Author + information
- Matthias Rindermann*,
- Ekkehard Grünig, MD†,
- Albrecht von Hippel*,
- Rolf Koehler, PhD*,
- Gabriel Miltenberger-Miltenyi, MD, PhD*,
- Derliz Mereles, MD†,
- Karlin Arnold†,
- Michael Pauciulo§,
- William Nichols, PhD§,
- Horst Olschewski, MD‡,
- Marius M Hoeper, MD∥,
- Jörg Winkler, MD¶,
- Hugo A Katus, MD†,
- Wolfgang Kübler, MD, FRCP†,
- Claus R Bartram, MD* and
- Bart Janssen, PhD*,* ()
- ↵*Reprint requests and correspondence:
Dr. Bart Janssen, Institute of Human Genetics, Im Neuenheimer Feld 328, 69120 Heidelberg, Germany.
Objectives The aim of our study was to identify genetic causes of primary pulmonary hypertension (PPH), to estimate the proportion of families with mutations in the BMPR2 (bone morphogenetic protein receptor type 2) gene, and to examine whether genetic heterogeneity might play a role.
Background The BMPR2 mutations have been identified in a substantial portion of patients with familial or sporadic PPH. However, the genetic cause of PPH remains unclear in at least 45% of families.
Methods We investigated 130 members of 10 families with at least 1 PPH patient, recruited without selection for familial disease. Manifest PPH was documented in 21 individuals. An increase in pulmonary artery systolic pressure (PASP) above 40 mm Hg during supine bicycle exercise was found in 46 healthy individuals. Their PASP increased from 21.0 ± 4.6 mm Hg at rest to 54.0 ± 9.8 mm Hg during exercise. In 51 relatives, PASP values were normal at rest and during exercise, and 12 members were classified as status unknown.
Results Two families showed a mutation in the BMPR2 gene. Three families with no BMBR2 mutation showed evidence for linkage to a more proximal location on chromosome 2q31 (odds ratio [OR] for linkage 1.1·106:1). This locus, designated PPH2, maps in-between the markers D2S335 and D2S2314. We obtained significant support for heterogeneity in PPH with an OR of 2.8·1011.
Conclusions We conclude that PPH may be a genetically heterogeneous disorder with at least two—and possibly more—causative genes.
Primary pulmonary hypertension (PPH) is a rare disease characterized by increased resistance of precapillary pulmonary arteries and a poor prognosis. Familial PPH was initially mapped to a single locus on chromosome 2q31-32 (1–3). Recently, autosomal dominant germline mutations in the bone morphogenetic protein receptor type 2 (BMPR2) gene localized on chromosome 2q33 (PPH1-gene) have been identified in approximately 55% of familial cases (4–6)and in 25% of patients with negative family history (6,7). However, the genetic cause remains unclear in almost half the patients.
The BMPR2 gene is a ubiquitously expressed receptor for secreted growth factors (bone morphogenetic proteins) and a member of the transforming growth factor-beta (TGF-beta) receptor family. Most of the identified BMPR2 sequence variants are truncating mutations, whereas the majority of missense mutations have been shown to be located in highly conserved domains, like the receptor kinase domain (4–6). The BMPR2 expression level has been shown to be reduced in the peripheral lung of patients with and without a BMPR2 mutation (8). Rudarakanchana et al. (9)reported that all recombinant constructs carrying a BMPR2 mutation caused increased cell proliferation, a process associated with the activation of the p38 protein. The p38 protein is a Smad-independent member of the TGF-beta–related signaling network. Hence, it is most likely that mutations in BMPR2 can cause PPH.
Recently, we demonstrated that a predisposition to PPH can be diagnosed at an early stage of disease using stress Doppler echocardiography (SDE) (10). A pathological rise of pulmonary artery systolic pressure (PASP) during supine bicycle exercise was observed in family members who shared the risk haplotype with the index patients. The microsatellite markers used in this previous study mapped to chromosome 2q31-32. The BMPR2 gene was isolated from the more distal band 2q33. Thus, an apparent discrepancy exists between the mapping data obtained in these studies. The aim of this investigation was to determine whether genetic heterogeneity might be an explanation for these findings.
Subjects and methods
The family panel
Between January 1998 and January 2000, we prospectively investigated the families of 39 index patients with PPH. All families (n = 10) with more than two potentially informative meioses were selected for linkage analysis. Two of these families, family 965 and family 1490, have been described before as “family A and B,” respectively (10). In all index patients the diagnosis of PPH was confirmed by right heart catheterization or autopsy and was made according to the criteria of the World Health Organization (11). For decreased relatives, medical records were reviewed when available, and the diagnosis of PPH was based on the criteria used for index patients. Pulmonary artery systolic pressure (PASP) was assessed at rest and during supine bicycle exercise (ER900EL, ergoline, Bitz, Germany) using SDE as described previously (10,12). In subjects with abnormal PASP values, secondary causes have been excluded on the basis of a cascade of clinical assessments including lung function tests with body plethysmography, carbon dioxide diffusing capacity, arterial blood gas analysis, chest X-ray, and in some cases heart catheterization. Left ventricular diastolic dysfunction was excluded by Doppler echocardiography at rest. All family members with possibly secondary reasons for an abnormal PASP such as blood pressures exceeding 220/110 mm Hg during exercise were classified as “status unknown” in the linkage analysis.
Right heart catheterization (RHC) was performed in all index patients at rest and in 13 relatives who complained of exercise-induced shortness of breath at rest and during exercise.
The control group consisted of 60 subjects (36 female, 24 male, mean age 27 ± 10 years, weight 66 ± 12 kg, height 175 ± 9 cm). Acute or chronic pulmonary or cardiac diseases were ruled out in all subjects. The clinicians were blinded to the results of genetic analysis. The Ethics Committee of the Medical Faculty of the University of Heidelberg approved the protocol of this study, and family members and controls gave their written informed consent.
Blood samples were collected for genetic analysis, and DNA was extracted from peripheral blood using a standard salting-out procedure. The BMPR2 mutation screening was performed in all index patients. The primer sequences used to amplify the gene fragments by polymerase chain reaction (PCR) can be obtained from the corresponding author. We performed cycle sequencing with Big Dye terminators on an ABI 3700 sequencer according to the manufacturer’s recommendations (M.P., W.N.). To confirm that no mutations had been missed owing to low-grade mosaicism or sequencing artefacts, we rescreened these 10 patients by denaturing high performance liquid chematography (dHPLC) (Transgenomics “WAVE” Nucleic Acid Fragment Analysis System) in a second independent laboratory (R.K.). Each PCR product was dHPLC-analyzed by using at least two different protocols per exon according to the manufacturer’s directions.
Microsatellite and linkage analysis
All families underwent genotyping for microsatellite markers covering the entire PPH candidate region (1): D2S222, D2S284, D2S156, D2S335, CHRNA1, D2S2314, D2S389, D2S309, and D2S307. Families 965, 1893, and 2135 were also typed for the additional markers D2S2307, D2S2188, D2S148, D2S364, D2S350, D2S2336, D2S324, D2S360, D2S2193, D2S1384, D2S369, and D2S143. The primer sequences were obtained from GDB (The Genomic Database), and PCR was performed according to GDB recommendations. We used CY5-labeled forward primers and analyzed the genotypes on an ALF-Express automatic sequencer.
The nine three-generation families were statistically analyzed by the GENEHUNTER program (13). As family 965 was too large for analysis by GENEHUNTER, we used the LINKAGE programs version 5.1 to calculate lod scores (logarithm of the odds) for this extended four-generation family (14). The standard markers were recoded to a three-, four-, or five-allele system and analyzed by overlapping runs of LINKMAP (5-point analyses) (family 965) or analyzed without recoding by GENEHUNTER (all other families). Carriers of PPH risk haplotypes were expected to show an abnormal response to exercise, with the exception of a small number of young individuals (age <20) (10). In accord with this, we calculated with 99% penetrance in adult individuals and 80% penetrance in persons <20 years of age. We used a rather conservative phenocopy rate of 20%. For individuals with manifest PPH we calculated with a phenocopy rate of 0.0001% (misclassified individuals with a secondary form). We used a disease allele frequency of 1 in 10,000. Intermarker distances were obtained from NCBI (National Center for Biotechnology Information). The lod scores were calculated at intervals of 0.5 cM and converted into heterogeneity-corrected lod scores by applying the Admixture test (15). At each map position the heterogeneity-corrected lod score was maximized by optimizing the proportion of linked families (α) in steps of 1%. Posterior probabilities (ω) were calculated by the formula: ω = (α·10Z1)/([α·10Z1] + [1 − α]·[10Z2]), where Z1and Z2represent the lod scores at locus 1 (BMPR2) and locus 2 (a locus with unknown position), respectively. Posterior probabilities for linkage were calculated using an α of 50%.
Statistics of clinical data
Data in figures and tables are given as mean values ± SD. Linear regression analysis was used to correlate invasive and noninvasive PASP measurements. Groups were compared by the Mann-Whitney-Wilcoxon analysis of variance test and (ANOVA). All p values <0.05 were considered statistically significant.
Clinical assessment of the families of PPH patients
The pedigrees of 10 PPH patients revealed more than two potentially informative meioses. Data on these 10 index patients and their families are presented here as the results of a prospective study. A total of 130 blood-related members (68 male, mean age 31.1 ± 18.4 years; range 6 to 80 years) were assessed clinically and genetically. Twenty-nine spouses were assessed only genetically. Manifest PPH was diagnosed invasively or by autopsy in 11 members. This translates to a total of 21 manifest PPH patients in the 10 families. Three index patients came from families 1490, 4452, and 0965 with confirmed familial disease. In these families manifest PPH was diagnosed in the respective index patient and in 11 additional members. The other seven index patients had a negative family history.
In 109 family members without manifest PPH, SDE was performed, and in 13 members who complained of symptoms RHC was also obtained at rest and during exercise. An abnormal PASP response to exercise (AR) (from 21.0 ± 4.6 mm Hg to 54.0 ± 9.8 mm Hg) without secondary reasons (AR members) was found in 46 of the 130 relatives (34.8%) with a mean age of 33 ± 19 years. In 51 of the 130 members (36 adults, 15 children ≤20 years) the PASP values were normal at rest (21.0 ± 5.0 mm Hg) and during exercise (37.3 ± 2.3 mm Hg) (NR). In the 13 individuals with simultaneous RHC (AR members, n = 8; NR members, n = 3, secondary pulmonary hypertension n = 2), the PASP values correlated highly to the values obtained by SDE (r = 0.9, standard estimation of error 7.5 mm Hg, p < 0.0001). In seven of the eight AR members diagnosed also by RHC, mean pulmonary artery pressure exceeded 30 mm Hg (mean maximal PAP = 36 ± 5 mm Hg). Twelve of the 109 relatives were classified as status unknown because secondary reasons for pulmonary hypertension could not be excluded (n = 7) or PASP could not be assessed due to inadequate Doppler signals (n = 5).
Clinical assessment of control subjects
Mean PASP of the 60 control subjects at rest was 20 ± 5 mm Hg and during exercise 37 ± 6 mm Hg, median 37 mm Hg. In 6 of the 60 controls (10%), PASP exceeded 40 mm Hg during exercise and reached a maximum of 53 ± 4 mm Hg. This exaggerated response to exercise occurred at low workloads (75 to 100 W) similar to that observed in the AR family members. The mean maximal PASP value during exercise in the remaining 54 controls was 36 ± 4 mm Hg. The number of AR members (members with PASP >40 mm Hg) was significantly higher in the 10 families with PPH (46/97 members = 47.4%) than in the control group (6/60 = 10%, p < 0.0001). The group of AR members (n = 46), NR members (n = 51), and controls (n = 60) did not significantly differ with respect to gender, age, body weight, and height. Both AR and NR groups had a similar mean maximal workload (137 ± 47 W vs. 129 ± 38 W; p = 0.34), mean cardiac output, heart rate (HR), and systemic blood pressures. The control group had a higher mean maximal workload (183 ± 50 mm Hg) and higher heart rates (mean maximal HR 167 ± 16) than did the other groups (p < 0.01).
The BMPR2 mutations and linkage to chromosome 2q33
In 143 of the 159 relatives, genetic analysis was performed. In 2 of the 10 families a BMPR2 mutation was identified. Family 4452 showed a Q82X nonsense mutation in exon 2. This will lead to a truncated protein, missing part of the ligand-binding domain and all further domains. In family 3771 an IVS8+1G>T splice mutation at the splice donor site of intron 8 was found in the index patient with manifest PPH and in his son with abnormal PASP response to exercise, but not in the relatives with normal PASP response to exercise. This mutation will at least disturb the function of the kinase domain and presumably also cause a frame-shift. By linkage analysis we identified only one additional family (1490) with a significant probability for being linked to the BMPR2 locus. In the other families, linkage to the BMPR2 region was excluded, or insignificant ⇓scores were obtained (Fig. 1, Table 1).
The PPH2 locus, a second locus on chromosome 2q31
In families 965, 2135, and 1893, the posterior probability of being linked to a more proximal locus on 2q31 (designated PPH2) was highly significant (p > 98%). None of these families showed a mutation in the BMPR2 gene. In family 965, manifest PPH was diagnosed in six family members. The haplotypes of the three PPH2-linked families are shown in Figure 2. The positions of crucial recombinations were checked by testing additional markers (see Methods). The highest lod score was found in family 965, the largest pedigree in this study. We obtained a maximum lod score of 5.22 for this family at a position in-between D2S335 and CHRNA1. The cumulative lod score for all families showed a maximum of 5.61 (odds ratio [OR]: 4.1·106) at a position 1 cM proximal to CHRNA1. A comparison of the haplotypes of the PPH2-linked families showed that the affected individuals, including the abnormal responders, share haplotype fragments flanked by the markers D2S335 and D2S2314. The relevant recombinations occurred in two individuals with manifest PPH and in six family members with a pathological PASP response during exercise (Table 2).
The cumulative lod score, which represents the evidence for linkage assuming homogeneity, was compared to the heterogeneity-adjusted lod score (see Fig. 1). The hypothesis that PPH might be a genetically heterogeneous disorder was unequivocally supported by our linkage data with an OR of 2.8·1011in favor of heterogeneity. This maximum was obtained with 20% of the families linked to BMPR2 (α = 0.2).
Comparison of clinical and genetic data
No cases of apparent nonpenetrance were identified in the three PPH2-linked families (965, 2135, and 1893), in the single family with significant linkage to BMPR2 (family 1490), or the families with a BMPR2 mutation (3771 and 4452). Because of the uncertain assignment to one of the loci, an evaluation of penetrance was not possible in the remaining families. In comparison to our previous study (10), we observed that two cases of nonpenetrance had progressed to an abnormal PASP response to exercise. Member IV-22 in family 965 progressed from a maximum PASP during exercise of 38 mm Hg at age 18, to 55 mm Hg at age 20 years. Subject IV-14 of family 965 progressed from a PASP of 37 mm Hg to 50 mm Hg two years later. Individual III-1 in family 1490 had progressed from an AR status to manifest PPH within three years.
In our linkage analysis, every abnormal responder (AR) was treated as a possible phenocopy in a fully unbiased manner, yet the results indicate that certain individuals are more likely to be phenocopies than are others. In family 4452, we observed four such phenocopies. These ARs do not share the BMPR2 mutation, nor the risk haplotype with the index patient. In family 965, two additional phenocopies were identified. Thus, 6 (18%) of the 34 ARs in these six families were likely to be phenocopies.
Locus heterogeneity in PPH
Our results strongly suggest genetic heterogeneity in PPH with one locus, BMPR2, mapping to 2q33 and a second locus, PPH2, located on 2q31. According to the haplotypes, the PPH2 locus maps in-between the markers D2S335 and D2S2314. The lod scores supporting this location are highly significant.
In a heterogeneity model, the AR status is likely to be a subclinical phenotype, not only present in carriers of BMPR2 mutations, but also in PPH2 mutation carriers. Although we cannot formally exclude PPH2 to be a modifier gene in an alternative (“two-locus”) model, our haplotype data do not support this alternative. The data contribute to a better understanding of the complexity of the pathogenesis: in some families a BMPR2 mutation is the primary cause of the disease, whereas in other families a PPH2 gene defect causes PPH. In other words, a single mutation causes a person to be at risk for PPH, independent of the genetic status of the other gene.
In addition, there will be environmental and genetic factors that trigger the onset of the disease as, for instance, suggested for appetite-suppressant drugs. An example of a genetic-modifying factor might be the 5-HTT transporter (16). Moreover, pulmonary hypertension can also occur as a component of other diseases, like hemorrhagic hereditary telangiectasia. Mutations in the gene encoding the activin receptor line kinase ALK1 have been found in patients with hemorrhagic hereditary telangiectasia and in relatives suffering from pulmonary hypertension only (17). The ALK1 is a member of the family of TGF-beta–related receptors, again pointing to the significance of the TGF-beta–related pathways in the development of PH.
The PPH2 locus and its role in PPH
Because the number of families reported here is rather small, it is not possible to give a reliable estimate of the proportion of families linked to the PPH2 and BMPR2 loci. We cannot disprove the suggestion that PPH1 (BMPR2) might be the major PPH locus. Alternatively—in the absence of reliable data obtained from a large set of families—equal importance could be attributed to both loci. In any case, we expect some of the familial and sporadic cases that were reported BMPR2 mutation-negative to exhibit mutations in the yet unknown PPH2 gene (5,7). Because both loci lie only 15 to 19 cM apart, many small families will show no recombination between the two regions. Therefore, it is not possible to classify small families by means of haplotype analysis. This might explain why Deng and colleagues (5)failed to find BMPR2 mutations in several families that showed marker data consistent—but not conclusive—with linkage to 2q33. In our own data set, we also found no recombination in a patient with manifest PPH. This phenomenon, which is a clear limitation of the study, can be explained by the small number of families with multiple cases of manifest PPH and the small number of affected individuals in these families.
Interpretation of high PASP values and the diagnostic value of SDE
A PASP of 40 mm Hg during supine bicycle exercise is generally accepted as the upper reference limit for normal individuals (18,19). This, together with the data presented here, suggests that the vast majority of individuals with a pathological PASP are carriers of BMPR2 or PPH2 mutations. The remaining ARs are to be classified as phenocopies. Stress Doppler echocardiography turned out to be very powerful with respect to the identification of carriers. We identified no cases of apparent nonpenetrance. Despite the high sensitivity of the method, the clinical relevance of a high PASP during exercise is still a matter of investigation. A yet unknown proportion of the carriers will never develop manifest PPH at all.
The region in-between the markers D2S335 and D2S2314 is quite large (5 to 6 Mb) and harbors several interesting genes. A promising candidate is the ATF2 gene, which maps in the center of the PPH2 region. The ATF2 transcription factor is a member of the TGF-beta pathway and therefore currently the most likely candidate for PPH2.
We thank Prof. Seeger, Dr. Kuecherer, Mr. Endris, Dr. Benz, and Dr. Borst for their assistance and helpful comments. We also thank the patients and members of the self-help group “PH eV” for their kind cooperation.
☆ The study was financially supported by a grant from the Medical Faculty Heidelberg. The first three authors contributed equally to this work.
- abnormal PASP response to exercise
- bone morphogenetic protein receptor type 2
- denaturing high performance liquid chromatography
- logarithm of the odds for linkage
- normal PASP response to exercise
- pulmonary artery systolic pressure
- polymerase chain reaction
- primary pulmonary hypertension
- stress Doppler echocardiography
- American College of Cardiology Foundation
- Morse J.H.,
- Jones A.C.,
- Barst R.J.,
- et al.
- Thomson J.R.,
- Machado R.D.,
- Pauciulo M.W.,
- et al.
- Atkinson C.,
- Stewart S.,
- Upton P.D.,
- et al.
- Rudarakanchana N.,
- Flanagan J.A.,
- Chen H.,
- et al.
- Grünig E.,
- Janssen B.,
- Mereles D.,
- et al.
- Grünig E.,
- Mereles D.,
- Hildebrandt W.,
- et al.
- Lathrop G.M.,
- Lalouel J.-M.,
- Julier C.,
- Ott J.