Validation of the EBMT Multiple Myeloma Early Relapse Score within Worldwide Network for Blood and Marrow Transplantation (WBMT) Global Study

Rationale: Early relapse (ER) within 12 months of Autologous Hematopoietic Stem Cell Transplantation (AHCT) is currently accepted as functional high risk among patients diagnosed with Multiple Myeloma (MM). Efforts to predict ER has led to different risk scores developed by CIBMTR, GIMEMA and EBMT. CIBMTR and GIMEMA score integrates parameters not always available (bone marrow plasma cell percentage prior to AHCT, lambda light chain, FISH and LDH), whereas the EBMT score includes very simple factors ISS (at diagnosis), performance and disease status (prior to AHCT) (Beksac et al BMT 2023). EBMT ER score was developed using data obtained from patients aged 40-70 with a first AHCT in mainly European transplantation centers. This retrospective study aims to validate the EBMT ER score using data from a worldwide cohort of AHCT MM patients. This large population differs from the one used in the original paper, by age range, the number of countries and ethnicities, features in favor of suitability for external validation.

Methods: We selected patients from the WBMT study, a collaboration of five MM transplantation registries around the world. Only registries with data available on all components of the EBMT score were selected. Patients included in WAUSTIM had a first AHCT between 2013-2017. EBMT data within this current study belonged to only 2013 which was not included in the original study. We used Cox proportional hazards regression and the c-index to measure the predictive discriminative performance of the EBMT ER score.

Results: In total 14,924 patients (EBMT (4.1%), CIBMTR (79%), Australia and New Zealand (2.8.%), Japan (14%) and EMBMT (0.2%) were included. Median age at AHCT was 61(25-83) years. IgG (56%), IgA (21%) and light chain (21%). Disease status at the time of AHCT was CR (16%), VGPR (39%), PR (39%), SD/MR (6%), Rel/Prog (0.3%): ISS at diagnosis was I/II/III: 38/36/27%; Karnofsky score prior to AHCT-1: ≤70/80/90/100: 12/29/41/18 %. Cytogenetic risk was standard/high (t(4;14),t(14;16),17pdel) in 69%/31% of those with data available (not available in 17%). 81%/16% received Mel200/Mel140. Maintenance treatment was unknown for 89%. EBMT ER score distribution was: 0/1/2/3/4: 18%/32%/31%/14%/3.5%.

Within the Mel200 only population restricted to the 40-70 age limit (as in the original study), after a median FU of 48 months 12-month PFS (PFS-12) was 84% (95% CI 83-84%) with an ER incidence of 14.7%. which is similar to that observed in the original EBMT study (Mel200 training: 14.7%, Mel200 validation: 11.6%, Mel140: 16.9%). The score 0/1/2/3/4 distribution was: 19%/34%/31%/13%/3.2%. Thus the prevalence of scores 0-4 within the original and the current study are similar as well. PFS-12 (95% CI) according to scores were as follows: score 0: 90 (89-91); score 1: 86 (85-87); score 2: 83 (82-84); score 3: 78 (76-80); score 4: 69 (64-74) resulting with HRs vs score 0: score 1: 1.42; score 2: 1.75; score 3: 2.32; score 4: 3.69 (all p values<0.001. The c-index was 0.58 without and 0.61 with cytogenetic risk included. EBMT ER score acts similarly within this worldwide population with a clear separation of curves between scores 0-4. Cytogenetic high risk vs standard risk HR: 1.88 (95% CI: 1.69-2.09) among Mel200 and 1.81(95% CI: 1.65-1.98) were among the whole population(p-value<0.001). Comparison of the original study HRs and score points with the current analysis will be presented at the meeting.

Similar analysis performed among the whole population not limited to age or conditioning regimen intensity, resulted with highly similar PFS-12 ranging between 90-69% for scores 0-4 with HRs 1.39-3.49 (scores 1-4, all p-values <0.001; high vs standard risk HR: 1.81; C index: 0.58 without cytogenetic, and 0.61 with cytogenetic).

Conclusion: In a population from various parts of the world, we have been able to validate the EBMT ER score among both Mel200 and Mel140 conditioned myeloma patients with a wider range of age at AHCT-1. Although the cytogenetic risk score provided additional predictive value in the original and current study, it had no modification effect on the EBMT ER score. Due to high rate of maintenance data missingness, we could not investigate the role of the score when patients were continued to be treated after AHCT. Based on our findings, EBMT ER score is a predictive tool for recognition of functional high risk MM patients at the time of AHCT-1.