AML is a malignancy with an average five-year survival rate of 50% in adults. Great strides have been made in deciphering the genetic heterogeneity of this disease and indentifying subgroups with favorable or unfavorable outcomes based on cytogenetic and molecular factors. We undertook this study to identify patient-specific germline determinants of treatment outcome and survival in AML.
Our cohort was comprised of 314 adult patients with AML in first or subsequent complete remission who underwent consolidation therapy with a busulfan-based autologous stem cell transplant (Bu-ASCT) at UCSF between 1986 and 2009. DNA samples were isolated from patients' peripheral blood mononuclear cell alliquots collected via apheresis following etoposide and high-dose cytarabine consolidation at the time of confirmed complete remission. Genotyping was performed using the Illumina Human OmniExpress Beadchip. To adjust for population substructure, EIGENSTRAT software was used to determine eigenvalues for the SNP correlation matrix and the corresponding eigenvectors were used as covariates. Statistical analyses were conducted with STATA and PLINK software. Association testing with DFS was based on Cox proportional hazards models. We utilized a multiplicative (log-additive) model with a genome wide-alpha set at 10−8. The subset of 168 genetically European patients (cau) was used for the primary analysis.
Overall, 9.7% of patients in our cohort had high-risk/relapsed acute promyelocytic leukemia, 12.9% core-binding factor leukemia [inv 16 or t(8;21)], 69.5% cytogenetically-normal leukemia, and 7.9% high-risk disease either by cytogenetic or molecular abnormalities. Disease-free survival in our cohort following Bu-ASCT was 59% at 5 years. As expected, baseline risk was significantly associated with DFS but treatment era was not. We identified several significant SNP clusters in our analysis. A 4-SNP cluster emerged on chr15 with a SNP ranked 2nd and one ranked 8th in the overall analysis. The SNP (rs933813) marked 2nd was upstream of the insulin-like growth factor-1 receptor (IGF1R). After adjustment for baseline risk, this SNP was significantly associated with DFS in cau (p-value= 6.9×10−8) as well as the overall population (p-value= 1.1×10−6). Other clusters are being evaluated.
We identified several significant SNPs and SNP clusters that are associated with DFS in adult patients with AML undergoing Bu-ASCT. So far a SNP upstream of IGF1R has emerged as highly significant in our analysis. Ongoing studies are focusing on imputation, fine mapping and functional validation of these results.
No relevant conflicts of interest to declare.
This icon denotes a clinically relevant abstract
Asterisk with author names denotes non-ASH members.