RBC-Enhance: A Randomized Pilot Feasibility Trial of Red Cell Transfusion Thresholds in Myelodysplastic Syndromes

Background

Red blood cell (RBC) transfusion dependence (TD) and anemia are associated with inferior quality of life (QOL) and survival in myelodysplastic syndromes (MDS). As part of a research programme to identify the optimal transfusion hemoglobin (Hb) threshold that translates into significantly improved QOL, we conducted a pilot randomized study of two transfusion algorithms (liberal versus restrictive) to determine the feasibility of a larger study. This trial was aligned with the REDDS study (ISRCTN26088319, Stanworth et al., BJH 2019) and followed the same pre-defined statistical analysis plan.

Methods

The study was undertaken at 3 cancer centers (NCT02099669). We included patients with MDS, CMML, low blast AML (20-30%) or myeloproliferative neoplasms who were TD (> 1 RBC/8 weeks x 16 weeks). Patients on disease modifying therapy for > 6 months were permitted only if they had remained stably RBC-TD with a life expectancy of > 3 months. Central randomization was to one of two RBC transfusion strategies over a 12-week period:

1. Restrictive strategy to maintain Hb between 85 and 105g/L. To achieve this, 2 RBC units were transfused when Hb < 85 g/L

2. Liberal strategy to maintain Hb between 110 and 120 g/L. To achieve this, 2 RBC units were transfused when Hb < 105 g/L and 1 unit for Hb 105-110 g/L.

The first 4 weeks were considered a run-in period to achieve the target Hb.

The primary outcomes of this feasibility study were

1. The percentage compliance of Hb being within or above the target range of the assigned transfusion threshold (after the 4 week run-in). Feasibility was set at 70%.

2. Achievement of at least a 15 g/L difference between the mean pre-transfusion Hb of the liberal and restrictive strategy groups.

Secondary outcomes included changes and differences in QOL scores, rates of transfusion reactions and alloimmunizations, overall blood utilization and visit numbers and impact on serum ferritin. The primary analysis was intention to treat. The study was closed early due to COVID-19, with 28 patients randomized of planned 30.

Results

30 patients were enrolled over 5 years and 28 patients randomized (n=15, liberal; n=13 restrictive). Median age was 74 (range 58-84), 26/28 had MDS or CMML and there were no significant imbalances in baseline disease and patient characteristics although patients in the restrictive group tended to have lower risk disease (IPSS-R very low + low 69% versus 53% liberal). Compared with the restrictive arm, liberal arm patients had more complete blood counts, were transfused more RBC units (only during the 4-week run-in) and were transfused at shorter intervals (Table 1). There were no differences in adverse events.

Over the 12 weeks, the mean Hb value was 90 ± 4 g/L (restrictive) versus 101 ± 4 g/L (liberal), (p <0.0001, Figure 1). The pretransfusion difference was 11.8 g/L (95% CI 9.4-14.3) with mean pre-transfusion Hb values of 98.6 (± 10.6) g/L and 86.7 (± 8.6) in the liberal and restrictive arms respectively. Percentage compliance of pre-transfusion Hb falling within or above the target Hb range was 48% for restrictive patients but only 14% for liberal patients.

Adherence with QOL completion (minimum of 4 serial) was 93%. While sample sizes are small and comparisons exploratory, the area under the curve for several serial quality of life scores was numerically higher and the curve more stable for the liberal arm (EQ5D (figure 2), EORTC cognitive and emotional functioning). Similarly, a higher % of patients in the liberal arm achieved pre-defined clinically meaningful increases in several symptom and function domains (emotional, social, fatigue, dyspnea, financial problems).

Discussion

Whilst our study results did not meet our feasibility endpoints, this small randomized trial in MDS patients demonstrated clinically important differences in mean Hb achievable with different transfusion thresholds. Poor rates of compliance with Hb targets were clearly identified for patients in the liberal arm, suggesting a need to understand the barriers to protocol adherence in this arm prior to embarking on larger trials.

Compared with the REDDS trial, we did not document a significant excess of RBC transfusions needed (post run-in period) to maintain the higher Hb threshold. Similar to REDDS, we did observe some signals of improved QOL with a liberal transfusion strategy. A formal analysis of pooled results with REDDS study patients is planned. (Funding, CCSRI grant QOLL-14 and MOSPI Fund 2014).

View largeDownload slide

View largeDownload slide

Close modal