Abstract

Introduction Pathogenic variants in the gene encoding for BMPR2 are a major genetic risk factor for heritable pulmonary arterial hypertension. Owing to incomplete penetrance, deep phenotyping of unaffected carriers of a pathogenic BMPR2 variant through multimodality screening may aid in early diagnosis and identify susceptibility traits for future development of pulmonary arterial hypertension.

Methods 28 unaffected carriers (44±16 years, 57% female) and 21 healthy controls (44±18 years, 48% female) underwent annual screening, including cardiac magnetic resonance imaging, transthoracic echocardiography, cardiopulmonary exercise testing and right heart catheterisation. Right ventricular pressure–volume loops were constructed to assess load-independent contractility and compared with a healthy control group. A transgenic Bmpr2Δ71Ex1/+ rat model was employed to validate findings from humans.

Results Unaffected carriers had lower indexed right ventricular end-diastolic (79.5±17.6 mL·m−2versus 62.7±15.3 mL·m−2; p=0.001), end-systolic (34.2±10.5 mL·m−2versus 27.1±8.3 mL·m−2; p=0.014) and left ventricular end-diastolic (68.9±14.1 mL·m−2versus 58.5±10.7 mL·m−2; p=0.007) volumes than control subjects. Bmpr2Δ71Ex1/+ rats were also observed to have smaller cardiac volumes than wild-type rats. Pressure–volume loop analysis showed that unaffected carriers had significantly higher afterload (arterial elastance 0.15±0.06 versus 0.27±0.08 mmHg·mL−1; p<0.001) and end-systolic elastance (0.28±0.07 versus 0.35±0.10 mmHg·mL−1; p=0.047) in addition to lower right ventricular pulmonary artery coupling (end-systolic elastance/arterial elastance 2.24±1.03 versus 1.36±0.37; p=0.006). During the 4-year follow-up period, two unaffected carriers developed pulmonary arterial hypertension, with normal N-terminal pro-brain natriuretic peptide and transthoracic echocardiography indices at diagnosis.

Conclusion Unaffected BMPR2 mutation carriers have an altered cardiac phenotype mimicked in Bmpr2Δ71Ex1/+ transgenic rats. Future efforts to establish an effective screening protocol for individuals at risk for developing pulmonary arterial hypertension warrant longer follow-up periods.

Introduction

Pulmonary arterial hypertension (PAH) is a rare disease marked by elevated pulmonary vascular resistance (PVR), resulting from pulmary vascular remodelling. This leads to increased right ventricular (RV) afterload, initiating extensive cardiac remodelling that ultimately results in right heart failure [1]. It is now well established that 80% of familial PAH cases and 11–40% of sporadic cases are associated with a pathogenic variant in BMPR2 [2]. Heritable PAH shows incomplete penetrance, and carriers have a lifetime risk of developing PAH ranging from 14% in males to 42% in females [3]. Current guidelines advise notifying PAH patients about the potential for a causative genetic condition and its presence in family members, allowing for early screening and diagnosis in carriers identified by cascade genetic screening [4, 5].

The 2015 European Society of Cardiology (ESC)/European Respiratory Society (ERS) guidelines recommended annual echocardiography for healthy carriers of a pathogenic variant in BMPR2 to aid in early identification and treatment of PAH [6]. However, the prospective DELPHI-2 study [7] showed through their longitudinal screening programme that unaffected carriers (UC) of a pathogenic variant in BMPR2 mutation, who were identified as having PAH at baseline, would not have been identified with echocardiography alone [7]. Updated directives emphasise the importance of a multimodal screening approach [4], but evidence-based screening programmes for UCs at risk for developing PAH are still lacking.

The presence of a pathogenic variant in BMPR2 is a significant risk factor but large variability in disease manifestation remains. It has not yet been established what sets apart BMPR2 carriers who develop PAH from those who do not. Experiments with induced pluripotent stem cells from UCs and familial PAH patients show that UCs have increased BMPR2 activators and decreased inhibitors allowing for preserved BMPR2 signalling, supporting the hypothesis that individual-specific protective mechanisms may be at play [8].

Our study outlines a comprehensive screening programme of UCs for a pathogenic variant in BMPR2 and of healthy control subjects aimed to aid in early diagnosis and offer a detailed phenotypic description of UCs. We additionally include a Bmpr2+/− transgenic rat model to explore whether potential phenotypic differences are reflected in animal models [9].

ResultsBaseline characteristics

In total, 28 UCs of a pathogenic variant in BMPR2 and 21 healthy non-carrier family members of participants with PAH from 14 families were included in the study. At baseline, 22 UCs underwent RHC. Six UCs refused RHC due to the invasive nature of the procedure. In the UC group, 57% were female with an average age of 44±16 years. In the healthy control group, 48% were females with an average age of 44±18 years. Two participants (7%) in the UC group and two (10%) in the control group had a medical history of controlled systemic hypertension. One participant among the control subjects had a history of a venous thromboembolic event. All UCs had a normal NT-proBNP level at baseline (19 ng·L−1, IQR 10–26 ng·L−1) (table 1). At baseline, echocardiography revealed a significantly lower but well within physiological range RV S’ wave in UCs when compared to controls (13.2±2.0 cm·s−1versus 12.1±1.7 cm·s−1; p=0.041). There were no significant differences between other echocardiographic measures, such as tricuspid annular plane systolic excursion (22.8±3.1 mm versus 23.5±2.9 mm; p=0.415), pulmonary arterial acceleration time (135.6±29.9 cm·s−1versus 132.1±17.7 cm·s−1) and tricuspid regurgitation velocity (TRV) (2.0±0.3 m·s−1versus 2.2±0.2 m·s−1). It is of note, however, that TRV measurements were not possible in every participant owing to a lack of signal. Echo-derived pulmonary artery (PA) flow notching was not observed at baseline. Subtle differences were observed using CPET. Both groups showed adequate effort during CPET with a mean maximum respiratory exchange ratio of 1.2±0.1 for both controls and UCs. Remarkably, UCs had a significantly lower end-tidal carbon dioxide partial pressure (4.7±0.3 kPa versus 4.3±0.4 kPa; p<0.001) and lower ventilatory efficiency, as shown by a higher ratio of minute ventilation to carbon dioxide production (V′E/V′CO2) at rest (30.0±3.4 versus 33.3±5.0; p=0.011) than controls (table 1).

TABLE 1

Baseline characteristics of DOLPHIN-GENESIS cohort

Smaller cardiac volumes and mass in UCs

To further elucidate potential differences in RV morphology in UCs, we performed cardiac MRI. Volumetric analyses revealed that UCs had significantly smaller indexed RV end-diastolic volume (EDVi) (79.5±17.6 mL·m−2versus 62.7±15.3 mL·m−2; p=0.001) and RV end-systolic volume (ESVi) (34.2±10.5 mL·m−2versus 27.1±8.3 mL·m−2; p=0.014) when compared to healthy control subjects (figure 2a, b). Interestingly, UCs also had a significantly lower indexed left ventricular (LV) EDVi (68.9±14.1 mL·m−2versus 58.5±10.7 mL·m−2; p=0.007) but no differences were observed for LVESVi (22.4±8.2 mL·m−2versus 19.1±7.3 mL·m−2; p=0.155). In addition to smaller volumes, we observed a lower mass index in both the right ventricle (23.5±4.8 g·m−2versus 20.2±5.0 g·m−2; p=0.031) and left ventricle (64.8±10.8 g·m−2versus 57.5±9.5 g·m−2; p=0.020) in the UC versus control population. This difference in volume was seen not only in the ventricles, but also in the indexed right atrial (RA) maximum volume (38.1±10.6 mL·m−2versus 29.5±10.8 mL·m−2; p=0.011). Indexed RA minimum volume and both minimum and maximum indexed left atrial (LA) volumes were similar in both groups. All volumetric and mass analyses are summarised in table 1. Linear regression analysis showed that carrier status was predictive of smaller RVEDVi, RVESVi and RV mass despite age and gender (supplementary table S2).

FIGURE 2

Cardiac magnetic resonance imaging indices for unaffected carriers of a pathogenic BMPR2 variant compared to control participants. a) Indexed right ventricular (RV) end-diastolic volume (EDVi). b) Indexed RV end-systolic volume (ESVi). c) RV global circumferential strain (GCS). Red dots indicate phenoconverters. *: p<0.05; **: p<0.01; ***: p<0.001.

To investigate the effect of smaller volumes and mass on functional parameters, we performed strain analyses. UCs had a higher peak RV global circumferential strain (GCS) (−12.8±2.4% versus −15.9±2.6%; p<0.001) than control subjects (figure 2c). No significant differences were seen in peak LV regional circumferential strain or GCS (−18.8±2.4% versus −19.6±2.1%; p=0.206). There were no significant differences observed in longitudinal strain. All strain analyses can be found in table 1. Again, carrier status was predictive of higher RV GCS on linear regression despite age and gender (supplementary table S2).

To validate our observations in human subjects using an experimental model, we conducted cardiac MRI on Bmpr2Δ71Ex1/+ transgenic rats. None of the rats showed signs of PAH at the time of analysis. Consistent with the findings from clinical cardiac MRI analysis, these transgenic rats exhibited a significant reduction in RVEDVi (p<0.015), RV stroke volume (p<0.05), RV mass (p<0.05), LVEDVi (p<0.05) and LV stroke volume (p<0.05) after correction for body surface area (figure 3; supplementary tables S3 and S4, supplementary figures S2 and S3).

FIGURE 3

Cardiac volumetric and functional values for wild-type (WT) and Bmpr2 (Bmpr2Δ71Ex1/+) rats. a) Mid-ventricular cardiac magnetic resonance imaging images from WT (n=5) and Bmpr2Δ71Ex1/+ transgenic (n=6) rats at end-diastole (ED) and end-systole (ES). b) Right ventricular (RV) and left ventricular (LV) indexed end-diastolic volume (EDVi), indexed stroke volume (SVi), ejection fraction (EF), ventricular–arterial coupling index (SV/ESV) and mass index (Mi) values, all corrected for body surface area. *: p<0.05.

Increased mean RA pressure and RV afterload resulting in altered RV–PA coupling in UCs in comparison to healthy controls

In total, 22 UCs underwent RHC upon inclusion. Haemodynamic characteristics, presented in table 1, were compared to a historical control group of healthy subjects. These subjects had been referred for RHC between 2013 and 2021 but did not receive a final diagnosis of PAH or have a history of cardiovascular disease. We compared our historical control group to the haemodynamic survey of healthy volunteers used by Kovacs et al. [11] and concluded that our control group adequately reflected a healthy population.

While none of the BMPR2 carriers was diagnosed with PAH at baseline, UCs had a significantly higher mPAP (14.1±4.3 mmHg versus 16.5±2.3 mmHg; p=0.036) (figure 4a), mean RA pressure (3.9±2.4 mmHg versus 5.7±2.1 mmHg; p=0.024) and a lower cardiac index (4.4±0.8 L·min−1·m−2versus 3.1±0.6 L·min−1·m−2; p<0.001) than healthy controls. To further assess potential load-independent changes in RV function, single beat pressure–volume analyses were performed in a subgroup of participants. RV afterload was increased in UCs in comparison to healthy controls (arterial elastance (Ea): 0.15±0.06 mmHg·mL−1versus 0.27±0.08 mmHg·mL−1; p<0.001) (figure 4b). Additionally, an increased RV end-systolic elastance (Ees) (0.28±0.07 mmHg·mL−1versus 0.35±0.10 mmHg·mL−1; p=0.047), a measure of RV contractility, and reduced RV–PA coupling were observed in UCs (2.24±1.03 versus 1.36±0.37; p=0.006) (figure 4c, d). Haemodynamic data and pressure–volume loop analyses can be found in table 2.

FIGURE 4

Comparison of haemodynamic parameters and pressure–volume relationships between unaffected carriers of a pathogenic BMPR2 variant and healthy control patients. a) Mean pulmonary artery pressure (mPAP). b) Arterial elastance (Ea). c) End-systolic elastance (Ees). d) Ees/Ea ratio. Red dots indicate phenoconverters. *: p<0.05; **: p<0.01; ***: p<0.001.

TABLE 2

Haemodynamic characteristics and pressure–volume loop analysis of BMPR2 carriers versus control group

Longitudinal follow-up

During the 4-year follow-up period, two UCs were diagnosed with PAH. At the date of this manuscript, a total of 15 UCs underwent RHC after 4 years. A comparison between the diagnostic modalities comparing the conversions with the overall UC population at baseline can be found in supplementary table S5. Excluding the two participants who developed PAH during the 4 years, volume and strain measurements were stable in all participants during the follow-up period (supplementary figure S4). MRI-derived PA flow notching, a nonspecific sign often observed in PAH, was only noted in two BMPR2 carriers throughout the study, both of whom developed PAH several years after it first appeared [12]. To further explore this signal, we compared acceleration time, ejection time and acceleration time/ejection time ratio based on PA flow curves between control subjects and UCs but did not find any significant differences between the groups [13].

Conversion 1

The first phenoconverter, a 62-year-old man, had a medical history of systemic hypertension and a family history of heart failure of unknown aetiology on the paternal side (suggestive of PAH) and a sibling with confirmed PAH. Both the participant and sibling with confirmed PAH were carriers of the pathogenic variant c.1454A>G (p.Asp485Gly) in BMPR2. There were no abnormalities on RHC, CPET or echocardiography at the time of inclusion. At baseline, cardiac MRI was refused due to claustrophobia. From the second year of follow-up onwards, he agreed to undergo cardiac MRI, which showed no signs of RA or RV dilation but did show PA notching on flow analysis.

RHC was performed at the fourth follow-up visit, at which an mPAP of 42 mmHg, PVR of 5.2 WU and a pulmonary capillary wedge pressure (PCWP) of 8 mmHg were measured. There were no signs of PAH on echocardiography and the plasma NT-proBNP was 24 ng·L−1. Flow analysis at the 4-year follow-up revealed distinct PA notching but no right heart dilation or hypertrophy. CPET showed an elevated V′E/V′CO2 at the anaerobic threshold. Initially, the participant denied any new-onset symptoms and was categorised as NYHA functional class I, but recalled mild exercise intolerance on a recent hiking trip after receiving the diagnosis. The participant was classified as low risk, according to the three-strata model, and started on macitentan and tadalafil. Although there was slightly worsened RV dilation at the 6-month follow-up, the patient remained in the low-risk category according to the four strata risk assessment model [6].

Conversion 2

The second conversion was a 56-year-old male UC carrying the pathogenic BMPR2 variant c.1978G>T (p.Glu660X). He had a previous diagnosis of epilepsy for which he used carbamazepine. At baseline, there were no abnormalities on CPET or echocardiography. During the follow-up period, a slightly decreased RV ejection fraction of 45% and PA flow notching were observed on cardiac MRI. At the 4-year follow-up visit, RHC was repeated and showed an mPAP of 31 mmHg, PCWP of 11 mmHg and PVR of 3 WU. TTE showed no signs of PAH and NT-proBNP was 4 ng·L−1. CPET showed a slightly decreased end-tidal carbon dioxide tension of 3.4 kPa at peak exercise. MRI-derived PA flow measures revealed persistent PA flow notching. The participant did not report any new-onset symptoms or exercise intolerance, and was thereby classified as NYHA functional class I. A risk assessment was performed, and the participant was categorised as low risk. The participant started on macitentan and tadalafil with an excellent response in exercise capacity. At 6-month follow-up, NT-proBNP remained low with a 6-min walk distance of 639 m and NYHA II, thereby categorised as low risk.

Discussion

Our study used an extensive, ongoing screening programme of unaffected BMPR2 carriers to reveal the following important findings:

UCs of a pathogenic BMPR2 variant had smaller cardiac volumes and mass alongside increased RV GCS. These findings were confirmed in a transgenic rat model. With regards haemodynamic parameters, PA and RA pressure in UCs were increased but still within physiological range.

Pressure–volume loop analysis revealed increased afterload and RV contractility, with decreased RV–PA coupling.

Echocardiography and NT-proBNP measurements were insufficient to effectively screen individuals susceptible to developing PAH. MRI-derived PA flow notching may be an early, nonspecific predictor of PAH.

Clinical relevance

Characterising and screening of UCs of pathogenic BMPR2 variants is of great clinical importance to identify potential risk factors for disease penetrance and find an effective screening method to aid in early diagnosis. Our approach included the incorporation of a healthy control group, revealing nuanced differences allowing for deep phenotyping of characteristics that have not been previously examined.

UCs of pathogenic BMPR2 variants have smaller cardiac volumes; PA and RA pressure is increased within physiological range

Reduced cardiac volumes and mass in UCs, along with smaller biventricular volumes and mass in transgenic Bmpr2+/− rats, further support existing research implicating BMPR2 in cardiac development. Moreover, it has been observed that Bmpr2Δ71Ex1/+ rats have reduced cardiac output, which has been attributed to the distinct changes in cardiomyocyte morphology, calcium (Ca2+) dynamics and cell contractility within the right ventricle. Specifically, adult RV cardiomyocytes from those rats exhibited diminished diameter, reduced sarcomere sensitivity to Ca2+, lower Ca2+ transient amplitude, decreased sarcoplasmic reticulum Ca2+ content and shorter action potential duration compared to those from WT rats [9]. How this precisely translates to humans is still not understood. It is well known that PAH patients carrying a pathogenic variant in BMPR2 have worse survival and show more impaired RV function than idiopathic PAH patients with a similar afterload [2, 14]. Pathogenic variants in BMPR2 have been shown to contribute to RV lipotoxicity in transgenic murine models, leading to impaired RV hypertrophy and triglyceride and ceramide deposition, which has also been observed in human right ventricle and cultured cardiomyocytes [15–17]. It is conceivable that pathogenic variants in BMPR2 in humans result in smaller cardiac volumes and mass during embryological development. Whether this altered phenotype makes the heart more susceptible to impaired adaptation in the setting of PAH has yet to be explored. Interpretation of these results in our human cohort remains challenging, because any volumetric and mass differences between healthy controls and UCs were still within the normal range and were not associated with functional limitations.

Increased afterload, RV contractility and decreased RV–PA coupling in carriers of pathogenic BMPR2 variants

We also observed an increased GCS and regional RV circumferential strain. Pressure–volume loop analysis revealed a significantly higher Ees and Ea among UCs, possibly indicating altered RV–PA coupling in UCs. Additionally, compromised arterial–ventricular coupling was observed in the transgenic Bmpr2+/− rats, again raising questions regarding the impact of pathogenic variants in BMPR2 on cardiac function. RV function is the primary prognostic factor in patients with PAH. While elevated afterload significantly influences RV function, notable variability in the adaptation of the right ventricle in PH persists [18–21]. A normal RV structure includes a superficial circumferential layer forming approximately a quarter of the wall thickness, followed by a longitudinally oriented subendocardial layer making up the bulk of the myocardium [22]. Animal models have suggested that increased pressure overload in the right ventricle can lead to RV hypertrophy with a large proportion of circumferentially oriented fibres [23]. Haemodynamic measurements revealed a slightly increased mPAP and mean RA pressure among UCs, which could possibly contribute to a subtle increase in circumferential strain through adaptive changes in the myocardium reflective of RV anatomy. Both an altered haemodynamic state and contractility results in altered RV–PA coupling, consistent with findings in transgenic Bmpr2 animals [9].

Increased RV afterload, characterised by higher mPAP and Ea, among pathogenic BMPR2 variant carriers could suggest subclinical changes in the pulmonary vascular bed. BMPR2 is highly expressed in the endothelium and smooth muscle cells of the pulmonary vasculature and has been shown to be one of the pathomechanistic drivers of PAH development via a pro-proliferative process led by endothelial and smooth muscle cell dysfunction [24, 25]. Previous animal studies have shown that Bmpr2 mutant rats without spontaneous PH exhibited higher muscularisation of both smaller and larger vessels as well as accelerated PA smooth muscle cell proliferation rates when compared to WT rats [9]. It is plausible that some degree of endothelial and vascular smooth muscle cell proliferation is present in human UCs of a pathogenic BMPR2 variant even in the absence of disease, leading to mild, but nevertheless distinct, haemodynamic aberrations. The combination of a more vulnerable pulmonary vascular bed and myocardium can potentially contribute to the lower survival and increased disease severity observed in patients with PAH due to a pathogenic variant in BMPR2 [2, 14]. Extended follow-up of UCs is essential to elucidate the clinical relevance of these altered cardiac phenotypes and borderline PA pressures towards the potential development of PAH.

Traditional screening methods are not sufficient for the effective screening of individuals susceptible to developing PAH

Our results further emphasise the difficulty of establishing an effective screening protocol for individuals at risk for PAH, as previously discussed by the investigators of the DELPHI-2 study [7]. NT-proBNP and conventional echocardiographic parameters seem insufficient for the effective screening of individuals susceptible to PAH, thereby confirming the poor discriminative ability of TTE for the early detection of PAH as established by Montani et al. [7]. While we did observe a significantly lower RV S’ among UCs, the values fell within the normal physiological range. It is also of note that NT-proBNP remained within the normal range for all participants, including those diagnosed with PAH during the course of the study. Analysis of PA flow curves revealed notching in both conversions up to 2 years prior to the diagnosis of PAH, suggesting that MRI-derived PA flow notching may be an early, nonspecific predictor of PAH. Expansion of PA flow analyses to integrate four-dimensional cardiac MRI PA flow may allow for the identification of vortex formation, an early sign of PAH which could potentially identify more subtle changes in flow patterns [12]. Exploration of the diagnostic potential of cardiac MRI-derived PA flow analysis warrants further evaluation in larger cohorts.

Additionally, participants who developed PAH displayed changes on CPET which were nonspecific but consistent with a pattern correlating to a circulatory limitation. While these changes on CPET and MRI may foreshadow a diagnosis of PAH, they often do not warrant sufficient suspicion for initiating more invasive testing. Grünig et al. [26] have shown in relatives of PAH patients that a pulmonary hypertensive response to exercise or hypoxia was correlated with but not caused by a BMPR2 mutation, which was also seen in a more recent study by Claessen et al. [27]. Although the presence of this trait was also present in some healthy controls, manifest PAH developed only in patients with both a pathogenic BMPR2 variant in addition to a hypertensive response [28]. An important finding in the DELPHI-2 study was the development of prevalent PAH in UCs with abnormal exercise haemodynamic profiles, indicating it might precede future development of prevalent disease [7]. Unfortunately, in our two conversions, exercising was not yet performed during RHC. The DELPHI-2 also used a noninvasive tool (ECG, low diffusing capacity of the lung for carbon monoxide and CPET probability) to address the likelihood of PH and to justify invasive testing, but exercise PH was still present in three participants with no suspicion of PH. Although this score might not be conclusive, further research should establish a clear set of characteristics of early-stage PAH or exercise PH to perform invasive diagnostics. For now, RHC remains the only diagnostic modality able to detect PAH at an early stage. Conversions could not be identified before significant remodelling of the vascular bed had occurred, further emphasising the importance of continued efforts to identify biomarkers of vascular remodelling. The prognostic implications of early diagnosis accentuate the importance of vigilance and the timely deployment of RHC to establish a definitive diagnosis.

Limitations

The following limitations in our study should be considered. The multimodal characterisation of UCs of pathogenic BMPR2 variants has allowed us to identify cardiac traits not present in healthy controls. However, despite our screening, the low incidence of conversions makes interpretation of these results challenging. The same applies to the 2 mmHg difference in mPAP, because this could be the consequence of limited sample size. However, a comparable small increase in resting state was seen for DELPHI-2 participants with exercise PH, individuals at increased future risk of developing prevalent PAH. Longitudinal assessment in this at-risk population is warranted to address relevancy, potentially in multicentre initiatives to increase power. Obtaining sufficient power through single-centre screening remains a challenge, especially because UCs are excluded in the presence of cardiovascular comorbidities. Second, a different control group was used for our pressure–volume loop analysis. Individuals suspected of PH were used as a control group for the assessment of RV contractility because it is not ethical to perform RHC in healthy volunteers. Also, cardiac phenotyping in this study was performed with MRI, a modality not widely available in other countries. However, PA flow notching might be an early sign of PH and could also be assessed using echocardiography. While we did employ echocardiography, we did not perform exercise echocardiography, which may have allowed for more subtle analysis of cardiac markers for PAH. Lastly, we chose to focus on BMPR2 because this is the most common site of a pathogenic variant causing PAH, resulting in a representative cohort. It remains unanswered whether our findings can be translated to UCs of different pathogenic variants because each affected gene harbours different phenotypic appearances. Overall, our findings suggest that UCs of a pathogenic BMPR2 variant have an altered cardiac phenotype. Future efforts in establishing an effective screening protocol for at-risk individuals warrant longer follow-up periods.