Genomic analysis of acute leukemia susceptibility

Genome-wide association studies have identified many common variants significantly associated with cancer risk. Most identified variants have modest effect sizes (OR < 1.2) and cumulatively account for a small proportion of estimated disease heritability. Common variants with larger effect sizes may explain more of the genetic basis for cancer risk but are seldom found in GWAS, possibly due to the presence of confounding biological or non-genetic heterogeneity between patients. Here we investigate the contribution of genomic variants to risk for heterogeneous cancers with subsets defined by prognostic recurrent cytogenetic abnormalities: pediatric B-cell acute lymphoblastic leukemia (B-ALL) and adult acute myeloid leukemia (AML). In therapy-related AML (t-AML), recurrent cytogenetic abnormalities are also linked to specific classes of chemotherapy. We hypothesize inherited genetic variation confers risk for acute leukemia irrespective of tumor karyotype and/or acquisition of different cytogenetic abnormalities has unique genetic susceptibilities. We aim to increase power to detect acute leukemia risk variants with moderate or larger effect sizes through overall meta-analysis of GWAS and by stratifying cases into more homogeneous cytogenetically-defined subsets.;Pediatric B-ALL is the most common childhood malignancy in which the vast majority of patients present with gross tumor karyotypic abnormalities. Recurrent abnormalities are used to stratify patients into "low-risk" and "high-risk" of relapse subsets. We performed meta-analysis of three pediatric B-ALL GWAS with 1225 total cases and 4103 controls. Cases included two low-risk subsets that were analyzed separately. In the overall meta-analysis we identified and replicated novel risk variants at the CDKN2B-ANRIL locus on 9p21.3. Functional analysis reveals the index and strongly linked SNPs are eQTLs for CDKN2B in whole blood and analysis in a multiethnic cohort suggests a candidate causal variant.;Adult de novo AML is the most common acute leukemia in adults. Cytogenetic abnormalities are used to stratify patients into favorable, intermediate, and adverse risk subsets. We performed meta-analysis of three adult de novo AML GWAS with 1175 total cases and 6347 controls. Cytogenetic subset analyses were conducted on karyotypically similar cases representative of the different risk groups. We identified a putatively significant variant at the LCP1 promoter on 13q14.13 unique to favorable risk APL cases. This finding suggests considerable power might be gained by controlling for biological heterogeneity among cases.;Therapy-related AML is a deadly complication of cytotoxic therapy. We hypothesized we could reduce biological and non-genetic heterogeneity using recurrent cytogenetic abnormalities as a proxy for different classes of chemotherapy. To test this hypothesis we performed GWAS on 230 t-AML cases and 1256 controls. No significant variants were found. We also hypothesized some t-AML patients may have undiagnosed cancer predisposition syndromes conferring heightened risk for t-AML or suggesting their leukemia is de novo AML. To test this hypothesis we performed whole-exome sequencing on 27 t-AML patients with clinical features consistent with inherited cancer predisposition and identified rare deleterious variants in cancer predisposition genes.;Our results indicate risk for pediatric B-ALL at 9p21.3 is more complex than previously described, adult de novo AML may have a small genetic risk component except in the case of APL and other favorable risk AML, and that rare deleterious variation may have a greater contribution to risk for t-AML than common variation. These studies demonstrate the benefits gained from acquiring large sample sizes with genome-wide information, reducing biological heterogeneity by conditioning on tumor karyotype, and expanding analysis to include common and rare variation in evaluating risk for acute leukemia.