Retreatment of multiple myeloma with previously refractory drugs

Advances in the understanding of the underlying disease biology, as well as the introduction of highly effective novel therapies over the past 2 decades, have led to a significant improvement in overall survival (OS) for patients with multiple myeloma (MM).^1,2 Despite these advances, MM continues to be a disease characterized by several relapses and remissions and remains without a reliable cure. Nevertheless, improvements in OS have led to MM behaving more like a chronic disease, with patients receiving several lines of therapy during the disease course and the disease becoming refractory to many of these therapies.³ Refractoriness and response to prior therapies are major factors in deciding the choice of therapy at each subsequent relapse for relapsed/refractory MM (RRMM).⁴ The current approach is to consider using at least 1 (or 2, if possible) new drugs that the disease has not been refractory to at the time of relapse.^3,4 As patients with MM continue to live longer and get exposed to most available drugs/drug classes, providers might be limited in their options to use a new drug at each subsequent relapse, especially for patients with advanced MM in later relapses. For MM that has been exposed to most available therapies, retreatment with a drug that the disease had previously been exposed/refractory to might be 1 viable option.⁵ This approach is also important for patients without access to novel therapies, in which retreatment with a drug that the disease had initially responded to but later become refractory to might still offer durable responses. MM outcomes with retreatment have previously been described for bortezomib- and lenalidomide-containing therapies and more recently with daratumumab and B-cell maturation antigen (BCMA)–directed chimeric antigen receptor (CAR) T-cell therapies.^6-14 However, there are limited data on the durability of responses with retreatment, as well as drug agnostic predictors of response with retreatment. In this study, we sought to describe the outcomes of patients with MM who received a drug that the disease had previously been refractory to and assessed disease and treatment patterns associated with favorable outcomes with retreatment.

After approval from the institutional review board, we retrospectively reviewed all patients with RRMM at our institution who were started on a new line of systemic therapy for disease progression between 1 January 2015 and 30 April 2022. We considered the first relapse after 1 January 2015, for which patients were started on a new line of systemic therapy, as the index relapse and the starting time of our study. For this index relapse, we assessed whether patients were treated with a drug that their disease had previously been refractory to (disease progression while receiving the drug or within 60 days of the last dose of the drug) any time during the disease course.¹⁵ Patients were included in the study if, at the index relapse, they were treated with a drug that the disease had previously been refractory to. This drug that the disease had previously been refractory to and was also used at relapse was considered as the index drug. In the case of 2 identified retreatment index drugs, the retreatment event with the shorter time gap between the last exposure to the index drug during the initial line of therapy and the beginning of retreatment line was included in the study. As a result, only 1 retreatment event per patient was included in the study. Characteristics collected at diagnosis included age, sex, international staging system (ISS) stage, revised ISS (R-ISS) stage, and interphase fluorescence in situ hybridization (FISH) abnormalities.^16,17 The Mayo Stratification of Myeloma and Risk-Adapted Therapy criteria were used for risk stratification based on FISH abnormalities.^18-20 

We calculated the duration of the initial line of therapy with the index drug that the patient was on as a marker of sensitivity of the disease to the initial index drug exposure. The duration of the initial line of therapy was calculated as the time from the start of treatment with index drug to disease progression on/after the initial line of therapy. We then divided the duration of the initial line of therapy into quartiles and assessed its impact on outcomes with retreatment. Similarly, we divided the duration of index drug–free period (time from the last dose during the initial line to the beginning of retreatment) into quartiles and assessed its impact on subsequent outcomes. Finally, we also assessed the impact of the number of intervening index drug–free lines of therapy between the initial therapy and retreatment on outcomes with retreatment. We classified retreatment regimens into the following mutually exclusive classes: triplet or quadruplet regimens containing a proteasome inhibitor (PI), immunomodulatory drug (IMiD), or an anti-CD38 monoclonal antibody; VDTPACE (bortezomib, dexamethasone, thalidomide, cisplatin, doxorubicin, cyclophosphamide, and etoposide)–like regimens; and other regimens (including other triplets and doublet regimens, etc). At index relapse, we also considered whether patients were treated with other nonrefractory “partner” drugs in addition to the index drug. We classified nonrefractory partner drugs into the following mutually exclusive classes: daratumumab, PI, or IMiD combinations (triplets or quadruplet regimens containing PI or IMiD as the nonrefractory partner drug), treatment with previously refractory drugs only, and other (VDTPACE, alkylators, and other classes).

Baseline characteristics were summarized using descriptive statistics. We used the χ² test for comparing categorical variables and Kruskal-Wallis test for comparing continuous variables. We defined progression-free survival (PFS) as the time from the start of therapy to disease progression or death; and OS as the time from the start of therapy to death due to any cause. PFS and OS were calculated using the Kaplan-Meier method, and differences between the groups were assessed using the log-rank test. Cox models were used to assess the prognostic significance of various parameters in predicting PFS and OS. We used the maximally selected rank statistics method to determine the optimal cutoff point for continuous variables (duration of initial line of therapy and time duration between initial treatment and retreatment) for the most significant differences in PFS between the 2 groups. Median follow-up was calculated using the reverse Kaplan-Meier estimator method. For all tests, 2-sided P values of < .05 were considered statistically significant. All analyses were performed using R version 4.3.1. All patients authorized the use of their medical record information for research purposes.

A total of 315 patients were included in the study. The median age was 61 years (interquartile range, 54-68) at retreatment. At diagnosis, 33%, 42%, and 25% of patients were classified as ISS stage I, II, and III, respectively. Baseline characteristics of all included patients are described in Table 1. Of the total 315 patients, 163 patients (52%) were retreated in the very next line and within 30 days after the disease became refractory to the index drug. The remaining 152 patients (48%) were treated with at least 1 intervening index drug–free line or, if the initial treatment was followed by a treatment-free period, at least 30 days between the initial treatment with the index drug and retreatment. For the majority of patients, the index drug that the disease had previously been refractory to and was subsequently retreated with was lenalidomide (n = 143 [45%]), followed by bortezomib (n = 95 [30%]), pomalidomide (n = 36 [11%]), daratumumab (n = 19 [6%]), ixazomib (n = 7 [2%]), and carfilzomib (n = 5 [2%]). The remaining 10 patients (3%) were retreated with cyclophosphamide, thalidomide, and selinexor.

Of the total 315 patients, responses achieved with retreatment were available for 285 patients. A total of 74 patients (26%) had a partial response (PR) with retreatment, and 86 patients (30.2%) had a very good PR (VGPR) or better. The overall response rate (ORR; PR or better) was 56.2%. Overall, 44 patients (15.4%) had only a minimal response with retreatment, whereas, for 81 patients (28.4%), the best response achieved was less than a minimal response or progressive disease. The PR or better and VGPR or better rates were successively lower with more prior lines at the beginning of retreatment (supplemental Figure 1). The retreatment regimens received and their respective response rates are described in supplemental Table 1. Based on the nonrefractory partner drug, 131 patients (42%) were treated with PI or IMiD combinations, 96 (30%) with daratumumab, and 63 (20%) with other drugs, whereas 25 patients (8%) were treated only with a previously refractory drug combination.

The median follow-up for the entire cohort from the beginning of retreatment was 56 months (95% confidence interval [CI], 48-64). The median number of prior lines before the retreatment line was 2 (range, 1-12). The median PFS with retreatment for the entire cohort was 11 months (95% CI, 9-13). Based on the number of prior lines before retreatment, the median PFS for patients with 1 prior line was 29 months (95% CI, 18-46), for 2 prior lines was 11 months (95% CI, 9-16), for 3 prior lines was 7 months (95% CI, 5-11), and for ≥4 prior lines was 4 months (95% CI, 3-8; P = .003 for 2 vs 3 prior lines; P = .48 for 3 vs 4 prior lines; P < .001 for all other pairs; supplemental Figure 2). The median OS with retreatment for the entire cohort was 53 months (95% CI, 46-66). Based on the number of prior lines before retreatment, the median OS for patients with 1 prior line was 78 months (95% CI, 64 to not reached [NR]), for 2 prior lines was 49 months (95% CI, 39 to NR), for 3 prior lines was 57 months (95% CI, 30 to NR), and for ≥4 prior lines was 26 months (95% CI, 22-48; supplemental Figure 3).

The median duration of the initial line of therapy with the index drug for all patients was 14.9 months (quartile 1 [Q1], 5.4 months; Q3, 32.9 months; range, 0.4-142.5; N = 315). For patients in Q1 of the duration of the initial line of therapy, the PR or better rate was 63.5% (VGPR or better rate, 41.9%); for patients in Q4, the PR or better rate was 56.2% and the VGPR or better rate was 30.2%. However, no statistical differences were found in the response rates between Q1 and Q4 (P = .47 for PR or better; P = .43 for VGPR or better rate; supplemental Figure 4). The median PFS with retreatment for patients in Q4 of the duration of the initial line of therapy was numerically higher than that of patients in Q1, but it was not statistically different (median PFS for Q4, 15.6 months [95% CI, 13.4-26.5] vs median PFS for Q1, 10.5 months [95% CI, 6.2-14.3]; P = .23; Figure 1A). On multivariable analysis after adjusting for high-risk FISH, R-ISS stage at diagnosis, number of prior lines, age ≥65 years, index drug retreated with, and type of nonrefractory partner drug received, no differences were found in terms of PFS with retreatment between Q1 (reference) and Q4 (hazard ratio [HR], 1.01; 95% CI, 0.56-1.82) of the duration of the initial line of therapy (supplemental Figure 5). For the duration of the initial line of therapy, we found that the optimal cutoff point was 28.4 months to find the most significant difference in PFS among the groups. Based on this, the duration of the initial line of therapy of >28.4 months was significantly associated with better PFS (median PFS, 16.9 months [95% CI, 14.6-26.8] vs 8.1 months [95% CI, 5.9-10.5]; P < .001; Figure 1B). In terms of OS, we did not find a significant difference among quartiles of the duration of the initial line of therapy on univariate analysis (median OS for Q4, 61.6 months [95% CI, 43.2-83.1]; median OS for Q1, 73.9 months [95% CI, 51.7 to NR]; supplemental Figure 6) and on multivariable analysis (supplemental Figure 7).

The median time gap between the initial line of therapy with the index drug and retreatment line was 17.1 months (Q1, 7.5 months; Q3, 34.4 months; range, 1.0-133.2; n = 152). For patients in Q1, the PR or better rate was 60% (VGPR or better rate, 22.9%), whereas for patients in Q4, the PR or better rate was 52.6 % (VGPR or better rate, 23.3%). No statistically significant differences were found in the response rates for patients in Q1 vs Q4 (P = .81 for PR or better; P = .43 for VGPR or better rate; supplemental Figure 8). The median PFS with retreatment for patients in Q4 of the gap time was numerically higher than that of patients in Q1, but it was not statistically different (median PFS for Q4, 13.3 months [95% CI, 7.0-26.8] vs median PFS for Q1, 5.9 months [95% CI, 3.3-13.0]; P = .14; Figure 2A). On multivariable analysis to adjust for high-risk FISH, R-ISS stage, number of prior lines of therapy, age ≥65 years, index drug retreated with, and type of nonrefractory partner drug, the HR favored Q4 for PFS, but these differences were not statistically significant (HR for Q4 vs Q1, 0.59; 95% CI, 0.25-1.4; supplemental Figure 9). The optimal cutoff point for time gap between initial treatment and retreatment was 46.1 months. Based on this cutoff, a time gap of >46.1 months was associated with a superior PFS with retreatment (median PFS, 28.2 months [95% CI, 8.1 to NR] vs 8.9 months [95% CI, 5.9-12.1]; P = .016; Figure 2B). In terms of OS, we did not find a significant difference among quartiles of time gap on univariate analysis (median OS for Q4, 61.8 months [95% CI, 61.6 to NR]; median OS for Q1, 21.3 months [95% CI, 17 to NR]; supplemental Figure 10) and on multivariable analysis (supplemental Figure 11).

To assess the impact of drug class refractoriness at index relapse, we tested the impact of the duration of the initial line of therapy and the time gap between initial treatment and retreatment in 2 additional multivariable models and found similar trends as the above models containing number of prior lines of therapy (supplemental Figures 15 and 16).

Based on the number of intervening lines between initial therapy and retreatment, patients who had a treatment-free duration (0 intervening lines) had superior PFS compared with patients with 1 intervening line, 2 intervening lines, and ≥3 intervening lines of therapy (median PFS for 0 intervening lines/treatment-free interval, 18 months [95% CI, 10.5-28.4]; 1 intervening line, 7.5 months [95% CI, 5.8-13.0]; 2 intervening lines, 8.4 months [95% CI, 4-16]; ≥3 intervening lines, 5.5 months [95% CI, 3.5-12]; for any number of intervening lines, 7 months [95% CI, 5.5-11.2]; P = .003 vs 0 intervening lines). We did not find an incremental relationship between the number of intervening lines between initial therapy and retreatment and PFS with retreatment (supplemental Figure 12).

In this retrospective study, we sought to describe the outcomes of patients with MM who were retreated with a drug that the disease had previously been refractory to, exploring disease- and treatment-related factors predictive of an improved outcome with retreatment. We found that retreatment with previously refractory drugs was a feasible and effective option, demonstrating an ORR of 56.2%, a median PFS of 11 months (95% CI, 9-13), and a median OS of 53 months (95% CI, 46-66). Patients with a longer time on initial therapy with the index drug (>28.4 months) had superior PFS with retreatment (median PFS, 16.9 vs 8.1 months). Similarly, patients with a longer time gap between the initial line of therapy with index drug and retreatment with index drug (>46.1 months) had better PFS with retreatment (28.2 vs 8.9 months).

In this study, MM that demonstrated good sensitivity to initial drug exposure (longer PFS with initial line of therapy) also demonstrated a longer PFS when the drug was reinstituted later in the disease course, even though the disease had become refractory to the drug at the time of initial exposure. Additionally, a longer time from prior drug exposure was seen to be a favorable factor influencing outcomes with retreatment. Our findings are consistent with prior genomic and clinical studies in MM. Several hypotheses regarding the mechanisms of response with retreatment have been previously discussed. First, PIs and IMiDs may modulate/enhance the effects of other cytotoxic agents, outside of their own antimyeloma effects.^21,22 Thus, even disease that is refractory to these drugs might benefit from these synergistic effects with a new drug. Second, under extended periods of drug exposure, drug targets (eg, CD38) might get downregulated, leading to a loss of efficacy. With enough time off the index drug, these targets on MM cells can recover, allowing for the possibility of retreatment.⁹ Third, as MM develops resistance to different therapeutic regimens, it undergoes clonal evolution characterized by the appearance and disappearance of different MM clones across the disease course.^21,22 However, these resistant clones can disappear over time when MM is treated with drugs with different mechanisms of action, allowing for an opportunity to reuse a previously exposed drug for clinical benefit.⁵ 

The findings of this study are in line with existing studies evaluating outcomes of MM retreatment with previously exposed and/or refractory drugs. A longer time from initial drug exposure has been associated with better outcomes with retreatment in existing studies, although data on predictors of outcomes with retreatment have been limited.^6-14,23 Studies evaluating outcomes with bortezomib retreatment in MM exposed but not refractory to bortezomib have reported ORR ranging from 40% to 71%, with a median PFS ranging from 8.3 to 15 months.^6,7,23 Nooka et al reported the results of 22 patients who had MM refractory to either daratumumab or pomalidomide and were retreated with a combination of daratumumab, pomalidomide, and dexamethasone. They found an ORR of 40.9%, and a median PFS of 5.7 months with retreatment for this cohort.¹³ Similarly, Abdallah et al in a retrospective study reported outcomes of 43 patients with MM who had disease refractory to and retreated with daratumumab-based regimens, demonstrating an ORR of 49% and a median PFS of 8 months, comparable with our study.¹¹ Kunacheewa et al reported 64 patients who received retreatment with lenalidomide-based therapy immediately after progression on lenalidomide maintenance and found an ORR of 58%, with a median PFS of 13.6 months.¹² More recently, studies have reported outcomes with retreatment with BCMA-directed therapies. Reyes et al reported outcomes of 9 patients who were retreated with BCMA-directed CAR T-cell therapy after a prior BCMA-directed CAR T-cell therapy and found a high ORR of 89% and a median PFS of 8.3 months with retreatment.⁸ Ferreri et al reported outcomes of 50 patients receiving idecabtagene vicleucel (ide-cel) after prior exposure to another BCMA-directed therapy. They found that patients who had a response with ide-cel tended to have a longer time from the last BCMA-directed therapy (209 vs 128 days; P = .052).¹⁴ However, they also found that patients who had a response with ide-cel tended to have a shorter duration of exposure to the initial BCMA-directed therapy (23 days for responders vs 63 days for nonresponders; P = .025). Although these findings contrast with our findings, Ferreri et al considered exposure to CAR T-cell therapy as 1 day of duration of the initial line of therapy, which is difficult to compare with our study in which we considered the initial exposure to the index drug as the duration of the entire initial line of therapy with the index drug, including days when the drug was not administered.¹⁴ 

Although we did observe favorable trends, we did not find a difference in PFS with retreatment based on quartiles of the duration of the initial line of therapy with index drug and the time gap between initial treatment and retreatment with index drug, which could be partly due to small numbers in Q4 and Q1 in both scenarios. Additionally, we did not find statistically significant differences in OS based on retreatment patterns, which could be explained by unaccounted for differences in the subsequent therapies received after retreatment.

It is important to note that the standard-of-care approach for RRMM remains using drugs that MM has not been previously exposed to or has not become refractory to.⁴ As the therapeutic landscape of RRMM continues to evolve with the increasing availability of newer therapies, we anticipate that the question of retreatment with previously refractory drugs would become less relevant over the coming years. Nevertheless, because MM remains without a cure, it is important to consider all possible therapeutic options for patients, including the possibility of reusing therapies that the disease might have been refractory to previously.

In this study, we found that patients with standard-risk disease and those treated in earlier lines of therapy had superior outcomes with retreatment. Not surprisingly, these patient/disease characteristics associated with a favorable outcome with retreatment are also the characteristics that would result in a favorable response with a new drug. At each relapse, the potential benefit of retreatment should be weighed against the potential efficacy of a new drug and should be individualized to the patient, number of available therapies already exposed to, and aggressiveness of the relapse. Although it remains challenging to predict the ideal disease phenotype likely to benefit with retreatment vs using a new drug, the benchmarks of initial duration of index drug therapy and the time gap between therapies can be valuable data points for such decisions.

These findings also highlight the complexities of patient selection for RRMM clinical trials. RRMM trials often enroll patients based on prior drug exposure/refractoriness but have typically not considered the time from the development of drug refractoriness as one of the factors affecting eligibility. We observed that MM, even if refractory after initial treatment, can have a good response with retreatment with the same drug after a long enough drug-free period. As our understanding of MM drug refractoriness and loss of refractoriness continues to evolve, these criteria might become important for determining RRMM trial eligibility.

Our study is prone to several limitations owing to its retrospective nature and data being from a single institution, which could bias the results in terms of patient population and practice patterns. The favorable retreatment outcomes seen in patients with a longer time on initial line of therapy and longer time gap between initial treatment and retreatment could be due to selecting for indolent disease biology in the group with favorable outcomes. Although we adjusted for the type of nonrefractory partner drug received during the index relapse, this is prone to residual bias, which limits the ability to completely attribute retreatment outcomes to retreatment itself vs the contribution of the nonrefractory partner drug. Our study comprises a heterogeneous cohort in terms of treatments received during the index relapse; this might limit comparisons across different groups, even after adjusting for index and partner drugs. Additionally, our study also does not answer the utility of retreatment with a previously refractory drug compared with a standard-of-care triplet regimen: information that might be more helpful in determining the choice of therapy for relapsed disease. Future prospective trials evaluating the impact of retreatment with previously exposed/refractory drugs in addition to standard-of-care therapy vs standard-of-care therapy alone can offer better insights into MM responses with retreatment.

In conclusion, patients with a longer duration on initial line of therapy and patients with a longer time between the initial line of therapy and retreatment had superior PFS with retreatment of MM with previously refractory drugs. Retreatment with previously refractory drugs is a viable therapeutic option for MM and can be considered for MM that has been exposed to most available therapies, as well as for patients who might not have access to novel therapeutic agents.

The authors acknowledge the Mayo Clinic hematologic malignancies program.

Contribution: U.G. and S.K.K. contributed to study concept and design, drafting of the manuscript, and acquisition, analysis, and interpretation of the data; and U.G., C.C., P.K., M.B., F.K.B., D.D., A.D., A.F., M.A.G., W.I.G., S.R.H., M.A.H., Y.L.H., T.K., M.Q.L., N.L., Y.L., R.M.W., R.A.K., S.V.R., and S.K.K. critically reviewed the manuscript and provided final approval for submission.

Conflict-of-interest disclosure: P.K. reports honoraria from AbbVie (institution), AstraZeneca, BeiGene, MustangBio, Pharmacyclics, and Sanofi (institution); consulting or advisory role with Sanofi (institution); research funding from AbbVie (institution), Amgen (institution), Bristol Myers Squibb/Celgene (institution), GlaxoSmithKline (institution), Ichnos Sciences (institution), Karyopharm Therapeutics (institution), Regeneron (institution), Sanofi (institution), Sorrento Therapeutics (institution), and Takeda (institution); and travel, accommodations, and expenses from GlaxoSmithKline, Janssen, and Sanofi. A.D. reports consulting or advisory role with Janssen Research & Development; research funding from Alnylam (institution), Celgene (institution), Janssen Oncology (institution), Pfizer (institution), Prothena, and Takeda (institution); and travel, accommodations, and expenses from Janssen Oncology, Pfizer, and Prothena. D.D. reports consulting or advisory roles with Alexion Pharmaceuticals, Apellis Pharmaceuticals, Bristol Myers Squibb, J&J Innovative Medicine, Legend Biotech, Novartis, and Sanofi; and research funding from K36 Therapeutics (institution). M.A.G. reports honoraria from AbbVie, Akcea Therapeutics, Alnylam, Amgen, Apellis Pharmaceuticals, Celgene, Janssen Oncology, Juno/Celgene, Med Learning Group, Prothena, Research to Practice, Sanofi, and Telix Pharmaceuticals; consulting or advisory roles with Bristol Myers Squibb/Sanofi and Prothena; and travel, accommodations, and expenses from Celgene, Novartis, and Prothena. W.I.G. reports consulting or advisory role with Amgen (institution); research funding from Bristol Myers Squibb Foundation (institution) and ORIC Pharmaceuticals (institution); and patents, royalties, and other intellectual property (patent number 10996224 for assessing and treating precursor plasma cell disorders). N.L. reports stock and other ownership interests with AbbVie, Senseonics, and Verrica Pharmaceuticals; and research funding from Omeros. Y.L. reports consulting or advisory roles with Adicet Bio (institution), Bristol Myers Squibb (institution), Caribou Biosciences (institution), Chimeric Therapeutics (institution), Fosun Kite (institution), Genentech (institution), Janssen Oncology (institution), Kite/Gilead (institution), Nektar (institution), NexImmune (institution), Pfizer (institution), Regeneron (institution), Sanofi (institution), and Vineti (institution); and research funding from Bristol Myers Squibb (institution) and Janssen Oncology (institution). S.K.K. reports consulting or advisory roles with AbbVie (institution), Bristol Myers Squibb/Celgene (institution), Genentech/Roche (institution), Janssen Oncology (institution), K36 (institution), Pfizer (institution), Regeneron (institution), Sanofi (institution), and Takeda (institution); research funding from AbbVie (institution), Allogene Therapeutics (institution), Bristol Myers Squibb/Celgene (institution), CARsgen Therapeutics (institution), GlaxoSmithKline (institution), Janssen Oncology (institution), MedImmune (institution), Novartis (institution), Regeneron (institution), Roche/Genentech (institution), Sanofi (institution), and Takeda (institution); and travel, accommodations, and expenses from AbbVie and Pfizer. The remaining authors declare no competing financial interests.

Correspondence: Shaji K. Kumar, Division of Hematology, Mayo Clinic, 200 1st St SW, Rochester, MN 55905; email: kumar.shaji@mayo.edu.