An emerging phenotype of pulmonary arterial hypertension patients carrying SOX17 variants

Discussion

We report the phenotype and outcomes of 20 childhood-onset and adult-onset PAH patients carrying a SOX17 variant. These patients are remarkable because they present with a variety of cardiovascular characteristics including severe pulmonary arterial remodelling with plexiform lesions, dilated systemic bronchial and nonbronchial arteries, and CHD. Indeed, ASD or VSD was observed in one-third of our patients, mainly in childhood-onset disease, underscoring that PAH patients presenting with an associated CHD may have a heritable disease caused by a SOX17 pathogenic variant. Lung HRCT of the chest and CTPA detected ground-glass opacities and abnormal pulmonary arteries, systemic bronchial and nonbronchial arteries. More than half of our patients presented with severe clinical and haemodynamic impairment refractory to PAH therapy justifying lung transplantation, which was associated with frequent complications (notably primary graft dysfunction and haemothorax). Lung histopathology showed severe pulmonary arterial remodelling, subpleural vessel dilation and numerous haemosiderin-laden macrophages.

The radiological and histopathological findings depicted in our cohort are likely to be related to the embryological role of SOX17. Members of the SRY-related HMG box (SOX) family are shown to be essential for the regulation of numerous developmental processes, where SOX17 acts as a crucial regulator in pulmonary vascular morphogenesis [16, 17] and cardiovascular development [18]. SOX17 is a transcription factor implicated in the regulation of various cell developmental processes, notably endoderm formation and tumour angiogenesis [16, 19], but is also required for the normal development of the pulmonary vasculature [17, 20]. Although SOX17 has a well-demonstrated role in embryogenesis, PAH occurrence at adult age in some patients suggests that SOX17 is also implicated in pulmonary vascular remodelling in adults. Approximately 20 SOX genes have been characterised and three SOX group F genes (SOX7, SOX17 and SOX18) are coexpressed in vascular endothelial cells [21]. In animal models, SOX7, SOX17 and SOX18 have overlapping roles in postnatal neovascularisation, and vascular endothelial-specific deletion of all three leads to massive oedema [22]. However, SOX17 plays a key role in developing and maintaining arterial endothelial cell specificity and integrity [16]. SOX17 is selectively expressed in arteries (but not in veins) and SOX17 knockout reduces the expression of arterial-specific genes and induces the expression of venous-specific genes [16]. In a mouse postnatal retina model, the inhibition of Sox17 expression in endothelial cells led to strong vascular hypersprouting, a loss of arterial integrity and large arteriovenous malformations [16]. These studies support the critical role of SOX17 in angiogenesis, maintenance of vascular homeostasis and arterial specificity [20]. The SOX family is characterised by a highly conserved HMG box, a sequence-specific DNA-binding domain, where most of the missense variants are located. Site-directed mutagenesis studies have shown that missense mutations within this region can impair both direct DNA binding [23] and SOX17/β-catenin protein complex interactions [24], demonstrating that sequence alteration within this domain has a strong functional impact on the SOX17 protein.

We identified a SOX17 pathogenic variant in two unrelated patients with familial PAH. Segregation analysis was then performed in seven additional family members of these two patients, and the SOX17 variant was found in the two PAH patients and in four unaffected relatives (figure 1a). Considering these family trees and the higher frequency of the sporadic PAH cases, we conclude that PAH due to SOX17 mutations has an autosomal transmission with incomplete penetrance. This genetic inheritance, as well as the female predominance, are consistent with what has been described previously for other PAH predisposition genes such as BMPR2, the leading cause of heritable PAH [5, 25, 26]. PAH due to SOX17 mutations was relatively rare in previously reported idiopathic or familial PAH cohorts, with SOX17 variants found in ∼0.9% and ∼0.7% in reports of 1038 [7] and 413 PAH patients [8], respectively. In our present study, ACMG class 4 and 5 variants were identified in 3.1% of PAH index patients. We intentionally included ACMG class 3 variants in our study in order to have a more exhaustive cohort, although their pathogenicity is not demonstrated. It appears, however, that one of the two patients harbouring a class 3 variant (Patient XV) had a severe haemodynamic impairment (PVR 10.3 WU) and underwent double lung transplantation 7 months after diagnosis (with pathological abnormalities similar to what was noticed in other patients). The role of SOX17 in PAH was only recently reported and its mutation frequency was undoubtedly underestimated until now. Genetic screening in patients with CHD-PAH was not routinely done before evidence of the role of certain development genes such as TBX4, SOX17 and KDR was demonstrated. In the French PH Referral Centre, children and adults with CHD-PAH are systematically screened with a genetic panel comprising those genes, which explains an increased identification of these mutations in our cohort.

Rare deleterious SOX17 variants have also been identified in ∼3.2% of patients with CHD-associated PAH in another cohort of 256 patients, suggesting an over-representation of SOX17 variants in this form of PAH [8]. These results are in accordance with those of our report, in which one-third of SOX17 variant carriers with PAH presented with an associated CHD, a finding reminiscent of that observed with other developmental genes associated with heritable PAH, such as TBX4 [25, 27]. Saba et al. [28] demonstrated that Sox17 mesoderm-specific loss of function in mouse embryos is associated with cardiac defects, which could explain this frequent association. Interestingly, SOX17 variant carriers with PAH and CHD were significantly younger at PAH diagnosis (childhood onset) compared with SOX17 variant carriers with isolated PAH (adulthood onset) (figure 2). We hypothesise that two non-mutually exclusive mechanisms could explain this difference. First, coexisting PAH and CHD may be the consequence of a major development impairment caused by dysfunction of SOX17, a key player of cardiac and pulmonary vascular development [18, 20]. Second, CHD could act as a second hit triggering accelerated pulmonary vascular remodelling in a predisposed dysfunctional pulmonary circulation related to SOX17 mutation.

At diagnosis, SOX17 variant carriers with PAH presented with severely compromised haemodynamic parameters, with a median mPAP of 67 mmHg and a median PVR of 14.0 WU. These parameters are in line with the haemodynamic severity reported in large series of PAH patients with a BMPR2 mutation [26, 29]. HRCT of the chest and CTPA identified a SOX17 variant carrier radiological phenotype with numerous vascular abnormalities (dilated and tortuous vessels associated with micronodular and ground-glass opacities), which are not usual findings of idiopathic PAH [30]. Pathological examination can explain these radiological findings, with evidence of a congested architecture, dilated vessels (notably in the subpleural zones) and severe pulmonary arterial remodelling with plexiform lesions. Moreover, ground-glass opacities seen at HRCT are most likely explained by the micro-haemorrhagic foci depicted in lung histopathology.

SOX17-associated vascular malformations are associated with recurrent haemoptysis, frequently requiring bronchial artery embolisation and eventually listing for lung transplantation. Increased haemoptysis risk was already described in BMPR2 mutation carriers [15]. Considering the risk of severe life-threatening haemoptysis in SOX17 mutation carriers, further research is needed to evaluate the value of preventive systemic artery embolisations once such vascular malformations are identified. However, such preventive procedures would not be exempt from potential risks, at least in part due to the presence of arteriovenous shunts. These shunts are seen on CTPA and further supported by identification of embolisation material in subpleural veins in a lung explant following embolisation, supporting the presence of an anatomical communication between bronchial arteries and pulmonary veins (figure 5).

The overall prognosis of PAH patients carrying a SOX17 mutation is poor, with more than half of our patients receiving transplantation at follow-up and one early death in a patient treated only with calcium channel blockers. Transplant-free survival is, however, difficult to evaluate since diagnosis was made in several prevalent PAH cases. Our data suggest a high risk of complications following lung transplantation, notably haemothorax and primary graft dysfunction, that should be firmly confirmed in larger multicentre studies. Periprocedural haemorrhagic complications may be explained at least in part by lung histopathology showing dilated and fragile vessels, notably within the subpleural space (supplementary table S1). Patients carrying a SOX17 pathogenic variant should thus be managed by a trained lung transplantation team with extreme caution before and immediately after lung transplantation, highlighting the relevance of genetic screening and imaging of the chest of such PAH patients.

Healthy carriers of pathogenic SOX17 variants had no or mild radiological abnormalities and they had no evidence of pre-symptomatic cardiac malformation. This wide phenotypic variability within the same family harbouring the same mutation has already been observed in PAH associated with mutations in developmental genes, especially in small patella syndrome associated with TBX4 mutations [25]. Genetic counselling and screening of first-degree relatives is recommended in PH guidelines [4, 6]. This is particularly true in healthy carriers of SOX17 variants when the risks of PAH and CHD are considered. Longitudinal follow-up of healthy SOX17 variant carriers will reveal whether pulmonary vascular anomalies and PH may occur with time. Screening procedures remain to be defined in these individuals since the penetrance of PAH and CHD in SOX17 variant carriers is unknown. We previously demonstrated that echocardiography and N-terminal pro-brain natriuretic peptide blood level analysis may be insufficient tools to effectively screen for PH in asymptomatic relatives harbouring BMPR2 mutations [31]. More exhaustive strategies, including cardiopulmonary exercise testing, D_LCO measurement and ECG, might be useful screening tools in first-degree relatives of patients with heritable PAH [31]. The vascular phenotype associated with SOX17 pathogenic variants suggests that CTPA should be encouraged in first-degree relatives. Echocardiography screening of CHD of first-degree relatives of patients carrying a SOX17 pathogenic variant should also be considered, allowing detection of asymptomatic CHD and timely management.

In conclusion, SOX17 pathogenic variants are associated with a severe form of heritable PAH that can be associated with a variety of cardiac and thoracic vessel anomalies. Although our cohort was of relatively limited size, our data clearly demonstrate a unique phenotype of heritable PAH related to SOX17 pathogenic variants. International prospective registries are warranted to further reveal and understand the phenotype of rare variants associated with PAH, such as SOX17 pathogenic variants. Identification of signs highly suggestive of SOX17 pathogenic variants is critical since this phenotype is associated with an augmented risk of various complications, such as recurrent haemoptysis. Furthermore, the role of SOX17 in pulmonary vascular development and remodelling is of growing interest, as restoring both its gene expression and signalling might constitute a future therapeutic solution for this severe disease.