https://orcid.org/0000-0002-9654-6299
TO THE EDITOR:
Congenital or hereditary thrombotic thrombocytopenic purpura (hTTP) is a rare autosomal recessive disease caused by biallelic severe defects in the ADAMTS13 gene, either homozygous or compound heterozygous. Three research groups, including ours, have attempted to estimate the prevalence of hTTP in the general population. In 2011, we analyzed 3200 individuals from the general population in Japan and suggested that 1.1 per 10⁶ individuals could be biallelic ADAMTS13 deficient.1 Subsequently, von Krogh et al2 estimated the prevalence in central Norway to be 16.7 per 106, whereas Zhao et al used the Genome Aggregation Database (gnomAD), version 2.1.1,3 to estimate a global prevalence of 0.43 to 1.1 per 106.4 These analyses, however, may over- or underestimate the hTTP frequency for various reasons discussed in a recent study by Seidizadeh et al, published in Blood Advances.5
Seidizadeh et al5 analyzed ADAMTS13 data from 807 162 individuals in gnomAD, version 4.1.0,6 to refine the global and regional prevalence estimates for hTTP. Using such large-scale genomic data sets is critical for accurately assessing inherited disease frequencies. Their comprehensive analysis that incorporated data from multiple databases and published case reports estimated a global hTTP prevalence of 40 per 106 individuals (based on all 758 inferred pathogenic variants; looser estimation) or 23 per 106 (considering only the 140 previously reported pathogenic variants; stricter estimation). These values are substantially higher than previous estimates.1,2,4 The stricter estimation revealed the highest prevalence among East Asians (42/106), followed by Finnish (32), non-Finnish Europeans (28), Admixed Americans (19), Africans/African Americans (6), and South Asians (4).
The unexpectedly high prevalence among East Asians prompted me to review the basis for these estimates. ADAMTS13 data for East Asians, which included 22 448 individuals (44 896 alleles), identified 294 alleles with previously reported pathogenic variants. This corresponded to a pathogenic allele frequency of 0.00655 and an estimated biallelic deficiency prevalence of 42.9 per 106. Notably, the missense variant p.Gln723Lys (c.2167C>A, rs138014548) had the highest allele frequency of 0.00557, accounting for 85% of the affected alleles. In the Japanese genetic variation database Tohoku Medical Megabank Organization (ToMMo) 54KJPN,7 this variant had a higher allele frequency of 0.0112. However, such a frequency suggests it may be a common single nucleotide polymorphism rather than a causative variant for a rare disease. If the p.Gln723Lys variant were pathogenic, it is estimated that there would be ∼15 000 patients with homozygous hTTP in Japan, which is unlikely.
The p.Gln723Lys variant was first and only reported in a Japanese patient with hTTP.8 In this report, the nucleotide sequences of all 29 exons of ADAMTS13, including intron-exon boundaries, were determined by polymerase chain reaction–direct Sanger sequencing, as described.9,10 The patient was heterozygous for 2 candidate pathogenic variants, p.Arg398Cys and p.Gln723Lys, with the former being present in the mother and 1 sibling and the latter being present in the father and the other sibling. Both variants were screened for in 96 individuals (192 alleles) from the Japanese general population and were not considered to be common polymorphisms. Plasma ADAMTS13 activity was <0.5% of normal in the patient and ranged from 23% to 46% in family members who carried 1 variant. Based on these findings and the absence of commonly available population-level genomic data at the time, both variants were reasonably considered pathogenic. Although current genomic data suggest that the p.Gln723Lys variant is likely a benign polymorphism, the conclusion drawn at that time cannot be faulted. It is possible that another pathogenic defect (eg, structural variant) in the p.Gln723Lys allele in this family was not detected by polymerase chain reaction–direct Sanger sequencing.
Our ADAMTS13 data from 64 Japanese patient families with hTTP, including unpublished information, found only 1 p.Gln723Lys allele among 128 alleles. Given its frequency in the general population, far more alleles would be expected if it were pathogenic. Among the 72 pathogenic variants identified, the most common were p.Arg193Trp (15 alleles), p.Cys908Tyr (14 alleles), p.Cys754Alafs∗24 (5 alleles), p.Gln449∗ (5 alleles), p.Tyr1074Alafs∗46 (4 alleles), c.414+1G>A (3 alleles), p.Ile673Phe (3 alleles), and the remaining variants were 1 or 2 alleles. All of these variants have allele frequencies of <0.001 in the ToMMo 54KJPN7 or gnomAD exomes6 databases.
Excluding p.Gln723Lys from Seidizadeh et al’s stricter estimation, which uses all variants previously reported in patients with hTTP5, reduces the pathogenic allele frequency in East Asians from 0.00655 to 0.00098, thereby lowering the estimated biallelic deficiency prevalence from 42.9 to 0.96 per 106. In the looser estimation, which includes all variants predicted to be pathogenic, the allele frequency decreases from 0.00918 to 0.00361 and the estimated prevalence drops from 84.2 to 13.0 per 106. We must be cautious when classifying relatively common variants as pathogenic because of their potential impact on the estimated increase in patient numbers.
In conclusion, accurately assessing the prevalence of rare genetic diseases is complex. Pathogenicity depends on previous reports and structure-function predictive models, but these methods are not definitive. Even cell-based expression experiments cannot lead to completely correct conclusions. Whether a variant is significantly enriched in the patient group, compared to the nonpatient group or the general population, is an important factor to consider. However, this is difficult to test in rare diseases. Further research is needed to understand the role of ADAMTS13 variants that have been reported in patients with hTTP.
Contribution: K.K. wrote the manuscript.
Conflict-of-interest disclosure: K.K. declares no competing financial interests.
Correspondence: Koichi Kokame, Department of Molecular Pathogenesis, National Cerebral and Cardiovascular Center, 6-1 Kishibe-Shimmachi, Suita 564-8565, Japan; email: kame@ncvc.go.jp.
