Bortezomib, Lenalidomide and Dexamethasone (VRd) vs Carfilzomib, Lenalidomide and Dexamethasone (KRd) as Induction Therapy in Newly Diagnosed Multiple Myeloma

Lenalidomide and dexamethasone with bortezomib (VRd) or carfilzomib (KRd) are commonly used induction regimens in the U.S. This single-center, retrospective study evaluated outcomes and safety of VRd and KRd. Primary endpoint was progression-free survival (PFS). Of 389 patients with newly diagnosed multiple myeloma, 198 received VRd and 191 received KRd. Median PFS was not reached (NR) in both groups; 5-year PFS was 56% (95%CI, 48%−64%) for VRd and 67% (60%−75%) for KRd (P = 0.027). Estimated 5-year EFS was 34% (95%CI, 27%−42%) for VRd and 52% (45%−60%) for KRd (P < 0.001) with corresponding 5-year OS of 80% (95%CI, 75%−87%) and 90% (85%−95%), respectively (P = 0.053). For standard-risk patients, 5-year PFS was 68% (95%CI, 60%−78%) for VRd and 75% (65%−85%) for KRd (P = 0.20) with 5-year OS of 87% (95%CI, 81%−94%) and 93% (87%−99%), respectively (P = 0.13). For high-risk patients, median PFS was 41 months (95%CI, 32.8–61.1) for VRd and 70.9 months (58.2-NR) for KRd (P = 0.016). Respective 5-year PFS and OS were 35% (95%CI, 24%−51%) and 69% (58%−82%) for VRd and 58% (47%−71%) and 88% (80%−97%, P = 0.044) for KRd. Overall, KRd resulted in improved PFS and EFS with a trend toward improved OS compared to VRd with associations primarily driven by improvements in outcome for high-risk patients.


Introduction
Multiple myeloma (MM) is the second most common hematologic malignancy with over 35,000 new individuals being affected annually in the United States with global incidence increasing sharply in recent decades (1,2). Over the past two decades, the therapeutic landscape and management of MM has evolved substantially, re ected in median overall survival (OS) of more than 10 years for newly diagnosed patients in select subgroups. Currently, there is no established curative treatment for multiple myeloma.
The use of proteasome inhibitors (PIs) and immunomodulatory drugs (IMiDs) in patients with newly diagnosed multiple myeloma (NDMM) has improved clinical outcomes, independent of transplant eligibility (3,4,5,6,7). Initially, bortezomib, lenalidomide, and dexamethasone (VRd) was developed as an induction regimen for patients with NDMM based on results from a phase 1/2 trial (8) and subsequently found to result in improved progression-free survival (PFS) and OS with an acceptable safety pro le (9,10,11,12). However, the high incidence of bortezomib-induced peripheral neuropathy, which is often irreversible, can result in signi cant morbidity and prevent its long-term use, despite use of the subcutaneous formulation or alteration in the dosing schedule. Car lzomib, a next-generation PI, in combination with lenalidomide and dexamethasone (KRd) has been demonstrated in multiple phase 2 studies in patients with NDMM to have high overall response rates (ORR) and deep, durable responses with tolerable side effects (13,14,15,16). Rare but serious side effects associated with car lzomib include cardiovascular events. Currently only one randomized study (ENDURANCE trial) has compared VRd and KRd in patients with NDMM (17). However, patients with high-risk cytogenetics and those intended for immediate transplant were excluded because a parallel phase 3 study, SWOG 1211, was concurrently enrolling patients with high-risk MM (18). In standard-risk NDMM, the ENDURANCE trial demonstrated no differences in PFS or OS between VRd and KRd. Single-center retrospective studies have also compared VRd and KRd regimens in high-risk MM (19).
Although the overall toxicity pro les of VRd and KRd are considered favorable, their prolonged use warrants a heightened vigilance for toxicity evaluation, and treating physicians need to carefully balance e cacy and toxicity pro les for each patient. Since patients enrolled on clinical trials, per eligibility criteria, are less frail and have fewer comorbidities than patients in the general population (20,21), we were motivated to conduct a retrospective study to de ne the e cacy and safety of VRd and KRd in patients with standard and high-risk MM.

Methods
Using an in-house data query platform, DataLine, we retrospectively identi ed patients with NDMM treated with VRd and KRd at Memorial Sloan Kettering Cancer Center (MSK) between January 1, 2015 and December 31, 2019. The last follow-up date was September 30, 2022. This study received approval from the Institutional Review Board at MSK. The research was performed in compliance with the terms from the Declaration of Helsinki and was waived from the obligation to obtain written informed consent. Data underwent peer-based quality check for completeness and internal consistencies. Patients with NDMM were included if they completed at least 1 cycle of VRd or KRd as induction therapy. Patients were also included if they received ≤1 cycle of a different induction regimen prior to receiving VRd or KRd. High-risk FISH/SNP signature was de ned as one or more of the following abnormalities: 1q+, t(4;14), t(14;16), t (14;20), and del(17p).
Response was assessed using the International Myeloma Working Group (IMWG) consensus criteria for response (22). Minimal residual disease (MRD) testing, if completed, was performed during or after completion of induction cycles using a validated ow cytometry assay with at least a limit of detection of 10 − 5 (23).
Patients' charts in the electronic health records (EHR) were reviewed to evaluate for speci c adverse events (AEs), including pulmonary and cardiovascular events, hypertension (worsening from baseline or requiring medication modi cations), renal complications, and peripheral motor or sensory neuropathy.
AEs were graded using Common Terminology Criteria for Adverse Events (CTCAE), version 5.0 retrospectively. AE data was consolidated for each patient and AE category by taking the worst grade across time and category (pulmonary and cardiovascular events, hypertension, renal events, and peripheral motor or sensory neuropathy) and counting patients documented to have multiple AEs within a category only once. All available data were reviewed to assess reversibility de ned as complete resolution or return to baseline of toxicities up to 6 months following completion of induction therapy.

Statistical Analysis
Descriptive statistics was used to summarize patient and disease characteristics and toxicities experienced across the VRd and KRd groups. These variables were compared between the two treatment groups using Fisher's exact test for categorical variables and Wilcoxon rank-sum tests for continuous variables. Event-free survival (EFS) was de ned as the time from start of induction therapy to death, progression of disease and change in the line of therapy following initial VRd or KRd induction, whichever came rst. Progression-free survival (PFS) was de ned as the time from the start of induction therapy until progression of disease or death from any cause. Overall survival (OS) was de ned as the time from start of induction therapy to death from any cause. Left truncation was used in patients who transferred and completed their induction treatment to MSK. EFS, PFS, and OS were estimated by the Kaplan-Meier method, and differences in time to event outcome between VRd and KRd were assessed by log-rank test.
Median follow-up was calculated using reverse Kaplan-Meier method. Multivariable Cox proportional hazard regression models were used to estimate the hazard ratios (HR) and 95% con dence intervals (CIs) for the comparison of effects of induction regimens on risk of time to event outcomes, adjusting for age at time of induction start (years, continuous), cytogenetic risk (standard, high), R-ISS Stage (I, II, and III), cardiac history (see table 1 for list of medical conditions included), and early autologous stem cell transplant (time-dependent covariate). Proportional hazards test based on weighted residuals was used to check for violations of the assumptions of Cox proportional-hazard regression (24). Since the assumptions were violated in the OS analysis, two separate Cox proportional-hazard regression were tted to: 1) all patients with follow-up time truncated at 5.5 years and 2) patients with follow-up time longer than 5.5 years (e.g. landmark analysis at 5.5 years). Different follow-up thresholds ranging from 5 to 6 years were checked in sensitivity analysis to qualitatively con rm the main ndings. To con rm our ndings, survival analysis with propensity weighting was additionally performed. We constructed propensity weights using logistic regression with the same set of covariates except for age at diagnosis, which was modeled by cubic splines. Variables with P values < .05 were considered statistically signi cant. All analyses were performed using CRAN R Version 3.3.0 (The R Foundation for Statistical Computing, Vienna, Austria).

EFS
Re ective of standard clinical practice in the United States, there were patients who changed therapy in the absence of standard de nition of progressive disease per IMWG criteria. To account for such potential bias, we estimated EFS and found the 5-year rates were 34% (95%CI, 27%-42%) and 52% (95%CI, 45%-60%) in the VRd and KRd groups, respectively (P < 0.001) (Fig. 1B).

OS
At the median follow-up of 51.6 months, 37 patients died in the VRd group and 22 died in the KRd group. During induction, 2 patients died while receiving VRd (one patient died from a hemorrhagic stroke and one patient from unknown cause during cycle 4). No patients died during KRd induction. Median OS has not been reached in either group. The Kaplan-Meier estimates for OS at 5 years were 80% (95%CI 75%-87%) in the VRd group and 90% (85% − 95%) in the KRd group (P = 0.053) (Fig. 1C).
After 5 years of follow-up, which included the coronavirus disease 2019 (COVID-19) pandemic, 3 patients in the VRd group and 7 patients in the KRd group died. In the VRd-treated patients, 1 patient died from a secondary primary malignancy and 2 patients died from progressive disease (plasma cell leukemia). In the KRd group, 2 patients died from progressive disease (PD), 1 patient died from infection, 1 patient had PD and disseminated fungal infection, and 3 patients had unknown cause of death. Since 7 out of the 22 deaths in the KRd group were observed after 5 years of follow-up, the main assumption of the Cox proportional-hazards model was not satis ed (P = 0.002). We accounted for observing an inconstant effect of induction over time by reporting an analysis of OS for all patients with follow-up truncated at 5.5 years and a landmark analysis of OS at 5.5 years. When follow-up was truncated at 5.5 years, KRd was associated with improved OS (P = 0.004); in contrast, for landmark analysis at 5.5 years, VRd was associated with longer OS (P = 0.019) (Supplement Fig. 2A and 2B).

OS
Estimated 5-year OS for patients with standard-risk cytogenetics was 87% (95%CI, 81%-94%) for VRdtreated patients and 93% (87%-99%) for KRd (P = 0.13) (Fig. 3A). Among high-risk patients, the 5-year OS rate for VRd and KRd induction was 69% (95%CI, 58%-82%) and 88% (80%-97%), respectively (P = 0.044) (Fig. 3B). Similar to the main analysis of OS, we analyzed OS for these subgroups based on follow-up period (truncated at 5.5 years and landmark point at 5.5 years). For standard-risk patients, there were no statistically signi cant differences in OS comparing VRd and KRd in both analyses (all patients with follow-up truncated at 5.5 years: P = 0.10 and landmark analysis at 5.5 years: P = 0.80) (Supplement Fig. 5A and 5B). For high-risk patients, KRd compared to VRd was associated with improved OS when follow-up for all patients was truncated at 5.5 years (P = 0.002); for the small group of patients used in the landmark analysis at 5.5 years VRd was associated with longer OS than KRd (P = 0.04) (Supplement Fig. 6A and 6B).

Multivariable analysis
We conducted a multivariate analysis for important clinical variable that may affect survival outcomes ( Multivariable analysis for OS with follow-up truncated at 5.5 years con rmed that KRd was associated with improved OS (HR 0.43, 95%CI 0.22-0.81; P = 0.01) after adjusting for important clinical variables (Table 3). Additionally, other variables associated with improved OS included standard-risk cytogenetics (HR = 2.06, 95%CI 1.13-3.77; P = 0.019) and R-ISS Stage I vs III (HR 3.47, 95%CI 1.19-10.1; P = 0.023). In the landmark analysis at 5.5 years, association with KRd and longer OS was no longer statistically signi cant (HR 9.88, 95%CI 0.88-111; P = 0.063) after adjusting for the same confounders (Supplement Table 4A).

Safety
Overall (including all grades), 15% of patients developed pulmonary and cardiovascular events and 30% developed new or worsening neuropathy during therapy (Supplement Table 2). Patterns of these toxicities are described in further detail below.

Cardiovascular and pulmonary adverse events
In the VRd group, 5% of patients experienced at least one grade (G) 2 or higher cardiovascular and pulmonary AE compared to 8% in the KRd group; these events were reversible in 67% and 88% of patients who experienced a grade 2 or higher event while receiving VRd and KRd, respectively. The incidence of grade ≥2 pulmonary and cardiovascular AEs are summarized in Supplement Table 3.

Peripheral neuropathy
For patients who received VRd as their frontline regimen, 30 (15%) patients had sensory alteration or paresthesia interfering with function and/or symptomatic weakness interfering with function, or more severe forms of neuropathy (grade ≥ 2). Three (2%) patients had improvement to baseline or resolution of symptoms within 6 months of completing therapy, and 27 (14%) developed chronic neuropathy. VRdtreated patients who developed peripheral neuropathy required various interventions, including opiates (N = 16), pregabalin/gabapentin/duloxetine (N = 35), rehabilitation/physical therapy (N = 4), assistive devices for ambulation such as canes, walkers, and wheelchairs (N = 5). None of the patients treated with KRd developed grade ≥ 2 peripheral neuropathy.

Discussion
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 e cacy, including depth of response, PFS, EFS, and OS, and characterize the safety pro le 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 con rmed KRd bene t 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 ndings suggesting that VRd and KRd induction can achieve better results than ndings 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 signi cantly 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 in uence 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 signi cant 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 bene t 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 bene t 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, re ected 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 ndings are an important re ection 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 nd any grade ≥ 2 peripheral neuropathy with KRd, and none of the patients required neuropathy-speci c interventions. Car lzomib 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 uid 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 signi cantly elevated rates of grade 3 or higher pulmonary and cardiovascular AEs associated with car lzomib, 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 re ective 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 strati ed 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 di cult 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.

Declarations
Contributors CT, BD, MH, OL conceived and designed the study. CT curated and managed the data. CT, BD, MH, AD, DN, OL, NK analyzed, accessed, veri ed, and interpreted the data. CT wrote the rst draft of the manuscript.
All authors reviewed the manuscript. All authors had full access to the data in the study and had nal responsibility for the decision to submit for publication.
Con ict of interest SG reports personal fees and advisory role (scienti c advisory board) from Actinium, Celgene, Bristol Myers Squibb, Sano , Amgen, P zer, GlaxoSmithKline, JAZZ, Janssen, Omeros, Takeda, and Kite, outside the submitted work. HH reports grants from Celgene, Takeda, and Janssen, outside the submitted work. NK reports research funding through Amgen and participates in advisory board with Medimmune. OBL reports serving on Advisory Board for MorphoSys. AL reports grants from Novartis, during the conduct of the study; grants from Bristol Myers Squibb; personal fees from Trillium Therapeutics; grants, personal fees and non-nancial support from P zer; and grants and personal fees from Janssen, outside the submitted work. AL also has a patent US20150037346A1 with royalties paid. SM reports research funding from Allogene Therapeutics, Juno/Bristol Myers Squibb, Takeda Oncology, and Janssen Oncology; personal fees from Plexus communication, and Physician Education Resource, outside the submitted work. US reports personal fees from Physicians Educations Resources; grants and other from Celgene/Bristol Myers Squibb; other from Janssen; and grants from Parker Institute for Cancer Immunotherapy and Myeloma Crowd, outside the submitted work. BD reports personal fees from Medscape, outside the submitted work. CT reports research funding from Janssen and personal fees from Physician Educations Resource. MH reports research funding from Amgen, Daiichi Sankyo, GlaxoSmithKline; has received honoraria for consultancy/participated in advisory boards for Curio Science LLC, Intellisphere LLC, Bristol Myer Squibb, and GlaxoSmithKline. OL has received grant support from LLS, Rising Tide Foundation, NIH, FDA, MMRF, IMF, Perelman Family Foundation, Amgen, Celgene, Janssen, Takeda, Glenmark, Seattle Genetics, Karyopharm; has received honoraria for scienti c talks/participated in advisory boards for Adaptive, Amgen, Binding Site, BMS, Celgene, Cellectis, Glenmark, Janssen, Juno, P zer; and served on Independent Data Monitoring Committees (IDMC) for international randomized trials by Takeda, Merck, Janssen, Theradex.