Undetectable MRD can change the deal

In this issue of Blood, Goicoechea et al provide an elegant demonstration that achieving an undetectable minimal residual disease (MRD) overcomes poor prognosis in high-risk (HR) multiple myeloma (MM) patients. By analyzing tumor reservoirs persisting after treatment, they also show that clonal selection-induced resistance results from a different mechanism in HR patients.¹ 

In MM, patients displaying HR cytogenetic features have similar levels of complete response as standard-risk (SR) patients, but shorter progression-free survival (PFS) and overall survival (OS), which is why response duration is more important than the depth of response.²  However, thanks to the recent MRD revolution, it is possible to evaluate the response to treatment with unprecedented sensitivity; consequently, it is possible to better identify patients who will progress soon among those in a conventional complete response.³  As expected, SR patients are more likely to achieve undetectable MRD, but importantly, some HR patients can also achieve undetectable MRD, and that may change the deal.^4,5 

In this issue of Blood, Goicoechea et al measured MRD by next-generation flow (NGF) in transplant-eligible patients after induction by the triplet of bortezomib, lenalidomide, and dexamethasone, in a clinical trial. Half of the SR patients achieved undetectable MRD vs just over one-third of the HR patients. Among the HR patients, those displaying translocation t(4;14) were more likely to achieve undetectable MRD than those displaying deletion 17p. Survival data confirmed that achieving an undetectable MRD (<3 × 10⁻⁶, namely <3 MM cells among 1 million analyzed bone marrow cells) overcomes poor prognosis of the HR myeloma patients, because their PFS and OS were similar to those of the SR patients, including the few patients with deletion 17p. Conversely, when MRD was persistent, SR patients did better than HR patients (see figure). Was this because the number of residual cells was lower in SR patients? Not really: persistent MRD levels were not so different according to risk group. Actually, similar levels of MRD cells translate into poorer PFS in HR patients, suggesting a different mechanism for resistance.

To address this apparent paradox, the authors took the advantage of NGF technology to sort residual tumor cells and compare them to diagnostic cells from the same patients. A couple of years ago, the same group was the first to explore clonal evolution by analyzing paired samples obtained at diagnosis and at the MRD endpoint. They showed that the different modalities of clonal selection (stable, linear, branched) were already detectable at the MRD stage, where resistant MM cells displayed a specific genetic expression profile characterized by an overexpression of integrins, adhesion molecules, and chemokine receptors, which could be interpreted as reflecting microenvironment hijacking by MM cells.⁶  Here, whole-exome sequencing was done for paired samples (diagnosis, MRD), and the data obtained suggest a greater clonal selection in SR patients, but more new mutations in HR patients, probably reflecting higher genomic instability. However, although there was similar therapeutic pressure in all the patients, no unifying loss or gain abnormalities driving MRD were identified, as previously reported.⁷  RNA sequencing suggests that there is reprogramming of unrelated genetic background residual cells, with enrichment of resistance transcriptional programs, including quiescence and specific resistance to proteasome inhibition. Given the biological heterogeneity of MM, focusing on distinct subgroups may be necessary to improve our understanding of resistance. For instance, RNA sequencing performed here showed a distinct “MRD phenotype” for HR patients with half the deregulated genes compared with SR patients, and a specific enrichment of the reactive oxygen species pathway. Accordingly, a high expression of 2 antioxidant genes, SOD1 and PRDX6, was independently associated with lower PFS, including in an external validation data set. The only slight disappointment in this study is that single-cell (sc) RNA sequencing is of limited value. In theory, sc analysis should give unprecedented clarity of the subclonal nature of tumors, including MM, and how it evolves after treatment.^8,9  However, in practice, the costs incurred and the technical difficulties inherent in the MRD endpoint drastically limit the number of samples analyzed in this way.

Hence, prognosis of MM is established at diagnosis, and MRD can change the deal during treatment. Does it mean that risk is a “dynamic concept”? The reality is that we are sometimes wrong because we miss the prognostic impact of additional, poorly characterized factors, such as the impact of the immune microenvironment.¹⁰  MRD gives us the opportunity to refine our prognostic models, until that faraway day when risk will be perfectly predicted from diagnosis, and provide the information needed to provide perfectly adapted treatment. However, back to 2020, where Goicoechea et al bring another argument that MRD should be the clinical endpoint in MM, especially for HR patients. If MRD remains or reverts positive, an early therapeutic intervention may be essential. This study also paves the way to understand how to target these residual cells, a key next step in the road to cure myeloma.