Article Text

Download PDFPDF

Original article
Phenotypic spectrum of the SMAD3-related aneurysms–osteoarthritis syndrome
  1. Ingrid M B H van de Laar1,
  2. Denise van der Linde2,
  3. Edwin H G Oei3,
  4. Pieter K Bos4,
  5. Johannes H Bessems4,
  6. Sita M Bierma-Zeinstra5,
  7. Belle L van Meer4,
  8. Gerard Pals6,
  9. Rogier A Oldenburg1,
  10. Jos A Bekkers7,
  11. Adriaan Moelker3,
  12. Bianca M de Graaf1,
  13. Gabor Matyas8,
  14. Ingrid M E Frohn-Mulder9,
  15. Janneke Timmermans10,
  16. Yvonne Hilhorst-Hofstee11,
  17. Jan M Cobben12,
  18. Hennie T Bruggenwirth1,
  19. Lut van Laer13,
  20. Bart Loeys13,
  21. Julie De Backer14,
  22. Paul J Coucke14,
  23. Harry C Dietz15,16,17,
  24. Patrick J Willems18,
  25. Ben A Oostra1,
  26. Anne De Paepe14,
  27. Jolien W Roos-Hesselink2,
  28. Aida M Bertoli-Avella1,
  29. Marja W Wessels1
  1. 1Department of Clinical Genetics, Erasmus Medical Center, Rotterdam, The Netherlands
  2. 2Department of Cardiology, Erasmus Medical Center, Rotterdam, The Netherlands
  3. 3Department of Radiology, Erasmus Medical Center, Rotterdam, The Netherlands
  4. 4Department of Orthopedic Surgery, Erasmus Medical Center, Rotterdam, Netherlands
  5. 5Department of General Practice, Erasmus Medical Center, Rotterdam, The Netherlands
  6. 6Department of Clinical Genetics, VU University Medical Center, Amsterdam, The Netherlands
  7. 7Department of Cardio-Thoracic Surgery, Erasmus Medical Center, Rotterdam, The Netherlands
  8. 8Institute of Medical Molecular Genetics, University of Zurich, Zurich, Switzerland
  9. 9Department of Pediatric Cardiology, Erasmus Medical Center, Rotterdam, The Netherlands
  10. 10Department of Cardiology, Radboud University Hospital Nijmegen, Nijmegen, The Netherlands
  11. 11Department of Clinical Genetics, Leiden University Medical Center, Leiden, The Netherlands
  12. 12Department of Pediatric Genetics, Academic Medical Center, Amsterdam, The Netherlands
  13. 13Center of Medical Genetics, University and University Hospital of Antwerp, Antwerp, Belgium
  14. 14Center for Medical Genetics, Ghent University Hospital, Ghent, Belgium
  15. 15McKusick–Nathans Institute of Genetic Medicine, Johns Hopkins University School of Medicine, Baltimore, Maryland, USA
  16. 16Howard Hughes Medical Institute, Baltimore, Maryland, USA
  17. 17Smilow Center for Marfan Syndrome Research, Baltimore, USA
  18. 18GENetic DIAgnostic Network, GENDIA, Antwerp, Belgium
  1. Correspondence to Ingrid M B H van de Laar, Department of Clinical Genetics, Erasmus MC, PO Box 2040, Rotterdam 3000 CA, The Netherlands; i.vandelaar{at}


Background Aneurysms–osteoarthritis syndrome (AOS) is a new autosomal dominant syndromic form of thoracic aortic aneurysms and dissections characterised by the presence of arterial aneurysms and tortuosity, mild craniofacial, skeletal and cutaneous anomalies, and early-onset osteoarthritis. AOS is caused by mutations in the SMAD3 gene.

Methods A cohort of 393 patients with aneurysms without mutation in FBN1, TGFBR1 and TGFBR2 was screened for mutations in SMAD3. The patients originated from The Netherlands, Belgium, Switzerland and USA. The clinical phenotype in a total of 45 patients from eight different AOS families with eight different SMAD3 mutations is described. In all patients with a SMAD3 mutation, clinical records were reviewed and extensive genetic, cardiovascular and orthopaedic examinations were performed.

Results Five novel SMAD3 mutations (one nonsense, two missense and two frame-shift mutations) were identified in five new AOS families. A follow-up description of the three families with a SMAD3 mutation previously described by the authors was included. In the majority of patients, early-onset joint abnormalities, including osteoarthritis and osteochondritis dissecans, were the initial symptom for which medical advice was sought. Cardiovascular abnormalities were present in almost 90% of patients, and involved mainly aortic aneurysms and dissections. Aneurysms and tortuosity were found in the aorta and other arteries throughout the body, including intracranial arteries. Of the patients who first presented with joint abnormalities, 20% died suddenly from aortic dissection. The presence of mild craniofacial abnormalities including hypertelorism and abnormal uvula may aid the recognition of this syndrome.

Conclusion The authors provide further insight into the phenotype of AOS with SMAD3 mutations, and present recommendations for a clinical work-up.

  • SMAD3
  • TGFβ
  • aortic aneurysm
  • arterial tortuosity
  • osteoarthritis
  • clinical genetics
  • cardiovascular medicine
  • connective tissue disease
  • congenital heart disease
  • diagnosis
  • diagnostics
  • diagnostics tests
  • epidemiology
  • osteoarthritis
  • molecular genetics
  • osteoporosis
  • genetics

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Aortic aneurysm is a common condition, with high mortality from dissections and ruptures.1 Whereas abdominal aortic aneurysms usually occur sporadically, thoracic aortic aneurysms and dissections (TAAD) can be inherited in an autosomal dominant manner with decreased penetrance and variable expression.2 Familial TAAD is subdivided into non-syndromic forms, sometimes associated with bicuspid aortic valve and/or persistent ductus arteriosus,3–5 and syndromic forms with features of a systemic connective tissue disorder. Non-syndromic familial TAAD can be caused by mutations in genes encoding proteins of the contractile unit of the vascular smooth muscle cell such as the ACTA2, MYH11 and MYLK genes.3–5 However, in the majority of patients, the genetic cause is still unknown.

Syndromic familial TAAD includes several systemic connective tissue disorders such as: Marfan syndrome (MFS), caused by mutations in the FBN1 gene; Loeys–Dietz syndrome (LDS), caused by mutations in the TGFBR1 or TGFBR2 gene; arterial tortuosity syndrome (ATS), caused by mutations in the SLC2A10 gene; and autosomal recessive cutis laxa type I (AR-CL), caused by mutations in the FBLN4 gene.6–11 As all these syndromes are characterised by increased transforming growth factor (TGF)-β signalling in the arterial wall, it has become evident that TGFβ signalling plays a central role in the pathogenesis of arterial aneurysms.6–11

Recently, we described a new syndromic form of autosomal dominant TAAD characterised by the presence of arterial aneurysms and tortuosity, mild craniofacial features, skeletal and cutaneous anomalies, and osteoarthritis at a young age.12 As arterial aneurysms and early-onset osteoarthritis are the cardinal features of this new disorder, the term aneurysms–osteoarthritis syndrome (AOS) was coined. Patients with AOS show aneurysms throughout the arterial tree and a high risk of early dissection/rupture, resembling patients with LDS. Interestingly, early-onset joint abnormalities, including osteoarthritis, intervertebral disc degeneration, osteochondritis dissecans (OCD) and meniscal anomalies, are present in almost all patients with AOS, whereas they are uncommon in LDS, MFS and ATS. This establishes early-onset joint abnormalities as a key feature of this new syndrome.

We previously showed that AOS in three different families is caused by heterozygous mutations in the SMAD3 gene encoding SMAD3, which is a key protein in the TGFβ pathway.12 Here we identified five novel SMAD3 mutations and present an extensive clinical description of 45 patients from eight families with SMAD3-related AOS.


Patient collection

DNA from 393 patients (95 Dutch, 158 Belgian, 133 Swiss and seven North American) with TAAD but without mutation in the coding region of the FBN1, TGFBR1 and TGFBR2 genes was analysed for mutations in the coding region of the SMAD3 gene. When a SMAD3 mutation was found, clinical data on the patient were collected, clinical investigations were performed, and a family tree was constructed or extended through family histories, whereby possibly affected relatives were studied and screened for the SMAD3 mutation found in the index.12

A total of 34 patients with a mutation in SMAD3 were interviewed and examined by a clinical geneticist, six of whom have subsequently died. All had extensive clinical investigations, with scoring of five major systems implicated in connective tissue disorders, including the cardiovascular, joint, skeletal, craniofacial and cutaneous systems. Medical records from 11 deceased patients were reviewed. This study was approved by the medical ethics committee of the Erasmus Medical Center Rotterdam (Erasmus MC), and all patients gave written informed consent for this study.

Cardiovascular studies

Extensive cardiovascular studies were performed in 29 patients with AOS with a SMAD3 mutation, and included physical examination, ECG, transthoracic echocardiography and imaging of the thorax and abdomen by CT angiography (CTA) or magnetic resonance angiography (MRA) as described previously.12 Aortic root dilatation was defined as a Z-score ≥2 at any level. Z-scores were calculated on the basis of body surface area-corrected normal values published by Roman et al.13 For the other arteries, aneurysm is defined as a 50% or greater increase in diameter compared with the expected normal diameter of the vessel. CTA of the cervical and intracranial arteries was performed in 17 AOS patients with a SMAD3 mutation. Tortuosity of the thoracic, abdominal and cerebral arteries was scored by a radiologist.

Joint studies

Twenty-five patients were evaluated by an orthopaedic surgeon. An extensive physical examination for signs of osteoarthritis, intervertebral disc degeneration, spondylolysis or spondylolisthesis, OCD, meniscal lesions and joint laxity was performed.

Nineteen patients filled out a questionnaire about joint complaints. A radiographic skeletal survey of the total spine, hips, knees, hands and feet was performed in 26 patients. Osteoarthritis in the extremities is characterised by the degradation of articular cartilage and subchondral bone of joints and was scored as described previously.12 In addition, the presence of spondylolysis or spondylolisthesis was scored. OCD, defined as separation of an articular cartilage subchondral bone segment from the remaining articular surface, was scored in all patients who were radiologically evaluated. MRI of the joint was performed when abnormalities were seen on radiography or if patients had symptoms. Every patient who had surgery for meniscal pathology, OCD or joint replacement because of osteoarthritis was considered to be affected for the respective feature.

Phenotypic studies

Physical examination was performed by a clinical geneticist. Hypertelorism was defined as an inner canthal distance ≥+2 SD without lateral displacement of the inner canthi.14 Dolichostenomelia was defined as an arm span/height ratio of ≥1.05. Arachnodactyly was scored when the middle finger length exceeded the palm length, as described by Hall.14 Scoliosis was radiographically defined as a lateral curvature of the spine greater than 20 degrees in the coronal plane accompanied by vertebral rotation in the axial plane measured on standing x-rays. Hypermobility was scored when the Beighton score was ≥5. Acetabular protrusion was scored on pelvic radiographs or CT scans when the acetabular line crossed the normal oval shape formed by the two iliopectineal lines.

Molecular studies

Genomic DNA was isolated from peripheral blood using standard procedures (Gentra Systems, Minneapolis-USA). DNA samples from deceased patients were obtained from stored autopsy tissue (frozen or paraffin-embedded tissue). Bidirectional sequencing of all coding exons and exon–intron boundaries of the SMAD3 gene was undertaken as previously described.12 For annotation of cDNA and protein changes, the Mutation Nomenclature Guidelines from the Human Genome Variation Society were followed (the A from the ATG start codon and Met of the reference sequence NM_005902.3 and NP_005893.1, respectively, were numbered 1).

If SMAD3 missense mutations were identified in patients with AOS, the possible presence in controls was investigated by direct sequencing in at least 342 ethnically matched control chromosomes. The putative pathogenicity of missense variants was investigated in silico using the prediction programs PolyPhen-2, HOPE and SIFT.


Identification of eight families with SMAD3 mutations

SMAD3 sequence analysis in 393 patients with TAAD (without mutations in the FBN1, TGFBR1 and TGFBR2 genes) revealed five novel heterozygous SMAD3 mutations: c.313delG (p.Ala105ProfsX11), c.539_540insC (p.Pro180ThrfsX7), c.788C→T (p.Pro263Leu), c.1045G→C (p.Ala349Pro), c.1080dupT (p.Glu361X) (figure 1A). Three other mutations have previously been reported by our group: c.741_742delAT (p.Thr247fsX61), c.782C→T (p.Thr261Ile) and c.859C→T (p.Arg287Trp) (figure 1A).12

Figure 1

SMAD3 mutations in eight families with aneurysms–osteoarthritis syndrome (AOS). (A) Schematic representation of the SMAD3 gene. Boxes represent exons 1–9 with the untranslated regions (UTRs). The three main functional domains MH1, MH2 and the linker region are indicated. Mutations previously identified in the AOS syndrome12 are depicted in black font, and mutations identified in this study are depicted in blue. (B) Simplified family trees of eight unrelated families with AOS. Squares indicate males, circles represent females. A horizontal line above the symbol indicates medical examination by one of us. Owing to lack of space, generation III from family 1 is split into two levels. An arrow points to the index patient. The upper right blue square indicates the presence of osteoarthritis, the lower right red square the presence of a thoracic aortic aneurysm, the lower left green square the presence of an aneurysm in any other artery, and the upper left yellow square the presence of arterial tortuosity. Open symbols are individuals with a normal or unknown phenotype. Four individuals with open symbols (family 1, patient II-10, V-5, V-12 and family 3, patient III-2) had other signs of AOS, not indicated in the legend. A question mark (?) indicates sudden cardiovascular death possibly from an arterial rupture or dissection without autopsy. Age of death is displayed below the symbol. The presence (+/−) or absence (−/−) of a SMAD3 mutation is indicated underneath.

The eight families with SMAD3 mutations are unrelated and originate from the Netherlands (four families), Belgium (two families), Spain (one family) and the USA (one family). After molecular screening, 45 patients with a SMAD3 mutation were identified. The genealogical trees of these AOS families are shown in figure 1B. In four families, multiple patients were reported (figure 1B, families 1, 2, 4 and 5). In three families, the parents were unavailable for testing and no medical records were available.

The mutations were located in exons 2, 4, 6 or 8 of the SMAD3 gene (figure 1A). Four mutations introduced a frame shift (p.Ala105ProfsX11, p.Pro180ThrfsX7 and p.Thr247fsX61) or stop codon (p.Glu361X), and were considered to be pathogenic. Four missense mutations (p.Thr261Ile, p.Pro263Leu, p.Arg287Trp and p.Ala349Pro) were probably pathogenic, based on the following observations: (1) all involved residues that are highly conserved throughout evolution (from primates to zebrafish, data not shown); (2) in silico analysis predicts that these missense variants are likely to be pathogenic; (3) in two familial cases the SMAD3 mutation co-segregated with AOS; (4) these four mutations were absent in at least 342 ethnically matched control chromosomes. All variants are absent in the 1094 individuals from the 1000Genomes project.

Initial clinical features

Clinical data for 45 patients with a SMAD3 mutation were collected. The mean age of these patients with AOS was 45 years, including six children aged 17 (n=3), 15, 13 and 9 years. The main clinical characteristics of all 45 individuals from the eight families are summarised in table 1. All patients with a SMAD3 mutation had one or more signs of AOS.

Table 1

Clinical findings in 45 patients with aneurysms–osteoarthritis syndrome


All but three adult patients had consulted different physicians because of AOS symptoms before this study. In 19/35 (54%) of the adult patients, joint complaints were the initial symptom for which medical advice was sought (age range 18–61 years). In none of them was a (aneurysm) syndrome suspected. In these patients with AOS, joint abnormalities mainly consisted of OCD, osteoarthritis and meniscal lesions.

Cardiovascular abnormalities were the presenting feature in 46% (16/35) of the adult patients (age range 20–66 years). Sudden death from aortic dissections, aortic aneurysms and severe mitral valve insufficiency was the most common presentation. In three patients, the diagnosis of MFS was made at the time of presentation on the basis of the revised Ghent criteria.15

One patient (figure 1, family 1, patient II-1) died suddenly at the age of 64 years from an unknown cause.


All six children (aged 9–17 years) were referred for initial check-up after AOS was diagnosed in the family. Radiological studies were performed in three patients (family 1, patients V-8, V-10 and V-11). A 12-year-old patient presented with knee and lower back pain. Radiography and MRI showed agenesis of the anterior cruciate ligaments, OCD of the knee and severe intervertebral disc degeneration. A 17-year-old boy with mild back pain had severe intervertebral disc degeneration at multiple levels. One 16-year-old boy had a tenodesis of the first metacarpophalangeal joint.

All six children had cardiovascular examinations, which revealed an aortic aneurysm at the level of the sinus of Valsalva in two patients. These aneurysms were first diagnosed at the age of 14 and 16 years. Two children had mitral valve prolapse.

Cardiovascular anomalies

Cardiovascular abnormalities were documented in 89% (40/45) of our patients with AOS. These included thoracic aortic aneurysm and/or dissection, aneurysm of other arteries, tortuosity of the arterial tree, left ventricular hypertrophy, atrial fibrillation and congenital heart malformation. Arterial anomalies were present in 83% of patients.

Thoracic aortic aneurysms were present in 28 of 39 patients who had aortic root measurements. They were mainly present at the level of the sinus of Valsalva and ranged from 36 mm (Z-score 2.9) to 63 mm (Z-score 13.2) with a mean age at diagnosis of 39 years (range 14–65 years) (figure 2A). Eleven patients had been successfully operated on by elective aortic root replacement at maximum aortic diameters between 40 and 63 mm. Mean age at surgery was 41 years (range 20–64). In four patients, an abdominal aortic aneurysm was reported, at ages 49, 50, 61 and 62 years (figure 2E).

Figure 2

Arterial anomalies, including aneurysms, dissections and tortuosity. (A) Three-dimensional (3D) reconstructed CT angiography (CTA) of a 20-year-old man (family 1, patient V-1) shows an aneurysm at the level of the sinus of Valsalva of 45 mm (arrow). (B) 3D reconstructed CTA of a 50-year-old woman (family 1, patient III-17) showing a Stanford type A aortic dissection (arrows) extending into the brachiochephalic trunk at a maximal aortic diameter of 40 mm. (C) CTA of a 29-year-old man (family 1, patient IV-19) shows an aneurysm of the truncus pulmonalis of 50 mm. (D) Magnetic resonance angiography (MRA) of a 38-year-old man (family 1, patient IV-4) shows an aneurysm of the splenic artery of 40 mm (arrow) and marked arterial tortuosity of the splenic artery. (E) 3D reconstructed CTA of a 50-year-old woman (family 1, patient III-17) showing tortuosity of the abdominal aorta, suprarenal aneurysm of the abdominal aorta of 30 mm, and aneurysms of the coeliac trunk, and left common iliac artery. (F) CTA of a 45-year-old woman (family 1, patient IV-1) shows a Stanford type B aortic dissection at a maximal abdominal aortic diameter of 24 mm with dissection flap extending into the left common iliac artery (arrow). (G) MRA of a 34-year-old man (family 1, patient IV-17) shows a saccular aneurysm of the right ophthalmic artery of 3.5 mm (arrow). (H) 3D reconstructed CTA of a 29-year-old man (family 1, patient IV-19) shows a fusiform aneurysm of the left vertebral artery of 11 mm (arrow). (I) MRA of a 41-year-old man (family 4, patient III-3) showing the cerebral arteries. The calibre of the basilar artery is similar to that of the internal carotid arteries, indicating fusiform dilatation.

In total, 13 patients had an aortic dissection. A Stanford type A dissection was present in 11 patients; in five of them, the aortic root diameters could be determined before dissection occurred and ranged between 40 and 63 mm (mean 51 mm). In two patients, aortic dissections occurred while the aorta was only mildly dilated (figure 1B, family 1, patients III-2 and III-17), with maximal ascending aortic diameters of 45 mm and 40 mm, respectively (figure 2B). Five patients with a Stanford type A dissection had a successful aortic root replacement at a mean age of 46 years (38–52 years). Four patients had a Stanford type B dissection, and in two of them the dissection occurred in only mildly or non-dilated abdominal aortas (figure 2F). Two patients had both a Stanford type A and B dissection. Three patients had dissections in other non-dilated arteries, namely the coronary, common and internal iliac, and superior mesenteric artery.

Fifteen patients with AOS died suddenly between 34 and 69 years of age. Autopsy was performed in six patients and confirmed a Stanford type A dissection in five patients and a Stanford type B dissection in one patient. In seven patients, no autopsy was performed, but three of them were previously known to have aortic aneurysms/dissections. Other arterial aneurysms were detected in nine of 25 (36%) patients studied, mainly involving the vertebral, pulmonary, splenic, iliac and mesenteric arteries (figure 2C–E). One patient (figure 1B, family 1, patient IV-4) had an aneurysm of the splenic artery of 40 mm (figure 2D), and another patient (figure 1B, family 2, patient II-7) showed bilateral internal iliac aneurysms of 80 mm, as well as an abdominal aortic aneurysm of 100 mm. Imaging of the cerebral arteries revealed both intra- and extra-cranial aneurysms in 38% of patients involving the vertebral, carotid, basilar and ophthalmic arteries (figure 2G–I). In two patients (figure 1B, family 1, patient II-10 and II-12), a stroke was reported, at 56 and 76 years. The family history of family 2 revealed two 50% risk carriers with a stroke at 52 and 67 years (figure 1B, family 2, patients II-2 and III-2).

Tortuosity of the large- or medium-sized arteries was present in the majority of patients. Aortic tortuosity was found in 38% (figure 2E), tortuosity of other thoracic and abdominal arteries (mainly the subclavian and splenic arteries) in 38% (figure 2D), and tortuosity of the cerebral arteries (including the vertebral, internal carotid, cerebral and pericallosal arteries) in 50% of our patients with AOS.

Left ventricular hypertrophy was diagnosed in 18% of patients. It was mild to moderate, mainly concentric, and was not the consequence of hypertension, as most patients were normotensive without treatment. Atrial fibrillation was a common finding, with 24% (8/33) of patients having at least one episode. The age at onset ranged between 23 and 76 years. Three of eight patients had a single episode of atrial fibrillation after surgery. Mitral valve abnormalities were reported in half of the patients, the youngest being 14 years old. These anomalies ranged from mild prolapse to severe regurgitation requiring valve replacement. Congenital heart malformations were found in 9% (3/33) of our patients with AOS, and included bicuspid aortic valve, pulmonary valve stenosis, persistent ductus arteriosus and atrial septal defect. Of 13 women having a total of 23 pregnancies, one had a severe postpartum haemorrhage, but no other vascular complications or uterine ruptures were reported. In more than 30% of patients who initially presented with cardiovascular anomalies, joint abnormalities were reported later in life.

Joint anomalies

Almost all (96%) patients studied had radiologically proven osteoarthritis, with 75% of these in two or more joint types. Eighty-five per cent of these patients had painful joints. The mean age at osteoarthritis diagnosis was 42 years, and the youngest patient with osteoarthritis was detected at 12 years of age. The joints that were mostly affected were spine, hands and/or wrists, and knees, but osteoarthritis was also reported in all other joints including feet and/or ankle, hip and shoulder (figure 3H–J). Hand/wrist osteoarthritis was present in 14 patients, and in half of them the first carpometacarpal joint was involved (figure 3H). Other affected joints were the scaphotrapeziotrapezoidal, distal interphalangeal, proximal interphalangeal and occasionally metacarpophalangeal I joints. Furthermore, intervertebral disc degeneration mainly involving the cervical and lumbar discs was present in 92% (34/37) of patients (figure 3E–G) on retrospective evaluation of x-rays and CT scans. In addition, vertebral bodies showed shape irregularities located in the region of the anterior growth plates. In some documented cases, these abnormalities were already present at a young age (youngest 12 years).12 Spondylolysis and/or spondylolisthesis (figure 3F) were common (38%).

Figure 3

Osteoarthritis and osteochondritis. (A) X-ray of the left knee of a 41-year-old patient (family 4, patient III-3) shows a large osteochondral lesion without separation in the lateral femoral condyle (arrow). (B) MRI of the right knee of a 12-year-old patient (family 1, patient V-10) shows congenital absence of the anterior cruciate ligament (ACL). Asterisk: no ACL is seen in its expected location. The arrow points to the normal posterior cruciate ligament. (C) MRI of the knee of a 48-year-old woman (family 2, patient III-1) shows a large osteochondral lesion without separation in the medial femoral condyle (upper arrow). There is also a horizontal tear of the medial meniscus (lower arrow). (D) MRI of the right knee of a 17-year-old man (family 1, patient V-1) with a loose intra-articular body (encircled) due to a large osteochondral lesion of the medial femoral condyle (not shown). (E) MRI of the thoracic and lumbar spine of a 17-year-old man (family 1, patient V-8) with marked irregularity and impression of the anteroinferior endplates at multiple levels (see arrows for examples). (F) CT scan of a 44-year-old woman (family 1, patient IV-6) with severe multilevel degenerative disc disease and a spondylolisthesis at the L4–L5 level due to a bilateral spondylolysis (arrow). (G) MRI scan of a 50-year-old man (family 3, patient II-4) with marked degenerative abnormalities of the lower cervical spine (arrow) with narrowing of the spinal cord. (H) X-ray of the right hand and wrist of a 31-year-old man (family 1, patient IV-18) with moderate osteoarthritis of the first carpometacarpal joint (arrow). (I) X-ray of the right foot of a 31-year-old man (family 1, patient IV-18) with moderate osteoarthritis of the first metatarsophalangeal joint (arrow). (J) X-ray of the ankle of a 40-year-old woman (family 1, patient IV-9) with marked degenerative changes of the talocrural joint with severe lateral joint space narrowing (arrow).

More than half of the patients (56%) had non-traumatic OCD even at a young age (figure 3A,C,D). OCD occurred mainly in the knee and occasionally in the ankle or hip. Patients with OCD were operated on before the age of 40 years—the youngest at the age of 10 years. OCD was asymptomatic in some cases (figure 3A). Seven patients with AOS (28%) had meniscal lesions, one of whom had bilateral meniscectomy at the age of 13 years. In one patient, a congenital absence/agenesis of the anterior cruciate ligament was seen on MRI of the knee at the age of 12 years (figure 3B). Three patients had an arthroplasty of the knee at an average age of 64 (range 61–68 years), and one patient had an arthroplasty of the thumb base at the age of 58 years. Joint laxity defined as a Beighton score of ≥5, was seen in a minority (10%) of patients.

In the 19 patients who initially presented with joint abnormalities, extensive cardiovascular work-up was performed in the following years because of their family history or cardiovascular symptoms. In 64%, cardiovascular abnormalities were reported. More importantly, four of the 19 died suddenly from an aortic dissection.

Skeletal anomalies

Approximately 40% of the patients had long and slender fingers and toes, but overt arachnodactyly (as defined above) was not present. A positive thumb sign was seen in seven patients, and a positive wrist sign in one patient. Dolichostenomelia was present in 21% of patients.

Twelve patients (36%) had pectus carinatum, pectus excavatum or asymmetry of the costosternal junction. Scoliosis was present in 61% of our patients, and three of them were operated on for severe scoliosis. One patient had foraminal stenosis requiring foraminotomy of L5–S1 with spondylodesis of L4–S1. Protrusio acetabulae was present in one-third (35%) of patients and was usually mild. Over 90% of patients had pes planus. Camptodactyly was present in four out of 30 (13%) patients.

Craniofacial abnormalities

Figure 4 illustrates the facial features of 20 patients with AOS. Facial characteristics included high forehead, hypertelorism, long face, flat supraorbital ridges and malar hypoplasia, but were generally mild. Uvular anomalies (raphe, broad or bifid) were common in our series (52%). Of the 13 patients with uvular abnormalities, 62% had a broad uvula with or without a raphe and 38% had a bifid uvula. Uvular abnormalities may be an easy diagnostic clue, as they only occur in LDS but not in other syndromic or non-syndromic forms of TAAD. High-arched palate was common, and one patient was operated for a cleft palate. Dental malocclusion and retrognathia were occasionally seen. No craniosynostosis was observed or reported. There was a marked inter- and intra-familial variability in facial features (figure 4).

Figure 4

Facial features of 20 patients with aneurysms–osteoarthritis syndrome from four different families. Photographs are boxed in red (family 1), green (family 2), blue (family 4) and yellow (family 5). Facial features include hypertelorism, a long face, flat orbital ridges, a high forehead and malar flattening. Overall, the most prominent facial feature is hypertelorism. Written consent was obtained for publication of these images.

Additional features

Some features that are common in connective tissue disorders are also common in AOS. Umbilical and/or inguinal hernias were present in 17/40 (43%) of patients (age range 1–50 years). Pelvic floor prolapse occurred in seven of 17 adult women and mainly involved the uterus (6/7) and occasionally the bladder (2/7) and bowel (1/7). The mean age at operation for pelvic floor prolapse was 50 years (range 43–64). Varices or thread veins were reported or observed in 18 of 31 patients, usually already present at a young age (youngest patient 17 years) and were therapy (surgery)-resistant. Velvety skin and striae were present in the majority (62% and 53%, respectively) of the patients. Other recurrent findings included easy bruising and atrophic scars. Recurrent severe headaches or migraine was present in half (15/30) of the patients and did not co-occur with the cerebrovascular abnormalities.

Some additional features occurred sporadically in the eight families, but were not systematically evaluated in all patients. Diverticulosis was reported in four patients, and dural ectasia in seven patients. Two patients had unexplained severe lung emphysema, at the age of 63 and 54 years; one patient did not smoke, but no details on smoking or other risk factors for emphysema were available for the other patient. In three patients, xanthelasmata around the eye were reported, although no dyslipidaemia was found. Almost 40% (11/28) of patients complained of chronic or intermittent increased fatigue. Ophthalmological examination in 29 patients revealed no lens luxation. One patient had cataract surgery and multiple procedures for retinal detachment; one patient had mild cataract at the age of 54 years, and amblyopia was present in two patients. Hydrocephaly was not found. No moderate or severe developmental delay was reported in any patient, although no IQ tests were performed.


SMAD3 mutations

We have identified here five new and private heterozygous SMAD3 mutations in five unrelated AOS families. As we screened 393 patients with TAAD (negative for FBN1, TGFBR1 and TGFBR2 mutations), SMAD3-associated TAAD represents a small fraction of TAAD. Because patients in our cohort were initially analysed for syndromic TAAD genes, this cohort may be enriched for patients with MFS and LDS features.

In total, we have identified eight SMAD3 mutations, six of which were located in the MH2 domain, which mediates oligomerisation of SMAD3/SMAD4 and Smad-dependent transcriptional activation. Two frame-shift mutations were located upstream within the MH1 or proline-rich linker region. They led to truncated transcripts, which were probably subjected to nonsense-mediated mRNA decay, as shown before for the p.Thr277ProfsX61 mutation.12 The most likely effect of these mutations is loss of function, with TGFβ signals not being propagated via SMAD3. Notably, we have previously shown that this leads to a paradoxical increase in TGFβ signalling in the aortic wall,12 which has also been found in other syndromes characterised by arterial wall anomalies, such as MFS, LDS, ATS and FBLN4-related AR-CL.7 10 11 16

Regalado et al17 recently described four different SMAD3 mutations (p.Ala112Val, p.Asn218fs, p.Glu239Lys and p.Arg279Lys) in patients with TAAD and aneurysms affecting other vessels, including cerebral arteries, and osteoarthritis. The frequency of SMAD3 mutations in their cohort of non-syndromic familial TAAD patients was 2%, which is comparable to that in our cohort of (not necessarily familial) TAAD patients.

In addition, a p.Asn197Ile missense variant was found in a patient with osteoarthritis who was not evaluated for other AOS anomalies.18

Patients with AOS

We present the clinical and molecular data for 45 patients with SMAD3 mutations from eight unrelated families with AOS. The patients come from families of Dutch, Belgian, Spanish and American ancestry.

All patients with a SMAD3 mutation exhibited symptoms or signs of AOS, with the youngest patient being 9 years old. Although not all families could be completely evaluated, the penetrance of the mutations is nearly 100%. The expression varied from very mild (isolated bifid uvula in a 9-year-old girl, family 1, patient V-12) to severe (multiple aneurysms and dissections in a 50-year-old woman, family 1, patient III-17) disease. Age-dependent progression of the phenotype is evident, as aneurysms and osteoarthritis were encountered mainly during adulthood, although this study only included six children. The cardiovascular abnormalities at a young age were generally mild and mainly included mitral valve prolapse or congenital heart malformations. The youngest patient diagnosed with an aortic aneurysm was 14 years old. All dissections occurred in adulthood—the youngest patient was 34 years of age.

AOS is mainly characterised by a combination of arterial anomalies with early-onset osteoarthritis, but mild craniofacial anomalies and other features reminiscent of connective tissue disorders are also present. The AOS phenotype with typical cardiovascular and orthopaedic anomalies was present in at least five of the eight families. Two families were not screened for joint abnormalities, and in only one family joint problems were not reported. Similarly, Regalado et al reported osteoarthritis or joint disease in four of their five families (37% of their cases) with SMAD3 mutations, although radiological investigations to assess osteoarthritis were not performed.17

Intrafamilial variability, as illustrated by the clinical findings in a large family of 33 patients with AOS, was significant: while some patients presented mainly with arterial aneurysms and dissections, others only had joint abnormalities. Therefore the genotype–phenotype correlation, if present, will be difficult to establish.

Cardiovascular anomalies

The vast majority (89%) of patients with AOS had cardiovascular abnormalities. Aneurysms and tortuosity were found throughout the complete arterial tree studied, in both large and medium-size vessels, including the cerebral arteries. Despite the presence of intracranial aneurysms, stroke has rarely been reported in AOS. Dissections occurred in the aorta and in medium-/small-sized arteries, including a coronary artery. Arterial dissection and rupture occurred occasionally in aortas that were only mildly dilated; therefore early preventive surgery with resection of the aneurysms is advised.

Apart from arterial aneurysms and tortuosity, there were also other cardiovascular anomalies present in many patients with AOS, including mitral valve anomalies, congenital heart malformations, ventricular hypertrophy and atrial fibrillation. The congenital heart malformations were significantly more common than expected in the general population (p<0.0001). It is very likely that SMAD3 mutations also lead to cardiac hypertrophy and atrial fibrillation via TGFβ upregulation. It is currently unclear how loss-of-function mutations in SMAD3 lead to a paradoxical increase in TGFβ signalling12 and the congenital and age-related cardiovascular anomalies described above.19

Joint anomalies

Most of the patients developed joint abnormalities, including OCD, meniscal lesions, intervertebral disc degeneration and osteoarthritis. These abnormalities were already present at a young age. Interestingly, joint complaints were the first symptom for which clinical advice was sought in the majority of patients.

OCD was present in more than half of patients, mainly in the medial, but also in the lateral, femoral condyle of the knee. Interestingly, mutations in the ACAN gene encoding the proteoglycan aggrecan have been described in families with autosomal dominant inheritance of OCD.20 Aggrecan is a downstream effector of the TGFβ signalling pathway, and may mediate SMAD3-associated OCD in AOS.

Intervertebral disc degeneration was present in most (92%) patients and mainly involved the cervical and lumbar spine. Mice studies have shown that TGFβ is essential for promoting and/or maintaining the intervertebral disc during development.21

Osteoarthritis was present in almost all patients with AOS, and primarily involved the joints of the knee, spine, hand and foot. In SMAD3-related disease, osteoarthritis could be secondary to OCD, joint laxity or disc degeneration, which are present in many patients with AOS. However, OCD of the medial femoral condyle (most commonly observed in patients with AOS) rarely results in osteoarthritis. This area is non-weight-bearing and therefore less prone to osteoarthritis.22 In addition, osteoarthritis is also present in joints not affected by OCD or meniscal lesions. Joint laxity may also play a role in development of osteoarthritis: pes planus, scoliosis and joint hypermobility may indicate ligamental insufficiencies. Therefore, in addition to intrinsic abnormalities of the hyaline cartilage of the joints, early-onset osteoarthritis in these families may be enhanced by overload based on abnormal menisci, intervertebral discs and/or ligaments. Spinal osteoarthritis at the intervertebral and uncovertebral joints may be the result of the early disc degeneration.

It is likely that osteoarthritis in AOS is due to secondary changes and abnormal development of the cartilage directly caused by the SMAD3 mutation. TGFβ has a dual role in chondrocytes, primarily acting as a stimulator in chondrocyte differentiation and, in later stages of development, blocking chondrocyte terminal differentiation.23–25 SMAD3 has an important function in this TGFβ-mediated growth inhibition and maintenance of the articular cartilage.25 26 This is corroborated in Smad3 knock-out mice, which show premature chondrocyte maturation and subsequent premature osteoarthritis.25 27 A direct role for SMAD3 in osteoarthritis is further supported by the identification of a SMAD3 mutation in a patient with knee osteoarthritis,18 the recent association between a single-nucleotide polymorphism in intron 1 of SMAD3 and the risk of both hip and knee osteoarthritis,28 and several in vitro studies.29

Other phenotypic anomalies

A MFS habitus with dolichostenomelia, long slender fingers, pectus deformity and scoliosis was present in a minority of patients. Aspecific cutaneous anomalies commonly seen in connective tissue disorders, including velvety skin with striae and easy bruising, were present in the majority of patients with AOS. Craniofacial features were often mild or absent and mainly included hypertelorism (figure 4) and a broad or bifid uvula. Overall, the phenotypic anomalies in many patients with AOS were discrete, and missed on consultation for cardiovascular anomalies, whereby the patients were classified as non-syndromic TAAD.

Comparison with other TGFβ-related syndromes

Although many cases of AOS were classified as non-syndromic TAAD, the phenotype overlaps with that of aneurysm syndromes such as MFS and LDS. Some patients had a MFS habitus, whereas others had craniofacial anomalies with features such as hypertelorism and broad/bifid uvula reminiscent of LDS.

The cardiovascular features in patients with AOS are similar to those of LDS, including thoracic aortic aneurysms at the level of the sinus of Valsalva and aneurysms and tortuosity throughout the arterial tree. However, involvement of the entire arterial tree, including the intracranial arteries, is rare in patients with MFS. Similarly to LDS, the aortic aneurysms of patients with AOS tend to rupture at smaller aortic diameters than in MFS. Aneurysms are less common in ATS than in AOS, although a similar tortuosity of the entire arterial tree is found.

Atrial fibrillation and ventricular hypertrophy (24% and 18%, respectively) have not yet been reported in LDS and are both uncommon in MFS,30 whereas some patients with ATS show ventricular hypertrophy.31 Mitral valve anomalies, mainly prolapse, are equally common in patients with AOS (50%) and MFS (54%)32 and less common in LDS (35%).33

Congenital heart malformations are found in only 9% of patients with AOS, in contrast with 22–35% in LDS.8 The nature of these defects—that is, persistent ductus arteriosus and atrial septal defect—are similar in both disorders.8 In only 1% of patients with MFS have congenital heart malformations been reported.33

Joint anomalies with osteoarthritis, OCD and meniscal lesions are key features of AOS, being present in almost all patients. Such anomalies are rarely described in LDS, MFS or ATS. None of the 90 patients with LDS type I or II described by Loey et al were reported to have osteoarthritis, OCD or meniscal abnormalities, although cervical dislocation or instability, spondylolisthesis and intervertebral disc degeneration have been occasionally described.8 34 Interestingly, arthralgia, osteoarthritis of the hand, hip and/or spine was reported in several patients from a large family with LDS due to a TGFBR2 mutation.35 Also in MFS, osteoarthritis, OCD and meniscal abnormalities have only sporadically been described,36 although spondylolisthesis is present in 6% of patients with MFS.37 Further joint studies in patients with MFS and LDS are warranted to establish the frequency of osteoarthritis and OCD in these related syndromes.

In conclusion, joint anomalies such as osteoarthritis, OCD and meniscal abnormalities may be a useful discriminating feature from other forms of TAAD. Therefore the syndrome is named AOS. X-ray examinations of knees, total spine and hands, particularly in TAAD patients with a medical or family history of joint complaints or abnormalities, is recommended. Furthermore, as these typical joint anomalies may be the presenting feature of AOS before symptoms or signs of the cardiovascular features become obvious, we recommend imaging of the heart and complete arterial tree, including cerebral arteries, to exclude arterial anomalies in patients with early-onset osteoarthritis in combination with OCD or a family history of aortic aneurysm or sudden death.


We thank all patients and family members for their enthusiastic participation in the study. We thank the referring physicians for sharing the data on the patients. We acknowledge T de Vries-Lentsch and R Koppenol (Department of Clinical Genetics, Erasmus Medical Center, Rotterdam, The Netherlands) for the photographic work.


View Abstract


  • Funding This work was partially funded by an Erasmus Fellowship 2009 (Erasmus Medical Center, The Netherlands) to AMB-A, a Research Foundation Flanders grant (Ghent University Hospital, Belgium) to BL, and the Swiss National Science Foundation grant 31003A-120504 to GM.

  • Competing interests None.

  • Patient consent Obtained.

  • Ethics approval Medical ethics committee of the Erasmus Medical Center Rotterdam (Erasmus MC) in The Netherlands.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement Anonymised clinical records are partially available upon request by physicians, genetic counsellors and other professionals in the field.