Background Progression-free survival (PFS) of lower-risk (LR) MDS patients treated with red blood cell transfusions (RBCT) is usually reduced, but whether the RBCT intensity is important for prognosis is unknown. The EUMDS Registry contains data on 2,192 newly diagnosed LR-MDS patients from 16 European countries and Israel, collected at diagnosis and at 6 months intervals.The aim of this analysis is to assess the effect of RBCT intensity on PFS of patients with LR-MDS in a prospective, observational setting.

MethodsData on the number of RBCT units received during each visit-interval were analyzed using proportional hazards regression with time-varying co-variates. The cumulative dose received at the end of each interval was divided by the time since the beginning of the interval in which the first RBCT was received rather than for each visit-interval, to give an overall dose density. Dose density rises over time when RBCT intensity increases over time, but declines when patient becomes RBCT independent after treatment. In Cox regression with time-varying covariates, PFS was adjusted by relevant confounders, including age, gender, patient condition (proxied by EQ-5D index), and number of cytopenias at diagnosis. We analyzed a cohort of 1,509 patients with at least 3 visit reports available. Within this group 476 patients have died and 118 patients have progressed to higher-risk MDS or AML, which were defined as events.

ResultsPatients with ring sideroblastic MDS (RS-MDS) received more frequently RBCT after diagnosis compared to non-RS-MDS (60% vs 54%). Very low-risk patients according to IPSS-R received RBCT less often compared to intermediate risk patients (35% vs 69%). Age, gender, presence of co-morbidities, and countries influenced RBCT dependency and RBCT rate. The transfused patients were subdivided in 3 groups according to the maximum RBCT rate over time: low rate (<0.75 units/month), intermediate rate (0.75-2 units/month) and high rate (>2 units/month) in order to analyse the impact of the RBCT rate in more detail (figure A). Unadjusted overall PFS stratified this way, is shown in figure B.

Univariate analysis of the 404 patients transfused at a high rate showed a strong association of age (p < 4 x 10-4) with PFS, but no difference by sex (p = 0.19). The EQ-5D index, baseline MDS diagnosis, bone marrow percentage category and number of cytopenias were all associated with PFS in univariate regression. No association was found between country and PFS (p = 0.26). RBCT dose density was associated with PFS (p<1 x 10-4) with a significant non-linear component (p<1 x 10-4). All the relevant variables were entered into multivariate regression analysis of RBCT dose density and retained statistical significance, except for the baseline cytopenias (p=0.06). The non-linear component of the dose density effect was retained (p = 0.0003). The dose density has an increasing effect until a dose density of about two units/month. Thereafter, the effect is leveled off (figure C). Replacing RBCT density by a cumulative dose of RBCT units over time in a similar multivariate regression model using the same variables resulted in a similar functional form as observed in the dose density analysis (figure 4). The impact of the cumulative dose RBCTs levels of around 25 units RBCTs received was examined, and all variables except sex (p = 0.52), baseline cytopenias (p = 0.45) and the number of pre-diagnosis RBCT, retained statistical significance. The same analyses were repeated in the group of patients who received maximally between 0.75 and 2 units/month. Transfusion dose density in the univariate analysis was associated with PFS (p = 0.005) with a significant non-linear component (p = 0.03) and this impact was confirmed in the multivariate analysis (p = 0.002) (figure D). Also the dose density in the group of patients who were transfused at a low density (<0.75 units/months) was associated with overall survival (p = 0.025) with a weakly significant non-linear component (p = 0.025) in univariate analysis, but this was not confirmed in multivariate analyses due to the small sample size in this group.

ConclusionThe negative effect of RBCT treatment on PFS already occurs at low RBCT densities below 2 units/month and before 25 units of RBCT have been transfused. This indicates that the RBCT dependency, even at relatively low intensities, may be considered as an indicator of poor prognosis for survival.


Fenaux: Amgen: Honoraria, Research Funding; Janssen: Honoraria, Research Funding; Astex: Honoraria, Research Funding; Amgen: Honoraria, Research Funding; Janssen: Honoraria, Research Funding; Novartis: Honoraria, Research Funding; Celgene: Honoraria, Research Funding; Novartis: Honoraria, Research Funding; Astex: Honoraria, Research Funding; Celgene: Honoraria, Research Funding. Sanz: Gamida Cell: Research Funding. Germing: Janssen: Honoraria; Novartis: Honoraria, Research Funding; Celgene: Honoraria, Research Funding. Almeida: Servier: Consultancy; Alexion: Honoraria; Novartis: Consultancy; Celgene: Consultancy; Bristol Meyer Squibb: Honoraria. Itzykson: Janssen: Research Funding; Novartis: Research Funding. Park: Novartis: Membership on an entity's Board of Directors or advisory committees, Research Funding; Hospira/Pfizer: Research Funding; Celgene: Research Funding. de Witte: Novartis: Consultancy, Honoraria, Research Funding; Celgene: Consultancy, Research Funding; Amgen: Consultancy, Honoraria, Research Funding; Janssen Cilag: Research Funding.

Author notes


Asterisk with author names denotes non-ASH members.

This icon denotes a clinically relevant abstract