LMO2 Expression and the Hans Algorithm in Predicting Germinal Center Phenotype and Survival in Diffuse Large B-Cell Lymphoma Treated with Rituximab
PN Meyer, DD Weisenburger, WL Choi, LM Smith, TC Greiner, P Aoun, J Delabie, RM Braziel, JM Vose, G Lenz, LM Staudt, WC Chan, K Fu. University of Nebraska Medical Center, Omaha, NE; University of Hong Kong Faculty of Medicine, Hong Kong, China; University of Oslo Medical Center, Oslo, Norway; Oregon Health and Science University, Portland, OR; National Cancer Institute, Bethesda, MD
Background: Diffuse large B-cell lymphoma (DLBCL) has wide variation in survival. Gene expression profiling (GEP) shows DLBCL derived from germinal center B cells (GCB) have better prognosis. Immunohistochemical techniques to predict the cell of origin, such as the Hans algorithm (Blood 103:275, 2004), have been developed. Recent reports show that LMO2 also predicts outcome in DLBCL. Our goal was to study LMO2 expression and the Hans algorithm in predicting GCB phenotype and survival in DLBCL treated with rituximab.
Design: Tissue microarrays (TMA) were created from 175 cases of DLBCL treated with rituximab in combination with CHOP-like therapies. LMO2 expression was determined by immunohistochemistry. Analysis of CD10, BCL6, and MUM1 expression divided the cases into GCB or non-GCB types (Hans algorithm). GEP by microarray analysis was performed on 57 cases. The relationships between LMO2 expression, GCB phenotype, GEP, as well as overall and event-free survival, were analyzed.
Results: LMO2 expression showed good agreement with the Hans algorithm (p<0.0001, kappa 0.54) or GEP (p=0.0001, kappa 0.50). Using GEP as the gold standard, the Hans algorithm showed greater sensitivity (90% vs. 84%), specificity (73% vs. 65%), positive predictive value (80% vs. 74%), negative predictive value (86% vs. 77%), and concordance (82% vs. 75%) when compared to LMO2; however, the differences were not statistically significant. LMO2 expression and the Hans algorithm were both predictive of overall (p=0.0023 and p=0.026, respectively) and event-free survival (p=0.0038 and p=0.027, respectively).
Conclusions: Our results demonstrate that the Hans algorithm and LMO2 are similar in predicting the GEP, and confirm recent reports that LMO2 is predictive of survival in DLBCL. Unlike some reports, we found that the Hans algorithm is also predictive of survival in patients treated with rituximab.
Wednesday, March 11, 2009 9:30 AM
Poster Session V # 178, Wednesday Morning