International prognostic index (IPI) has been widely used for predicting outcome in diffuse large B cell lymphoma. Although the introduction of rituximab to chemotherapy has dramatically improved the outcome of DLBCL, it also changed the prognostic value of IPI. The National Comprehensive Cancer Network IPI (NCCN-IPI) and age-adjusted IPI (aaIPI) were used to evaluate the prognosis for DLBCL in the rituximab era. However which one of them is more powerful in predicting survival remains unknown in Chinese patients.

Patients and Methods

A total of 334 patients with de novo diffuse large B-cell lymphoma diagnosed from 2003 to 2012 were included. All patients were treated with CHOP with or without rituximab. They were divided into CHOP and R-CHOP groups. IPI, NCCN-IPI and aaIPI score were recorded. Survival was performed according to the Kaplan-Meier (K-M) curves using the log-rank test. The predictive abilities of PI, NCCN-IPI and aaIPI were investigated by Harrell's C-statistics.


Compared with the IPI, the NCCN-IPI and the aaIPI had better discrimination in different scores in all patients treated with or without rituximab. Both the NCCN-IPI and aaIPI had power in discriminating low, low-intermediate and high-intermediate risk groups. In comparison, the NCCN-IPI discriminated low-intermediate and high-intermediate risk groups (5-year overall survival [OS]: 74% vs 50%) better than the aaIPI (5-year OS: 80% vs 62%) in all patients and in the R-CHOP group with the NCCN-IPI 5-year OS: 85% vs 67% while the aaIPI: 87% vs 77%. In the CHOP group, the aaIPI discriminated low-intermediate and high-intermediate risk groups better (5-year OS: 70% vs 25%) than the NCCN-IPI (5-year OS: 60% vs 24%). According to the Harrel C statistic, the NCCN-IPI showed higher discrimination of the different scores (C_index: 0.669 ) than the aaIPI (C_index: 0.663 ) in all the patients (C_index: 0.669 vs. 0.663), especially in the R-CHOP group (C_index: 0.681 vs. 0.636). In the CHOP group, the aaIPI had stronger power (C_index: 0.711 ) than the NCCN-IPI (C_index: 0.683 ) in discriminating different scores.


The NCCN-IPI is more powerful than the IPI for predicting survival in the rituximab era. In patients aged 60 or younger and treated with CHOP without rituximab, aaIPI is a preferable tool for evaluating prognosis.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.