Prognostic Impact of Elevated Serum Ferritin at Diagnosis of Diffuse Large B-Cell Lymphoma Treated In the Rituximab Era

Combination of rituximab (R) and anthracyclin-based chemotherapy (CT) regimens had greatly improved the overall survival (OS) of Diffuse Large B-Cell Lymphoma (DLBCL). It is challenging to identify more accurately patients who will progress during therapy or relapse after initial treatment. Currently, International Prognostic Index (IPI) is the better tool in daily practice to define at diagnosis the prognostic of DLBCL patients. New biological parameters in relation with the pathogenesis of DLBCL were developed but were not routinely used. In this context, iron homeostasis was not well explored in the pathogenesis of DLBCL and the markers of iron metabolism had not been analysed as prognostic markers for DLBCL patients treated with R-CT.

In this study, we assessed the prognostic role of serum ferritin on complete response (CR) after initial treatment, OS and Progression Free Survival (PFS) in a cohort of DLBCL patients treated by R-CT in first line therapy. Between March 2005 and December 2008, 87 DLBCL patients could have at diagnosis a serum ferritin dosage. Outcomes were estimated using the Kaplan-Meier method and compared between patients with or without elevated ferritin and in each quartile of ferritin dosage using the log-rank test.

The median age of patients was 67 years old (range, 23 – 96). Initial characteristics of whole series were as follow: male gender, 41%, stage III-IV 69%, elevated LDH 84%, number of extra-nodal sites >1 24%, Performance Status (PS) >1 22% and 69% of patients had an age-adjusted IPI (aaIPI) 2–3. Median ferritin was 268μ g/l (range, 27 – 3343), 45 patients (52%) had an elevated ferritin level. Ferritin level had no prognostic impact on CR at the end of initial treatment whether patients were compared with or without elevated ferritin or by quartile groups. The 2-year OS was 69% (95% confidence interval (CI) 53–81) for patients with elevated ferritin and 88% (CI 74 – 95) for patients with normal ferritin (p=0.04). The 2-year PFS was 58% (CI 43–73) for patients with elevated ferritin and 75% (CI 59 – 86) for patients with normal ferritin (p=0.09). The 2-year OS for patients in the first quartile (27-115 μ g/l) was 86% (CI 67–95); in the second quartile (116-267 μ g/l) 94% (CI 74–99), in the third quartile (268-465 μ g/l) 74% (CI 51–89) and in the fourth quartile (≥ 466 μ g/l) 58% (CI 38–76) (p=0.02). The 2-year PFS for patients in the first quartile was 72% (CI 51–87); in the second quartile 82% (CI 56–94), in the third quartile 58% (CI 36–77) and in the fourth quartile 54% (CI 34–73) (p=0.14). Patients could be classified using the fourth quartile (≥ 466 μ g/l) versus the lower 3 quartiles. The 2-year OS was 85% (CI 74–92) for patients with low ferritin (<466 μ g/l) and 58% (CI 38–76) for patients with elevated ferritin (≥ 466 μ g/l) (p=0.01). The 2-year PFS was 71% (CI 58–81) for patients with low ferritin (<466 μ g/l) and 54% (CI 34–73) for patients with elevated ferritin (≥ 466 μ g/l) (p=0.09). Interestingly, median age, sex, Ann Arbor stage, LDH level, PS, aaIPI, number of extra-nodal sites >1, CRP level (normal vs. abnormal) distribution were not statistically different between patients with low and elevated ferritin. Three prognostic classes integrated the aaIPI could be proposed: Low ferritin (<466 μ g/l) and aaIPI 0–1 (Group 1, 20 patients), Low ferritin and aaIPI 2–3 or high ferritin (≥ 466 μ g/l) and aaIPI 0–1 (Group 2, 52 patients), high ferritin and aaIPI 2–3 (Group 3, 15 patients) with respectively a 2-year OS of 100%, 77% (CI 63–87) and 53% (CI 30–75) (p=0.003) and a 2-year PFS of 90% (CI 70–97), 63% (CI 48–76), 47% (25-70) (p=0.03).

This exploratory study identifies a subgroup of patients with high aaIPI 2–3 and a high ferritin (≥ 466 μ g/l) who presented an immediate poor outcome. These results need to be confirmed in a large prospective series of DLBCL to know if serum ferritin could add prognostic information to IPI. These results also incline to decipher the role of iron homeostasis in the pathogenesis of DLBCL.

No relevant conflicts of interest to declare.