Improvements in multiple myeloma treatments have extended patient survival to a decade or more. Treatment response rates >90% have introduced new challenges for drug development, including a need for early endpoints with greater sensitivity. The FDA, based on data from two independent academic research groups and industry, evaluated minimal residual disease (MRD) negativity as an intermediate endpoint for progression-free and overall survival, culminating in a unanimous vote by the Oncologic Drugs Advisory Committee in April 2024 supporting MRD-negative complete response as an early endpoint reasonably likely to predict clinical benefit in multiple myeloma that may be used to support accelerated approval.

Significance:

The acceptance of MRD-negative complete response as an endpoint that is reasonably likely to predict clinical benefit will allow for the design of streamlined clinical trials for accelerated approval, enabling significantly faster patient access to novel therapies. Cooperative efforts were required to obtain and analyze clinical trial data from multiple sponsors and to determine the best approach to analysis with a relatively limited number of available datasets. The process to evaluate MRD as an intermediate endpoint, undertaken jointly by myeloma researchers and industry, with feedback from the FDA, serves as a roadmap for other areas of oncology to develop intermediate endpoints.

With approximately 20 drug approvals in the past 20 years (1), clinical outcomes in patients with multiple myeloma have significantly improved, and it is not uncommon in practice to see patients living with the disease for more than 15 years after diagnosis (2). However, there is not yet an established cure for multiple myeloma, and the majority of patients achieve relatively long survival by transitioning from one therapy to the next when the disease progresses. Based on available data from the NCI Surveillance, Epidemiology, and End Results Program database, patients diagnosed with multiple myeloma are confronted with a 5-year relative survival rate (measured from 2014 to 2020) of 61.1% (3). Thus, there is an ongoing need to develop new therapies and combinations to further extend survival and ultimately find a cure.

Types of Endpoints and Clinical Shortcomings

True clinical trial endpoints, such as survival, show direct clinical benefit to the patient (4). Surrogate endpoints, on the other hand, predict but do not directly measure clinical benefit. Meeting a primary clinical benefit endpoint—or an established primary surrogate endpoint—is required for regular FDA approval of a new drug. Treatments for serious or life-threatening illnesses may be granted accelerated approval, though based on meeting a surrogate or intermediate endpoint that is reasonably likely to predict clinical benefit upon further evaluation.

Most endpoints used in oncology today are intermediate clinical endpoints due to the prolonged survival times seen with modern therapies. In multiple myeloma, regular FDA approval of a new drug requires positive results for progression-free survival (PFS); significantly improved overall survival (OS) is not mandatory to show because salvage therapies introduce the potential for biased OS outcomes between the randomized clinical trial arms (1). Importantly, OS is assessed as a safety measure when PFS data are reviewed, to ensure there is no excess risk of dying. As a consequence of improved survival times in clinical trials, the use of intermediate endpoints has become more important to support timely access and approval. In 1992, the FDA instituted its Accelerated Approval Program to allow for earlier approval of drugs that treat serious conditions and fill an unmet medical need based on a surrogate endpoint (5). Drugs receiving accelerated approval still need to be tested in confirmatory clinical trials using endpoints that demonstrate clinical benefit. If the drug later proves unable to demonstrate clinical benefit to patients, the FDA may withdraw approval. Currently, for multiple myeloma drugs, accelerated approval can occur based on overall response rate (ORR), as defined by International Myeloma Working Group (IMWG) criteria (6), and with durability. Although ORR has been a useful surrogate endpoint in phase II trials aimed at accelerated FDA approval, its utility is waning as therapeutic efficacy improves. For example, when bortezomib—the first in the modern era of multiple myeloma drugs—was approved in 2003, the median OS was 16 months and ORR was 35% in patients with relapsed or refractory disease (7). Similarly, in 2006, the initial study of once-daily lenalidomide in multiple myeloma yielded a 24% ORR, a median PFS <8 months, and a median OS of 28 months (8). Today, the median PFS reaches 3 years in the relapsed/refractory multiple myeloma (RRMM) setting (9, 10) and >5 years in the newly diagnosed multiple myeloma (NDMM) setting (1114); the estimated timeline for a readout of PFS in a phase III trial for NDMM exceeds 10 years because of these improvements in multiple myeloma therapy. Additionally, continued improvements in the efficacy of newer drugs and combinations have resulted in trials with ORRs in excess of 85% (11, 1317), leaving little room for improvement when the next generation of multiple myeloma drugs will be developed. In the recent IMROZ study, ORRs were 91% and 92% in the experimental and control arms, respectively, whereas the 60-month PFS rates were 63% and 45%, illustrating the diminished ability of the intermediate endpoint (ORR) to detect differences in longer-term treatment effects (11). Thus, a new surrogate endpoint that is more sensitive and can read out earlier than PFS is urgently needed. The lengthy timeline for multiple myeloma drug development puts patients currently living with multiple myeloma (an estimated 179,000 in the United States; ref. 3) at risk of dying before new therapies are made available.

Minimal Residual Disease Negativity

Minimal residual disease (MRD) positivity is defined by the IMWG as the persistence of malignant plasma cells after therapy (18). MRD testing is performed on bone marrow aspirate samples by either multiparameter flow cytometry (MFC) or next-generation sequencing (NGS). In the MFC assay, bone marrow samples are analyzed for expression of known tumor antigens relative to consensus plasma cell markers. The NGS technique involves detection of myeloma-specific rearrangements in the variable diversity junction region of the immunoglobulin gene. When evaluating healthy plasma cells, this region has a broad range of variability that is reflective of various immune triggers. In contrast, outgrowth of malignant plasma cells is represented by excess copies of one (or a few) unique VDJ(s) (19). Depending on the sensitivity of current modern assays, MRD can be measured as 1 malignant cell per 100,000 to 1,000,000 healthy cells (Table 1). Currently, the FDA, National Comprehensive Cancer Network Clinical Practice Guidelines in Oncology (NCCN Guidelines), and IMWG guidelines define the minimum sensitivity threshold for MRD analysis as 1 malignant cell per 105 cells (denoted as sensitivity of 10−5; Table 1; refs. 18, 20, 21).

Table 1.

MRD definitions/concepts and comments.

Definition/conceptComment
MRD negative 10−5 Using an assay that can rule out 1 myeloma cell per 100,000 tested cells. There are no detectable myeloma cells taking this approach 
MRD negative 10−6 Using an assay that can rule out 1 myeloma cell per 1,000,000 tested cells. There are no detectable myeloma cells taking this approach 
MRD-negative CR Using a validated MRD assay that can rule out at least 1 myeloma cell per 100,000 tested cells (10−5). There are no detectable myeloma cells. Furthermore, when evaluating peripheral blood using serum protein assays, there are no detectable abnormal serum proteins (i.e., based on bloodwork, the patient fulfills criteria for CR) 
MRD negative not in CR Using a validated MRD assay that can rule out at least 1 myeloma cell per 100,000 tested cells (10−5). There are no detectable myeloma cells. However, when evaluating peripheral blood using serum protein assays, there is evidence of detectable abnormal serum proteins, which in many cases is due to delayed clearance of secreted abnormal proteins. This is frequently seen when using modern effective therapies which can deliver rapid MRD negativity (e.g., CAR T-cell therapy and four-drug combinations). In some cases, it may represent residual protein secreting disease cells that are not detected by MRD assays 
Definition/conceptComment
MRD negative 10−5 Using an assay that can rule out 1 myeloma cell per 100,000 tested cells. There are no detectable myeloma cells taking this approach 
MRD negative 10−6 Using an assay that can rule out 1 myeloma cell per 1,000,000 tested cells. There are no detectable myeloma cells taking this approach 
MRD-negative CR Using a validated MRD assay that can rule out at least 1 myeloma cell per 100,000 tested cells (10−5). There are no detectable myeloma cells. Furthermore, when evaluating peripheral blood using serum protein assays, there are no detectable abnormal serum proteins (i.e., based on bloodwork, the patient fulfills criteria for CR) 
MRD negative not in CR Using a validated MRD assay that can rule out at least 1 myeloma cell per 100,000 tested cells (10−5). There are no detectable myeloma cells. However, when evaluating peripheral blood using serum protein assays, there is evidence of detectable abnormal serum proteins, which in many cases is due to delayed clearance of secreted abnormal proteins. This is frequently seen when using modern effective therapies which can deliver rapid MRD negativity (e.g., CAR T-cell therapy and four-drug combinations). In some cases, it may represent residual protein secreting disease cells that are not detected by MRD assays 

Currently, the FDA, NCCN Clinical Practice Guidelines in Oncology (NCCN Guidelines), and IMWG guidelines define the minimum sensitivity threshold for MRD analysis as 1 malignant cell per 105 cells (denoted as sensitivity of 10−5).

Abbreviation: CAR, chimeric antigen receptor.

Early meta-analyses evaluating the relationship between MRD and PFS or OS suggested that MRD is a strong predictor of clinical outcomes. The first of these, published in 2016, analyzed results from four studies evaluating MRD and PFS; two of these also examined OS (22). Three studies used MFC, and one used NGS for the detection of MRD at a sensitivity of at least 10−4. The analysis found that MRD positivity after treatment was associated with worse PFS [HR, 2.85; 95% confidence interval (CI), 2.17–3.74; P < 0.001] and OS (HR, 2.08; 95% CI, 1.44–3.01; P < 0.001) relative to MRD negativity (22). A second meta-analysis published a year later included results from 14 studies that reported on PFS and 12 that reported on OS (23). This analysis also found that MRD negativity was associated with both better PFS (HR, 0.41; 95% CI, 0.36–0.48; P < 0.001) and better OS (HR, 0.57; 95% CI, 0.46–0.71; P < 0.001) compared with MRD positivity at the same minimum sensitivity. These analyses provided early support for the potential role of MRD in clinical prognostication.

In 2020, two meta-analyses were published that provided more support for the potential utility of MRD as a surrogate marker for PFS and OS (2, 24). These analyses found a strong, positive association between MRD negativity and improved PFS across studies. One analysis looked at trial-level association between MRD and PFS across six randomized trials (n = 3,283) and found an adjusted correlation coefficient of 0.97 (2). The other analysis included reconstructed patient-level data, which collectively demonstrated significant improvements in both PFS (44 studies; HR, 0.33; 95% CI, 0.29–0.37; P < 0.001) and OS (23 studies; HR, 0.45; 95% CI, 0.39–0.51; P < 0.001) among patients who achieved MRD negativity relative to those who remained MRD positive (24). In 2022, a meta-analysis using patient-level data from four studies (n = 2,510) from a single sponsor was published (25). It showed that in patients with transplant-ineligible (TIE) NDMM or RRMM, PFS was improved in patients who achieved MRD-negative complete response (CR) versus those who did not (HR, 0.a20; P < 0.0001). Lastly, another meta-analysis published in 2023 showed moderate trial-level association between MRD negativity and PFS across 13 trials, with a correlation coefficient of 0.53 (95% CI, 0.21–0.77; ref. 26). These studies from 2016 to 2023 showed various methods for conducting meta-analyses and incremental improvements in the quantity and quality of data available.

Beginning in 2009, the FDA undertook an interagency initiative with the NCI and the National Heart, Lung, and Blood Institute to enhance endpoint development for oncology drugs. To better understand the state of the science on MRD in hematologic malignancies, the FDA co-sponsored public workshops from 2012 to 2014, and in 2016, a workshop was held to discuss the clinical, statistical, and technical barriers to implementing MRD assessments in clinical trials (21). A 2017 analysis of MRD data submitted to the FDA in New Drug Application or Biologics License Applications across chronic myeloid leukemia, chronic lymphocytic leukemia, acute lymphoblastic leukemia, and multiple myeloma showed that although 40% of studies included MRD data, the data were uninterpretable in one third of those studies, highlighting the need for additional support to clarify the approaches to MRD data collection in clinical trials (27).

The workshops and the inconsistent quality of MRD data submitted to the FDA led the agency to publish guidance on the use of MRD as a biomarker in regulatory submissions in January 2020. Since publication of this guidance, more pharmaceutical companies and academic groups have been routinely including MRD as a biomarker in clinical trials for multiple myeloma conducted under an Investigational New Drug (IND) Application, using a more standardized approach.

The FDA Guidance for Industry outlining regulatory considerations for the use of MRD in the development of drug and biological therapies provided a framework for researchers interested in conducting meta-analyses (21). Key elements of the guidance provided by clinical reviewers at the FDA included that the use of MRD in trials should include all multiple myeloma disease settings—transplant-eligible NDMM (TE-NDMM), TIE-NDMM, and RRMM—and that MRD should only be assessed in patients who are in CR (Table 1).

The FDA described two potential uses of MRD: as a validated surrogate endpoint for traditional approval or as a surrogate endpoint reasonably likely to predict clinical benefit for accelerated approval. In both cases, the guidance explained that the strength of evidence required for a surrogate endpoint is based on the biological plausibility of the relationship, demonstration of the prognostic value of the surrogate endpoint for the clinical outcome, and evidence from clinical trials that treatment effects on the surrogate endpoint correlate with effects on the long-term clinical outcome. The two types of endpoints are distinguished by the amount of clinical data available to support surrogacy (Fig. 1). The FDA guidance also outlined the statistical principles relevant to validating MRD as a surrogate endpoint using meta-analysis.

Figure 1.

Definitions of surrogate endpoints. In multiple myeloma, ORR is currently an accepted intermediate endpoint, and PFS is an accepted surrogate for OS.

Figure 1.

Definitions of surrogate endpoints. In multiple myeloma, ORR is currently an accepted intermediate endpoint, and PFS is an accepted surrogate for OS.

Close modal

Meta-analysis Methodology

Individual-level association refers to the association between the surrogate endpoint (i.e., MRD) and the true clinical endpoint (PFS or OS; ref. 21). This analysis requires individual patient data (IPD). Trial-level association refers to the treatment effect on the surrogate endpoint and its correlation with the treatment effect on the true clinical endpoint. Analysis of trial-level associations requires the use of randomized trials. A sufficient number of trials need to be included, and their poolability should be justifiable (21). Analysis should include trials with both positive and negative results with respect to the primary endpoint and must include long-term clinical data.

Criteria to establish surrogacy must be prespecified. Statistical handling and potential effects of unevaluable samples, missing data, and confounding factors should be included in the statistical analysis plan (SAP), and sensitivity analyses should be run. Global OR is used for statistical quantification of individual-level associations. This value, estimated by the bivariate Plackett copula model, is the ratio of the odds of the long-term endpoint being greater at a fixed timepoint (e.g., 4 years) for MRD-negative patients compared with MRD-positive patients. Statistical significance is reached if the 95% CI excludes 1.0. For statistical analyses at the trial level, two methods—R2WLS and R2copula—can be used. The first of these (R2WLS) separately estimates the treatment effect on MRD and on the long-term endpoint using logistic and proportional hazard regression within each trial. Using weighted least squares, this value estimates the correlation between the log OR and log HR across all studies. Weights are based on each study’s sample size or the SE estimates for the log OR treatment effect. R2copula, on the other hand, uses the bivariate Plackett copula to estimate the treatment effect on MRD and the long-term endpoint while accounting for patient-level correlation. For both methods, values closer to 1.0 indicate stronger trial-level correlations.

Despite early literature supporting the use of MRD negativity as a clinical biomarker, IPD needed to support FDA decision-making were needed. To support acceptance as a surrogate endpoint, the FDA requires that models demonstrating the proposed association between endpoints be internally validated on IPD to ensure the relationship holds true regardless of patient population, disease setting, or treatment history. Partnership between academia and industry was needed to coordinate gathering IPD from industry-sponsored trials, discuss and test various approaches to the meta-analysis, and to perform independent analyses. In 2024, the results of these collaborative efforts over 15 years were presented at a dedicated meeting of the Oncologic Drugs Advisory Committee (ODAC) focused on MRD as an early endpoint for drug development in multiple myeloma.

Parallel Paths

Achieving FDA acceptance of a new surrogate endpoint can be accomplished through either (i) the biomarker qualification process or (ii) discussion with the Center for Drug Evaluation and Research or Center for Biologics Evaluation and Research review division (21). To evaluate the use of MRD as a surrogate endpoint in multiple myeloma, two research groups separately worked in consultation with the FDA to conduct meta-analyses of clinical trials using the Center for Drug Evaluation and Research review pathway for acceptance. The EValuating mInimal resiDual disEase as aN intermediate Clinical End point for multiple myeloma (EVIDENCE) meta-analysis was initiated in 2009 between investigators at the NCI and National Heart, Lung, and Blood Institute (Fig. 2A). At the FDA’s 2012 roundtable event on MRD in myeloma (28), external groups were invited to join the ongoing effort. Subsequently, the International Independent Team for Endpoint Approval of Myeloma Minimal Residual Disease (I2TEAMM) study was formed, and the two research groups decided to file independent INDs to initiate pre-IND meetings between the FDA and the researchers. An “MRD in Myeloma” meeting first held in 2014 brought together key academic leaders in the field, organizations for patients with myeloma, industry, and the FDA. Since inception, these well-attended meetings have occurred every year, with participation by the FDA, the Multiple Myeloma Research Foundation, International Myeloma Foundation, and HealthTree Foundation. In parallel, the two research groups (i.e., EVIDENCE meta-analysis and I2TEAMM study) received feedback from the FDA to ensure that their studies were designed appropriately to meet regulatory requirements and to develop their own formal SAPs. The two groups also met frequently with the FDA to evaluate preliminary data and models and to answer complex questions that needed to be resolved in order to optimize the analysis, such as which datasets to use and how to gain access, the need for data pooling given the limited number of available datasets, which timepoints to use for MRD assessment, how to handle missing data, which MRD sensitivity level to use, which treatment indications to include, how to handle pooling of indications, and which statistical approaches are needed. Modulation of these variables affected the amount and quality of data available for analysis. For example, increased flexibility of the timepoint at which MRD was measured by adding a ±3-month window around the 12-month landmark increased the amount of data available for analysis.

Figure 2.

A, Timeline of FDA initiative and EVIDENCE meta-analysis on MRD in myeloma. B, Studies analyzed in the EVIDENCE meta-analysis and I2TEAMM study in NDMM. Registry identifiers of each trial are as follows: ALCYONE (NCT02195479), CASSIOPEIA (NCT02541383), CLARION (NCT01818752), FORTE (NCT02203643), GEM2010MAS65 (NCT01237249), GEM2012MENOS65 (NCT01916252), GMMG-MM5 (EudraCT number 2010-019173-16), GRIFFIN (NCT02874742), MAIA (NCT02252172), OCTANS (NCT03217812), and TOURMALINE (NCT01564537). *Included in 9-month MRD-negative CR analysis for I2TEAMM and sensitivity analysis in EVIDENCE. NHLBI, National Heart, Lung, and Blood Institute; PIND, pre-IND.

Figure 2.

A, Timeline of FDA initiative and EVIDENCE meta-analysis on MRD in myeloma. B, Studies analyzed in the EVIDENCE meta-analysis and I2TEAMM study in NDMM. Registry identifiers of each trial are as follows: ALCYONE (NCT02195479), CASSIOPEIA (NCT02541383), CLARION (NCT01818752), FORTE (NCT02203643), GEM2010MAS65 (NCT01237249), GEM2012MENOS65 (NCT01916252), GMMG-MM5 (EudraCT number 2010-019173-16), GRIFFIN (NCT02874742), MAIA (NCT02252172), OCTANS (NCT03217812), and TOURMALINE (NCT01564537). *Included in 9-month MRD-negative CR analysis for I2TEAMM and sensitivity analysis in EVIDENCE. NHLBI, National Heart, Lung, and Blood Institute; PIND, pre-IND.

Close modal

The independent work of the two research groups led to differences in approach. For example, the EVIDENCE team decided to pool patients with TE-NDMM and patients with TIE-NDMM, analyzing all patients with NDMM independent of transplant eligibility status. This decision was based on their clinical experience that modern combination therapies for multiple myeloma provide equally deep responses in these two groups of patients. Patients with RRMM were evaluated separately. In contrast, the I2TEAMM study pooled all nontransplanted patients (TIE-NDMM and RRMM) and evaluated TE-NDMM separately. The FDA clinical reviewers eventually endorsed both approaches as subanalyses; however, these different approaches highlight the complexity of these collaborative efforts.

In preparation for the ODAC meeting in April 2024, the EVIDENCE meta-analysis and the I2TEAMM study aligned their analytic approaches to ensure a consistent message. For example, the I2TEAMM study team agreed to use the EVIDENCE team’s statistical model (to pool patients with TE-NDMM and patients with TIE-NDMM and analyze all patients with NDMM independent of transplant eligibility status and to evaluate patients with RRMM separately). Also, the importance of high-quality MRD data was discussed to ensure robust assessments between MRD and PFS/OS and to allow for comparisons across the two studies. The EVIDENCE meta-analysis only included datasets with an MRD assay validated to a 10−5 sensitivity threshold as a prespecified endpoint. Taking this approach, a total of 6,965 randomized patients were included in the EVIDENCE meta-analysis. The I2TEAMM statistical original analysis plan (primary analyses) included studies with data at any MRD sensitivity level (including 10−4); however, in preparation for ODAC, the I2TEAMM agreed to focus on 10−5 level datasets. Taking this approach, a total of 6,325 randomized patients were included in the I2TEAMM analysis. This cooperation between two parallel, partly competitive efforts was key for increasing the probability of overall success at ODAC and is a model for collaborative science to advance the field and ensure that benefiting patients remains the primary goal.

On April 12, 2024, the FDA convened an ODAC meeting to discuss the adequacy of available data to support the use of MRD as a surrogate endpoint reasonably likely to predict clinical benefit for accelerated approval of new therapies for multiple myeloma (29). The EVIDENCE meta-analysis, the I2TEAMM study, and the FDA presented their results, followed by a discussion by the ODAC Advisory Committee.

EVIDENCE Meta-analysis

The EVIDENCE meta-analysis, with data submitted to the FDA, was designed to assess the use of MRD for the prediction of long-term clinical benefit, as measured by PFS and OS, in patients with multiple myeloma and examine the potential role of MRD negativity as an intermediate clinical endpoint reasonably likely to predict long-term clinical benefit in patients with multiple myeloma (30, 31). Eligible studies were phase II or III trials that enrolled patients with NDMM or RRMM and included MRD negativity using an assay validated to a 10−5 sensitivity threshold as a prespecified endpoint. Twelve studies were included, with eight in NDMM (n = 5,130 randomized; Fig. 2B) and four in RRMM (n = 1,835 randomized). Studies with an MRD assay below the 10−5 sensitivity level were not included as MRD negativity is currently defined by a sensitivity of 10−5 or better, and inclusion of less-sensitive assays would bias the results.

In NDMM, individual-level associations between MRD-negative CR at 12 months and survival outcomes were strong, with ORs of 4.72 (95% CI, 3.53–5.90) for PFS and 4.02 (2.57–5.46) for OS (Table 2). Associations were observed for both TIE [PFS, 6.15 (95% CI, 4.27–8.03); OS, 4.08 (95% CI, 2.44–5.72)] and TE patients [PFS, 2.45 (95% CI, 1.40–3.51); OS, 3.78 (95% CI, 0.78–6.78)]. In the RRMM setting, the individual-level associations were strong between MRD and PFS, with ORs of 7.67 (95% CI, 4.24–11.10) for PFS and 6.03 (3.12–6.23) for OS. Trial-level correlations in the combined NDMM group were moderate to high, with correlation (R2WLS) between the treatment effect on MRD negativity and the treatment effect on PFS at 0.67 (0.43–0.91). For the TIE-NDMM subgroup, it was 0.83 (0.71–0.96); trial-level correlations for TE-NDMM and RRMM could not be estimated because of the low number of studies. The authors concluded that these results support the consideration of MRD as an early clinical endpoint reasonably likely to predict clinical benefit in multiple myeloma that may be used to support accelerated approval.

Table 2.

Summary of results: associations between MRD-negative CR at 12 months and clinical benefit endpoints.

PFSOS
EVIDENCE (30, 31)I2TEAMM (33)aFDA (1)EVIDENCE (30, 31)I2TEAMM (33)aFDA (1)
Individual-level association, global OR (95% CI)  
 TE-NDMM 2.45 (1.40–3.51) 3.86 (2.79–4.93) 3.39 (2.87–3.92) 3.78 (0.78–6.78) 4.81 (2.62–7.00) 3.83 (3.00–4.67) 
 TIE-NDMM 6.15 (4.27–8.03) 10.01 (6.15–13.87) 7.30 (5.21–9.38) 4.08 (2.44–5.72) 6.45 (2.60–10.31) 4.75 (2.91–6.58) 
 RRMM 7.67 (4.24–11.10) 12.09 (4.36–19.83) 7.67 (4.24–11.1) 6.03 (3.12–6.23) N/A 6.03 (2.48–9.59) 
Trial-level association, R2copula (95% CI)  
 NDMM 0.84 (0.64–0.99) 0.82 (0.60–1.00) 0.58 (0.29–0.86) 0.32 (<0.01–0.86) 0.68 (0.32–1.00) 0.29 (<0.01–0.63) 
Trial-level association, R2WLS (95% CI)  
 NDMM 0.67 (0.43–0.91) 0.85 (0.70–1.00) — 0.21 (<0.01–0.53) 0.80 (0.52–1.00) — 
PFSOS
EVIDENCE (30, 31)I2TEAMM (33)aFDA (1)EVIDENCE (30, 31)I2TEAMM (33)aFDA (1)
Individual-level association, global OR (95% CI)  
 TE-NDMM 2.45 (1.40–3.51) 3.86 (2.79–4.93) 3.39 (2.87–3.92) 3.78 (0.78–6.78) 4.81 (2.62–7.00) 3.83 (3.00–4.67) 
 TIE-NDMM 6.15 (4.27–8.03) 10.01 (6.15–13.87) 7.30 (5.21–9.38) 4.08 (2.44–5.72) 6.45 (2.60–10.31) 4.75 (2.91–6.58) 
 RRMM 7.67 (4.24–11.10) 12.09 (4.36–19.83) 7.67 (4.24–11.1) 6.03 (3.12–6.23) N/A 6.03 (2.48–9.59) 
Trial-level association, R2copula (95% CI)  
 NDMM 0.84 (0.64–0.99) 0.82 (0.60–1.00) 0.58 (0.29–0.86) 0.32 (<0.01–0.86) 0.68 (0.32–1.00) 0.29 (<0.01–0.63) 
Trial-level association, R2WLS (95% CI)  
 NDMM 0.67 (0.43–0.91) 0.85 (0.70–1.00) — 0.21 (<0.01–0.53) 0.80 (0.52–1.00) — 
a

Values shown with imputation of missing MRD data as MRD positive, for comparability with other analyses.

I2TEAMM Meta-analysis

The initial analysis of the I2TEAMM was published and included 15 studies of multiple myeloma with more than 50 patients in each treatment arm that described PFS and MRD (10−5) data (32). The trial-level correlation between treatment effects on PFS and treatment effects on MRD-negative rates across all trials was 0.70 (95% CI, 0.41–0.98).

A subsequent analysis, with data submitted to the FDA, was performed to evaluate whether currently available data can support MRD as an early endpoint that is reasonably likely to predict clinical benefit in future multiple myeloma clinical trials (33). These analyses included IPD from 20 studies (n = 12,316) and investigated both 9-month (primary analysis) and 12-month MRD-negative CRs. Studies with data at any MRD sensitivity level (including 10−4) were included per the I2TEAMM SAP, but in preparation for ODAC, the authors focused on those at the 10−5 level to preserve data quality. A total of 6,325 randomized patients were included in the analysis based on 10−5 threshold. Other differences between the EVIDENCE meta-analysis results and the I2TEAMM study results presented to the FDA included handling of missing data, which were imputed as MRD positive (i.e., an intention-to-treat approach) in the EVIDENCE meta-analysis and removed from the primary analysis of the I2TEAMM study (Table 3). However, in preparation for ODAC, the I2TEAMM group reanalyzed their data, taking the intention-to-treat approach for comparability with EVIDENCE.

Table 3.

Comparison of EVIDENCE and I2TEAMM meta-analyses (primary analyses; ref. 30).

VariableEVIDENCEI2TEAMM
Patient population NDMM and RRMM NDMM and RRMM 
Timepoint to evaluate MRD status Prespecified at 12 ± 3 months; jointly agreed by the FDA and study collaborators Data driven 
Patients in CR without MRD evaluation ITT approach MRD positive (i.e. all patients were kept in the analysis; patients without MRD evaluation done within the prespecified time window were counted as MRD positive) Removed from analysis 
Sensitivity cutoff for MRD assays 10−5 or bettera Anya 
VariableEVIDENCEI2TEAMM
Patient population NDMM and RRMM NDMM and RRMM 
Timepoint to evaluate MRD status Prespecified at 12 ± 3 months; jointly agreed by the FDA and study collaborators Data driven 
Patients in CR without MRD evaluation ITT approach MRD positive (i.e. all patients were kept in the analysis; patients without MRD evaluation done within the prespecified time window were counted as MRD positive) Removed from analysis 
Sensitivity cutoff for MRD assays 10−5 or bettera Anya 

Abbreviation: ITT, intention-to-treat.

a

In preparation for the ODAC meeting, the importance of high-quality MRD data was discussed to ensure robust assessments between MRD and PFS/OS and to allow for comparisons across the two studies. The EVIDENCE meta-analysis only included datasets with an MRD assay validated to a 10−5 sensitivity threshold as a prespecified endpoint. Taking this approach, a total of 6,965 randomized patients were included in the EVIDENCE meta-analysis. The I2TEAMM statistical original analysis plan (i.e., primary analyses) included studies with data at any MRD sensitivity level (including 10−4); however, in preparation for ODAC, the I2TEAMM agreed to focus on 10−5 level datasets. Taking this approach, a total of 6,325 randomized patients were included in the I2TEAMM analysis.

In the analysis at 9 months (with missing MRD data imputed as MRD positive), the global ORs (95% CI) for individual-level associations with PFS were 2.74 (1.88–3.61) for TE-NDMM, 8.17 (4.29–12.05) for TIE-NDMM, and 6.70 (3.61–9.78) for RRMM (33). For OS, the ORs were 2.57 (1.41–3.73), 9.25 (0.86–17.63), and 5.63 (2.02–9.23), respectively. Trial-level associations (R2WLS) for NDMM were 0.77 (0.49–1.00) for PFS and 0.79 (0.49–1.00) for OS. At 12 months, global ORs for individual-level correlations with PFS were 3.86 (2.79–4.93) for TE-NDMM, 10.01 (6.15–13.87) for TIE-NDMM, and 12.09 (4.36–19.83) for RRMM (Table 2). For OS, the values were 4.81 (2.62–7.00) for TE-NDMM and 6.45 (2.60–10.31) for TIE-NDMM; the OR for RRMM could not be calculated. Trial-level correlations (R2WLS) for NDMM were 0.85 (0.70–1.00) for PFS and 0.80 (0.52–1.00) for OS. Pooling patients with NDMM and patients with RRMM provided correlations of 0.72 (0.46–0.99) for PFS and 0.69 (0.45–0.94) for OS. The researchers concluded that the high individual-level associations provided strong evidence that MRD-negative CRs at the 9- and 12-month timepoints were reasonably likely to predict clinical benefit to PFS (33).

The FDA analyzed all data submitted by both the EVIDENCE and I2TEAMM groups, in total comprising 18 trials with 25 two-arm comparisons in 11,019 patients (1). Their final results were in agreement with the conclusions of the two academic groups—individual-level associations of MRD-negative CRs at 9 months and at 12 months with PFS and OS were significant for all multiple myeloma populations, indicating a high prognostic value for MRD, and global ORs were similar for MRD measured at 9 versus 12 months. For instance, ORs were 6.55 and 7.30 for associations of 9- and 12-month MRD-negative CRs, respectively, with PFS in patients with TIE-NDMM. Trial-level associations were weak to moderate for PFS and weak for OS and varied across multiple myeloma populations. For 12-month MRD-negative CR, trial-level correlation coefficients with PFS were 0.35 for TE-NDMM, 0.83 for TIE-NDMM, and 0.00 for RRMM. For OS, correlation coefficients were 0.36, 0.34, and 0.12, respectively. Overall, the FDA agreed with the approaches and interpretation of results submitted by the EVIDENCE meta-analysis and the I2TEAMM study groups (Table 2).

Strengths and Limitations of the Meta-analyses

The meta-analyses presented had many strengths, including the use of randomized trials across multiple disease settings (1). Additionally, the assays used to measure MRD (i.e., NGS or MFC with a sensitivity of ≥10−5) were relatively consistent across all trials. Importantly, IPD were analyzed for all trials, and analysis methods were prespecified in collaboration with the FDA. Limitations existed though and included heterogeneity in trial designs, conduct, and patient populations, as well as the limited number of trials available, particularly within the RRMM setting. Finally, the analyses were not able to establish a surrogate threshold effect, defined as the minimum treatment effect on MRD-negative CR needed to predict a positive effect on PFS with 95% CI, because the data were too limited.

Advisory Committee Discussion

The Committee agreed that the available patient-level data and the biological plausibility supported the use of MRD as an intermediate endpoint, particularly in the first-line setting (29). More data may be needed to analyze MRD as an endpoint in RRMM, but the committee generally felt that the data were supportive in both the NDMM and RRMM disease settings. Both the 9- and 12-month timepoints were acceptable for MRD assessment, according to the committee. The choice of timepoint should be flexible and based on the trial, therapy, and setting. It was suggested that measuring sustained MRD negativity may be helpful to assess durability, though repeated bone marrow biopsies would pose a burden to patients. Because clinical trials utilizing MRD may not capture the full safety profile of the drugs under investigation, follow-up for long-term outcomes will continue to remain important.

The meeting concluded with a unanimous (12–0) vote agreeing that the evidence supports the use of MRD as a surrogate endpoint reasonably likely to predict clinical benefit and thus suitable for accelerated approval in multiple myeloma clinical trials. Although MRD negativity may not correlate perfectly with clinical outcomes, the committee concluded that its use in supporting accelerated approval is a reasonable approach to support timely access to new medications in the current treatment era.

The acceptance of MRD as a surrogate endpoint for accelerated drug approvals has implications for not only patient access to novel drugs but also clinical trial design and standardization of the collection of MRD data in clinical trials (Fig. 3A). Updated FDA guidance on the use of MRD in clinical trials is anticipated in the future to help guide these processes. As more datasets are collected, analyses will be rerun to answer remaining questions, including the need to establish a surrogate threshold effect for MRD. The recently published PERSEUS, CEPHEUS, and IMROZ trials will inform such analyses (11, 13, 14). These trials have shown large differences in MRD-negative status between treatment arms as well as large differences in PFS. For example, in PERSEUS, MRD negativity rates were 75% versus 48% in the experimental and control arms, respectively, and the 48-month PFS rates were 84% versus 68% (13).

Figure 3.

A, Future directions for drug approval in myeloma. B, Impact of MRD as a surrogate endpoint likely to predict clinical benefit on hypothetical clinical trial timelines. SOC, standard of care.

Figure 3.

A, Future directions for drug approval in myeloma. B, Impact of MRD as a surrogate endpoint likely to predict clinical benefit on hypothetical clinical trial timelines. SOC, standard of care.

Close modal

An MRD surrogate endpoint allows for more streamlined clinical trial designs. The current requirement for randomized trials in NDMM to show superior PFS can take over 10 years; ORR is no longer useful as most patients in both arms will have a response. In the RRMM setting, accelerated approval is typically supported by a single-arm phase II study, showing predefined minimum ORR levels followed by a confirmatory randomized controlled trial showing superior PFS. For both NDMM and RRMM, there is interest in condensing this design into a single, two-arm trial that would include both the short-term endpoint and the longer-term clinical benefit endpoint (34). Under this model, a single trial would support both accelerated approval and regular approval. The validation of MRD as a surrogate endpoint will allow such a trial design in novel drugs for which ORR cannot practically serve as an early endpoint, ultimately shortening the timelines of drug development and accelerating patient access. To appreciate the impact on clinical trial timelines, we can use an example of a trial in NDMM in which the expected 4-year PFS rate is 84% for standard of care and the expected HR is 0.70. At an accrual rate of 30 patients per month, MRD would read out at 3.7 years after study start, whereas PFS would not read out until 11.5 years after study start (Fig. 3B). Hence, there is potential for eight additional years of patient access to a novel therapy.

We feel that it is appropriate to briefly address the abbreviation “MRD.” In the literature, some authors refer to MRD as “minimal residual disease,” whereas others refer to it as “measurable residual disease.” Some people argue that “measurable” is better because it embeds the idea of relativity, and therefore, invites details on the depth and method of measurement. On the other hand, others argue that “minimal” is better because it has been used for a long time and therefore is well established, and by indicating, for example, MRD negativity 10−5 (or 10−6) by NGS, the details on the depth and method are clearly provided. Just like life in general, we respect differences in opinion. In this article, we refer to MRD as “minimal residual disease” as that is the terminology FDA used at ODAC as well as in their FDA approval notice.

Moving forward, it will be important to work with drug regulatory agencies worldwide to gain their acceptance of clinical trials using MRD as an endpoint, both in multiple myeloma and other disease states. The process that has unfolded to validate MRD as an early endpoint and the potential it has for redesigning clinical trials for faster drug approvals present an opportunity for similar approaches to be taken in other areas of oncology.

O. Landgren reports grants from Janssen, Sanofi, and Amgen during the conduct of the study, as well as grants from Amgen, Menarini, and Sebia, grants and personal fees from Janssen, Pfizer, and AbbVie, personal fees from GSK and Bristol Myers Squibb, and other support from Takeda and Novartis outside the submitted work. S.M. Devlin reports grants from Johnson and Johnson Innovative Medicine during the conduct of the study, as well as other support from Miltenyi Biotec outside the submitted work.

O. Landgren is supported by the Sylvester Comprehensive Cancer Center NCI Core Grant (P30 CA 240139), the Riney Family Multiple Myeloma Research Program Fund, The Tow Foundation, the Myeloma Solutions Fund, and the Cannon Guzy Family Fund. Medical writing support was provided by Valerie P. Zediak, PhD, of Eloquent Scientific Solutions and funded by Janssen Global Services.

1.
US Food and Drug Administration
.
FDA presentation at April 12, 2024 oncology drugs advisory committee meeting
.
[cited 2024 Dec 4]. Available from:
https://www.fda.gov/media/177736/download.
2.
Avet-Loiseau
H
,
Ludwig
H
,
Landgren
O
,
Paiva
B
,
Morris
C
,
Yang
H
, et al
.
Minimal residual disease status as a surrogate endpoint for progression-free survival in newly diagnosed multiple myeloma studies: a meta-analysis
.
Clin Lymphoma Myeloma Leuk
2020
;
20
:
e30
7
.
3.
Zhu
DT
,
Park
A
,
Lai
A
,
Zhang
L
,
Attar
H
,
Rebbeck
TR
.
Multiple myeloma incidence and mortality trends in the United States, 1999–2020
.
Sci Rep
2024
;
14
:
14564
.
4.
US Food and Drug Administration
.
Surrogate endpoint resources for drug and biologic development
.
[cited 2024 Dec 4]. Available from:
https://www.fda.gov/drugs/development-resources/surrogate-endpoint-resources-drug-and-biologic-development.
5.
US Food and Drug Administration
.
FDA Oncology Center of Excellence
.
Project Confirm
.
[cited 2024 Dec 4]. Available from:
https://www.fda.gov/about-fda/oncology-center-excellence/project-confirm#:∼:text=Accelerated%20Approval%20Background,Act%20(FDASIA)%20in%202012.
6.
Rajkumar
SV
,
Harousseau
JL
,
Durie
B
,
Anderson
KC
,
Dimopoulos
M
,
Kyle
R
, et al
.
Consensus recommendations for the uniform reporting of clinical trials: report of the International Myeloma Workshop Consensus Panel 1
.
Blood
2011
;
117
:
4691
5
.
7.
Richardson
PG
,
Barlogie
B
,
Berenson
J
,
Singhal
S
,
Jagannath
S
,
Irwin
D
, et al
.
A phase 2 study of bortezomib in relapsed, refractory myeloma
.
N Engl J Med
2003
;
348
:
2609
17
.
8.
Richardson
PG
,
Blood
E
,
Mitsiades
CS
,
Jagannath
S
,
Zeldenrust
SR
,
Alsina
M
, et al
.
A randomized phase 2 study of lenalidomide therapy for patients with relapsed or relapsed and refractory multiple myeloma
.
Blood
2006
;
108
:
3458
64
.
9.
Martin
T
,
Usmani
SZ
,
Berdeja
JG
,
Agha
M
,
Cohen
AD
,
Hari
P
, et al
.
Ciltacabtagene autoleucel, an anti-B-cell maturation antigen chimeric antigen receptor T-cell therapy, for relapsed/refractory multiple myeloma: CARTITUDE-1 2-year follow-up
.
J Clin Oncol
2023
;
41
(
Suppl 16
):
1265
74
.
10.
Martin
T
,
Dimopoulos
MA
,
Mikhael
J
,
Yong
K
,
Capra
M
,
Facon
T
, et al
.
Isatuximab, carfilzomib, and dexamethasone in patients with relapsed multiple myeloma: updated results from IKEMA, a randomized phase 3 study
.
Blood Cancer J
2023
;
13
:
72
.
11.
Facon
T
,
Dimopoulos
MA
,
Leleu
XP
,
Beksac
M
,
Pour
L
,
Hájek
R
, et al
.
Isatuximab, bortezomib, lenalidomide, and dexamethasone for multiple myeloma
.
N Engl J Med
2024
;
391
:
1597
609
.
12.
Richardson
PG
,
Jacobus
SJ
,
Weller
EA
,
Hassoun
H
,
Lonial
S
,
Raje
NS
, et al
.
Triplet therapy, transplantation, and maintenance until progression in myeloma
.
N Engl J Med
2022
;
387
:
132
47
.
13.
Sonneveld
P
,
Dimopoulos
MA
,
Boccadoro
M
,
Quach
H
,
Ho
PJ
,
Beksac
M
, et al
.
Daratumumab, bortezomib, lenalidomide, and dexamethasone for multiple myeloma
.
N Engl J Med
2024
;
390
:
301
13
.
14.
Usmani
SZ
,
Facon
T
,
Hungria
V
,
Bahlis
N
,
Venner
CP
,
Braunstein
M
, et al
.
Daratumumab SC + bortezomib/lenalidomide/dexamethasone in patients with transplant-ineligible or transplant-deferred newly diagnosed multiple myeloma: results of the phase 3 CEPHEUS study
.
Rio de Janeiro (Brazil)
;
2024
.
[cited 2024 Dec 4]. Available from:
https://events.jspargo.com/ims24/Public/mainhall.aspx.
15.
Berdeja
JG
,
Madduri
D
,
Usmani
SZ
,
Jakubowiak
A
,
Agha
M
,
Cohen
AD
, et al
.
Ciltacabtagene autoleucel, a B-cell maturation antigen-directed chimeric antigen receptor T-cell therapy in patients with relapsed or refractory multiple myeloma (CARTITUDE-1): a phase 1b/2 open-label study
.
Lancet
2021
;
398
:
314
24
.
16.
Moreau
P
,
Dimopoulos
MA
,
Mikhael
J
,
Yong
K
,
Capra
M
,
Facon
T
, et al
.
Isatuximab, carfilzomib, and dexamethasone in relapsed multiple myeloma (IKEMA): a multicentre, open-label, randomised phase 3 trial
.
Lancet
2021
;
397
:
2361
71
.
17.
Voorhees
PM
,
Kaufman
JL
,
Laubach
J
,
Sborov
DW
,
Reeves
B
,
Rodriguez
C
, et al
.
Daratumumab, lenalidomide, bortezomib, and dexamethasone for transplant-eligible newly diagnosed multiple myeloma: the GRIFFIN trial
.
Blood
2020
;
136
:
936
45
.
18.
Kumar
S
,
Paiva
B
,
Anderson
KC
,
Durie
B
,
Landgren
O
,
Moreau
P
, et al
.
International Myeloma Working Group consensus criteria for response and minimal residual disease assessment in multiple myeloma
.
Lancet Oncol
2016
;
17
:
e328
46
.
19.
Maclachlan
KH
,
Came
N
,
Diamond
B
,
Roshal
M
,
Ho
C
,
Thoren
K
, et al
.
Minimal residual disease in multiple myeloma: defining the role of next generation sequencing and flow cytometry in routine diagnostic use
.
Pathology
2021
;
53
:
385
99
.
20.
Kumar
SK
,
Callander
NS
,
Adekola
K
,
Anderson
LD
,
Baljevic
M
,
Baz
R
, et al
.
Multiple myeloma, version 2.2024, NCCN clinical practice guidelines in oncology
.
J Natl Compr Canc Netw
2023
;
21
:
1281
301
.
21.
US Food and Drug Administration
.
Hematologic malignancies: regulatory considerations for use of minimal residual disease in development of drug and biological products for treatment: guidance for industry
.
[cited 2024 Dec 4]. Available from:
https://www.fda.gov/regulatory-information/search-fda-guidance-documents/hematologic-malignancies-regulatory-considerations-use-minimal-residual-disease-development-drug-and.
22.
Landgren
O
,
Devlin
S
,
Boulad
M
,
Mailankody
S
.
Role of MRD status in relation to clinical outcomes in newly diagnosed multiple myeloma patients: a meta-analysis
.
Bone Marrow Transplant
2016
;
51
:
1565
8
.
23.
Munshi
NC
,
Avet-Loiseau
H
,
Rawstron
AC
,
Owen
RG
,
Child
JA
,
Thakurta
A
, et al
.
Association of minimal residual disease with superior survival outcomes in patients with multiple myeloma: a meta-analysis
.
JAMA Oncol
2017
;
3
:
28
35
.
24.
Munshi
NC
,
Avet-Loiseau
H
,
Anderson
KC
,
Neri
P
,
Paiva
B
,
Samur
M
, et al
.
A large meta-analysis establishes the role of MRD negativity in long-term survival outcomes in patients with multiple myeloma
.
Blood Adv
2020
;
4
:
5988
99
.
25.
Cavo
M
,
San-Miguel
J
,
Usmani
SZ
,
Weisel
K
,
Dimopoulos
MA
,
Avet-Loiseau
H
, et al
.
Prognostic value of minimal residual disease negativity in myeloma: combined analysis of POLLUX, CASTOR, ALCYONE, and MAIA
.
Blood
2022
;
139
:
835
44
.
26.
Ficek
J
,
Kalaitzaki
E
,
Yuan
SS
,
Tosolini
A
,
Du
L
,
Kremer
BE
, et al
.
Association of minimal residual disease negativity rates with progression free survival in frontline therapy trials for newly diagnosed multiple myeloma: a meta-analysis
.
Clin Lymphoma Myeloma Leuk
2023
;
23
:
e213
1
.
27.
Gormley
N
,
Bhatnagar
V
,
Ehrlich
LA
,
Kanapuru
B
,
Lee
HZ
,
McKee
AE
, et al
.
FDA analysis of MRD data in hematologic malignancy applications
.
J Clin Oncol
2017
;
35
:
2541
.
28.
Landgren
O
,
Gormley
N
,
Turley
D
,
Owen
RG
,
Rawstron
A
,
Paiva
B
, et al
.
Flow cytometry detection of minimal residual disease in multiple myeloma: lessons learned at FDA-NCI roundtable symposium
.
Am J Hematol
2014
;
89
:
1159
60
.
29.
Food and Drug Administration Center for Drug Evaluation and Research
.
Final summary minutes of the oncologic drugs advisory committee meeting: April 12, 2024
.
[cited 2024 Dec 4]. Available from:
https://www.fda.gov/media/180108/download.
30.
US Food and Drug Administration
.
University of Miami presentation at April 12, 2024 oncology drugs advisory committee meeting
.
[cited 2024 Dec 4]. Available from:
https://www.fda.gov/media/177737/download.
31.
Landgren
O
,
Prior
TJ
,
Masterson
T
,
Heuck
C
,
Bueno
OF
,
Dash
AB
, et al
.
EVIDENCE meta-analysis: evaluating minimal residual disease as an intermediate clinical end point for multiple myeloma
.
Blood
2024
;
144
:
359
67
.
32.
Paiva
B
,
Zherniakova
A
,
Nuñez-Córdoba
JM
,
Rodriguez-Otero
P
,
Shi
Q
,
Munshi
NC
, et al
.
Impact of treatment effect on MRD and PFS: an aggregate data analysis from randomized clinical trials in multiple myeloma
.
Blood Adv
2024
;
8
:
219
23
.
33.
US Food and Drug Administration
.
i2TEAMM presentation at April 12, 2024 oncology drugs advisory committee meeting
.
[cited 2024 Dec 4]. Available from:
https://www.fda.gov/media/177738/download.
34.
Costa
LJ
.
MRD accelerating myeloma drug development
.
Blood
2024
;
144
:
345
7
.
This open access article is distributed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0) license.