Impact of Mutational Burden and IPSS-M on Response to ESAs in Lower Risk Myelodysplastic Neoplasms

Abstract

INTRODUCTION Lower risk myelodysplastic neoplasms (LR-MDS), as defined by IPSS-R, are frequently presenting with anemia which is primarily treated with erythropoiesis stimulating agents (ESAs). However, response rates are suboptimal due to primary resistance or subsequent loss of response, leading to transfusion dependency and poor quality of life. In 2022 a new prognostic system has been developed considering number and type of somatic mutations in MDS (IPSS-M), showing a stronger prognostic power in MDS. AIMS We aimed to assess whether IPSS-M risk score could predict ESA response in IPSS-R LR-MDS patients. METHODS In this multicentric European study we evaluated baseline IPSS-M score in 397 LR-MDS cases from 14 different centers (Italy, Germany, France and Spain) and correlated with ESA response assessed with IWG criteria. Mutational data were obtained by different next generation sequencing (NGS) panels targeting genes recurrently mutated in myeloid neoplasia and necessary to calculate IPSS-M. RESULTS Median age of patients was 74 years with male preponderance (54.5%). Serum EPO levels (sEPO) were available in 277/397 cases (mean 157U/L, <200 U/L in 80% of the cases). When transfusion history (TH) was known (252/397), 44.6% of cases were transfusion dependent (TD). The majority of the cases received erythropoietin-α and darbepoetin-α with response rate of 52.6% (HI-E) according to IWG criteria and a median duration of response of 30 months (data from 165 responders with follow up information). We observed that IPSS-M score, but not the number of mutations, was significantly lower in ESA responders compared to non-responders (median -1,08 vs -0,63 respectively , p<0.0001). The very low (VL) IPSS-M risk group was enriched in ESA responders (74 vs 26% p=0.036), while groups from moderate low (ML) to very high (VH) were more enriched with non-responders (37.5 vs 62% p=0.001; 45.4 vs 54.6% p=0.045; 29.6 vs 70,4% p=0.0017; 25 vs 75% p=0.021 respectively, Fig.1). In cases with known TH (n=252), non-TD (NTD) ESA responders had lower IPSS-M scores (mean -1,17) compared to i) NTD ESA non-responders (mean -0,61 , p<0.0001), ii) TD ESA responders (mean -0,93, p=0,025) and iii) TD ESA non-responders (mean -0,54, p<0.0001). As observed in other settings, 17% of the cases (n=67) were re-stratified to higher risk IPSS-M categories (8%, 13,4% and 44,3% of IPSS-R very low, low and intermediate risk respectively). Among upstaged cases, most were ESA non-responders especially in the moderate high (MH, 67%), high (H, 61%) and very high-risk (VH, 100%) groups. Only 4% of cases (n=17) were down-staged, all from IPSS-R intermediate risk group, and showed response rates similar to IPSS-R intermediate cases who remained ML or MH. Finally, IPSS-M score, sEPO and TD correlated with response in multivariate logistic regression analysis ( p=0.005, p=0.004 and p=0.014 respectively), but only IPSS-M score and TD were predictive (OR 0.51 and 0.34 respectively). Based on this model, we designed a ROC curve (AUC 0.79) where a cut-off of 0.46 showed a sensitivity of 0.9 and specificity of 0.67 in discriminating responders from non-responders (Fig.2). CONCLUSIONS To our knowledge, this is the first report correlating IPSS-M with ESA response in a large number of LR-MDS cases. IPSS-M scores exhibited predictive power, with lower values in ESA responders compared to non-responders and in NTD compared to TD cases. We also observed that higher IPSS-M categories were inversely correlated with ESA response. A fraction of cases with scarce ESA response was upstaged to IPSS-M higher risk groups. The number of mutations did not correlate with ESA response, while IPSS-M score, which considers the prognostic weight of specific mutations, showed predictive significance. In the multivariate analysis IPSS-M score and TD were strong predictors of ESA response, irrespective of baseline sEPO, potentially defining at diagnosis LR-MDS for whom novel agents like luspatercept could be considered as first line therapeutic option.