In addition, there was better tolerability with KRd compared to VRd, reflected in an absence of severe neuropathy and substantially lower rates of cardiovascular and thromboembolic events compared to select prior studies (17, 28, 29). In our study, 16% of the patients treated with VRd developed grade ≥ 2 peripheral neuropathy with 14% having persistent symptoms more than 6 months after completion of induction. Our findings are an important reflection of clinically impactful bortezomib-induced peripheral neuropathy since the most common reason for treatment-discontinuation due to AEs in the VRd group was peripheral neuropathy. In contrast, we did not find any grade ≥ 2 peripheral neuropathy with KRd, and none of the patients required neuropathy-specific interventions. Carfilzomib is known to have a cardiovascular signal. In our study, we captured patients who experienced grade ≥ 2 cardiovascular and pulmonary AE (VRd 5% vs KRd 8%). These events were reversible in the majority of patients (VRd 66% vs KRd 87%). The lower rates of cardiopulmonary AEs associated with KRd in our study is likely driven by optimized intravenous fluid management and modern anticoagulation therapy given at our institution (30). The ENDURANCE trial reported a composite of treatment-related grade ≥ 3 cardiac and pulmonary disorders occurring in 16% of KRd-treated patients (17, 30), while several other clinical trials have not found significantly elevated rates of grade 3 or higher pulmonary and cardiovascular AEs associated with carfilzomib, ranging from 2%-5% in the FORTE and MMRC trials (13, 26, 31, 32, 33).
Strengths of the study include large sample size and uniform treatment administration and supportive therapies at a high-volume myeloma program in the United States during a 5-year period. Our real-world study design allowed for inclusion of patients with various comorbidities reflective of the general population, who are frailer than those treated on clinical trials. We acknowledge that because of these factors, there are varying treatment schedules and doses given in the real-world setting to mitigate side effects and personalize treatment among the included patients. Weaknesses include retrospective study design without randomization leading to inherent selection bias. To overcome inherent biases, we adjusted and stratified for confounders. However, while a Cox-regression analysis was used to adjust for a heterogenous patient population and propensity score weighting was performed, it is likely that patient heterogeneity and treating physician bias had some uncompensated impact on the analysis. When the proportional-hazards assumption in the Cox regression model was not met, truncated analysis at change point of 5.5 years of follow-up was performed. There were only a small group of patients (61 VRd and 54 KRd) with longer than 5.5 years of follow-up in both groups, making it difficult to draw any conclusions. We failed to see a difference in overall survival between VRd and KRd in the landmark analysis at 5.5 years after adjusting for confounding variables, and longer follow-up is needed. Moreover, given the retrospective nature of the study, AEs were not prospectively collected with same rigor as in clinical trials and only AEs documented in the EMR were captured for analysis.
In summary, this real-world data analysis involving almost 400 patients with NDMM – including both standard-risk and high-risk patients – from a high-volume treatment center treated with VRd or KRd combination therapy provide clinically important information for treating physicians and patients with newly diagnosed multiple myeloma. Future studies are needed to investigate the role of added monoclonal antibodies to these combination therapies and to investigate the role of MRD testing for clinical decision making.