SMM, an asymptomatic precursor condition of MM, has significant heterogeneity in disease progression [17]. The paradigm shift in strategies to manage SMM in the last decade has highlighted the importance of early identification of patients with SMM at the greatest risk of progression to active MM [18, 19].

While the role of treatment in high-risk SMM is being evaluated in various clinical trials, interpretation of results is complicated by variations in the risk-stratification criteria used [7]. Several risk-stratification models have been developed, including the AQUILA trial [10], Mayo 2018 [11], IMWG 2020 [13]. While accurately characterizing patients with high-risk SMM allows for appropriate care and management of these patients, identifying patients with low-risk SMM also avoids oversurveillance and overtreatment [20].

Historically, the standard of care for SMM has been observation until patients develop active MM, regardless of their risk status [2, 7, 8]. However, results from two randomized controlled clinical trials in high-risk patients with SMM support the benefit of early therapy with lenalidomide, alone and in combination with dexamethasone, in reducing the risk of progression to active MM [21, 22].

This study leveraged data from the Flatiron Health database to assess baseline characteristics and outcomes in patients with SMM who were stratified by their risk for progression to active MM. The IMWG 2020 criteria build upon the Mayo 2018 criteria to further characterize patients as high risk through the incorporation of genetic factors, including t(4;14), t(14;16), +1q, and/or del13q/monosomy 13 abnormality [13]. The use of multiple criteria from several models to stratify risk enhanced the robustness of our results as it considered a wide array of factors comprehensively, such as clonal BMPC; serum M-protein; immunoglobulin isotype and immunoparesis; FLCr; and t(4;14), t(14;16), +1q, and/or del13q/monosomy 13 abnormality.

Depending on the stratification criteria used, 21.3%–61.0% of patients with SMM in the Flatiron Health database who could be risk-stratified were identified as having a high risk of progression to active MM. The variation in the proportion of patients who were identified as high-risk could be due to the IMWG 2020 criteria requiring the inclusion of genetic abnormalities as a stratification criterion, which is often unavailable in real-world databases. The inclusion of such information would require real-world databases to provide information not only on the receipt of the relevant genetic testing but also on the test results, whether negative or positive. While each of the three risk-stratification models used different approaches to identify high-risk patients, TTP, PFS, and PFS2 in these patients were generally similar, although time to each of these outcomes was somewhat longer in patients stratified using the AQUILA trial criteria at data cut-off. This may have been the result of the broader definition of high risk with this model, and the model may require further optimization to provide the pertinent treatment to the appropriate patients. High-risk patients with SMM were 3.2–4.9 times more likely than non-high-risk patients to progress to MM, 2.4–3.9 times more likely to progress to MM or die, and 2.1–4.0 times more likely to die or progress on 1 L MM treatment. The consistency of outcomes between the risk models provides additional confidence to these findings. Similar results have been reported previously by Visram et al., who assessed outcomes in patients who were risk-stratified using the Mayo 2018 and IMWG 2020 criteria up to 5 years post-diagnosis. In addition, a three times higher risk of progression was reported in patients who evolved to a higher-risk category during the follow-up period compared with patients with an unchanged or decreased risk category [19]. More recently, the Mayo 2018 criteria were used to risk stratify patients with SMM from the Czech Myeloma Group Registry of Monoclonal Gammopathies, and results of that study demonstrated that patient outcomes, including PFS and PFS2, were significantly worse in high-risk versus non-high-risk patients [23].

Similar results were seen in the sensitivity analyses wherein patients were adjusted for age, CCI, time between SMM diagnosis and index data, race, creatinine, and hemoglobin levels, and when data were censored from patients who died more than 180 days and 365 days after the last observation. It was considered a possibility that some patients who were classified as SMM in the Flatiron Health database may have had MM. However, when these patients were excluded in the sensitivity analyses (had received early treatment, had early progression, and/or had bone disease diagnosis), consistent results were observed.

This study does have a few limitations. Firstly, the Flatiron database does not systematically capture risk factors over time. While a small number of patients had received treatment for SMM, this information was not curated; SMM-specific treatments were not provided. As such, no definitive conclusions can be drawn about SMM treatment. In addition, the cause of death was unavailable. Finally, there was a high proportion of patients with SMM with data in the Flatiron database who were not able to be classified, and, as a result, these data may not fully represent the entirety of this patient population.

Each risk-stratification model evaluated in this study identifies a cohort of patients with SMM with an increased risk of progression to active MM and patients who overlap across the models. As has been stated previously, this is not considered a limitation. Patients identified as high risk by any of these models (e.g. IMWG identified high-risk as patients with 50% progression risk at 2 years from diagnosis based on the levels of M-protein, BMPC infiltration, or sFLC ratio) should be considered to have an increased risk of progression to active MM. Even though the models assessed in this study define high risk differently and may not be completely comparable, the benefit of the parameters used in the risk-stratification models is that they are widely available and can be used in practice and in clinical trials worldwide. We expect that models predicting risk in patients with SMM will continue to evolve, and as they do so, patients will need to be tested according to these criteria to be captured in real-world studies.

To our knowledge, this is the first real-world study evaluating risk in patients with SMM using the criteria defined in the ongoing AQUILA trial in comparison with other commonly used models. The AQUILA criteria were able to distinguish well between patients with and without a high risk of progression, and the results suggest a need for intervention in high-risk patients.