Association of BAX G(−248)A and BCL2 C(−717)A polymorphisms with outcome in diffuse large B-cell lymphoma patients

Disruption of the physiological balance between cell proliferation and apoptosis is an important feature of diffuse large B-cell lymphoma (DLBCL) (Muris et al, 2007). The crucial regulators proteins of apoptosis are the pro-apoptotic BAX and anti-apoptotic BCL2. The ability to induce apoptosis varies in humans, and this variability may alter susceptibility to DLBCL. A single nucleotide polymorphism (SNP) located in the 5′-untranslated region of the BAX gene with a G→A substitution at −248 nucleotide position (rs4645878) has its variant allele A associated with lower protein expression when compared with the G allele (Starczynski et al, 2005). The AA genotype produced by a C→A substitution at −717 nucleotide position located in the promoter region of the BCL2 gene (rs2279115) (Park et al, 2009) is associated with increased BCL2 protein expression in comparison with CC genotype (Nuckel et al, 2007). Low BAX (Sohn et al, 2003) and high BCL2 (Perry et al, 2014) expression were associated with unfavourable prognostic in DLBCL, but BCL2 C(−717)A SNP did not influence outcomes in disease in a unique study (Park et al, 2009). The present study aimed to determine whether these SNPs affect clinicopathological features and outcome in DLBCL patients.

The study comprised 150 de novo DLBCL patients seen at diagnosis between December 2007 and June 2014. All procedures were carried out according to the Declaration of Helsinki, and the study was approved by the local Ethics Committee. DLBCL was classified as germinal centre B-cell (GCB) or non-GCB histological subtype (Meyer et al, 2011). Patients were treated with curative intent with six cycles of R-CHOP (rituximab, cyclophosphamide, doxorubicin, vincristine, prednisolone). Genotypes were obtained in DNA of peripheral blood by polymerase chain reaction and enzymatic digestion. The differences between groups were analysed by the Chi-square (χ²) or Fisher exact test. Multivariate analysis using the logistic regression model served to assess associations between genotypes and clinicopathological features. Progression-free survival (PFS) and overall survival (OS) were calculated from the date of diagnosis to date of progression, death from disease or last follow-up, and date of death resulting from any cause or lost to follow-up, respectively. Survival probabilities were estimated using the Kaplan–Meier method and curves were compared by the log-rank test. The Cox hazards model was used to identify prognostic variables influencing the PFS and OS in univariate and multivariate analyses.

B symptoms were more common in patients with elevated serum lactate dehydrogenase (LDH), an inflammatory protein, compared to those with normal LDH levels (61·4% vs. 38·6%, P = 0·001). The BCL2 AA genotype was less common in patients who obtained complete response (CR) than in those with other responses to chemotherapy (21·7% vs. 45·0%, P = 0·03) (Table 1). Patients with the BCL2 AA genotype had a 2·95 [95% confidence interval (CI): 1·09–7·98]-fold increased risk of not achieving CR than those with BCL2 CC or CA genotype. The activation of apoptotic proteases is a critical step in treatment response and so, carriers of the BCL2 AA genotype, related to high expression of the anti-apoptotic protein, could be more resistant to chemotherapy. This hypothesis is corroborated by a study conducted in lymphocytic leukaemia cell lines, in which BCL2 expression was directly associated with chemoresistance (Reed et al, 1994).

The median follow-up time of overall DLBCL patients enrolled in the study was 35 months (1–77). The 2-year PFS and OS for all patients were 73·3% and 76·2%, respectively. At 2 years of follow-up, PFS and OS were shorter in patients with B symptoms (68·0% vs. 84·0%, P= 0·01; 71·1% vs. 86·8%, P = 0·04), non-GCB subtype (57·6% vs. 77·1%, P = 0·02; 62·8% vs. 78·4%, P = 0·02), Ann Arbor stage IV (61·5% vs. 80·8%, P = 0·004; 65·3% vs. 83·2%, P = 0·01), National Comprehensive Cancer Network International Prognostic Index (NCCN-IPI) of high intermediate or high risk (63·7% vs. 84·7%, P = 0·009; 67·7% vs. 86·4%, P = 0·002), and BCL2 CA or AA genotype (69·1% vs. 84·7%, P = 0·02; 72·1% vs. 87·4%, P = 0·01) (Kaplan–Meier estimates). The significance of differences between groups remained the same in Cox analyses. Univariate and multivariate analyses showed that patients with the BCL2 CA or AA genotype had a 2·45- and 3·12-fold chance of relapse, and a 2·98- and 3·86-fold higher risk of progressing to death other patients, respectively. Patients with the variant A allele (GA or AA genotype) of BAX G(−248)A SNP had a 4·01- and 3·86 fold chance of relapse and fatal evolution, respectively, than those with the GG genotype, only in multivariate analysis (Table 2).

In this study, lower PFS and OS were seen in patients with non-GCB subtype, Ann Arbor stage IV and NCCN-IPI of high intermediate or high risk, as previously reported. Patients with B symptoms also presented lower PFS and OS in this study, and this finding was occasionally described in rituximab era. We hypothesized that B symptoms may have negatively altered our patients’ outcome due to their association with LDH, as inflammatory status may favour malignant cell proliferation (Seymour et al, 1995).

BAX immunonegativity was associated with unfavourable prognosis in DLBCL patients treated with CHOP or CHOP-like regimen (Sohn et al, 2003). This finding indicates that reduction of BAX proapoptotic protein, favouring proliferation of tumour cells, negatively influences outcome in DLBCL patients.

Increased BCL2 expression was associated with worse prognosis in a group of DLBCL patients treated with radiotherapy and CHOP or CHOP-like regimens (Sohn et al, 2003), and in those treated with R-CHOP (Perry et al, 2014), possibly due to the fostering of tumour cell proliferation. In DLBCL, the BCL2 C(−717)A SNP alone did not alter OS in Korean DLBCL patients treated with R-CHOP (Park et al, 2009). Distinct ethnic origin of populations (BCL2 AA genotype frequency) may constitute a plausible explanation for the differences found by Park et al (2009) and our study.

In conclusion, our data present preliminary evidence that inherited abnormalities in the intrinsic apoptosis pathway, related to BAX G(−248)A and BCL2 C(−717)A SNPs, are associated with treatment response and act as independent prognostic factors in DLBCL. If these findings are confirmed in larger studies, they might help to determine optimal patient choice for treatment with apoptosis-targeted drugs (Davids et al, 2014).

Author contributions

BAB carried out the genetic study, analysed the data and wrote the manuscript. DMT, FMF, SFA and SCA contributed with clinical data. OC carried out the genetic genotyping. LCS and VJ designed and supervised the study.

Conflict of interest

The authors have no competing interests.