Identification of genetic factors associated with myeloid neoplasms

• Main Author: Galata, Gabriella
• Format: Dissertation
• Language: English
• Published: University of Southampton 2022
• DOI: 10.5258/SOTON/T0048

Myeloid neoplasms are clonal haematopoietic disorders characterised by the abnormal proliferation of specific myeloid cell types. The first part of this thesis focuses on mastocytosis, a rare haematological neoplasm characterised by the uncontrolled proliferation of mast cells. To test the hypothesis that germline variants can alter the risk of developing mastocytosis, a two stage case-control genome-wide association study was conducted in five European populations with 1,035 KITD816V-positive cases and 17,960 controls. This analysis identified three genome-wide significant SNPs: rs4616402 (Pmeta=1.37×10-15, I2=4.2), rs4662380 (Pmeta=2.11×10-12, I2=0) and rs13077541 (Pmeta=2.10×10-9, I2=0). Expression and methylation quantitative trait loci analysis were used to identify candidate genes located near the SNPs, specifically CEBPA, TEX41 and TBL1XR1. Statistical analysis with available clinical data, showed that rs4616402 was associated with age at presentation (P = 0.009; beta = 4.41; n = 422) in patients with non-advanced disease. Additional focused analysis identified suggestive associations between mastocytosis and genetic variation at TERT, TPSAB1/TPSB2, and IL13. Finally, a gene-based analysis was performed using the summary statistics of the stage 1 meta-analysis and multiple regression which suggested that the VEGFC gene is also associated with mastocytosis. The findings described in this thesis demonstrate that multiple inherited common risk variants predispose to KITD816V positive mastocytosis and provide novel avenues for functional investigation. In the second part of this thesis, the genetics of somatically acquired uniparental disomy (aUPD) in myeloid malignancies was investigated. Several regions of recurrent aUPD have been identified in patients affected with haematological neoplasms, many of which harbour somatic mutations that drive clonal proliferation. Similar regions of aUPD have also been identified in apparently healthy individuals, especially the elderly, which confer a tenfold increased risk of developing haematological malignancies. Large-scale sequencing initiatives of individuals unselected for cancer therefore represent a valuable resource to identify novel regions of aUPD and the underlying somatic mutations which drive clonal haematopoiesis (CH). Whole-exome sequence (WES) data for 49,996 individuals from the UK biobank (mean age = 56.5 years) was used to develop an automated pipeline for identifying aUPD regions and a new scoring system (gg score) to select aUPD regions with high confidence for manual review. Precision and recall were used to evaluate the gg score. The recall (or sensitivity) showed that it correctly identifies 55% of the predicted aUPD regions, although the model can also produce false negatives. On the other hand, the score performed well in term of precision and indicated that 90% of the aUPD regions were correctly classified. The methodology was then applied to WES data from a Swedish Case-Control study of Schizophrenia consisting of 12,380 samples and with a mean age of 65. Genes targeting the aUPD regions identified in the Swedish cohort are known (MPL, 1p; TET2, 4q; EZH2, 7q; JAK2, 9p; FLT3, 13q; MEG3-DLK1, 14q). Regions of aUPD were screened for somatic mutation if they were overlapping in two or more samples. However, only JAK2V617F was confirmed in all five samples with UPD9p and new aUPD regions with unknown gene target were not identified. This work showed that the frequency of sample with aUPD regions identified by WES data is lower (0.2-0.3%) than expected (1-2%) and provides an estimate what is needed in term of sample size to detect aUPD regions from WES data.