Serum albumin (SA) has been shown to be a prognostic marker in many hematological malignancies, and in diffuse large B-cell lymphoma (DLBCL) before the introduction of rituximab. SA may be a surrogate for age, comorbid status, and disease severity. Here, we aimed to assess the role of SA as prognostic marker for overall survival (OS) in patients with newly diagnosed DLBCL treated with and without rituximab (R) containing regimens (CHOP and CHOP-like).

Patients and method

We collected 738 patients complete for IPI factors [age, LDH, extra nodal sites, stage and performance status (PS)] and SA recorded in the GISL database from 1998 to 2012. OS was estimated using the Kaplan-Meier method, with Cox proportional hazard model used to identify potential risk factors for time-to-event data. Optimal cut-off for SA was calculated by means of ROC curve at 5 years of follow-up. Stability of the cut-point was analyzed using bootstrap techniques. The role of SA was explored interacting SA with R and adjusting by IPI. The strength of SA as prognostic factor was evaluated counting the bootstrap frequency of inclusion (BIF, 1000 re-samples) of SA, adjusted by the IPI factors, in Cox PH regression.


Of the 738 patients included in the study 322 (44%) were treated with R-CHOP and 416 (56%) were treated without regimens containing R. The median age at diagnosis was 60 years (range 21-89) and 45%, 25% and 30% were categorized in IPI 0-1, 2 and 3/5 respectively; and the median SA was 3.8 g/dL (range 1.0-6.4). Patients treated with R showed a greater % of high IPI 3/5 (43%) compared to those not treated with R (20%, P<0.001) and a lower SA level (3.7 vs 3.9 g/dL, P=0.001). SA showed correlation with IPI (4.0, 3.8 and 3.5 g/dL in IPI 0-1, 2 and 3/5, respectively, P=0.0001).

Overall, the median follow-up was 64 months (range 1-157) with 241 dead. The 5-ys OS was 66% (95 CI 62-70) and the best cut-off for SA was 3.7 g/dL with ROC=68.5%. OS at 5-yrs in patients not treated with R was 74% and 52% for SA ≥3.7 and <3.7 g/dL, respectively (P<0.001). In patients treated with R the 5-yrs OS was 78% and 50% for SA ≥3.7 and <3.7 g/dL (P<0.001). The SA <3.7 g/dL, interacted by R and adjusted by IPI, had HR = 1.71 (95 CI 1.22-2.40, P=0.002) and HR = 1.99 (95 CI 1.31-3.01, P=0.001) in cases not treated and treated with R, and the HRs did not vary between treatments (P=0.582). The BIF of SA<3.7 g/dL adjusted by IPI factors was 88% and 79% in patients without and with R chemotherapy.


The SA is a simple variable collected from the clinical-chemistry, that showed a good prognostic property in DLBCL. Although correlated with the IPI, SA has maintained a strong prognostic value both before or during the rituximab era. The biologic role of SA over the survival is unclear. It has been hypothesized that low level of SA may be associated with 1) increased inflammatory response to the tumor, 2) increased cytokine release and 3) poor nutritional status, and due to this condition the patients would not be able to receive the full cycles of chemotherapy.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.

Sign in via your Institution