DP113 | COULD EARLY CLINICAL AND NON-GENETIC FEATURES PREDICT FUNCTIONAL HIGH RISK STATUS IN NEWLY DIAGNOSED MULTIPLE MYELOMA PATIENTS?

Background. A not negligible rate of non-high risk multiple myeloma (MM) patients unexpectedly relapse early after frontline therapy, being defined functional high risk (FHR). Cytogenetics and genomics are not always performed in clinical practice, so the aim of this study was to search clinical features to predict FHR probability. Methods. MM patients treated from 2019 to 2024 in 12 hematological Italian centres were retrospectively analysed. Patients with a PFS duration ≤ 18 months if transplant eligible (TE) and ≤ 12 months if not-transplant eligible (NTE), independently of traditional staging systems, were defined as FHR. Laboratory parameters significantly associated with FHR status by linear regression were combining in formula like EASIX score. Disease and patients factors were then analysed by Cox regression univariate and multivariate analysis to search those associated with FHR. Results. Out of 959 patients analysed, we found 220 (23%) FHR patients, 40 (11%) in TE group and 180 (30.5%) in NTE one. FHR patients had a significantly shorter OS than non-FHR (27 months vs NR; p<0.001; 22 months vs NR in NTE and 35 months vs NR in TE). FHR status was associated with higher LDH (p=0.004) and creatinine (p=0. 032), lower Hb (p=0.005) and platelets count (p=0.05) as continuous variables and ISS 2-3 (p<0.001) and ECOG-PS ≥ 2 (p=0.001), as categorized variable whereas cytogenetic risk, R-ISS, simplified frailty score and first-line therapy did not. By significantly continuous variables, we built the following formula: Lg2[(LDH x creatinine)/(Hb x PLT) x 100] showing a score > 2.0 and < 2.0 in 89% vs 11% of FHR patients, respectively (p<0.001). Multivariate Cox analysis selected the above formula and ECOG-PS ≥ 2 as factors affecting FHR probability (25% vs 8%, HR: 2.2, p<0.001; 35% vs 12%, HR: 2.1, p<0.001, respectively). Scoring these variables with the same weight according to similar HR, 3 groups of patients with none (LRS: low risk score), 1 (IRS: intermediate risk score) and 2 adverse prognostic factors (HRS: high risk score) were segregated. FHR probability was 4.5%, 17.5% and 62% in the 3 groups, respectively (p<0.001) (Fig.1). However, the ROC AUC was 66%. Conclusions. Our results suggest that FHR population is very heterogeneous. Simple clinical features can help to prognosticate groups at higher risk of becoming FHR but very in-depth biological tools, not available in clinical practice, are needed to identify singularly risk of becoming FHR.