Abstract
Follicular lymphoma (FL) is the most common type of low-grade non-Hodgkin’s lymphoma. A subset of FL patients shows a favorable treatment response, and remains in complete remission with long-term follow-up, while another subset of patients has a clinical course characterized by frequent relapses and a shorter survival. Survival predictors for FL are mainly based on clinical data, and they seem to lack accuracy enough predicting survival among patients with advanced stage disease. The aim of this study is to build a survival predictor based on a set of biological markers using Tissue Micro Arrays (TMA). To this purpose, we have retrospectively analysed the expression of a group of 40 selected genes - related with apoptosis control, cell cycle, B-cell differentiation and signaling- in a series of 192 FLs using TMA. The association of these molecules with survival, and their usefulness to discriminate among FLIPI groups was evaluated. Results were quantified using different tools; singularly nuclear markers were scanned using the Bliss system and the quantitative expression was measured using the TMAscore v.1.0 image analysis software (Bacus Laboratories, Inc.). The mean overall survival (OS) was 74 months, and 38 months for progression-free survival (PFS). Statistically significant differences in OS were found with the Follicular Lymphoma International Prognostic Index (FLIPI) score (p < 0.01). No significant OS or PFS differences were observed between FL grades 1–3, between grades 3a and 3b, or using Ki67 expression. Univariate analysis revealed several TMA markers with capacity to predict OS and PFS. After multivariate analysis, a set of 4 apoptosis and cell-cycle markers was integrated into a FLIPI-independent clinical predictor, with the capacity to recognize two groups of FL patients with statistically significant differences in OS (83% versus 43% of OS at 120 months; p< 0.001). Then, patients were classified into low-risk (FLIPI: 0–2) and high-risk groups (FLIPI: 3–5), and the protein-based predictor model was used in both groups. High-risk FLIPI patients were stratified by the protein-based predictor into two groups with OS probability of 79,3% versus 14,2% at 120 months, p < 0.001, and low-risk FLIPI patients were also separated into two groups with OS probability of 100% versus 60,4% at 120 months, p < 0.01. The model is now being validated in a blind set. These results suggest that an integrated use of the FLIPI and the protein-based model could reach a higher accuracy predicting survival in FL.
Author notes
Corresponding author
