Impact of the host genetic background on prognosis of chronic lymphocytic leukemia

To the editor:

Sellick et al¹  showed the prognostic role of single nucleotide polymorphisms (SNPs) in chronic lymphocytic leukemia (CLL). The primary end point of the study was progression-free survival (PFS) after first treatment.¹  Several factors may limit the general applicability of these results. (1) PFS after first treatment applies to a CLL population biased for unfavorable characteristics compared with a consecutive CLL cohort. (2) The search for SNPs relevant for PFS after first treatment may be enriched with pharmacogenetic SNPs more related to chemotherapy than to CLL kinetics. (3) One of the SNPs identified by Sellick et al¹  remained significantly associated with CLL outcome after adjusting for multiple comparisons. (4) Validation in an independent CLL cohort was not performed.¹ 

Our aim was 2-fold: to explore the prognostic impact of SNPs identified by Sellick et al¹  on time to first treatment (TTFT) and overall survival (OS) from CLL diagnosis; and to validate the prognostic role on PFS and OS from first line treatment.

Forty-two SNPs from the study by Sellick et al¹  were selected for genotyping according to the following criteria: (1) involvement in immune regulation or DNA damage-response; and (2) functional relevance according to PolyPhen and SIFT algorithms.^2,3  Three additional SNPs belonging to the MDM2, CD38, and IRF4 genes and previously associated with CLL prognosis were also analyzed.^4-6 

After a median follow-up from diagnosis of 75 months, 163 of 331 patients were treated and 86 of 331 died. By Cox analysis adjusted for age at diagnosis, sex, stage, immunoglobulin gene (IGHV) mutations, CD38 expression, and TP53 deletion/mutation, SNPs significantly associated with TTFT were THBS1 rs2228262 (P = .007), ERBB2IP rs3213837 (P = .020), BARD1 rs2229571 (P = .026), LRP2 rs4667591 (P = .027), and LILRA4 rs2241384 (P = .034). In the same series, SNPs significantly associated with OS were MGMT rs12917 (P < .001), FHL5 rs9373985 (P = .042), and BARD1 rs2229571 (P = .017; Table 1). After correction for multiple testing by false discovery rate (FDR), a significant association with OS (q = .022) was demonstrated for MGMT rs12917 (Table 1), whereas no SNPs associated with TTFT (q ≥ .2 in all instances).⁷ 

After a median follow-up since first treatment of 60 months, 118 of 163 patients progressed, and 63 of 163 died. By Cox analysis adjusted for age at first treatment, sex, stage, IGHV mutations, CD38 expression, TP53 deletion/mutation, and type of treatment, SNPs significantly associated with PFS after first treatment were APOE rs7412 SNP (P = .011) and BARD1 rs2229571 (P = .041). In the same series, SNPs significantly associated with OS after first treatment were APOE rs7412 (P = .003), XRCC2 rs3218536 (P = .037), BARD1 rs2229571 (P = .043), KLRC4 rs1841958 (P = .047), and THBS1 rs2228262 (P = .047; Table 1). After correction for multiple testing, a significant association with OS after first treatment (q = .183) was documented for APOE rs7412 (Table 1), whereas no SNPs associated with PFS after first treatment (q ≥ .2 in all instances).

This study confirms in an independent CLL series the predictive value of selected SNPs identified by Sellick et al¹  and corroborates their relevance by controlling statistical associations for multiple testing. Our results expand the observations by Sellick et al¹  by documenting that selected SNPs also predict TTFT and survival from diagnosis in addition to PFS from treatment.