Estimating the Population-Based Prevalence of Congenital Thrombotic Thrombocytopenic Purpura Using Large-Scale Sequencing Data

Abstract

Introduction Thrombotic thrombocytopenic purpura (TTP) is a life-threatening disorder caused by severe ADAMTS13 deficiency. This severe deficiency (<10%) is classified into two forms of acquired or hereditary TTP. Acquired TTP is either due to circulating autoantibodies inhibiting ADAMTS13 activity or increasing its clearance, whereas congenital TTP results from homozygous or compound heterozygous ADAMTS13 mutations (cTTP, Upshaw-Schulman syndrome). Because of ADAMTS13 defects, an abnormal amount of ultra-large von Willebrand factor multimers is concentrated, resulting in the formation of platelet-rich microthrombi in the circulation. Given that the true prevalence of cTTP is not established, we sought to leverage the availability of exome or genome sequences of very large numbers of individuals (genome Aggregation Database, gnomAD) to estimate the worldwide and within-population prevalence of cTTP. We also aimed to examine the global mutational landscape of ADAMTS13. Methods Exome and genome sequencing data of 141,456 subjects of gnomAD were extracted and analyzed for ADAMTS13. We comprehensively reviewed literature to identify all ADAMTS13 variants that have been associated with cTTP so far. In addition, all variants reported to be associated with cTTP in HGMD and/or LOVD datasets were included. Finally, we searched for ADAMTS13 variants in ClinVar that have been classified as “pathogenic” or “likely pathogenic”. The following gnomAD variants were considered pathogenic: variants associated with cTTP in HGMD/LOVD; nonsense, frameshift, inframe deletion/insertion variants; splicing variants affecting the first/last two intronic nucleotides; splicing variants affecting the first/last eight intronic nucleotides predicted deleterious by 4 different in silico tools; missense variants predicted deleterious by 7 different in silico tools. Using the frequencies of known or predicted pathogenic variants in ADAMTS13 of the gnomAD population (n= 141,456), we estimated the prevalence of individuals homozygous or compound heterozygous for cTTP (assuming Hardy-Weinberg equilibrium). Results Following quality controls and removing variants with no clear association with cTTP, we found that 302 different ADAMTS13 variants have been so far reported to cause cTTP in the literature, HGMD and LOVD as well as ClinVar. We then collected high-quality data from gnomAD population including 141,456 subjects with different ethnicities: 12,487 Africans/African Americans, 17,720 Latinos/Admixed Americans, 5,185 Ashkenazi Jewish, 9,977 East Asians, 12,562 Finnish Europeans, 64,603 non-Finnish Europeans, 15,308 South Asians, and 3,614 persons without an assigned ethnicity. We identified a total number of 2212 variants within ADAMTS13 among 282,912 alleles analyzed in the gnomAD population. Of these, following a conservative approach, 250 different variants were pathogenic; 83 reported to be associated with cTTP and 167 predicted to be pathogenic. Among a maximum number of 282912 alleles analyzed, 2166 alleles were affected by novel or reported variants. Of which, 53% of the affected alleles were due to known pathogenic variants and the remaining 47% were due to novel predicted variants. Considering all pathogenic variants (n=250) and only those reported (n=83), we estimated an overall cTTP prevalence of 17 and 59 individuals per million people, respectively (Table 1). Considering reported variants only, the highest prevalence of cTTP was estimated in Europeans and Finnish population (30 and 26 per 10 6), followed by Latinos/Admixed Americans (14 per 10 6) and Africans/African Americans (6 per 10 6). East and South Asian ethnicities had a lower estimated prevalence of 4 and 3 per 10 6, respectively. Ashkenazi Jewish population had the lowest estimated prevalence of cTTP (0.6 per 10 6) among all populations (Table 1). Conclusion We have attempted to estimate the worldwide and within-population prevalence of cTTP using available genome/exome sequencing data of 141,456 individuals. Our results reveal that putative disease alleles and cTTP-associated variants are more prevalent than expected and the true cTTP prevalence is more than ten times higher than was reported. This finding suggests that a large number of patients might be undiagnosed. We also established the mutational burden of ADAMTS13 by identifying 167 novel variants in addition to the 302 previously reported.