Defining the Impacts of Iron Overload in Children Receiving Hematopoietic Stem Cell Transplantation for Non-Malignant Hematologic Disorders: A Multicenter Analysis

Introduction: Children with non-malignant hematological disorders (NMHD) often receive numerous red blood cell (RBC) transfusions prior to hematopoietic stem cell transplant (HSCT), leading to iron overload. It has been hypothesized that pre-HSCT iron overload leads to poor post-HSCT outcomes, and studies have demonstrated a prognostic link between portal fibrosis, hepatomegaly, and post-transplant survival. Ferritin and liver iron concentration (LIC) by MRI are used as surrogate markers of iron overload to determine eligibility for HSCT, but neither has been definitively associated with worse HSCT outcomes in this population. Our preliminary single-center data demonstrated that elevated ferritin is associated with worse HSCT outcomes, while LIC is not.

Objectives: Validate the use of PEDSnet for automated data extraction in children with NMHD receiving HSCT. Determine whether high pre-HSCT ferritin and LIC are associated with worse post-HSCT outcomes including overall survival (OS), graft failure, bacteremia, and veno-occlusive disease (VOD) in a multicenter cohort.

Methods: We conducted a multicenter retrospective cohort study of children with sickle cell disease (SCD), transfusion-dependent thalassemia (TDT), severe aplastic anemia, and other bone marrow failure disorders treated with chronic RBC from six pediatric academic centers from 2010 to 2023 using PEDSnet. Patients were identified using codesets that incorporated procedure, laboratory, and billing codes. We included patients with NMHD who underwent HSCT and had either a ferritin or LIC measured before HSCT. Disease, date of HSCT, LIC values, and additional covariates were manually validated. OS, graft failure, and bacteremia were compared using multivariable Cox proportional hazards models. VOD rates were compared using descriptive analyses.

Results: After manual validation, 557 of 636 (87.6%) of patients identified by PEDSnet codesets met criteria for this analysis. In the validated cohort, 450 patients had a pre-HSCT ferritin, including 183 patients with SCD and TDT. Patients were divided into cohorts by pre-HSCT ferritin level. The low pre-HSCT ferritin group (<1000 ng/mL) included 257 patients, the moderate ferritin group (1000-2500 ng/mL) included 145 patients and the high ferritin group (>2500 ng/mL) included 48 patients. In the low ferritin group, 3-year OS was 90.3%, with graft failure occurring in 6.2%, and bacteremia occurring in 16.7%. In the moderate ferritin group, OS was 94.5%, graft failure occurred in 9.0%, and 17.2% had bacteremia. In the high ferritin group, OS was 87.5%, 10.4% experienced graft failure, and bacteremia occurred in 31.3%. Differences in OS and graft failure were not statistically distinct between ferritin groups. Rates of bacteremia were statistically different between high and low ferritin groups, hazard ratio 2.15 (95% CI 1.19-3.89).

The validated cohort included 149 patients with a pre-HSCT LIC, including 86 patients with SCD and TDT. Seventy-eight patients had a low pre-HSCT LIC (<5 mg Fe/g dry weight) and 71 patients had a high LIC (≥5 mg Fe/g dry weight). Three-year OS was 93.6% in the low LIC group, 12.8% experienced graft failure, and 19.2% had bacteremia. In the high LIC group, OS was 85.9%, 5.6% had graft failure and 16.9% had bacteremia. Outcomes were not statistically distinct between LIC groups.

Within the ferritin cohort, 36% of patients received busulfan conditioning. In this group, VOD occurred in 9.3% of patients with low ferritin, 9.7% with moderate ferritin, and 4.2% with high ferritin. In the LIC cohort 43.6% of patients received busulfan. VOD occurred in 24.3% of patients with low LIC and 42.9% with high LIC.

For all outcomes, similar trends were observed between the overall cohort and the SCD and TDT subgroup in both ferritin and LIC analyses.

Conclusions: PEDSnet can identify patients with NMHD undergoing HSCT, though improvement in codeset accuracy is still needed. Patients with pre-transplant high ferritin have a significantly higher risk of bacteremia post-transplant. Patients with high LIC who received busulfan have higher rates of VOD than those with low LIC, though future work adjusting for confounding is needed. Both patients with high ferritin and those with high LIC show trends towards decreased survival. Both metrics of transfusional iron overload negatively influence transplant outcomes through differing mechanisms in children with NMHD.