Validation of the Elderly IPI (E-IPI) for Patients with Diffuse Large B Cell Lymphoma Using An Independent Data Set From the RICOVER-60 Trial of the German High-Grade Non-Hodgkin Lymphoma Study Group (DSHNHL)

The International Prognostic Index (IPI) is a robust clinical prognostic tool in diffuse large B cell lymphoma (DLBCL) and identifies four risk groups with predicted five-year survival rates. We have previously reported that an alternative index, the E-IPI (age cut off 70 years rather than 60 as used in the IPI) provided better discrimination in outcome for patients >60 years with DLBCL treated with RCHOP in the US Intergroup trial E4494 (Advani et al BJH 2010). The aim of this study was to validate our findings in an independent data set.

We compared the conventional IPI and the E-IPI in the RICOVER-60 data set of the German High-Grade Non-Hodgkin Lymphoma Study Group (DSHNHL) in which patients with DLBCL >60 years were treated with RCHOP-14. Patient distribution and overall survival (OS) were assessed according to the standard IPI and E-IPI. OS was estimated by the Kaplan-Meier method and compared across prognostic groups of the IPI and E-IPI using the log-rank test. Cox proportional hazards model was used to estimate the hazard ratio (HR) with the low risk group as the reference group. Performance of indices was compared by a measure of global fit (Akaike's information criteria, AIC), concordance measures CPE (concordance probability estimate) and C-statistic (or AUC). Low values of AIC indicate a better fit, whereas high values of CPE and C-statistics indicate better discrimination. The AUC (area under the receiver operator characteristic curve) over time was calculated for both the IPI and the E-IPI.

Six hundred and ten patients treated with R-CHOP-14 had a median observation time of 34.3 months. Patient characteristics were: stage III/IV, 50%; >1 extranodal site, 18%; elevated LDH, 50%; ECOG performance status ≥2, 15%. The median age was 68.5 years with 37% greater than 70 years. On univariate analysis all of these characteristics were significant for 3 year OS. The IPI and E-IPI both distinguished 4 prognostic groups (low, low intermediate, high intermediate and high risk). In E-IPI versus IPI more patients were categorized under the low risk group (47% versus 30%) and fewer as high intermediate (15% versus 25%). Both the IPI and E-IPI provided prognostic discrimination for OS for the 4 groups (Table I). The C-statistic at 2 y for OS are 0.659 (95% CI: 0.596, 0.721) for IPI and 0.679 (95% CI: 0.616, 0.742) for E-IPI. The AUC ranked the E-IPI higher than the IPI over all event times. Similar rankings were obtained using the global model fit criterion (AIC) and discrimination measure (CPE).

For patients >60 years treated with R-CHOP in the RICOVER-60 study, both the IPI and E-IPI provided prognostic discrimination for OS; however, the E-IPI outperformed the conventional IPI validating our previous findings. The E-IPI is a useful index to incorporate in studies for risk stratification of patients > 60 years with DLBCL.

Pfreundschuh:Roche: Membership on an entity's Board of Directors or advisory committees, Research Funding.