Measurable residual disease in multiple myeloma: ready for clinical practice?

The outcome of patients with multiple myeloma (MM) has improved significantly in the last 20 years. This was the result of more than eight novel agents incorporated into the treatment armamentarium of MM, which led to unprecedented rates of complete remission (CR) and prolonged survival. In fact, we are now in a position to discuss whether MM may become a curable disease, which was beyond imagination a few years ago. Eradicating all tumor cells is a prerequisite to cure most malignancies, which raises the need of using high-sensitive tools to evaluate treatment efficacy. Although the definition of CR in MM is very useful in clinical practice, its sensitivity is suboptimal in many patients since current criteria relies on traditional techniques such as serum immunofixation and plasma cell (PC) enumeration by morphology that does not discriminate between normal and tumor cells. Adding immunohistochemistry or immunofluorescence does not improve cytological analysis and its sensitivity is low (10⁻²) due to the recovery of normal PCs after therapy that normalize kappa/lambda ratios. Furthermore, the serum free light-chain ratio has proven to be of limited value to discriminate patients in CR at different risk of progression and in fact, the stringent CR definition has failed to improve risk-stratification beyond conventional CR [1‐3]. Therefore, the words “complete”–“remission” are misleading for many patients because they may interpret that, once achieved such status, the disease has been eradicated. Thus, it becomes evident that more sensitive techniques are needed to detect measurable (formerly called minimal) residual disease persisting below CR. Ideally, this would contribute to evaluate treatment efficacy with exquisite resolution (one that matches the high efficacy of new regimens) and to avoid both over and under treatment. Unfortunately, there is still a marked imbalance between the extraordinary therapeutic progress and the use of laboratory tests to monitor patients and, accordingly, to individualize treatments decisions in MM.

If response to therapy is one of the most, if not the most, effective marker to predict survival, who would not want to know with high precision, the quality of patients’ response to therapy? Should we ignore biological information with clear correlation with outcome?. We are now in 2020, but almost 20 years ago there was already evidence about the prognostic impact of persistent MRD in CR patients; should we wait for another two decades or should we implement this information to investigate innovative therapeutic interventions and to individualize patients’ management?

MRD techniques can be divided into those identifying extramedullary disease (e.g., positron emission tomography/computed tomography (PET/CT)) and those detecting intramedullary disease by either multiparameter flow cytometry (MFC) immunophenotyping or molecular assessment of immunoglobulin gene rearrangements. Using MFC, we can identify myelomatous PC based on light-chain clonality of phenotypically aberrant tumor cells. Initial MFC approaches (with a sensitivity of 10⁻⁴ and no standardization) [4, 5] have evolved into next-generation flow (NGF) cytometry developed by EuroFlow, which is based on optimized monoclonal antibodies combinations and sample preparation protocols that overcome blocking or internalization of monoclonal antibodies targeting PC antigens such as CD38, the acquisition of ≥ 10⁷ nucleated cells per sample, and novel software tools allowing for automated analysis with an expected sensitivity of 2 × 10⁻⁶ [6]. A similar evolution was observed on molecular grounds, where clonal immunoglobulin gene rearrangements (the unique ID of myelomatous PC) were initially identified by laborious and low-applicable ASO-PCR techniques and are now detected by next-generation sequencing (NGS) that performs millions of reads of DNA fragments in a standardized fashion with a sensitivity of 10⁻⁶ [7]. Both NGF and NGS have advantages and disadvantages for MRD detection that have been enumerated elsewhere [8, 9], but yield similar clinical results [10, 11] if used according to the guidelines of the International Myeloma Working Group (IMWG) [9]. NGS has been standardized through commercial kits developed by some companies and can be performed in frozen samples, which is an advantage for large multicenter clinical trials; NGF does not require baseline samples, allows evaluation of the whole bone marrow (BM) cellularity (e.g., hemodilution) and results are available in few hours. While both NGF and NGS supersede the performance of previous immunophenotypic and molecular methods, patients with undetectable MRD by any of these technologies continue to show a linear risk of relapse [12]. Thus, further improvement in the sensitivity of NGF and NGS are warranted to optimize risk-stratification based on patients’ MRD status. PET/CT is currently the optimal method to evaluate the disease outside the BM and there are ongoing efforts for its standardization [13]. Fluorodeoxyglucose is the most widely used radiolabeled compound but others such as methionine are under investigation [14]. PET/CT evaluation of treatment efficacy correlates with patients’ PFS [15‐17]. Furthermore, studies from the IFM and University of Arkansas demonstrated complementarity between PET/CT and flow cytometry for risk-stratification [16, 18]. A recent analysis of PETHEMA/GEM uncovered that approximately half of patients with undetectable MRD developing early progression, some of them with extra-osseous plasmacytomas at diagnosis, presented new plasmacytomas as an isolated criterion of disease progression, without detectable M-protein or BM infiltration. Thus, it appears that these were true false-negative MRD results, reinforcing the need to combine NGF or NGS with PET/CT to monitor treatment efficacy, particularly in patients presenting with extramedullary or macro-focal disease, as well as elevated LDH levels [19].

Here, we will discuss critical aspects related with the implementation of MRD in clinical practice.

We considered the following prerequisites to evaluate if undetectable MRD can be used as treatment endpoint in MM: (1) must supersede the prognostic value of CR; (2) must provide reproducible results irrespectively of methodology and disease setting; and (3) must be applicable to all patients.

All the above suggests that MRD meets the key requirements to become a treatment endpoint in MM. However, the potential pitfalls of MRD techniques should be recognized and have been summarized below in four items:

Since MRD is one of the most (if not the most) relevant prognostic factor, we should take advantage of this information to improve patient management (including both for innovative clinical trial—e.g., Table 1—design and in clinical practice). Naturally, MRD assessment should be performed only when a BM aspirate is collected to confirm CR, in accordance to the IMWG guidelines [9]. From thereafter, MRD testing should be performed whenever such results could help on clinical decisions (e.g., in between treatment stages) and repeated periodically (e.g., every 1 or 2 years) to confirm patients’ MRD status. First, it is important to clarify that while, as discussed above, an MRD-negative result still has a certain degree of uncertainty, persistence of MRD is a strong adverse prognostic feature, even among CR patients. Accordingly, it will be safer to make clinical decisions based on persistent MRD than on undetectable MRD.

How to take advantage of the higher sensitivity of modern MRD techniques to evaluate treatment efficacy and to guide therapeutic decisions? Table 1 shows the ongoing clinical trials that use MRD assessment using next-generation techniques. As a first example, if a patient is in CR before ASCT, how to evaluate the efficacy of subsequent high-dose therapy? In this context, you may be guided by the effect of high-dose therapy on persistent MRD. Similarly, if the patient is reluctant or is not a candidate for ASCT and has achieved CR, why not continue with additional cycles of consolidation until MRD becomes undetectable before moving to maintenance? Second, in high-risk patients with persistent MRD following an optimized induction plus ASCT, we know median PFS will be very short (typically less than two years) [21, 22]; accordingly, it could be envisioned that the introduction of novel agents such as monoclonal antibodies plus second/third generation of PI/IMIDS after ASCT may produce benefit; noteworthy, this “risk-adapted therapy approach” is being tested in some trials. Third, if we know that treatment “A” induces three-fold higher MRD-negative rates as compared to treatment “B,” should this influence my clinical practice? Fourth, to adapt maintenance intensity and duration. Several clinical trials are investigating this concept; for example, the RADAR study from the UK group segregates MRD-positive and MRD-negative patients after ASCT: in the first cohort, they will compare 1 vs 2 vs 3 drugs (IMID-PI-MoAb), while in MRD-negative cases, they will explore treatment until disease progression versus fixed duration. Similarly, in the Spanish GEM2014MAIN trial, after 2 years maintenance patients were randomized according to MRD: if positive, they continued for 3 years but if negative, they stopped treatment [19]. These are selected examples out of many other trials with similar conceptual design, all oriented to stop maintenance if the patient is MRD-negative and continue if positive. However, it can be argued that if patients remain MRD-positive after optimal intensive treatment (including 3–4 drugs), maintenance with a single agent will be of limited value and probably, these cases may benefit from an experimental approach (e.g., individualized immunotherapy according to tumor and immune cell biology). By contrast, if patients have undetectable MRD, standard maintenance approaches may effectively maintain immune surveillance and sustain undetectable MRD for long periods of time. Accordingly, data from most recent studies suggest that patients with undetectable MRD are the ones that (as opposed to cases with persistent MRD) benefit the most from maintenance therapy [19, 22, 25, 26]. We believe this is the surrogate biomarker of cure in MM, and trials designed to address these concepts are of utmost importance (Table 1). In fact, the notion that MRD can act as surrogate biomarker for survival and thereby accelerate drug development is evolving based on consistent and positive results observed in recent years (Table 2), and a progressive number of clinical trials are using MRD rates as primary endpoint (Fig. 1).

The longer an undetectable MRD status is sustained, the higher its impact in reducing the risk of progression and prolonging survival of MM patients. This requires periodic MRD assessment and invasive BM aspirates pose a challenge. Thus, further methodological innovation is warranted to monitor MRD in blood as frequent as possible. Of note, promising results have been recently reported using NGS [35] and NGF [36] in peripheral blood (PB). Namely, the EuroFlow consortium as recently reported that with NGF, it was possible to detect MRD in PB of 17% of patients in CR; most importantly, presence of MRD in PB identified a subgroup of patients in CR with dismal outcome (median PFS of 9 months) [37]. Conversely, both studies using NGS and NGF showed that approximately 40% of patients displayed MRD in BM that was undetectable in PB [35, 36]. There is also growing evidence supporting circulating tumor DNA (ctDNA) for liquid biopsies in MM. However, this approach suffers from a conundrum between applicability and extent of genetic information: while targeted sequence of a few genes or hotspot mutations is highly applicable, comprehensive whole-exome sequencing of cfDNA is possible in a small number of patients with high ctDNA burden [38‐40]. Thus, these approaches do not seem to be powered for sensitive MRD assessment in all MM patients. By contrast, matrix-assisted laser desorption ionisation time-of-flight mass spectrometry that detect M-proteins in serum has shown to be more sensitive compared to current electrophoretic methods [41, 42]. In fact, most recent observations suggest that this method may provide complementary information to MRD assessment in BM [43, 44]. More studies are needed to define if this concept is ready for prime time. We believe that a minimally-invasive MRD test will foster its use in clinical practice, particularly for preemptive therapeutic approaches upon MRD reappearance in PB. However, at the moment, we consider that BM remains the gold-standard sample for MRD testing.

We must not forget that under the pressure of enormous drug costs, the best way to make our health systems sustainable is by curing MM. We have experienced great progress and now we need to optimize the use of highly effective drugs developed including emerging immunotherapeuties. This should be implemented early in the course of the disease in order to overcome the poor prognosis of high-risk patients, including those with persistent MRD after optimal frontline treatment. In other words, “early detection of the problem guided by sensitive methods to allow early intervention.”