Iron Overload in Allogeneic Stem Cell Transplantation Outcome: A Meta-Analysis

An elevated serum ferritin before allogeneic stem cell transplantation (HSCT) is an adverse prognostic factor for overall survival (OS) and non-relapse mortality (NRM); but because ferritin is an imperfect surrogate of iron stores, the exact prognostic role of iron overload remains unproven. Four prospective studies using liver MRI to estimate liver iron content (LIC) have examined the impact of elevated LIC on HSCT outcome (Armand et al. Am J Hematol. 2012;87(6):569, Wermke et al. Clin Cancer Res. 2012;18(23):6460, Virtanen et al. Eur J Haematol. 2013;91(1):85, Trottier et al. Blood. 2013;122(9):1678), but have reached different conclusions on the prognostic relevance of LIC in HSCT. We therefore conducted an individual patient data meta-analysis of those 4 studies, using a random effects model. The outcomes of interest were OS and NRM. 276 patients were included in this analysis. The median serum ferritin pre-HSCT was 1523 (range, 20-8,878) ng/ml, and the median LIC was 5.0 (range, 0.3-25.4) mg/gdw. 38% of patients received a myeloablative conditioning regimen. With a median follow-up of 22 months for survivors, the estimated 2-year OS was 62% (95CI, 55-68), and the NRM was 25% (95CI, 20-31). The pooled hazard ratio (HR) for mortality associated with a pre-HSCT LIC>7 mg/gdw (the primary endpoint) was 1.39 (95CI, 0.85-2.27), p=0.18 (Figure 1A).  The pooled hazard ratio (HR) for mortality associated with a pre-HSCT ferritin >1000 ng/ml was 1.68 (95CI, 1.03-2.73), p=0.036 (Figure 1B).  

Figure 1. Forest plot of the hazard ratios for mortality associated with LIC>7 mg/gdw (Panel A) and ferritin>1000 ng/ml (Panel B).

There was no significant association between LIC>5 mg/gdw and OS, or between ferritin>2500 ng/ml and OS. The significant association between ferritin >1000 ng/ml and OS persisted in the subgroup of patients with MDS or acute leukemia, as did the absence of significant association between OS and LIC. There was also no significant association between LIC and OS among patients conditioned with a myeloablative regimen. In analyses of NRM, there was a trend towards increased NRM for LIC>7 mg/gdw (HR=1.74, p=0.09) and for ferritin>1000 ng/ml (HR=1.55, p=0.14) but neither reached statistical significance. There was no significant association between elevated ferritin or LIC and NRM among patients who received a myeloablative conditioning or among acute leukemia/MDS patients. In conclusion, despite the confirmed adverse prognosis conferred by pre-HSCT hyperferritinemia, our meta-analysis suggests that iron overload, as assessed by MRI-determined liver iron content, is not a significant factor for OS or NRM. This raises the possibility that the deleterious prognostic impact of ferritin on survival is related to other issues such as inflammatory status, infection, or disease characteristics. These results may inform future research on the use of chelation therapy in HSCT.