Leveraging Whole Genome Sequencing to Define the Mutational Landscape in Paroxysmal Nocturnal Hemoglobinuria

Background: Paroxysmal nocturnal hemoglobinuria (PNH) is a prototypic disorder of hematopoietic stem cells leading to recurrent complement-mediated intravascular hemolysis. The development of PNH has been linked to somatic mutations in the PIGA gene, coding for an enzyme critically involved in GPI anchor synthesis. While mutations in PIGA are pathognomonic for PNH, other mutations of GPI biosynthetic pathways have been described in isolated cases. Moreover, additional genetic determinants of clonal advantage and other genetic modifiers of the clinical course have not been sufficiently clarified.

Aim: This study was performed to investigate the molecular pathogenesis of PNH in a large patient cohort.

Methods: Using a typical myeloid targeted panel, mutations in PIGA gene can be identified 60-70% of cases and thus we performed whole genome sequencing (WGS, mean. coverage 100x). To further clarify the molecular pathogenesis of PIGA WT cases. A cohort of 94 patients (irrespective of PIGA mutations) with a confirmed hemolytic PNH, primary or evolved from antecedent AA was studied. The average clone size at diagnosis was 73.5%, measured by granulocyte flow cytometry. Before extracting all exomic variants, including small variants and large-scale insertions/deletions, primary analysis was performed for initial variant calling and filtering of putative sequencing artifacts (1A).

Results: As expected the majority of these patients (73.4%) harbored genetic lesions or mutations in PIGA with a mean variant allele frequency (VAF) of 29% and clone size of 69% by flow. Of these, 45% carried frameshift mutations, while 19% had splice variants and 14.5% carried missense mutations. Lastly, 14.5% of these mutations were based on gained stop codons or in-frame deletions while in 7% large scale genomic deletions in the PIGA locus were found (1B). The majority of these patients (67%) harbored a single genetic lesion, but >1 mutation was found in 23% of patients. Thus flow cytometric clone size may be a result of multiple genetically diverse PIGA mutant clones as indicated by the correlation between PNH clone size and number of PIGA mutants (P=4.14x10^-7) (1C). In patients with PNH in whom PIGA mutations were not found, the average clone size was smaller (average 15.0%). We thus subjected them to targeted deep sequencing (coverage mean 1000x) to ascertain that inability to detect mutations was not due to technical sensitivity. Indeed, 8/25 cases turned out to have a PIGA mutation. In addition we studied expression of PIGA in PIGA^mut versus WT cases, but no significant decrease in PIGA RNA was found. To further clarify the pathogenesis, we analyzed a gene panel of 33 proteins currently known to be involved in GPI anchor biosynthesis. We found a PIGO nonsense mutation (c.2593C>T) and a PIGV missense variant (c.467G>A) in two WT patients. These two variants were unique in PNH compared to a large cohort of myeloid neoplasias (N=2630) and appeared to be functionally consequential according to stringent in silico predictor and quality filter pipelines.

Since the progression of PNH into frank myeloid neoplasia can occur and the dynamics of the expansion of PNH clones is unpredictable and thus may be influenced by additional hits, we queried if additional lesions could explain these two phenomena. We also investigated the concomitant presence of myeloid mutation in these cases. We found additional somatic mutations (panel of 77 genes) in 47.8% of all PNH patients including ASXL1 (N=10), BCOR (N=8), KMT2D (N=6), and RUNX1 (N=4) (1D). Analyzing the VAF revealed that the majority of these events were subclonal in relation to PIGA mutations. Since we studied the patients with already highly expanded clones and due to assertion bias no difference in clone size and additional mutations was found but an ongoing study in early PNH will determine whether subclonal hits predict progression to full blown PNH (1E).

Conclusion: Our study shows that WGS can faithfully detect PIGA defects in a majority of PNH patients but this detection may be low in patients with smaller clones. Second, subclonal mutations linked to myeloid neoplasia are frequent events in PNH. Lastly we found two novel mutations in GPI synthetic pathway. Since none of these genes have been implicated in the pathogenesis of PNH, these findings extend our understanding of the pathogenesis of PNH and make the case for a targeted PIG pathway-wide sequencing of affected patients.