Adverse Prognostic Impact of Pretransplant Serum Ferritin and Hepcidin on Survival Outcomes After Allogeneic Hematopoietic Stem Cell Transplantation

Allogeneic hematopoietic stem cell transplantation (allo-HSCT) remains limited by significant morbidity and mortality. Scoring systems, such as the Hematopoietic Cell Transplantation-Comorbidity Index (HCT-CI) have been introduced to estimate the risk of transplant-related mortality. Serum ferritin (SF) is frequently elevated in transplant candidates [1, 2] and further peaks around day 42 posttransplant, reflecting iron overload (IO) due to red blood cell (RBC) transfusions and conditioning-induced erythropoietic suppression [3]. High SF has been associated with inferior overall survival (OS) or progression-free survival (PFS) and higher non-relapse mortality (NRM) [4], but a meta-analysis combining four studies investigating more specifically liver iron concentration (LIC) found no significant association between LIC and OS, relapse or NRM [5]. This indicates that high SF might not be an accurate IO marker but may also be influenced by inflammation, hepatic cytolysis, and renal failure. This multifaceted significance of SF and its association with transplant outcomes has never been addressed comprehensively. Hepcidin, the key regulator of systemic iron balance, increases with IO and inflammation and decreases with anemia or erythropoiesis. We demonstrated that hepcidin strongly increased early after allo-HSCT, paralleling the kinetics of SF and transferrin saturation (Tsat), contrarywise that of erythroferrone (ERFE) and soluble transferrin receptor (sTfR), reflecting the respective impact of IO, inflammation and myelosuppression [3, 6]. The only four hepcidin studies reported conflicting results on outcomes [7]. In this study, we investigated in 502 allo-HSCT patients the determinants of pretransplant SF and hepcidin, and their impact on OS, PFS, relapse and NRM with at least 5 years of follow-up. Materials and methods are provided as Supporting Information. Significance of pre-conditioning serum ferritin and hepcidin. Pretransplant characteristics of the 502 patients according to hepcidin and SF quartiles are described in Table S1 and Table S2, respectively. Median pre-conditioning hepcidin and SF were 25.13 ng/mL (0.4–345.1) and 1083 ng/mL (6–21 466), respectively, and correlated strongly (Figure 1A). Patients were divided into quartiles (Q1-Q4). Patients in Q4-hepcidin had higher PS and HCT-CI, more treatment lines and transfusions, and more frequently acute leukemias (AL), NMA conditioning, and HLA-mismatched donors (Table S1). Similar differences were observed between SF quartiles, except for HCT-CI and donor relationships (Table S2). Q4-SF included more high-risk Disease-Risk Index (DRI), and Q1-SF had larger use of MMF rather than MTX and longer time between diagnosis and transplantation. Among patients in Q1-, Q2-, Q3-, and Q4-SF quartiles, 60%, 37%, 35%, and 53% were in the same hepcidin quartile. Pretransplant biological characteristics are described in Table S3. Multivariate analyses (MV) of clinical and biological parameters associated with hepcidin and SF or posttransplant SF peak are displayed in Table 1. OS, PFS and NRM. After median follow-up of 9.3 years, OS was 81% at 100 days, 60% at 1% and 32% at 10 years (median 25.1 months; Figure 1B). Patients in Q3–Q4 SF and Q4-hepcidin groups had significantly lower OS (Figure 1C,D). In univariate analyses (UV), high hepcidin and SF were associated with decreased long-term OS (Table S4a). In bivariate analyses, SF was a significantly stronger predictor (Table S4b). In MV, hepcidin lost significance, whereas SF remained associated with poor long-term OS (Table S4c). PFS was 75% at 100 days, 50% at 1% and 26% at 10 years (median 12.1 months; Figure 1F). Patients in Q4-SF and Q4-hepcidin groups had inferior PFS (Figure 1G,H). In UV, high hepcidin and SF were associated with reduced PFS (Table S5a). In bivariate analyses, SF remained the best predictor (Table S5b). In MV, both hepcidin and ferritin lost significance for long-term PFS (Table S5c). Relapse incidence at 100 days, 1 and 10 years was 10%, 30%, and 44%, respectively, with no correlation with SF or hepcidin. In MV, DRI (HR 2.88, p < 0.0001), age (HR 1.02, p = 0.006), treatment lines (HR 1.11, p = 0.03), and total bilirubin (HR 1.44, p = 0.046) were significantly associated with long-term relapses. Among 185 AL patients, median pretransplant hepcidin was 36 ng/mL (63% in Q3-4) and SF 1548 ng/mL (73% in Q3-4). While hepcidin did not correlate with relapse, higher SF did (30% vs. 18%; Figure 1E). In MV, only DRI remained predictive. NRM was 14% at 100 days, 22% at 1% and 34% at 10 years, and higher in Q4 hepcidin (p = 0.02) and Q3–4 SF groups (p = 0.03). In UV, only pretransplant SF was significantly associated with long-term NRM, whereas hepcidin was not (Table S6a), hence no bivariate model was constructed. Neither of them emerged in MV as significant risk factor of NRM (Table S6b). Determinants of SF and hepcidin and their combined prognostic impact on outcomes remained underexplored. Previous studies exploring hepcidin were limited by small patient numbers (16–166), use of less reliable ELISA, and lack of MV analyses. Our study, the largest with 502 patients, confirmed elevated hepcidin by mass spectrometry and its strong positive correlation with SF [7], and MV demonstrated its dependence on RBC transfusions, HCT-CI, PS, and inflammation, SF also remaining an independent predictor. This confirms our previous finding that the stimulating impact of IO dominates the inhibitory effect of erythropoiesis on hepcidin [3, 6] Many studies reported elevated SF pretransplant, often linked to LIC, serum iron or Tsat, CRP or interleukin-6, ALAT, anemia, male sex, age, AL, previous treatment lines, advanced disease, RBC transfusions, high HCT-CI, or poor PS. None included MV analyses. We performed comprehensive MV, and SF correlated with anemia, Tsat, liver and renal dysfunction, previous RBC and platelet transfusions, AL, and PS, hepcidin being also associated with it. Hence, pre-conditioning SF reflects not only transfusional IO but also the cumulative burden of comorbidities, organ dysfunction, and disease severity that shapes transplant outcomes. Our next objective was to evaluate the influence of SF and hepcidin on survival outcomes, the literature reporting conflicting results [1, 2, 4]. High hepcidin and SF were significantly associated with poorer OS and PFS or increased NRM in UV. However, only SF retained predictive value in MV for OS, whereas neither was predictive of PFS or NRM, consistent with previous studies. Although our study is by far the largest, only Sakamoto observed a significant relationship between high hepcidin, but not SF, and poor OS in UV and MV [7]. Conversely, our results align with others, supporting the negative impact of SF on OS, PFS and NRM [8]. A meta-analysis of 25 studies (4545 patients) found pretransplant SF strongly associated with poor OS (HR 1.82; p < 0.001) and PFS (HR 1.72; p < 0.001) and increased NRM (HR 2.28; p < 0.001) [1]. Relapse rates were not influenced by SF or hepcidin overall but, in AL patients, higher SF correlated with relapse, agreeing with other reports, although this was not confirmed in MV. Elevated hepcidin and SF in AL likely reflect disease-related iron dysregulation, while the potential contribution of HFE mutations warrants further investigation. The array of parameters analyzed in our study unveiled predictors rarely considered by others. The literature shows that limiting analyzed factors to a few known predictors in addition to SF itself may result in SF being incorporated too often while ignoring potentially better clinical or biological markers. We took a less restrictive approach for selecting potential outcome predictors. Pretransplant RBC transfusions probably excluded SF from MV of NRM (Table S6b). Other studies identified DRI or similar parameters (diagnosis, disease status, treatment lines, advanced disease, prior transplant) alongside SF as powerful predictors of relapse, PFS, or OS [7]. Prognostic score proposed by Vaughn [9], based on age, disease type and stage, cytogenetics, and SF, resulted in three groups with 5-year OS of 56%, 22%, and 5%. Artz constructed a score based on SF, CRP, and albumin in 784 patients, leading to three groups with 1-year OS of 74%, 67%, and 56% [8]. As SF reflects multiple determinants, the role of iron overload per se remains uncertain. A meta-analysis of four MRI-based studies of LIC (276 patients) before allo-HSCT showed no significant impact of liver iron stores on OS or NRM, except for higher NRM among RIC-conditioned patients [5]. However, two of the four studies showed that high LIC was associated with higher NRM and poorer OS. In conclusion, our study shows that pretransplant SF and hepcidin are influenced by multiple factors, affect allo-HSCT outcomes, and require comprehensive assessment for accurate prediction, highlighting the complexity of iron metabolism and the need for further research [10]. This work was supported by the FNRS (Fonds National de la Recherche Scientifique)—Télévie and the Léon Fredericq Foundation at the University of Liège in Belgium. The study was approved by the Ethics Committee of the Faculty and the University Hospital at the University of Liège under # 2016-119. Patients (or their guardian if of minor age) signed an Ethics Committee-approved informed consent form for the collection of clinical data as well as blood samples for research purposes. The authors declare no conflicts of interest. The data that support the findings of this study are available from the corresponding author upon reasonable request. Data S1: ajh70272-sup-0001-Supinfo.docx. 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.