Real-World Impact of Isovolemic Hemodilution on Blood Utilization for Chronic Red Cell Exchange

Background: Automated red cell exchange (RCE) is recommended for patients with sickle cell disease (SCD) requiring chronic red blood cell (RBC) therapy. However, the high RBC requirements of RCE present challenges related to scarcity of matched RBC units and increased risk of alloimmunization. In isovolemic hemodilution RCE (IHD), a red cell depletion with normal saline or 5% albumin replacement occurs prior to RCE. The intent of IHD is to decrease the number of RBC units needed to attain the target hemoglobin S (HbS)%. However, the real-world effect of IHD on RBC usage is unclear, and less than half of recently surveyed institutions use IHD (Karafin et al, 2020). Thus, we conducted a retrospective cohort study to compare RBC usage in patients with SCD receiving standard vs. IHD RCE.

Methods: Patients with SCD-SS at our institution undergoing a period of chronic standard RCE followed by chronic IHD RCE between 2003-2022 were identified. Periods of at least 6 consecutive procedures of the same RCE type (standard vs. IHD) with ≤ 45 days between procedures were labeled as RCE “clusters”. Subjects were included if they had at least one standard cluster concluding within 1 year of the start of an IHD cluster, with a consistent HbS goal across all procedures. Cluster duration was trimmed to achieve two clusters (one IHD, one standard) of equal duration per subject.

Subject and RCE characteristics in IHD vs. standard RCE were described with paired t-tests. To account for repeat measures per subject, linear mixed effects regression models were used for multivariate analyses. Primary outcomes were RBC units per RCE and annualized cumulative RBC units per cluster. Baseline covariates, including year, age, sex, and HbS goal, were assessed at the index RCE procedure for inclusion in the multivariate model. Baseline hematocrit, baseline total blood volume (TBV), change in TBV between clusters, and cluster duration (for annualized analysis) were assessed as both covariates and interaction terms. Postestimation pairwise comparisons were used to model IHD effects at different TBV values.

Results: Our final analysis included 40 patients with 1213 RCE procedures. Median baseline age was 12.5 (range 7-29) years, 55% of subjects were male, and 85% had a HbS goal of <30%. Median cluster duration was 9.2 (range 3.3-35.7) months, with a median of 12 transfusions per cluster and 3 RBC units per RCE (range 2-10).

In paired analyses, baseline TBV was significantly larger in IHD clusters vs. standard (3240 mL vs 2851 mL, p<0.001), as was baseline hematocrit (28.3% vs 26.4%, p<0.001). Mean inter-procedure interval was 25 days in both clusters. Target fraction of cells remaining (FCR), representing the proportion of subject RBCs remaining after RCE, was lower in the IHD clusters (38.3% vs. 45.0%, p<0.001). RBC units/mL TBV did not differ between cluster categories, with 0.0012 RBC units/mL in both groups. In RCEs with HbS goal < 30%, mean pre-RCE HbS was 29.9% in IHD vs. 30.9% in standard clusters. This difference was not statistically significant (p = 0.22).

Our multivariate mixed effects model included HbS goal, hematocrit, year, and TBV, with interactions between RCE type and both baseline TBV and TBV change between clusters. Results were modeled to simulate no TBV change between clusters. TBV modified the effect of IHD on RBC units per procedure; partial unit savings with IHD were predicted in patients with a larger TBV. In individuals with TBV values < 3900 mL (at baseline), units per procedure did not differ between standard and IHD clusters. However, in those with TBV ≥ 3900mL, IHD conferred a statistically significant but small RBC unit savings that increased with larger TBVs. The peak saving with the largest TBV in our sample, 5055 mL, was 0.78 units/procedure (95% CI: 0.1 - 1.5 units saved, p =0.026). IHD was not associated with a significant change in annualized RBC unit exposure. However, there was a similar trend towards RBC unit savings with increasing TBV, with a predicted 5.69 units/yr saved with IHD at a TBV of 5055 mL (95% CI: 18.2 units saved to 6.8 units gained, p = 0.373).

Discussion: Our study suggests that IHD has limited utility in patients with small TBV receiving chronic red cell exchange. In patients with a lower TBV, IHD did not result in meaningful savings of RBC units. Conversely, larger patients may benefit from IHD. Prospective, multi-center studies are needed to standardize approaches to RCE and identify appropriate IHD candidates.