De novo mutations in antithrombin deficiency: high frequency and heterogeneous mechanisms

More than 78 de novo mutations (DNM) arise in the human genome per generation.¹ Most DNM are single-nucleotide variants (SNV), but small insertions or deletions (INDEL) and structural variants (SV) can also occur de novo.¹ This mutation rate may be elevated in certain individuals or concentrated in specific genomic regions due to risk factors such as advanced paternal age or particular genomic architectures.¹

While neutral or beneficial mutations may drive human evolution, DNM affecting essential genes can disrupt biological systems and lead to diseases.² Identifying and characterizing pathogenic DNM can help elucidate disease mechanisms and highlight genomic regions prone to mutagenesis. This information has important diagnostic and clinical implications for patients and their families. In the context of hemostasis, studies on DNM are scarce and typically limited to case reports, especially in hemophilia.³ Antithrombin deficiency (ATD) is a rare autosomal dominant disorder caused by variants in the SERPINC1 gene that significantly increase the risk of thrombosis due to the key anticoagulant role of this serpin.⁴ There are two main types of ATD: type I (with reduced antigen levels) are rare and often present with a severe clinical phenotype; type II (with normal antigen levels but impaired function) are more common and typically have milder thrombotic consequences, especially in cases involving impaired heparin binding (type II Heparin Binding Site [HBS] deficiency).⁴ We studied 433 consecutive, unrelated individuals with ATD recruited over 26 years (1998-2024). All participants gave informed consent to participate in this study, which was approved by the Institutional Review Board of the Hospital Universitario Morales Meseguer, Murcia, Spain, and was conducted in accordance with the principles of the Declaration of Helsinki of 1964 and its subsequent amendments. Functional and biochemical analyses performed to characterize the disorder included anti-FXa and anti-FIIa functional assays, determination of antigen levels, biochemical evaluation of plasma AT by western blot analysis, evaluation of heparin affinity, and, in some cases, recombinant expression.⁵ Molecular analysis of SERPINC1 found 158 different variants (112 SNV, 25 small INDEL, and 21 SV) in 347 individuals. Next, we identified 210 patients with SERPINC1 variants who had family history data and/ or parental samples available. In 11 of the 210 screened cases (5.2%), 10 different SERPINC1 variants were identified despite the fact that ATD was discarded in the parents, whose paternity/maternity was confirmed by genotyping 16 short tandem repeats. These findings indicate a de novo origin for these variants (Figure 1).

Remarkably, most cases with DNM (10/11; i.e., all except P11) presented a severe clinical phenotype, including idiopathic or early-onset thrombosis (median age at first event: 18 years; range: 3 months-36 years) and recurrent events in 50%. Median AT activity was 48%. Eight patients had type I, and three had type II Pleiotropic Effect deficiencies. None had Reactive Site or HBS deficiencies (Table 1). Eight patients had de novo SNV scattered throughout SER-PINC1, with no regional clustering. Transitions were more common than transversions (7:1), and none occurred at CpG dinucleotides (Figure 2). Six variants had been previously reported in other ATD cohorts, supporting multiple independent mutational events. Notably, 2 unrelated individuals (P8 and P10) had the same DNM: c.394C>T, p.Glu132*. One patient carried a novel dinucleotide insertion c.1318_1319insTT leading to a frameshift and premature stop codon in the C-terminus of AT (Table 1). Two patients had gross de novo deletions involving SERPINC1. The length and characteristics of these deletions have been previously described in detail by our group.⁶ Briefly, in P5, the deletion encompassed the entire gene and covered 29 additional loci. In P7, the deletion included exon 1 and two neighboring genes (Figure 2). In both cases, nanopore sequencing revealed repetitive elements flanking the breakpoints (L1PA2 and A-rich, respectively).

Figure 1.Flow chart of cases with de novo mutations in SERPINC1 identified in our cohort of patients with antithrombin deficiency.

Table 1.Characteristics of cases with antithrombin deficiency caused by a SERPINC1 de novo mutation.

The low AT activity observed in DNM carriers (Table 1) argued against a somatic mosaicism as the underlying mechanism for the emergence of these variants. Evaluation of the electropherograms of the de novo SNV and INDEL also supported a germline origin in the probands. However, since Sanger sequencing lacks the sensitivity to detect and quantify low-level mosaicism or confirm a true germline DNM, deep sequencing by nanopore was performed in patients carrying these DNV SNV/INDEL. For each case, over 50,000 reads covering the mutation were analyzed, showing an approximately equal distribution of mutant and wild-type alleles in all probands, consistent with a germline origin. The same approach was applied to evaluate potential mosaicism in the parents. In all instances, the presence of the variant allele in parental samples was <1% of total reads (<0.1% in 5 cases). We attempted to determine the paternal or maternal origin of the mutant allele for all carriers of SNV or the small insertion DNM by analyzing SERPINC1 intragenic haplotypes in each trio using nanopore sequencing of long-range PCR products covering the gene.⁷ However, likely due to the low genetic variability of SERPINC1, we were only able to identify the allele carrying the DNM (paternal or maternal) in 5 cases. Three DNM had a paternal origin (60%), and two maternal (40%) (Table 1).

To date, 546 different SERPINC1 variants have been reported (http://www.hgmd.cf.ac.uk/ac/gene.php?gene=SERPINC1) but only 10 (9 SNV and one small deletion) have been previously described as DNM causing ATD (Figure 2). Here, we provide the most comprehensive analysis to date of SERPINC1 DNM in one of the largest ATD cohorts. Our results reveal a relatively high proportion of disease-causing DNM in SERPINC1, accounting for up to 5.2% of all cases screened. However, these results may only show the tip of the iceberg, and the true prevalence of DNM in SERPINC1 might be underestimated. The severe clinical phenotype observed in most individuals with SERPINC1 DNM, both in our cohort and in previously reported cases, indicates a likely diagnostic selection bias. Indeed, all DNM carriers had severe type I or type II Pleiotropic Effect defects, often presenting with early and/or recurrent thrombosis. Milder cases, such as those with type II RS, or especially type II HBS variants, may have been under-represented and missed from our study, as these typically have a lower thrombotic risk, delayed onset, or non-typical manifestations (e.g., arterial thrombosis), and often lack a family history of thrombosis,⁴ features that commonly lead to exclusion from thrombophilia screening.⁸

As in other disorders caused by DNM, most SERPINC1 DNM are SNV. However, the expected enrichment of C>T transitions at CpG sites, attributed to higher methylation and spontaneous deamination in the male germline,⁹ was not observed in SERPINC1, although these findings may be attributable to chance due to the small sample size and should be validated in further studies. Regarding genomic distribution, de novo SNV in SERPINC1 show some clustering in exons 6 and, particularly, exon 7, where 4 DNM were identified within an 18bp region. This finding aligns with prior evidence suggesting exon 7 is a hotspot due to repetitive DNA sequences.¹⁰ However, the majority of SERPINC1 DNM are scattered across the gene without clear regional susceptibility. An exception is c.394C>T, p.Glu132*, a recurrent mutation site, as it was found as a DNM in 2 unrelated cases in our cohort and was previously reported in a Japanese patient with type I deficiency.¹¹

Notably, we report the first cases of gross DNM causing ATD, both deletions involving repetitive elements, which are well known to mediate genomic rearrangements.¹² SERPINC1 is flanked and interspersed with a high proportion of repetitive elements, some of which have been implicated in ATD through SV.⁶

Finally, our study explores the origin of the DNM in ATD with the following considerations. 1) Potential germline origin: deep sequencing found <1% of somatic mosaicism for SNV and INDEL in both patients and parents, supporting germline origin. However, recent studies showed that apparent DNM occasionally originate from undetected parental mosaicism, sometimes with variant allele frequencies <1%.¹³ Moreover, gonadal mosaicism cannot be ruled out by peripheral blood testing. In addition, the absence of a detectable mosaic fraction does not exclude a very early post-zygotic origin, particularly for SV, due to the instability of the early zygotic genome associated with aneuploidy and gross rearrangements,¹⁴ which could still manifest as a constitutional pattern. 2) Paternal bias: since 1947, it has been hypothesized that the male germline may be more mutagenic than the female germline.¹⁵ However, whole-genome analyses of parent-offspring trios have revealed substantial inter-family variation.¹ In our study, 40% of the DNM occurred on the maternal allele, suggesting that, for SERPINC1, there is no strong male bias. This finding is clinically relevant, as maternal DNM carry a risk of recurrence in subsequent pregnancies.¹ 3) Parental age: increasing paternal age is associated with higher DNM rates due to replication errors during the cell divisions required for continuous sperm production.¹⁶ In modern societies, where delayed parenthood is increasingly common, this trend raises concern about the rising incidence of de novo genetic disorders.¹⁶ In our study, the average age of fathers (32.2 years) and mothers (29.8 years) was not markedly high, suggesting parental age has a minimal impact on the generation of SERPINC1 DNM. The small sample size strongly encourages evaluating both paternal bias and age in further studies.

Figure 2.Localization in SERPINC1 of de novo mutations identified in this and other studies. This study (in red) and other studies (in black). The blue rectangle indicates the extension of the deletion.

In conclusion, we report a relatively high frequency of SERPINC1 DNM causing ATD: 5.2%, a value that may be underestimated due to clinical selection bias. DNM in SERPINC1 exhibit features distinct from those in other disorders with high DNM rates, including a notable proportion of SV and a lack of enrichment for C>T transitions at CpG sites. Our data suggest that most SERPINC1 DNM originate during gametogenesis in parents of non-advanced age. These findings support systematic thrombophilia screening, including molecular analysis, in patients with ATD, regardless of family history.

Footnotes

• Received August 4, 2025
• Accepted January 21, 2026

Correspondence

*PG-R and BM-B contributed equally as first authors.

^#MEM-B and JC contributed equally as senior authors

J. CORRAL - javier.corral@carm.es

Disclosures

No conflicts of interest to disclose.

Contributions

Funding

The study was funded by the Instituto de Salud Carlos III (ISCIII) through projects “PI24/00429” and “PMP21/00052”, which were co-funded by the European Union and by the Next Generation EU, respectively. MEM-B has a Ramon y Cajal contract at the University of Murcia. CB-P has a Juan Rodes fellowship from the ISCIII co-funded by the European Union (JR22/00041).