Background:

MDS are characterized by ineffective hematopoiesis, cytopenias, and clonal expansion, often progressing from lower-risk to higher-risk MDS or acute myeloid leukemia, leading to death. The newer IPSS-Molecular (IPSS-M) categorizes patients by progression risk through the assessment of hematologic parameters, bone marrow blasts, cytogenetics, and gene mutations. Red blood cell (RBC) transfusion dependency (TD) is a prognostic factor in MDS; thus, therapies that reduce TD may improve OS. In the phase 3 COMMANDS trial (NCT03682536), luspatercept demonstrated superiority for the primary endpoint (PE) of transfusion independence (TI) vs ESA epoetin alfa, including a trend for improved OS. We have reported that patients treated with luspatercept had more frequent IPSS-M risk downstaging during the treatment (tx) period vs ESA due to improved hemoglobin (Hb) levels. Luspatercept tx did not impact gene mutations or variant allele frequency changes, highlighting the need to understand the broader mechanistic attributes related to tx response.

Aim:

To characterize exploratory biomarkers linked to hematologic response to luspatercept and its potential association with improved OS vs ESA.

Methods:

The COMMANDS study design was previously described (Platzbecker, Lancet. 2023). The PE was RBC-TI ≥12 wks with a concurrent Hb increase ≥1.5 g/dL (Wks 1-24). A total of 206 baseline parameters, including cytomorphology, reticulocyte count, Hb, soluble proteins, cytokines, and molecular features (82-panel targeted sequencing [≥400× exon coverage; 3% sensitivity] and whole genome sequencing [80×]), were analyzed. For each tx arm, logistic and Cox univariate regression were used to identify predictors of achieving the PE and longer OS, respectively (statistically significant variables at P<0.05). The most significant predictors (lowest P-values) were selected for the initial multivariable model. A stepwise selection process was then used to add additional predictors and build the final model (using R).

Results:

Key baseline variables including IPSS-M, number of mutations, mutational burden, soluble proteins, and patient characteristics were balanced between tx arms (luspatercept, n=182; ESA, n=181). Luspatercept demonstrated superiority vs ESA for RBC-TI (PE; 60% vs 35%) with a trend for improved OS (median, not estimable vs 46 mo).

Univariate analysis identified significant explanatory variables across both tx arms (P<0.05) for the PE (n=35) and OS (n=40). For the luspatercept arm, significant PE predictors were lower IPSS-M and erythropoietin (EPO) levels, and higher alpha-2 macroglobulin (P<0.001). In the ESA arm, significant PE predictors were lower plasma iron, EPO, and GDF-15 levels (P<0.001); lower IPSS-M was not a predictor for the PE (P=0.05). Significant univariate predictors for improved OS with luspatercept included lower IPSS-M (HR, 0.27), CBL WT status (HR, 0.12), and SF3B1 mutation (HR, 0.35). For ESA, significant features were lower IPSS-M (HR, 0.34), higher platelets (HR, 0.3), and ASXL1 WT status (HR, 0.24).

Stepwise regression analysis showed IPSS-M did not reach significance for either tx arm, as other variables outweighed its predictive power due to the composite nature of this score. For luspatercept, significant predictors of improved PE were higher alpha-2 macroglobulin (P=0.007) and transferrin saturation (P=0.04), and lower plasma iron (P=0.01) and hepcidin (P=0.03). Lower NT-proBNP (P<0.001), higher white blood cell count (P=0.0004), CBL WT status (P=0.002), and higher VEGF (P=0.01) were associated with improved OS with luspatercept. For ESA, significant predictors of improved PE were higher Hb (P=0.02) and myoglobin (P=0.04), and lower iron (P=0.01). Lower NT-proBNP (P<0.0001), IDH2 WT (P=0.0001), ASXL1 WT (P=0.0005), higher platelets (HR, 0.54; P=0.01), and lower hepcidin (P=0.01) were significant predictors for improved OS with ESA.

Summary:

Predictive biomarkers for the PE and OS, such as NT-proBNP and iron levels, were shared across both tx arms. Luspatercept response was linked to stage of erythropoiesis and iron metabolism biomarkers including alpha-2 macroglobulin, hepcidin, and transferrin, consistent with the mechanism of action, better iron regulation, and lower inflammation. ESA response was associated with higher Hb and platelets, indicating a greater dependency on effective hematopoiesis. Additional longitudinal analyses are warranted to refine these predictive biomarkers.

This content is only available as a PDF.
2025
Sign in via your Institution