Yan Qin, Xiaohui He, Shiyu Jiang, Haizhu Chen, Shengyu Zhou, Liqiang Zhou, Peng Liu, Lin Gui, Yuankai Shi, Hongxin Jiang, Changgong Zhang, and Jianliang Yang
e19549 Background: Unlike BCL2 translocation ( BCL2TR), limited data about the prognostic significance of BCL2 mutations ( BCL2MUT) and BCL2 copy number variations in diffuse large B-cell lymphoma (DLBCL) is available, especially in the era of immunochemotherapy. Predicting outcomes at the time of diagnosis is of extreme importance. Methods: In this study, we performed probe capture-based high resolution sequencing on 191 Chinese patients with de novo DLBCL, attempting to comprehensively describe BCL2 genetic alterations and several other genes. We also examined correlation between genetic alterations and outcomes in 164 patients treated with R-CHOP. BCL2, MYC and BCL6 protein was determined by immunochemistry. Results: Of all the 191 patients, BCL2 gain/amplificatio n ( BCL2GA/AMP) and BCL2MUT occurred in 9.4% and 8.9% of patients, respectively, and only 4.2% of cases harbored BCL2TR. As a result, BCL2 alteration, comprised of the above three genetic alterations, was observed in 18.3% of patients. Compared to patients with the absence of BCL2 alteration, the 5-year progression-free survival (PFS) (13.7% vs 40.8%, P= 0.003) and overall survival (OS) (34.0% vs 70.9%, P= 0.036) were significantly inferior in cases that harbored BCL2 alteration. Importantly, patients who harbored BCL2GA/AMP also had remarkably inferior PFS (5-year PFS, 11.1% vs 38.3%, P< 0.001) and OS (5-year OS, 22.1% vs 69.6%, P= 0.009) compared to those without BCL2GA/AMP. By contrast, neither BCL2MUT nor BCL2TR were significantly prognostic for survival. Multivariate analysis showed that the presence of BCL2 alterations, especially BCL2GA/AMP, TP53 mutation and International Prognostic Index (IPI) were significantly associated with inferior PFS and OS. Novel prognostic models were constructed based on 3 risk factors, including BCL2 alteration (Model 1) or BCL2GA/AMP (Model 2), TP53 mutation and IPI, stratifying patients into three risk groups with different outcomes. According to Model 1, the 5-year PFS were 65.9%, 25.9%, and 12.9% ( P< 0.001), and 5-year OS were 89.4%, 65.0% and 16.4% ( P< 0.001), respectively, for the low-risk, intermediate-risk and high-risk groups. In Model 2, the 5-year PFS was 56.9% for the low-risk, 24.1% for the intermediate-risk and 14.3% for the high-risk group ( P< 0.001), and 5-year OS were 86.6%, 60.9% and 16.9% for the three risk groups ( P< 0.001), respectively. Conclusions: This study demonstrated that, in patients treated with R-CHOP, the presence of BCL2 alteration, especially BCL2GA/AMP, and TP53 mutation were significantly associated with inferior outcomes independent of IPI. The novel prognostic models we proposed could aid in individual risk prediction for DLBCL patients treated with R-CHOP. Further validation of this prognostic model is warranted.