Explain the unexplained Free

In this issue of Blood, Beer et al¹ assess risk in patients with myeloma through both genetic and transcriptional profiling and demonstrate that a comprehensive evaluation delivers a knockout blow to the concept of functional high risk.

A 57-year-old woman was diagnosed with myeloma. She had no comorbidities or clinical or molecular high-risk factors and presented with classical hyperdiploid myeloma. After optimal quadruplet-based induction, she received an intensive high-dose melphalan and autologous stem cell transplant, followed by quadruplet-based consolidation. She achieved complete response and began lenalidomide maintenance. After 6 cycles, she suddenly presented with disease progression. This is the story of “functional high-risk” patients. Despite therapeutic progress, high-risk myeloma remains an unmet medical need. Cooperative groups recently proposed clinical trials dedicated to this population which have yielded encouraging results.^2-4 However, these risk-adapted strategies make sense only if we are able to correctly identify high-risk patients from the time of diagnosis. When this is not the case, we refer to these patients as “functional high risk,” an elegant way of describing what we are unable to explain, as no molecular or clinical high-risk factor were detected at the time of diagnosis. The proportion of functional high-risk patients varies considerably from one study to another (roughly one-third of the early relapses) and depends heavily on how risk is assessed at diagnosis, which may be comprehensive and homogeneous. Recently, the International Myeloma Society and the International Myeloma Working Group (IMS/IMWG) proposed a genomic staging consensus to define high-risk patient.⁵ Thanks to the incorporation of rare high-risk subgroups such as TP53 mutation without any 17p deletion, the number of functional high-risk cases may decrease. Along with DNA exploration, gene expression profile (GEP) has been widely explored in myeloma and some studies already demonstrated that transcriptome analysis captures some poor prognosis patients who are missed by other methods.⁶ 

Beer et al assessed risk by GEP (SKY92) and by the new IMS/IMWG consensus criteria in newly diagnosed transplant eligible patients with myeloma from the Myeloma-XI trial. They observed that out of 135 patients, 25 (18.5%) experienced early relapse, defined here as disease progression occurring before 18 months after maintenance initiation. Among these 25 patients, 15 were classified high-risk according to the IMS/IMWG genomic staging consensus, including 8 patients (all with 1q gain), who were also characterized by GEP–high risk (HR). Six additional patients were GEP-HR only. Two other patients were standard risk (SR) by both methods, but were not strictly speaking functional high-risk, because they had only 1 “intermediate” cytogenetic abnormality (namely those that must be associated to each other to confer poor prognosis): a gain 1q for one patient, and a t(4;14) for the other one. For this former, one hypothesis is that some isolated t(4;14) may be high-risk due to a late breakpoint in NSD2.⁷ It could also be interesting for these patients to have their risk reassessed at the time of relapse: could a little HR subclone, undetectable at bulk level, have been selected quickly by treatment and changed the risk?⁸ 

In the past decade, measurable residual disease (MRD) assessment in myeloma has become essential, given its strong prognostic value. When a patient with SR remains MRD-positive during treatment, this can be considered as a form of functional high-risk. Interestingly, the 2 remaining “true” functional high-risk patients from this study (only 1.5% of the all cohort) both had a t(11;14), which is considered neutral or even favorable from a prognostic standpoint. Recently, this subgroup of patients has attracted attention of the myeloma community because of its reduced ability to achieve MRD-negativity, particularly after induction, suggesting a slower response to treatment.⁹ This observation led us to modulate the prognostic weight of MRD in myeloma according to the (genomic) context. Here, both patients relapsed early, suggesting that some t(11;14) patients have poor outcomes despite the absence of known high-risk genomic or transcriptomic factors. Either we are missing something that confers high-risk, or, for reasons we need to better understand, some of these patients may need a specific first-line treatment such as BCL2 inhibitor.

When they classified the entire cohort of 135 patients according to IMS/IMWG consensus and GEP, Beer et al demonstrated the clinical utility of transcriptome in myeloma stratification: 10% of patients were IMS/IMWG SR but GEP-HR and potentially misclassified. In addition, the combination of both methods predicted 80% of early relapse and resulted in the best compromise between sensitivity and specificity. In all prognostic models, there is usually a trade-off to be reached: is it “better” to incorrectly identify standard-risk patients as high-risk (lack of specificity), or to miss high-risk patients (lack of sensitivity)? The second option is probably the least desirable.

In this study, patients were not exposed to anti-CD38 therapy, and we know very well how much this class of drugs significantly improved patient outcomes. This is an unsolvable problem when studying prognostic factors; clinical follow-up is needed because we are always lagging behind with prognostic models based on patients treated with therapies that are already somewhat obsolete. Hence, this study should be repeated in patients treated with the current best therapy.

Although not foolproof, MRD assessment can help us continue to identify functional high-risk patients in real time, preventing them from falling through the net. Nevertheless, many avenues continue to be actively explored to improve risk assessment at the time diagnosis, one of the most promising areas likely being the evaluation of circulating tumor cells.¹⁰ In the near future, if every single high-risk patient can be identified at the time of diagnosis, thereby eliminating the concept of functional high-risk, it will clearly be a step forward. However, true progress will consist of finding therapeutic strategies to abrogate the poor prognosis of these well-identified high-risk patients.

Conflict-of-interest disclosure: J.C. reports consultancy, honoraria, and travel fees from Janssen, Sanofi, Bristol Myers Squibb, Pfizer, and Adaptive and research support from Sanofi and Bristol Myers Squibb.