Kinetics of Recipients with Thrombocytopenia and Its Impact on Outcomes after Allogeneic Hematopoietic Stem Cell Transplantation: A Nationwide Retrospective Study

Background: Thrombocytopenia is a common complication after allogeneic hematopoietic stem cell transplantation (allo-HCT), which can lead to increased mortality and impaired quality of life with prolonged transfusion dependency and high risk of bleeding. However, it is not clear how many patients survive without sufficient platelet recovery after allo-HCT. Here, we visualized the kinetics of neutrophil and platelet recovery and the impact on outcomes using multistate models with transplant registry data. We also compared the effects of different graft sources.

Methods: We analyzed Japanese nationwide registry data of adult patients with acute myeloid leukemia (AML) and acute lymphoblastic leukemia (ALL) in complete remission who underwent first allo-HCT from 2010 to 2022. Second transplantation was treated as a censoring event. Transitional changes of the population with neutrophil <0.5×10⁹/L (state 0), neutrophil ≥0.5×10⁹/L and platelet <20×10⁹/L (state 1), 20×10⁹/L≤ platelet <50×10⁹/L (state 2), and platelet ≥50×10⁹/L (state 3), death, relapse, and non-relapse mortality (NRM) were estimated by multistate model analysis using transition probability plot and proportional transition hazard model. The effects of hematopoietic recovery on mortality, relapse, and NRM were estimated by Simon-Makuch plot and proportional hazard model with time-dependent covariates. Influences of the background factors (age, recipient sex, performance status, comorbidity, diagnoses, disease risk, graft sources, ABO major mismatch, conditioning intensity, and year of transplantation) were adjusted.

Results: A total of 9105 patients (5802 AML and 3303 ALL) were extracted from the database. The number of the transplant sources were: 2398 HLA-matched unrelated bone marrow (uBM), 782 HLA-matched related BM (rBM), 326 HLA-matched unrelated peripheral-blood stem cells (uPB), 1283 HLA-matched rPB, 622 HLA-haploidentical PB (Haplo), and 3694 single-unit cord blood (CB). The median age at allo-HCT was 49 (range, 16-85) years. Neutrophil ≥0.5×10⁹/L at day 60, platelet ≥20×10⁹/L and ≥50×10⁹/L at day 100 were 99%, 91%, and 85% in uBM, 99%, 96%, and 92% in rBM, 99%, 95%, and 90% in uPB, 99%, 96%, and 93% in rPB, 98%, 91%, and 84% in Haplo, and 94%, 87%, and 82% in CB, respectively. Multistate model analysis with the entire patient population showed that the proportions in states 0, 1, 2, 3, relapse, and NRM were 1.4%, 9.9%, 7.4%, 77%, 1.3%, and 3.4% at day 60; and 0.1%, 3.8%, 4.3%, 80%, 5.4%, and 6.6% at day 120, respectively. Patients surviving without adequate platelet recovery (the sum of states 0, 1, and 2) was significantly higher in CB (25%) than in other sources (11–13%) at day 60. Interestingly, the difference became marginal at day 120 (8.3% in CB, and 6.5–8.7% in other sources). Survival probabilities by the Simon-Makuch method in states 0, 1, 2, and 3 were 83%, 93%, 98%, and 100% at day 60, and 44%, 71%, 91%, and 98% at day 120, respectively (P<0.001). Time-dependent covariate analysis showed that the hazard ratios (HRs) of the patients in states 0, 1, and 2 (vs. state 3) were 40, 5.3, and 2.4 for mortality; 57, 7.6, and 3.1 for NRM (all P<0.001), respectively, after adjustment for the background factors. The NRM causes of patients in states 1 and 2 were infection (24%), organ failure (19%), SOS/TMA (12%), and acute GVHD (10%), with the proportion of SOS/TMA and acute GVHD being more than double that of patients in state 3. Multistate model analysis showed that although the transition probabilities (TPs) to mortality from states 1 and 2 were higher in CB (HR 1.4 and 1.7, respectively) compared with uBM, the TP to mortality from state 3 was significantly lower in CB (HR 0.86; P=0.008; vs. uBM) with lower trends for both relapse and NRM.

Conclusions: In this largest retrospective study on hematopoietic recovery to date, the kinetics of survivors without platelet recovery and the detrimental impact of thrombocytopenia on overall mortality and NRM were clearly demonstrated. SOS/TMA and acute GVHD had a greater influence on the NRM of thrombocytopenic patients. Although the platelet recovery was delayed in CB, the proportion of survivors with thrombocytopenia at day 120 was comparable to other sources. The higher risk for mortality in CB was transient, and the risk in CB became less than in uBM after platelet recovery (≥50×10⁹/L). This analysis provides baseline data for future studies of hematopoietic recovery.