Congenital antithrombin deficiency in patients with splanchnic vein thrombosis

Splanchnic vein thromboses (SVT) are a rare condition that can be life‐threatening. The most severe thrombophilia associated to SVT is antithrombin (AT) deficiency, usually caused by SERPINC1 mutations. Although transitory AT deficiencies and congenital disorders of the N‐glycosylation pathways (CDG) have been recently reported as causes of AT deficiency, the current AT clinical screening still only includes anti‐FXa activity. This study aims to (a) improve the detection of AT deficiency in SVT and (b) characterize the features of AT deficiency associated with SVT.


| INTRODUC TI ON
Thrombosis of the splanchnic veins (SVT) is a rare condition with a prevalence of less than 5 cases every 10 000 inhabitants. However, despite being rare, SVT are a complex condition that can be lifethreatening and lead to severe complications such as portal hypertension or intestinal ischemia. 1 The most frequent causes of SVT are major thrombophilic disorders and local factors, the latter being mainly abdominal inflammatory processes such as pancreatitis and surgical interventions. It is remarkable that in up to 30% of patients with a local factor there is also an underlying thrombophilic disorder, acting the local inflammation as trigger for the development of thrombosis. 2 Regardless availability of better diagnostic tools such as detection of Calreticulin or JAK2 mutations, 3,4 in around 30% of cases it is not possible to identify the thrombosis' etiology and patients are finally classified as having an idiopathic thrombosis. Correctly classifying SVT thrombosis according to its etiology (major thrombophilic disorders vs local inflammation/idiopathic thrombosis) is highly relevant. Indeed, current guidelines only recommend longterm anticoagulation to prevent re-thrombosis in those patients with underlying thrombophilic disorders. On the contrary, patients with idiopathic thrombosis or those associated only to local factors will only receive anticoagulation in the setting of acute thrombosis. 5,6 Interestingly, however, despite following these current guidelines, recent evidence shows a relevant incidence of new thrombotic events in patients with idiopathic thrombosis not treated with anticoagulants. 7 Antithrombin (AT) deficiency is one of the most severe hereditary thrombophilia. [8][9][10] Homozygote AT mutations are lethal disorders that cause intrauterine foetal death while heterozygote mutations lead to a hypercoagulable state with high probability of developing thrombosis. 11 Antithrombin deficiencies have been classified into two types: in type I, levels of AT are low; in type II, AT levels are within the normal range but the AT molecule has impaired or null anticoagulant activity. 12 Several AT deficiency mechanisms have been described, being the most frequent genetic variants in SERPINC1 (the gene encoding antithrombin) 13 that account for up to 80% of the cases of AT deficiencies.
The real incidence of SVT among carriers of AT deficiency has not been previously established. 14 However, from a clinical point of view, it is striking that despite the high prothrombotic risk of AT deficiency, the reported incidence of this alteration is very low among patients with splanchnic venous thrombosis, thus suggesting that AT deficiencies might be underestimated. There are increasing evidences on the low sensitivity of the current functional diagnostic tests to detect pathogenic mutations causing AT deficiency. 13 Moreover, it has been reported that the pathogenic effect of some SERPINC1 mutations might be exacerbated by environmental factors, leading to transient AT deficiency that would only be detected by functional methods. This would be the case of SERPINC1 mutation p.Val30Glu. 15 Patients with this mutation would have normal levels and function of AT in baseline conditions but with a higher conformational susceptibility to stress, which would lead to deficiency of AT. 15 Similarly, post-traslational defects such as alterations and reduction of the N-glycosylation pathways [16][17][18] caused by congenital disorders of glycosylation (CDG), which might be not detected by conventional functional methods, are also known to impair AT function. 17 Rendering the interpretation of AT levels even more difficult, acquired causes of low AT are also a common cause of low AT levels, but importantly do not lead to hypercoagulable states. Therefore, when finding low levels of AT (<80%) it is essential to exclude potential acquired etiologies before diagnosing hereditary AT deficiency. Several conditions can lead to low levels of AT and must be taken into account in the differential diagnosis (ie liver disease can decrease AT levels because of decreased hepatic synthesis; recent or active thrombosis lead to a general decrease in coagulation factors).
If AT deficiency is undetected and misdiagnosed, patients requiring anticoagulation could remain untreated with the consequent risk of re-thrombosis, not only SVT but also in other vascular territories. Diagnosing these patients would allow a correct risk stratification which could change the decision of prescribing chronic anticoagulation.
In this framework, we have designed the current study with 2 objectives: (a) to evaluate if it is possible to improve the detection of AT deficiency in patients with SVT using new diagnostic techniques (b) to report the features of AT deficiency that associate with SVT.

| MATERIAL S AND ME THODS
The study was performed in accordance with the International

Key points
• Splanchnic vein thrombosis (SVT) are a rare condition.
• The most severe thrombophilia associated to SVT is antithrombin (AT) deficiency.
• Although new pathways for AT deficiency have been recently reported, clinical screening still only includes AT anti-factor Xa activity.
• We report and characterize AT mutations most frequently associated with SVT and we show how current screening fails to detect masked AT deficiencies.

| Patients and samples
This study is based in two different cohorts: Cohort 1. Since 2003, all consecutive patients with SVT seen at Hospital Clínic de Barcelona (n = 284) were asked for permission to obtain a blood sample for research purposes. Blood samples were stored at IDIBAPS Biobank facilities. As shown in Figure   S1, of these 284 patients 5 were diagnosed with AT deficiency

| Genetic analysis
Genetic SERPINC1 variants were identified by sequencing the 7 exons and flanking regions, or the whole gene by Sanger's or NGS methods as described before. 17

| Biochemical and functional characterization of plasma antithrombin
Antithrombin anti-FXa and anti-FIIa activity were determined in citrated plasma by chromogenic methods. Values were expressed as percentages referred to a pool of 100 healthy subjects. AT deficiency was considered when values were below 80%. Antigen levels were measured by rocket immunoelectrophoresis and/or ELISA (enzyme-linked immunosorbent assay). These methods were described in previous manuscripts. 19 Analysis of plasma AT forms included crossed immunoelectrophoresis and polyacrylamide gel electrophoresis under different conditions (denaturing and not denaturing) followed by immunological detection using conditions and reagents previously described. 20 Hypoglycosylation of AT and other hepatic proteins (α1-antitrypsin and transferrin) in plasma was evaluated by western blot or HPLC (high-performance liquid chromatography) as described elsewhere. 17 The reported results were performed in samples collected long after the acute event in the cohort of patients with congenital AT deficiency (mean 42 ± 76 months).

| Alignment of antithrombin protein sequences among species
The alignment of antithrombin protein sequences with up to 49 species was performed by uniprot tool (https ://www.unipr ot.org/align/ ) and visualized by Jalview version 2. 21

| Statistical analysis
Quantitative data are expressed as median and range. Qualitative data are expressed as percentages.

| RE SULTS
Cohort 1 Report of antithrombin deficiency in a large cohort of patients with SVT (PVT and BCS) in which this disorder had been originally ruled out. Table 1 shows baseline characteristics of Cohort 1, including the etiology and the different thrombophilia present in this cohort.
Overall, the 3 main etiologies were similarly represented: 28 patients (31.5%) had a known thrombophilia, 26 (29.2%) were associated to a local factor with no underlying thrombophilia and in 35 patients (39.3%) neither thrombophilia nor local factor were identified and were considered idiopathic.
Antithrombin deficiency had been, in the initial clinical screening, ruled out in all patients based on normal AT anti-FXa activity (or a reduction in anti-FXa activity in the setting of similar reduction in other liver synthesized factors) and the lack of familial history of neither thrombosis nor AT deficiency. The median values of AT activity measured as anti-FXa activity were 86.8% ± 17%. Of note however, in this cohort there were 25 patients who presented levels of AT below 80% (median 68%, range: 48-79), which isolated could had been indeed considered AT deficiency. Nevertheless, these patients had a concomitant decrease in other coagulant factors such as protein C (median 72%, range: 33-99), protein S (median 82%, range: 68-130), FII (median 77%, range: 57-92) or FX (median 73%, range: 44-143). Therefore, these decreased AT levels were considered acquired and clinically not relevant, 22 assuming that they were due to either consumption or to a global decrease in liver factors synthesis.
Using the biobank samples of these 89 cases, AT function was reassessed by measuring both functional activity (by anti-FXa and anti-FIIa tests) and by immunological methods (Laurell and western blot). Similar to what had been found in the clinical setting, this secondary screening showed that 21 cases had low AT levels in at least two tests (either AT anti-FXa, anti-FIIa activity or the two immunological assays). In these 21 cases with two positive findings PVT (n = 68) BCS (n = 21)

SVT (PVT + BCS) n = 89
Baseline characteristics n (%) or mean ± SD  cf.ac.uk/ac/all.php). Interestingly, the residue is highly conserved (80%) in ATs from 49 species (Figure 1). All these data support that this mutation caused a type II deficiency with impaired reactivity (type II RS). No long-term anticoagulation was administered and no re-thrombosis was observed after 9 years of follow-up.
Afterwards, since mutations that are not usually detected by current functional assays 23 and that are responsible for transient AT deficiencies 15,24 are usually located in exons 2 and 7, these exons were sequenced in the whole cohort of 89 patients. One mutation was found: • Patient 3 was a 51 year-old-woman with a known MPN that during follow-up developed a portal and splenic vein thrombosis. She presented strictly normal levels of AT anti-FXa (96%) and anti-FIIa (99%) and no AT deficiency according to immunological assays.
Since the development of thrombosis she was on anticoagulation and has not presented new thrombotic events after 7 years of follow-up. The present study, however, has enabled to detect that she was a heterozygous carrier of the c.89T>A; p.Val30Glu mutation (AT Dublin). 15 The presence of this mutation was confirmed in a new blood sample. Summing up, overall four patients (4.5%) in whom AT deficiency had been previously clinically discarded, in the present study were found to have different AT mutations or deficiencies.    (Table 3).

| D ISCUSS I ON
The present study is the first to evaluate the incidence of undetected AT deficiencies among patients with SVT in whom a first initial screening had ruled out AT disorders. Moreover, the study of a ity as a diagnostic test, 13 we hypothesized that these new diagnostic techniques could play a role in facilitating the diagnosis of masked AT deficiencies in the setting of SVT. In Cohort 1 we identified that 4.5% of patients previously considered to have a normal AT function presented an underlying AT disorder, which is especially significant when contrasted with the general population prevalence of AT deficiency of 1/500-5000. 25 Importantly, in all these four patients the development of SVT had been exclusively attributed to either a concomitant local inflammation or to a MPN, but it was afterwards proved that they also had an AT disorder. These findings reinforce the concept that having two synergic prothrombotic conditions probably leads to an even higher hypercoagulable state, and that therefore a complete etiologic work-up should be performed in all patients, even if they already present a plausible explanation for the thrombotic event.
The analysis of Cohort 2, one of the largest worldwide cohort of patients with AT deficiency and a previous thrombotic event, demonstrates that thrombotic events in the portal and mesenteric veins is not an unusual localization among carriers of this severe thrombophilia.
Indeed, in 4.1% of AT deficiency cases, the first thrombotic event was in the splanchnic territory. Although it is highly relevant if we take into account that the general population prevalence of SVT is <5 in 10 000 inhabitants, it is not possible to infer from this cohort the incidence of SVT among patients with AT, as they were only diagnosed with AT de- Finally, disorders affecting the efficacy of the N-glycosylation can impair AT function and therefore increase the risk of thrombosis. 17 In this study we have identified two cases with SVT due to hypoglycosylation, one in each cohort. This finding strongly supports that this aberrant post-translational modification causes a severe pro-thrombotic state.
We must acknowledge that one of our main limitations is the difficulty in differentiating the new mutations described in the present study from mere polymorphisms. However, taking into account the clinical setting of thrombosis and the low frequency in general population (MAF) of the mutations reported that reinforces their pathogenic role, it seems reasonable to assume causality between our findings and the development of thrombosis.
According to current evidence, it is recommended to screen patients with SVT for underlying thrombophilia and, given its high frequency among SVT, AT should be carefully studied. Figure 3 suggests an AT screening algorithm depicting when to sequence SERPINC1 and to assess N-glycosylation defects. Even though there is not enough evidence nor cost-effectiveness studies to suggest systematic screening of SERPINC1 mutations and CDG in all SVT patients (instead of performing only anti-FXa activity determinations), we consider that the findings of the present study, together with the small size of SERPINC1 and the development of massive sequencing methods that allow a relatively easy, fast and cheap sequencing of this gene, make it appropriate to recommend a comprehensive AT evaluation in cases of rethrombosis or doubtful interpretation of anti-FXa activity levels ( Figure 3).

ACK N OWLED G EM ENTS
We are most indebted to Lara Orts, Isabel Requejo and Leticia AT deficiency ** * In these cases it would be also interesƟng to sequence SERPINC1 and screen for CDG. However, as it is probably not cost-effecƟve, more studies are warranted before recommending this aƫtude in clinical pracƟce ** Consider sequencing SERPINC1 and screen for CDG to further caractherize the type of AT deficiency Decreased levels of liver sinthesized factors?

Yes No
No acƟon needed * No acƟon needed *