Germline Variant Burden Warrants Universal Genetic Testing in Pediatric AML

Acute myeloid leukemia (AML) is the most lethal of pediatric leukemias. Due to its rarity, the contribution of germline genetics to pediatric AML etiology is poorly understood. Here we investigated deleterious germline variants in the largest whole-genome sequenced (WGS) pediatric AML cohort to date, combining TARGET-21 (N=29) and COG-AAML1031 (N=336) for a total of 365 patients. WGS data from 29 subjects included non-hematopoietic fibroblasts expanded from marrow; 336 subjects had matched diagnosis/remission/relapse marrow samples. Variants were considered likely germline if present at variant allele frequency (VAF) > 0.3 in all sequenced samples for the patient.

We analyzed likely germline variants occurring in established genes for hereditary hematopoietic malignancy/immunodeficiency, and identified pathogenic/likely pathogenic (P/LP) variants according to the American College of Medical Genetics (ACMG) Guidelines. We found 5.5% of patients had at least one clinically-relevant P/LP variant in genes associated with cancer predisposition (CHEK2, TP53) or germline hematologic syndromes: BRIP1 (Fanconi anemia), ALAS2 (X-linked sideroblastic anemia), SAMD9 (MIRAGE syndrome), SBDS (Shwachman-Blackfan-Diamond syndrome), GATA2 (GATA2-deficiency disorder), CASP10 (autoimmune lymphoproliferative syndrome), and TNFRSF13B (common variable immunodeficiency). To identify additional risk loci, we annotated variants implicated in other malignancies and hematopoietic disorders, and found a P/LP variant in 21.6% of patients. In total, 26.3% of patients harbored a P/LP variant in an established or provocative gene.

We compared the prevalence of established P/LP germline variants in pediatric AML to that of acute lymphoblastic leukemia (ALL) and myelodysplastic syndrome (MDS) by meta-analysis of 10 published studies with a total of 4622 patients. A robust Poisson test for trend showed germline burdens significantly (p-value = 0.00027) higher (8.6%, N=519) in pediatric AML than in B-ALL (1.9%, N= 2449) or T-ALL (2.5%, N=531), but lower than in MDS (15.3%, N=287). A documented causal mechanism for the structural variants which dominate pediatric ALL is off-target activity of AID and RAG genes, both of which are mandatory for B and T cell maturation. By contrast, pediatric MDS is tightly linked to inherited bone marrow failure syndromes. Pediatric AML has neither high AID/RAG activity, nor high prevalence of known germline risk loci seen in pediatric MDS, nor activation of transposase genes (THAP9 and PGBD5) seen in pediatric solid tumors. This suggests a mechanistic role for interactions of heritable risk and exposures in pediatric AML etiology.

This missing heritability may be explained in part by P/LP variants of provocative genes, found in 21.6% of patients. To better understand these variants, we conducted analyses of epigenetic and somatic second hits, genetic and epigenetic signatures, copy-neutral loss of heterozygosity, and burden testing (which has only recently become feasible, due to the size of our expanded WGS cohort). We further analyzed variants predicted as deleterious by AlphaMissense. Although computational predictions alone are insufficient for clinical relevance, they provide insight into commonly disrupted biological pathways. We performed network analysis of protein-protein interactions disrupted by predicted deleterious germline variants, revealing the central hubs to be the chromatin modifiers EP300 and CREBBP and the inflammation regulator STAT3. Of note, EP300 and CREBBP are also central hubs of somatic disruption in pediatric AML, whether by structural, genetic, or epigenetic mechanisms, in contrast to disrupted TP53 interactions, central to adult AML. The results highlight leukemogenic processes distinct to pediatric AML.

Benefits to germline genetic testing in pediatric AML reach far beyond disease risk and mechanisms. Late effects are common among survivors; pharmacogenomic profiling stands to reduce this burden. Stem cell transplantation is the treatment of last resort in AML, and germline variants can profoundly influence donor selection. Modern long-read platforms have rapid turnaround and plummeting costs, facilitate sequencing of patients and parents at first encounter, and can resolve both genetic and epigenetic variants. We propose that universal germline genetic testing is mandatory to advance care in pediatric AML.