Pre-transplant serum ferritin is prognostic but is it useful?

Refined prognostication in hematopoietic cell transplant (HCT) has generated considerable attention because it might enable better patient selection, allow comparisons across studies, and perhaps even permit optimization of transplant strategies. Serum ferritin has emerged as a leading candidate to advance prognostication. Armand et al. initially observed that in patients with myelodysplastic syndromes (MDS) and acute myeloid leukemia (AML), serum ferritin in the top quartile (generally above 2500 ng/mL) was associated with an adverse prognosis [1]. Numerous subsequent reports confirmed that elevated pre-transplant ferritin confers an adverse prognosis, primarily in AML and MDS [2–4].

The biologic underpinnings of the association between high ferritin and inferior survival are likely multifactorial. In routine practice, ferritin represents a measure of body iron stores. Iron overload could affect transplant outcomes by impairing hepatic or cardiac function, or excessive iron may promote oxidative damage. However, inflammation also increases ferritin levels, and high inflammatory levels also confer an adverse prognosis [5]. Finally, excessive iron usually derives from red cell transfusions. Incremental blood transfusion requirements constitute by themselves a measure of more unfavorable disease, particularly for myelodysplastic syndromes. The adverse impact of serum ferritin on outcome has been linked in some studies to increased disease relapse, whereas other series suggest an effect on toxicity or transplant-related mortality, intimating that iron overload, inflammation and disease may all be relevant.

In the article that this Commentary accompanies, Tachibana and colleagues report a multicenter observational study to validate the utility of serum ferritin in patients with MDS and AML prior to allografting [6]. Previously, Tachibana created a simple risk stratification system including high-risk disease (present or absent) and high serum ferritin (present or absent), generating three prognostic categories in a single-center experience [7]. They also showed that serum ferritin remained prognostic when stratifying by disease risk [8]. The current study consisted of 153 recipients from three institutions transplanted over a one-decade period. The donors and conditioning regimens varied. Presumably the ferritin assay was performed in the clinical laboratory. The median ferritin value was 1139 ng/mL. Serum ferritin above 1000 ng/mL retained an adverse prognosis (hazard ratio = 1.8, 95% confidence interval 1.1–2.9, p = 0.018) in multivariate models, as did advanced disease. The three-point system previously described identified distinct survival patterns. The poor prognosis group with advanced disease and higher ferritin only achieved 20% survival at 5 years (n = 36), compared to 42% for the intermediate risk group (n = 82) and 68% for the group with no risk factors (n = 35). Most of the adverse prognosis appeared to be driven by relapse based on the profound effect on disease-free survival (p = 0.001) as opposed to non-relapse mortality (p = 0.083).

It is worth noting that the analysis was restricted to those patients for whom a serum ferritin was available. It is conceivable that this set of patients were a subgroup with a high transfusion burden or elevated liver enzymes. This type of selection represents a limitation, and limits generalization of the conclusions. That said, a large number of reports now point to ferritin as an adverse prognostic factor.

So why has the transplant community not adopted ferritin for routine use? Several questions remain unanswered. We still lack studies of entire unselected populations (rather than those where a ferritin assay was performed for clinical reasons). The optimal threshold values for ferritin require validation, and variability of laboratory assays must also be considered. Prior thresholds have varied from 500 ng/mL to above 2500 ng/mL. More confusing may be a proliferation of potential biomarkers and other prognostic markers among which the independent role, if any, of ferritin must be established. For example, will ferritin retain prognostic value when comorbidity scores, C-reactive protein or other prognostic tools are taken into account?

Even more controversial are the potential therapeutic implications of an elevated serum ferritin. Armand et al. reported seemingly conflicting results in that serum ferritin correlated with hepatic and cardiac iron as assessed by magnetic resonance imaging (MRI), but in a follow-up study found ferritin to be prognostic, but not iron content by MRI [9]. Iron chelation to mitigate the adverse impact of ferritin on survival seems premature.

The paper by Tachibana and colleagues provides at least a reasonable way to integrate ferritin into prognostication. Still, a 5-year survival rate of 20% in the worst risk category, which by definition includes patients with advanced disease, may still constitute a reasonable outcome for many with otherwise incurable leukemia or MDS. Finally, to the extent that high serum ferritin levels correlate better with disease relapse than with transplant related mortality, reducing regimen intensity for those with high ferritin levels could be harmful.

In summary, higher ferritin predicts a worse outcome for patients with MDS and AML after allografting, but more studies will be needed to determine how to integrate ferritin in prognostication and in treatment decisions.

Potential conflict of interest

Disclosure forms provided by the authors are available with the full text of this article at www.informahealthcare.com/lal.