Abstract
The prevalence of germline mutations in paediatric pulmonary hypertension (PH) is poorly documented. The objective of this study was to determine the mutation frequency in PH genes in a paediatric cohort and describe the clinical characteristics of mutation carriers.
The study involved 66 index cases with PH: 35 children with idiopathic pulmonary arterial hypertension (IPAH); five children with familial PAH (FPAH); three children with pulmonary veno-occlusive disease (PVOD); and 23 children with PAH associated with congenital heart disease (APAH-CHD).
No mutations were found in the 23 children with APAH-CHD. In the 40 children with IPAH or FPAH, 12 mutations were found: five on BMPR2; four on ACVRL1; and three on TBX4. In the three PVOD cases, two carried the EIF2AK4 mutation. Mutation carriers had a more severe disease at diagnosis and more aggressive first-line therapy was required. The three patients with PVOD had a very severe disease at diagnosis and required a lung transplantation.
The genetic architecture of paediatric PAH is enriched in ACVRL1 and TBX4 mutations compared to adult PAH, but further studies are required to confirm these results. Childhood-onset PAH in children carrying a mutation in one of the genes tested has a more severe presentation at diagnosis but a similar outcome to that observed in non-carriers.
Abstract
Paediatric pulmonary hypertension has a specific genetic architecture http://ow.ly/54yP301hCQi
Introduction
Pulmonary arterial hypertension (PAH) is a rare and devastating disease, resulting from progressive obliteration of small calibre pulmonary arteries by proliferating vascular cells and intimal fibrosis, and leading to cardiac failure, with an untreated mean survival <3 years in adults and <10 months in children [1–3]. PAH can be a complication of well-identified pathological conditions, it can occur in the context of family history or genetic mutations (heritable PAH; HPAH), or it is considered as idiopathic (IPAH) in the absence of an identified predisposing factor [4]. IPAH has an unknown origin and represents around 70% of PAH in adults [5]. In the prospective TOPP (Tracking Outcomes and Practice in Pediatric PH) registry, which collects data on paediatric and adolescent pulmonary hypertension, 88% of children had PAH, of which 57% were characterised as IPAH or HPAH, and 36% as PAH associated with congenital heart disease (APAH-CHD) [6]. HPAH is a genetically heterogeneous disease. The major gene associated with adult HPAH is BMPR2, a gene encoding a type 2 receptor of the bone morphogenetic protein signalling pathway. Low penetrance dominant BMPR2 mutations were found in >70% of familial PAH (FPAH) and in 15% of IPAH [7, 8]. Penetrance of these mutations is different in the two sexes (14% in males versus 42% in females according to Larkin et al. [9]). Mutations in other genes of the same pathway, ACVRL1 and in rare cases ENG, may cause PAH associated with hereditary haemorrhagic telangiectasia (HHT) in most cases [10, 11]. Few mutations in the SMAD family of genes, and in the KCNK3 and CAV1 genes have been reported [12–14]. Recently, TBX4 mutations (small patella syndrome) have been identified as being associated with childhood-onset PAH [15]. Biallelic recessive mutations of the EIF2AK4 gene have been found at the origin of heritable pulmonary veno-occlusive disease (PVOD) and in pulmonary capillary haemangiomatosis that present with closely related phenotypes predominating either on venous lesions or on capillary proliferation, respectively [16, 17].
A few studies were specifically dedicated to identifying gene mutations in paediatric PAH. Harrison et al. [18] identified two BMPR2 mutations arising de novo and two HHT-provoking mutations in ENG and ACVRL1 in 16 children with IPAH. Rosenzweig et al. [19] found 8 BMPR2 mutation carriers (10%) in 78 children with PAH, most of them with FPAH (7/8). In a series of 29 FPAH and IPAH children, Pfarr et al. [20] found BMPR2 mutations in four children with IPAH (14%), two mutations in ACVRL1, and one variant of unknown significance (VUS) in ENG. Fujiwara et al. [21] studied a series of 21 probands under 16 years of age with PAH and found four mutations in BMPR2 (19%) and five mutations in ACVRL1. Identifying mutations in PAH genes has a prognostic significance as adult patients with BMPR2 or ACVRL1 mutations are younger at diagnosis and time to death is shorter in these two groups of mutation carriers versus non-carriers [10]. In paediatric forms of PAH, Chida et al. [22] showed a similar age at onset and a worse prognosis restricted to BMPR2 carriers for whom time to death was significantly shorter, but this difference was not significant in children carrying the ACVRL1 mutation.
In APAH-CHD, the causal relationship between the left-to-right shunt and PAH is considered obvious in the majority of cases. The last international classification defined type 3 APAH-CHD as PAH with coincidental CHD that includes small shunt defects not causing severe PAH and following a course similar to IPAH [23]. The role of PAH genes mutations in these patients is unknown, nor is it known in children with post-operative PAH (type 4) after CHD repair of any type who go on to develop unexpected pulmonary vascular disease. Limited genetic data is available for paediatric patients with APAH-CHD. Pfarr et al. [20] described one ENG unlikely pathogenic VUS in a patient with type 2 APAH-CHD (left-to-right shunt) and one BMPR2 mutation in a familial case of PAH with CHD. Harrison et al. [18] did not identify any mutation in two cases of APAH-CHD. Roberts et al. [24] described three BMPR2 VUS in three children with type 2 APAH-CHD.
Limited data are available on the genetic architecture of childhood-onset PAH, taking into account the investigation of recently identified genes. This study sought to determine the proportion of pulmonary hypertension (PH) gene mutations in children with FPAH, IPAH, PVOD and in type 3 and 4 APAH-CHD, and describe their baseline clinical characteristics at time of diagnosis and their respective outcomes.
Methods
Study population
From January 2007 to December 2013, index paediatric patients (n=66) were included in the study; 66 unrelated patients aged 6.8±5.13 years (mean±sd) with a PH diagnosis were included after their parents gave written informed consent for this study. All patients were studied in the same paediatric cardiac department. PH diagnosis was determined during right-heart catheterisation, and defined as mean pulmonary arterial pressure >25 mmHg at rest, pulmonary vascular resistance index (PVRI) >3 Wood units·m2, and mean pulmonary capillary wedge pressure >12 mmHg. A complete diagnostic examination was performed in all children to detect comorbidities that may predispose to the development of PAH. HHT symptoms were searched for in all cases and their parents. Patients were classified according to the international classification agreed in Nice, France, in 2013 [4]. PAH was considered to be idiopathic after clinical and biological investigation eliminated all known causes. Cases were considered as FPAH when at least two cases were reported in the family extending up to third-degree relatives. Patients with type 1 (Eisenmenger's syndrome) and type 2 (left-to-right shunt) APAH-CHD [23] were not included because haemodynamic status was considered sufficient to explain PAH occurrence. Patients with type 3 (n=13) and 4 (n=10) APAH-CHD were included. Dysmorphic syndromes or inborn errors of metabolism, known aneuploidies (trisomy 21) and other cytogenetic anomalies (on chromosomes 1 and 14) were excluded (n=18). Patients had an echocardiography, and N-terminal pro-brain natriuretic peptide (NT-proBNP) and circulating endothelial cells measurements at diagnosis (tables 1 and 2) [25]. Incident cases of PAH were those genotyped at diagnosis within the period of inclusion (33/40), whereas prevalent cases (7/40) were alive at inclusion but genotyped after a variable delay (7±4.5 years; mean±sd). Similarly, patients with APAH-CHD were both incident (11/23) and prevalent (12/23).
Clinical and haemodynamic characteristics of children with IPAH and FPAH at diagnosis according to the presence or the absence of mutation in the PAH genes
Characteristics at diagnosis and treatment initiation according to mutation type
For children under the age of 18 years, genetic testing was proposed to symptomatic individuals after parent approval (no case of parent refusal). When a mutation was identified, genetic testing was performed in parents to differentiate inherited from de novo mutations. Genetic testing and detailed clinical examination was proposed to all relatives for PH diagnosis. Unaffected siblings were genotyped after parent approval and according to French laws for screening unaffected children.
Our institutional review board and ethics committee approved the protocol.
Genetic analysis
All patients were screened for BMPR2 and ACVRL1 point mutations and large rearrangements. Patients with IPAH or FPAH, and without a mutation in BMPR2 or ACVRL1, were further screened for EIF2AK4, TBX4 and KCNK3 point mutations. ENG and SMAD9 screening was performed for patients with FPAH for whom no mutation was identified in the other genes (figure 1). Since TBX4, KCNK3 and EIF2AK4 were not known to be involved in PH at the beginning of the study, included patients were retrospectively tested for these additional genes.
Flow chart of genetic analysis scheme. Each gene tested is indicated together with the technique used for detecting mutations or rearrangements by multiplex ligation-dependent probe amplification (MLPA®). The number of subjects carrying a mutation is indicated for each gene. PAH: pulmonary arterial hypertension; PVOD: pulmonary veno-occlusive disease; IPAH: idiopathic PAH; FPAH: familial PAH; APAH-CHD: PAH associated with congenital heart disease.
Screening of point mutations and large rearrangements was performed as described in Sztrymf et al. [7]. Primers used for BMPR2, ACVRL1, ENG, EIF2AK4, SMAD9 and KCNK3 have been previously described [7, 10–12, 14, 16]. Primers used for TBX4 screening are provided in supplementary table S1.
Nucleotide numbering follows the Human Genome Variation Society recommendations. All identified mutations were analysed with the Alamut software, version 2.2 (Interactive Biosoftware, Rouen, France).
Outcome
All patients were followed up with a standardised protocol at least twice a year, including World Health Organization functional class, NT-proBNP and echocardiography measurements (table 2). The occurrence of the following events was taken into account for analysis, either in isolation or in combination: death; lung or heart-lung transplantation; Potts shunt.
Statistics
Data are presented as mean±sd. The XLSTAT 2014 software (Addinsoft, New York, NY, USA) was used to perform a non-parametric Mann–Whitney U test, to compare patients with or without a mutation, or to calculate a Chi-squared test when appropriate; p<0.05 was considered statistically significant. The MedCalc statistical software, version 12.7.7 (MedCalc Software bvba, Ostend, Belgium) was used to construct Kaplan–Meier survival curves (figure 2) and calculate the log-rank test, which was used to test for a difference in the free probability distribution between mutation carriers and non-carriers (incident cases only) for three major events (death, transplantation, Potts shunt).
Distribution of probability of event-free survival according to mutation carrier status in pulmonary arterial hypertension patients (incident cases only). Events taken into account: death; transplantation; Potts shunt. The log-rank test was used to compare the two distribution curves: Chi-squared=0.6680; d.f.=1; p=0.4137.
Results
Mutations in PH genes
Following pedigree analysis and clinical examination, 35 children were classified as IPAH, five as FPAH, 13 as type 3 APAH-CHD, 10 as type 4 APAH-CHD and three as PVOD. A mutation was identified in 14 of the 66 children: eight in IPAH patients (23% of IPAH), four in FPAH patients (80% of FPAH) and two in PVOD patients (67% of PVOD). No mutations in the BMPR2 and ACVRL1 genes were identified in children with type 3 and type 4 APAH-CHD (table 3).
Mutations in pulmonary hypertension genes
Five BMPR2 (three IPAH and two FPAH), four ACVRL1 (three IPAH and one FPAH), three TBX4 (two IPAH and one FPAH) and two biallelic EIF2AK4 mutations (two PVOD) were identified (table 3).
Ten mutations have been previously described and classified as deleterious mutations. Four new mutations were identified: one large rearrangement of BMPR2 (deletion of exons 2–7) and three TBX4 mutations which are all non-sense mutations (all mutations are described in supplementary table S2).
No mutations were identified in the ENG, KCNK3 or SMAD9 genes. The four ACVRL1 mutation carriers showed symptoms of HHT. In the 14 index cases with an identified mutation, parental genetic screening for 11 cases was performed. In two IPAH cases, the BMPR2 mutations identified were de novo mutations. The remaining mutations were inherited and in all cases the parent carrying the mutation was asymptomatic for PAH or a single allele carrier for the EIF2AK4 mutations for PVOD.
In the three families with the TBX4 mutation, the mutation was inherited. The parent carrying the mutation was asymptomatic for PAH with normal pulmonary arterial pressure (PAP) at echocardiography, but symptomatic for bone abnormalities associated with TBX4 mutations.
Mutations in the EIF2AK4 gene were identified in two patients (previously reported in Eyries et al. [16]), who presented with the clinical, haemodynamic and chest computed tomography (CT) features of PVOD, while no EIF2AK4 mutation was identified in children with IPAH or FPAH. An 8-year-old girl was suspected to have PVOD on chest CT at the time of diagnosis, but no mutation in EIF2AK4 was found, nor in other genes tested. Subsequent CT scans, lung biopsy and evolution confirmed the diagnosis and her clinical status required transplantation during follow-up.
Clinical status at time of diagnosis in IPAH and FPAH patients
At time of diagnosis, IPAH and FPAH patients who had a mutation in one of the PAH genes were in a worse functional class, had more frequently associated signs of right ventricular dysfunction (high right atrial pressure) and had more frequent syncope (table 1). Although statistically they did not have different mean PAP, PVRI and cardiac index, values for these parameters tended to be worse in mutation carriers. Only one TBX4 mutation carrier was considered a responder at the acute vasodilation test according to the Sitbon criteria [26]. Consequently, upfront combination therapy (oral combination of endothelin receptor antagonists and phosphodiesterase type 5 inhibitors) or triple therapy with a prostanoid analogue, was more frequently used as the first-line therapy in this group. Of note, female predominance was observed as in the adult population, but the sex ratio was similar in mutation carriers and non-carriers. There was no difference in age at diagnosis between these two groups.
The more severe clinical and haemodynamic status at diagnosis in the two groups was overrepresented in the group of children younger than 2 years (four with mutation and eight without).
Table 2 shows the baseline characteristics of patients for each type of mutation.
Outcomes
The mean duration of follow-up of the mutation carriers and non-carriers was similar (table 4). In IPAH and FPAH mutation carriers, 11/12 children were in functional class I or II and PVRI was significantly improved by treatment at the last follow-up. None of the mutation carriers died during follow-up. By contrast, in the non-carrier group, four patients died, two immediately after diagnosis despite upfront triple therapy, and the other two 4 years after diagnosis from class III heart failure. The three patients with PVOD had rapid functional deterioration and underwent a transplant.
Clinical characteristics and major events at last follow-up in patients with idiopathic and familial pulmonary arterial hypertension
The number of patients in both groups who had events and were on triple therapy during follow-up was similar in the two groups (table 4). The event-free (death, transplantation, Potts anastomosis) survival curve in the two groups showed that the occurrence rate in mutation carriers and non-carriers was not significantly different (figure 2).
Discussion
This study looked at the genetic status of a cohort of children with various forms of PH to better define the genetic architecture of PAH in childhood and evaluate the clinical conditions associated with different gene mutations. Patients with IPAH, FPAH, PVOD and type 3 and 4 APAH-CHD were included and screened for mutations in known PAH and PVOD genes, including recently identified genes (TBX4, KCNK3 and EIF2AK4). Patients for whom PAH was part of a systemic disease or associated with chromosomal anomalies were not included in the study.
No mutation in PAH coincidental with CHD (type 3 APAH-CHD) and in post-operative PAH (type 4 APAH-CHD) in the BMPR2 and in ACVRL1 genes was found, suggesting that the genetic background is different from IPAH. In these two groups, the absence of a predisposing mutation suggests either that haemodynamic factors are sufficient to induce PAH, or that other, yet-to-be identified genetic factors are involved in this setting. A sample of this size had a 94% power (data not shown) to detect any susceptibility variant in the genes tested if the genetic background was the same. Previously published results on a small series of patients with APAH-CHD did not detect pathogenic mutations in either the BMPR2 or ACVRL1 genes; only missense VUS (most of them probably benign according to in silico tools) were found but no deleterious mutations, in the same context [18, 20, 24]. Therefore, both the results of this study and those already published do not justify screening these patients with the genes used in this study.
In FPAH, BMPR2 mutations were identified in less than half of index cases, whereas BMPR2 mutations are found in more than 80% of adult familial cases [27] and other mutations were found in ACVRL1 and TBX4 that represented nearly 50% of mutations found in children. Similarly, in IPAH, beside the BMPR2 mutations found in similar proportions in the adult PAH population, mutations were also found in the ACVRL1 and TBX4 genes that were as frequent as BMPR2 mutations if considered together. This study lacks the statistical power to reach significant differences in mutation distribution in adults with FPAH because the sample size was small; further studies are needed to confirm these differences. ACVRL1 missense mutations found in this study have already been reported in HHT patients and were found in children from families where HHT symptoms were present either in affected children or in the carrier parent since PAH can precede other symptoms of HHT in young children but are usually present in the carrier parent [10]. Indeed, Girerd et al. [10] previously showed that, in adults, the age of onset of PAH associated with a mutation in ACVRL1 was earlier than in BMPR2 mutation carriers and this age of onset was earlier than in adults without a detected mutation.
TBX4 mutations were detected in 3/40 (7.5%) of IPAH/FPAH patients in our study, a proportion lower than that reported by Kerstjens-Frederikse et al. [15] where TBX4 mutations were detected in 21% of PAH children without developmental anomalies. Routine testing of this gene in paediatric and adult PAH will help identify the importance of TBX4.
TBX4 mutations were transmitted from parents presenting with skeletal malformations typical of small patella syndrome, but without symptoms of PAH. Indeed, a mutation in the TBX4 gene rarely causes PAH in adults (<2% according to Kerstjens-Frederikse et al. [15]).
Currently, the major change in lung penetrance of TBX4 mutations from one generation to the following and the severity of PAH observed in some children cannot be explained. Interaction with a genetic factor inherited from one parent and/or involvement of epigenetic mechanisms are possible hypotheses that require further studies. The absence of adults affected by PAH and carrying TBX4 mutations could be because of high mortality in this type of childhood-onset PAH, but no follow-up and family data are currently available to support such a hypothesis. Our limited experience (n=73) in routinely searching for TBX4 mutations in adult PAH patients shows a frequency of mutations similar to what has been already reported (data not shown) [15].
All three children with PVOD had very rapid clinical worsening and underwent a lung transplant. Although the three patients had a similar evolution leading to lung transplantation, no mutation of the EIF2AK4 gene was found in one of the three patients who had no affected siblings. No genetic heterogeneity was detected in the study by Eyries et al. [16], but other unidentified genetic factors might have been involved in this patient.
In this study, mutation carriers had a more severe clinical presentation at diagnosis and this is in accordance with previously reported experience in adult patients with BMPR2 and ACVRL1 mutations [7, 10]. However, differences with regard to haemodynamic status did not reach significance, probably because of the small number of patients involved. Because of this worse clinical presentation at diagnosis, initial therapy was more aggressive in this group (table 2) but this difference was no more significant at the outcome (table 4). Interestingly, in adult and paediatric PAH, BMPR2 mutation carriers are nearly never responders to calcium channel blockers. This is also the case in the series of childhood-onset PAH in this study, except for one patient with a TBX4 mutation (tables 2 and 3) [19, 28, 29].
In this study, identification of the mutation did not influence the initial treatment; this was based on the usual criteria before the results of the genetic screening were obtained. This holds true if patients with PVOD, in whom lung transplantation was indicated in all cases, were excluded.
In adults, BMPR2 and ACVRL1 mutation carriers also have a more severe presentation at diagnosis and die younger than mutation non-carriers [7, 10, 29]. The sample size in this study does not allow for any significant difference in outcome to be detected according to mutation carrier status (table 4 and figure 2) Patients with a more severe disease presentation, who required lung transplantation or Potts anastomosis immediately after diagnosis, were PVOD patients (EIF2AK4 mutation) or ACVRL1 mutation carriers as also observed in adults [10]. In the study by Kerstjens-Frederikse et al. [15], there was no increased mortality in TBX4 carriers compared to other childhood-onset PAH patients. This was also the case in this study.
This study has some limitations because it involves a rare disease and the sample size is small. The lack of power might explain the absence of significant differences between groups. Allowing prevalent cases, who can be considered as survivors, within the inclusion period could have modified the mutation carrier frequency. However, the mutation carrier frequency was similar in prevalent (2/7; 29%) and in incident cases (10/33; 30%). This is not relevant for APAH because no mutation was detected in the whole group. Although unlikely, some mutations may have been missed because some patients were not included or because some genes were not screened in all the patients included in the study (figure 1).
In conclusion, this study shows that screening patients with type 3 and 4 APAH-CHD for BMPR2 and ACVRL1 is not advantageous and this result corroborates the absence of deleterious mutation found in other studies [18, 20, 24]. The proportion of PAH patients with a mutation is comparable to that of adults but with a very different distribution of the involved genes. The important contribution of TBX4 mutations points out that the penetrance of these mutations is different between adults and children.
Finally, more severe clinical status at diagnosis in patients with a mutation appears to be comparable to what has been previously observed in the adult population.
The series in this study does not allow specific recommendations for initial therapy in the population of children with a mutation. The presence of a mutation in one of the genes tested did not predict clinical outcome in our study, except for the EIF2AK4 gene. Screening for these mutations in a large population of paediatric IPAH and FPAH patients may help to further define the predictive value of these mutations with regard to clinical outcome.
One key issue of genetic diagnosis in paediatric PAH is the possibility to test relatives for their carrier or non-carrier status, such as siblings in the case of PVOD transmitted as a recessive disease. The benefit of an early diagnosis and careful follow-up should exceed the prejudice of a diagnosis given for a severe disease before symptoms have appeared, and for which the probability of appearance has not been determined precisely. Knowing the poor prognosis of some forms of paediatric HPAH will be a source of high emotional stress for parents, but one that would help them in their parental decisions. In these cases, genetic counselling delivered by geneticists and genetic counsellors should help parents to clearly understand the genetic situation and inform their decision-making.
Acknowledgements
We are grateful to Anne Leroy, Marie-Christine Waill and Salim Bakas, Genetics Dept, GH Pitié-Salpêtrière, Assistance Publique Hôpitaux de Paris, Paris, France, for genotyping the patients. We thank David Tregouët, INSERM, Institute of Cardiometabolism and Nutrition, UMR_S 1166-ICAN, Paris, France, for advice with the statistical data analysis, and Professor David Smadja, Hematology Department, AP-HP, Hôpital Européen Georges Pompidou, Paris, France, for measuring circulating endothelial cells.
Authors' contributions: M. Levy, D. Bonnet: clinical care of patients, study design, article writing. I. Szezepanski: care of patients, data collection. F. Soubrier, M. Eyries: genetic data, study design, article writing. S. Nadaud, M. Ladouceur: statistical analysis of data, critical review of the article.
Footnotes
Editorial comment in Eur Respir J 2016; 48: 987–989.
This article has supplementary material available from erj.ersjournals.com
Conflict of interest: Disclosures can be found alongside this article at erj.ersjournals.com
- Received January 28, 2016.
- Accepted June 7, 2016.
- Copyright ©ERS 2016
References
