Prognostic assessment of BCL2-938C>A polymorphism in chronic lymphocytic leukemia

To the editor:

The recent report by Nückel et al pointed to the prognostic role of the BCL2-938C > A polymorphism in chronic lymphocytic leukemia (CLL).¹  In this report, the BCL2-938 AA genotype emerged as an independent predictor of shorter treatment-free-survival (TFS) both by univariate and multivariate analysis.¹  We aimed at testing the prognostic impact of BCL2-938C > A in an independent series of 182 consecutive chronic lymphocytic leukemia (CLL) from our institution.

Genotyping of BCL2-938C > A was performed by single nucleotide polymorphism (SNP) minisequencing (ABI Prism SNaPshot Multiplex kit; Applied Biosystem, Foster City, CA), after validation of this approach by DNA direct sequencing in a pilot panel of cases (n = 30; data not shown).²  The genotype distribution of BCL2-938C > A in CLL was 47 of 182 (25.8%) AA, 96 of 182 (52.7%) AC, and 39 of 182 (21.5%) CC. Allele frequencies were in accordance with Hardy-Weinberg equilibrium (A: 0.522; C: 0.478; P = .749, χ)² . Biologic and clinical variables at diagnosis distributed without significant differences among AA, AC, and CC genotypes (Table 1). In our series, TFS and overall survival (OS) did not differ in AA, AC, and CC genotypes when analyzed separately or after pooling C-allele carriers (Tables 1, 2).

The independent prognostic value of BCL2-938C > A was assessed separately for biologic and clinical variables by Cox multivariate analysis. Biologic covariates were CD38 expression, immunoglobulin heavy chain variable gene (IGHV) mutation status, IGHV3-21 usage, and FISH karyotype. Clinical covariates were age, sex, Rai and Binet stage, number of nodal lesions, largest lymph node size, splenomegaly, lymphocyte count, hemoglobin, platelet count, lymphocyte doubling time (LDT), bone marrow lymphocytes, bone marrow infiltration pattern, beta-2-microglobulin, LDH, alkaline phosphatase, and albumin. When tested along with biologic covariates, BCL2-938C > A in our CLL series was not an independent predictor of TFS (Table 2). In addition, when tested along with clinical covariates, BCL2-938C > A in our CLL series was not an independent predictor of TFS (Table 2). As observed for TFS, multivariate analysis did not identify BCL2-938C > A as a predictor of OS in our series.

The discrepancy between our results and the study by Nückel et al cannot be ascribed to differences in BCL2-938C > A genotype distribution (this study: AA = 25.8%, AC = 52.7%, CC = 21.5% vs Nückel et al: AA = 34.1%, AC = 44.7%, CC = 21.1%; P = .263, χ²) or allele frequency (this study: A = 0.522 vs Nückel et al: A = 0.57; P = .301, χ²) between the 2 CLL series.¹  Other potential explanations may account, at least in part, for the observed discrepancy. In the study by Nückel et al, 17 of 37 (45.9%) AA patients were in Binet stages B or C versus 17 of 77 (22.1%) AC/CC patients (P = .005, χ², causing a potential bias in the distribution of unfavorable Binet stages among BCL2-938C > A genotypes.¹  Conversely, in our study, 12 of 47 (25.5%) AA patients were in Binet stages B or C versus 38 of 135 (28.1%) AC/CC patients (P = .729, χ²) (Table 1). Excess of unfavorable Binet stages among AA patients may have powered the prognostic relevance of BCL2-938C > A observed by Nückel et al.¹  This potential bias is notable because Nückel et al reported that unfavorable Binet stage was an independent predictor of TFS stronger than BCL2-938C > A.¹  Alternatively, the discrepancy between this study and that of Nückel et al may be the consequence of the different genetic background of the 2 populations investigated. In fact, genetic variability due to ethnicity may influence the prognostic value of specific polymorphisms in CLL and other cancer settings.^3,,,–7  In this respect, it is remarkable that the prognostic assessment of polymorphisms of other genes, namely P2X7 and BAX, has yielded conflicting results in CLL.^5,7,,–10