Minimal Residual Disease as a Primary Endpoint in Multiple Myeloma: Implications for Clinical Trials and Everyday Clinical Practice

In a given clinical trial, demonstrating an OS benefit in most cancers has required not only a very favorable efficacy and safety profile, but also quite large sample sizes and prolonged follow-up. Now, in MM, with ever-increasing salvage options — including immunomodulatory agents, proteasome inhibitors, naked monoclonal antibodies, XPO inhibitors, and the recent explosion of chimeric antigen receptor T-cell therapy and bispecific antibodies — any OS impact of a given intervention could be completely altered by differential access and application of such salvage therapies. Therefore, while any potential negative impact on OS should always be carefully monitored, the ability to demonstrate an OS benefit — particularly early on in the disease course — is an increasingly elusive goal.

Health authorities have accepted PFS as a primary endpoint for the vast majority of MM studies (with smoldering myeloma and maintenance therapies after stem cell transplantation being notable exceptions that are beyond the scope of this discussion). Recent studies have demonstrated unprecedented results, such as the 61.9-month PFS achieved in the arm of the phase III MAIA study in which patients received a combination of daratumumab, lenalidomide, and dexamethasone.³  In the phase III IMROZ trial, PFS was not reached at five years in the study arm receiving quadruplet therapy comprising isatuximab, bortezomib, lenalidomide, and dexamethasone.⁴  However, even with these regimes, the vast majority of myeloma patients are not being cured. If PFS remains the primary endpoint, novel drug development (particularly in NDMM and early relapsed MM), as well as the search for cures, will take a prohibitively long time.

MRD testing provides the opportunity to precisely and quickly compare novel drugs and combinations. Currently, novel agents seeking accelerated approval are often first studied in single-arm trials in RRMM, with a primary endpoint of overall response rate (ORR) and PFS, before being evaluated in a randomized clinical trial (RCT) in earlier lines of therapy. While the vast majority of MM agents receiving accelerated approval have gone on to receive full registrational approval after phase III RCTs, the current model has several limitations. The push for single-agent efficacy via ORR and PFS results in increasing the dose intensity of these novel agents to a level that is actually not practically deliverable in most real-world patients. It is also impossible in a single-arm study to determine whether the efficacy and safety profile of a given agent is due to certain patient and disease factors or the investigational treatment at hand. With the new ODAC decision, we can move away from this model and instead directly design randomized controlled trials using MRD as a primary endpoint for accelerated approval.

The International Myeloma Working Group (IMWG) defines MRD negativity as the absence of clonal plasma cells at a sensitivity threshold below 10^-5, as assessed by flow cytometry or next-generation sequencing (NGS). There are key differences between these techniques. Notably, NGS requires the presence of a baseline clone, which can be absent in 3 to 20% of samples depending on the NGS platform used.⁵  Flow cytometry needs to be processed quickly, typically within 48 hours.⁶  Further, as myeloma is inherently patchy, bone marrow biopsy results can vary depending on the location and quality of the specimen obtained. Hemodilution can result in false-negative MRD results; therefore, the first pull off the aspirate should be used for MRD testing.⁷ 

The sensitivity threshold chosen is also of critical importance. A large meta-analysis demonstrated that while achieving MRD at any sensitivity threshold (10^-4, 10^-5, or 10^-6) was associated with improved survival, both PFS and OS were most improved with MRD negativity at the 10^-6 threshold.⁸ 

Some patients who achieve MRD negativity in the bone marrow may still have detectable serological, urine, or radiologic findings. A retrospective review demonstrated that the presence of detectable serum immunofixation in patients deemed MRD negative via EuroFlow testing correlated with shorter time to progression.⁹  Particularly in patients with focal bony lesions or extramedullary disease and minimal marrow burden, MRD testing should be used in conjunction with sensitive imaging techniques such as positron emission tomography/computed tomography or magnetic resonance imaging.¹⁰  We anticipate that improvements in techniques to measure paraproteins via novel technologies such as mass spectrometry, circulating tumor plasma cells, or cell-free DNA may decrease the need for premature and invasive bone marrow sampling for confirmation of MRD status, as well as help bridge the gap in accurately understanding disease burden in patients primarily having bony or extramedullary disease.⁶  Further, the fact that even MRD-negative patients relapse tells us that this is not a surrogate for a cure. In particular, MRD negativity does not address the immune microenvironment in patients that likely also plays a role in long-term disease control.

MRD resurgence can occur after achieving MRD negativity. In the MAIA and ALCYONE trials in transplant-ineligible NDMM patients, those who achieved sustained MRD negativity at six and 12 months had the longest time to next therapy.¹¹  In an analysis of 306 evaluable patients in the FORTE trial who achieved MRD negativity, 39% had MRD resurgence at a median follow-up of 50.4 months, with the presence of high-risk cytogenetics, baseline high-circulating tumor cells (>0.07%), and delayed time to first MRD negativity predicting whether MRD would be sustained.¹²  In MRD analysis from the GEM2012MENOS65 and GEM2014MAIN trials, 42% of patients had MRD resurgence, with high-circulating tumor cells (≥0.01%), International Staging System grade 3, and failure to achieve MRD negativity after induction predicting MRD resurgence.¹³  Therefore, striving for MRD negativity at a single time point may be short-sighted. Whether we are designing clinical trials or caring for patients in the clinic, we should aim to sustain MRD as we work towards a myeloma cure.

As demonstrated by the unanimous ODAC vote, study after study has demonstrated MRD negativity as a prognostic variable to predict PFS and OS. However, not all MRD is created equal. MRD results need to be used in conjunction with broader clinical context, such as those with the monoclonal gammopathy of undetermined significance-like phenotype (where residual positive MRD clone may not have a negative prognostic impact) versus those with high-risk cytogenetic abnormalities/extramedullary disease. In the Myeloma XI trial, high-risk MRD-negative patients had shorter PFS and OS compared to standard-risk MRD-negative patients, suggesting that achieving MRD negativity was insufficient to reverse the impacts of high-risk cytogenetics.¹⁴  Perhaps more concerning is the MASTER study, in which despite having sustained MRD negativity, high-risk patients relapsed upon discontinuation of therapy.¹⁵ 

While MRD testing has clear prognostic value and can inform us on expected outcomes when comparing two different arms of an RCT, its utility in supporting treatment decisions for an individual patient at the bedside remains unclear. For example, in the phase II GRIFFIN and phase III PERSEUS studies, which respectively compared the use of bortezomib, lenalidomide, and dexamethasone with and without the addition of daratumumab in transplant-eligible NDMM patients, the PFS curves did not separate during the induction period but only later with continued therapy.^16,17  The rates of MRD negativity also increased during maintenance, and the difference in MRD negativity rates between the study arms widened over time during maintenance, favoring the addition of daratumumab. Therefore, it is currently unclear if quadruplet-treated patients who attain MRD negativity at one point after stem cell transplant would enjoy the same PFS demonstrated in the studies if they discontinued daratumumab. In fact, updated data from the CASSIOPEIA trial recently demonstrated that the best PFS was experienced with the continuous use of daratumumab both before and after transplant.¹⁸  Another appealing option is to use risk status as an additional deciding factor regarding ongoing anti-CD38 therapy. However, IMROZ study data suggest that standard-risk patients benefited more than high-risk patients from the addition of frontline isatuximab.⁴ 

Therefore, to take MRD from a prognostic tool to a predictive one, there must be more prospective response and/or risk-adapted randomized studies with prespecified landmark analyses, such as DRAMMATIC and AURIGA.^19,20  Until then, caution should be used in applying MRD testing (particularly at a single time point) in everyday clinical practice to escalate or de-escalate therapy. As the body of evidence builds on using MRD to guide clinical decision making, we hope to one day confidently decrease therapy duration or intensity in appropriate patients and escalate therapy when clinically indicated.

Dr. Kumar has received research funding from Bristol Myers Squibb and Janssen and will be receiving an honorarium from Sanofi for a speaking engagement. Dr. Chung has received research funding from Abbvie, Bristol Myers Squibb, Caelum, CarsGen, Cellectis, Janssen, K36 Therapeutics, and Merck Consulting, and he reported consulting activity for Janssen. Dr. Chari has received research funding from Janssen, and he reported consulting activity for Abbvie, Adaptive, Amgen, Antengene, Bristol Myers Squibb, Forus, Genentech/Roche, GlaxoSmithKline, Janssen, Karyopharm, Millenium/Takeda, and Sanofi/Genzyme.