Analyses of large cohorts of data are needed to understand how genetic aberrations can affect treatment outcomes in patients with hematologic malignancies (HMs). With this in mind, the HARMONY Alliance has developed a large platform for HMs called “Big Data for Better Outcome,” and aims to analyze approximately 100,000 cases of acute myeloid leukemia (AML), acute lymphocytic leukemia, chronic lymphocytic leukemia, multiple myeloma, myelodysplastic syndromes, non-Hodgkin lymphoma, and pediatric HMs. This report focused on results from the first approximately 5000 AML cases and served as proof of principle.
The Observational Medical Outcomes Partnership (OMOP) common data model was used to implement de facto anonymization and data harmonization. Thereafter, the following analyses were performed: gene–gene interaction analyses for co-occurrence and mutual exclusivity, a hierarchical Dirichlet process for class discovery, and a Bradley-Terry analysis to estimate clonal evolution. Prognostic multistage models were fitted to assess the effects of genomic and clinical data on rates of remission, relapse, and survival.
Data from 2941 patients with AML with available clinical and molecular information were analyzed. The median age was 52.4 years, and 53% of patients were male. A total of 808, 1193, and 940 patients had favorable, intermediate, and adverse cytogenetics, respectively, according to the European LeukemiaNet (ELN) 2017 risk classification. A total of 1251 patients received an allogeneic stem-cell transplantation (allo-SCT); 1690 patients received conventional consolidation.
Analysis of gene–gene interactions confirmed known patterns of co-occurrence and mutual exclusivity. Furthermore, cluster analysis enabled segmentation of “unique” ELN risk groups. Overall, these data were able to establish that epigenetic driver mutations in genes affecting DNA methylation occur very early, whereas histone-modifying enzyme mutations occur later. Finally, these data confirm that the presence of high-risk mutations differentially influences overall survival (OS) following an allo-SCT in a complex manner. Indeed, patients whose disease exhibits many high-risk genotypes, such as the TP53 mutation, seem to derive modest benefit from an allo-SCT (median OS of 54.4 weeks and 12.8 weeks with vs without allo-SCT, respectively; P <.001). However, patients whose disease exhibits other higher-risk genotypes, such as DNMT3A in combination with PTPN11, benefit more significantly from an allo-SCT (median OS of 212.7 weeks and 60.8 weeks with vs without allo-SCT, respectively; P = .027).
Results from this proof-of-principle study combining large data cohorts of clinical outcomes and disease genomic profiling from patients with AML confirm the benefit of similar approaches using the HARMONY Alliance platform, and demonstrate that analyses of OMOP-harmonized big data sets will help inform individualized therapy for optimal outcomes in other HMs.
Reference
Bullinger L, Martinez Elicegui J, Strang E, et al. Novel Insights into Genomic Classification and Prognosis in Acute Myeloid Leukemia Based on a Pan-European Public-Private Partnership, the HARMONY Alliance. Presented at: 25th European Hematology Association Congress Virtual; June 11-21, 2020. Abstract S130.