Page 127 - Haematologica March 2020

Iron chelation in lower risk MDS patients
using a propensity-score matched model. Of 2,200 patients, 224 received iron chelation. The hazard ratio and 95% confidence interval for OS for chelated patients, adjusted for age, sex, comorbidity, performance status, cumulative RBC transfusions, Revised-International Prognostic Scoring System (IPSS-R), and presence of ringed sideroblasts was 0.50 (0.34-0.74). The propensity-score analysis, matched for age, sex, country, RBC transfusion intensity, ferritin level, comorbidity, performance status, and IPSS-R, and, in addition, corrected for cumulative RBC transfusions and presence of ringed sideroblasts, demonstrated a significantly improved OS for chelated patients with a hazard ratio of 0.42 (0.27-0.63) compared to non-chelated patients. Up to 39% of chelated patients reached an erythroid response. In conclusion, our results suggest that iron chelation may improve OS and hematopoiesis in transfused lower-risk MDS patients. This trial was registered at clin- icaltrials.gov identifier: 00600860.
Introduction
Myelodysplastic syndromes (MDS) comprise a hetero- geneous group of clonal hematopoietic stem cell disorders characterized by abnormal differentiation and maturation of hematopoietic cells, bone marrow (BM) failure and genetic instability, with an enhanced risk of progressing to acute myeloid leukemia (AML).1 Iron overload (IOL), as a consequence of frequently administered red blood cell transfusions (RBCT) and/or ineffective erythropoiesis, is a common finding in MDS. The effects of toxic iron species in other iron loading diseases, such as primary hemochro- matosis, thalassemia and sickle cell anemia are well known, but the consequences in MDS are less clear.2-4 With an expected median survival of 2.4-11.8 years in lower risk MDS (LR-MDS) patients,5 these patients are prone to long-term accumulation of iron due to RBCT as well as direct iron toxicity due to the formation of reac- tive oxygen species (ROS).6
Several studies have reported beneficial effects of iron chelation therapy (ICT) on overall survival (OS) and other clinical outcomes in MDS patients with IOL.7-10 However, valid data on the effect of ICT are limited, as most studies are carried out in small or highly selected patient groups, or suffer from serious methodological problems such as confounding by indication. Performing a randomized, controlled trial to explore this is cumbersome due to patients' widespread belief in the beneficial effects of ICT and also the personal opinion of many treating physi- cians, which may negatively affect enrollment. Likewise, patients included in a randomized, controlled trial do not generally reflect the actual LR-MDS patient group, which is usually made up of elderly patients with multiple comorbidities.
In addition to the possible beneficial effects of ICT on OS, there is increasing evidence to indicate hematologic improvement in patients during treatment with iron chelators.11-16 Alongside improvement in hemoglobin, platelet, and neutrophil levels, transfusion independence is achieved in a minority of chelated patients.11,12,14 The underlying mechanisms are still unclear.17
The aim of this study was to evaluate the effect of ICT on OS, hematologic improvement, and ferritin levels in lower risk MDS patients in the European MDS (EUMDS) Registry.
Methods
The EUMDS registry prospectively collects observational data on LR-MDS patients from 142 centers in 16 countries in Europe
and Israel. Patients were included within 100 days of MDS diag- nosis according to the World Health Organization 2001 classifi- cation, restricted to patients with a low or intermediate-1 score according to the International Prognostic Scoring System (IPSS).18 IPSS was the current prognostic indicator at the start of the registry, in accordance with the currently used prognostic score; the Revised-IPSS (IPSS-R) was reconstructed afterwards. The ethics committees of all participating centers approved the protocol and all patients provided written informed consent. Data were collected at baseline and at each 6-monthly routine outpatient follow-up visit. Data were collected on: comorbidity, transfusion history, use of iron chelators (agent, time frame; no drug doses or schedules were collected), peripheral blood values, conventional iron parameters (e.g. serum ferritin), bone marrow pathology, and progression to higher-risk MDS or AML. Subjects were prospectively followed until death, loss to follow up, or withdrawal of informed consent.
In Europe, three iron chelators are available for treatment of secondary IOL, but availability varies between countries. We analyzed all patients, chelated or non-chelated, who are eligible for receiving ICT based on at least one criterion for starting ICT (cumulative ≥15 RBC units, RBCT intensity of ≥1 unit/month during a 6-month period, or serum ferritin level >1000 mg/L), thereby preventing immortal time bias. As chelated and non- chelated patients may differ in characteristics that affect out- come, two different approaches were performed in order to con- trol for potential bias: 1) analysis of all eligible chelated and non- chelated patients using receipt of ICT as a time-varying co-vari- ate, adjusting for co-variates related to both receiving ICT and OS: sex, age, comorbidity, performance status, RBCT intensity, number of units transfused, IPSS-R, and presence of ringed sider- oblasts; 2) Propensity Score (PS), i.e. conditional probability for being treated with ICT on the basis of patient characteristics, matching of the same group. Variables included in the PS were: age, sex, country, RBCT intensity, ferritin level, MDS comorbid- ity index, performance status, and IPSS-R. A 3-to-1 nearest neighbor matching method with replacement and caliper (0.2) was applied.19 In addition, we used a robust sandwich estimator to correct for intra-individual correlation of multiply used con- trols. Further details on the PS matching are provided in the Online Supplementary Methods.20-22 OS was defined as the time from eligibility for ICT to death; subjects still alive were cen- sored at the last follow-up date. Cox proportional hazards regression models and Kaplan-Meier survival curves were applied and hazard ratios (HR) with 95% confidence intervals (95%CI) were reported.23
Erythroid responses were defined as a reduction in RBCT den- sity (number of RBCT over time; see Online Supplementary Methods for definition and details) or as transfusion independen- cy at least once as the transfusion density was reduced to zero, platelet responses were assessed according to the modified
haematologica | 2020; 105(3)
641