Red cell transfusion thresholds in outpatients with myelodysplastic syndromes: Combined results from two randomized controlled feasibility studies

The associations between RBC transfusion dependence (TD) in patients with myelodysplastic syndromes (MDS) and lower survival, quality of life (QOL), iron overload and cardiac morbidity were recently summarized in a systematic review.1 However, red blood cell (RBC) TD remains the reality for many patients with MDS with a paucity of data to inform optimal transfusion policy for these patients. With only inpatient data2 or expert opinion to guide decisions,3 transfusion thresholds are heterogeneous4, 5 ranging from hemoglobin (Hb) 70 to 90 g/L and 26% of patients wishing they were transfused at higher thresholds.5 Two parallel blinded randomized feasibility trials of similar design comparing liberal with restrictive transfusion thresholds were conducted in the United Kingdom/Australia/New Zealand (REDDS)6 and Canada (RBC-Enhance) (in press, Transfusion). We combined our data sets in a secondary analysis to better evaluate feasibility, explore the impact of higher Hb transfusion thresholds on QOL, and obtain insights in improving the design of future trials. Methods have been previously published.6 Patients with MDS and MDS/MPN including chronic myelomonocytic leukemia (CMML) with a life expectancy of more than 6 months who were stably transfusion dependent (minimum of 1 unit in 8 weeks) were randomized 1:1 using computerized block randomization to a liberal (maintain Hb 110–125 g/L) or restrictive (maintain Hb 85–100 g/L) transfusion strategy with minor differences summarized in Table S1. The transfusion strategy was applied for 12 weeks. The schedule and number of transfusions were dictated by the transfusion algorithms created for the trials (Table S1). Upon enrollment, patient and disease characteristics were captured including WHO classification, risk scores, iron status, transfusions, and treatment. REDDS and RBC-ENHANCE recorded QOL using EQ-5D-3L7 and EORTC-QLQ-C308 at baseline, pre-transfusion, 5–7 days later, and monthly. The primary outcome for both studies was feasibility. For the combined analysis, feasibility outcomes were defined as (1) more than 70% compliance of the pre-transfusion Hb being below the target range of the assigned transfusion threshold; and (2) achievement of at least a 20 g/L (REDDS) or 15 g/L (RBC-ENHANCE) difference between the mean pre-transfusion Hb in the liberal and restrictive strategy groups. Secondary outcomes included enrollment rates, transfusion utilization and visits, adverse events (transfusion reactions and iron assessment), completion of QOL questionnaires, and magnitude of changes in QOL. The statistical considerations, CONSORT diagrams, definitions of clinically significant differences in QOL scores are summarized in the Supplement. Both trials received ethics committee approval at participating institutions. All patients provided informed consent and studies were registered in clinical trial registries (ISRCTN26088319 and NCT02099669). After screening 252 patients in both studies, 66 patients (n = 33 restrictive and n = 33 liberal) were randomized between January 2015 and February 2020. The enrollment rate was 0.46–1 patients/month. There were no significant differences in baseline characteristics between the two transfusion arms (Table S2). Of the 66 randomized patients, 61 received transfusions while on study. The mean (SD) pre-transfusion Hb for the restrictive and liberal arms were significantly different at 85.9 ± 9 and 98.8 ± 9 g/L, respectively, with a mean difference of 12.9 g/L (95% confidence interval [CI] 11.2–14.6; p < .0001) as was the mean Hb of all measured Hb values according to transfusion strategy (liberal 100.4 ± 4.6 vs. restrictive 89.9 ± 6.8 g/L; Figure S1a). There was a significant difference in Hb over time (p = .0025) with a net positive slope in the liberal (slope = 0.49) and a net negative slope for the restrictive arm (slope = −0.32). In the restrictive arm, the percentage of pre-transfusion Hb falling below the allocated transfusion target range (85–100 g/L) was 58%. In the liberal arm, the percentage of pre-transfusion Hb falling below the allocated transfusion target range was 81%. In a post-hoc sum of squares analysis, we found that the mean individual amplitude of variation in Hb was 42% higher in the restrictive versus the liberal arm at 74 ± 59 and 43 ± 31, respectively. In total, 232 and 471 units of RBC were transfused in the restrictive and liberal arms, respectively. Patients in the liberal arm had more CBCs (13.8 vs. 10.3, p = .001), a mean of 3.1 ± 2.9 more transfusion visits, and a mean of 6.3 ± 5.9 extra units of blood during the 12-week study. The mean number of days between transfusions was 18.9 ± 9.9 in restrictive and 13.0 ± 7.5 in liberal arms, respectively (Table 1). There was one episode of allo-immunization and one febrile non-hemolytic transfusion reaction in the liberal arm. Mean ferritin was similarly elevated at baseline but increased by 926 μg/L (95% CI 1446–2095) and 28 μg/L (95% CI −864 to 922) above baseline in the liberal and restrictive arms, respectively (Figure S2). Rates of QOL completion were high: pre- (91%) and post-transfusion (89%). There appeared to be greater stability in the EQ-5D health utility scores for patients in the liberal arm (Figure S1b). The QOL scores measured immediately pre- and 5–7 days post-transfusion were compared between both arms. The pre-transfusion EQ5D health utility scores (0.77 vs. 0.7, p = .003), visual analog scores (71.4 vs. 62.7, p = .001), EORTC QLQ-C30 emotional functioning (84.3 vs. 75.1, p = .002), fatigue (35.7 vs. 43.1, p = .01), nausea/vomiting (3.3 vs. 6.8, p = .01), and appetite loss (12.3 vs. 20.0, p = .02) scores were superior for patients in the liberal arm. Post-transfusion, EQ-5D health utility (0.77 vs. 0.71, p = .01), visual analog scores (72.2 vs. 66.6, p = .03), emotional functioning (84.7 vs. 77.8, p = .01), and constipation scores (17.2 vs. 22.4, p = .03) remained superior in the liberal arm. Pre-transfusion, a higher percentage of patients treated in the liberal arm experienced improvement in physical and social functioning, fatigue, dyspnea, and insomnia while a higher percentage of patients in the restrictive arm experienced improvement in emotional functioning, and financial problems compared with baseline (Figure S3a). Post-transfusion, some but not all of these differences were maintained (Figure S3b). Individual changes in patient mean pre-transfusion QOL scores compared with baseline are presented in waterfall plots for EQ5D health utility, EQ-5D visual analog, physical functioning, social functioning, global QOL, fatigue, and dyspnea (Figure S4a–g). While there were more patients in the liberal arms experiencing benefit (above the x-axis), there were patients who had stability or decrements in selected symptoms and function domains in both treatment arms. We present a combined analysis of two feasibility trials of similar design, providing data on 66 RBC TD patients with MDS randomized to restrictive versus liberal transfusion thresholds. We found a mean pre-transfusion Hb difference of 12.9 g/L between liberal and restrictive arms. Maintaining this mean Hgb difference for 12 weeks required the transfusion of 6.3 extra units of blood over 3.1 extra transfusion visits. This differs from the findings of a small Nordic group study in which the achievement and maintenance of a Hb >120 g/L did not confer a higher transfusion rate once the hemoglobin target was reached.9 Overall, we observed less variability in Hgb levels in the liberal arm with patients reporting clinically important improvements pre- and post-transfusion (compared with baseline) in selected symptom and functional domains. However, many patients in both transfusion arms experienced stability or declines in their scores. This underscores the complex interplay of factors that contribute to QOL in MDS patients such as frailty, comorbidity, disability, and disease-related inflammation. Further research into the impact of transfusions on functional outcomes in sufficiently large cohorts of patients is needed. We thank all participating sites and investigators in RBC-ENHANCE, REDDS, and the NHSBT CTU. We also acknowledge the agreement of the sponsor (NHSBT) to share REDDS trial data. REDDS was funded by NHSBT R&D, the Australian and New Zealand Society of Blood Transfusions (ANZSBT), and the Wellington Division of New Zealand Cancer Society. RBC-ENHANCE was funded by Canadian Cancer Society Research Institute Grant QOLL-14 and MOSPI Fund 2014. All of this research is original and has not been published elsewhere. No medical writers were used to compose this manuscript. The authors declare no conflicts of interest. The data that supports the findings of this study are available from the corresponding author upon reasonable request. Data S1: Supporting information. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.