Types and locations of DCVs

A total of 211 DCVs in BMPR1a were identified, 82 from the literature, 17 from LOVD and 112 from ClinVar classified as pathogenic or likely pathogenic (see Additional File 4). 178 of these DCVs occur in coding regions, and 33 occur in non-coding regions of the gene. The DCVs include missense, nonsense and frameshift variants, as well as large deletions (see Fig. 2). Missense variants classified as DCVs were included if they were identified from the literature in patients with JPS, or from LOVD or ClinVar and classified as likely pathogenic or pathogenic by submitters to the respective databases. At this stage, assessment by these submitters has been completed prior to any consideration by the InSiGHT ClinGen Variant Curation Expert Panel based on gene specific modifications of the ACMG criteria.

Fig. 2

Location and type of DCVs in BMPR1a identified in the literature and pathogenic or likely pathogenic variants from LOVD and ClinVar [3, 9, 10, 14, 16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34]

Of the DCVs occurring in BMPR1a, frameshift DCVs predominated (38.86%), followed by nonsense DCVs (24.17%), then missense DCVs (18.01%), with large deletions being the least common (2.80%). DCVs were seen across all functional domains of the gene, with most occurring in the Intracellular domain (46.92%) and the MH1 domain (29.38%). Across all functional domains, frameshift DCVs were the most common (see Table 1). There was no apparent genotype–phenotype correlation.

Table 1 Frequency of DCV types across functional domains of BMPR1aProportion of JPS accounted for by BMPR1a DCVs

The proportion of JPS cases accounted for by DCVs in either BMPR1a or SMAD4 is 45% to 60% [16, 35] with BMPR1a DCVs accounting for 17% to 38% of cases [3, 10, 16,17,18,19,20].

However, there are still a number of JPS cases with no identifiable DCV, which could be attributed to additional genes which have not yet been identified. MacFarland et al. [14] proposed any additional genes are likely to be low penetrant autosomal dominant or autosomal recessive, due to the lack of family history and younger age at diagnosis seen in the DCV negative group. Alternatively, epigenetic changes may be responsible for driving the phenotype in JPS patients without an identified pathogenic variant [14].

Other reasons could include older studies using more limited gene sequencing techniques which did not detect large deletions. The use of Multiplex Ligation-dependent Probe Amplification (MLPA) has allowed for identification of large deletions in BMPR1a causing JPS that were not detected with Sanger sequencing techniques in the past [19, 21]. Contemporarily, Next Generation Sequencing almost always detects both point variants and large deletions. This is demonstrated in a cohort where Sanger sequencing identified 40.9% of JPS cases to have a DCV in either BMPR1a or SMAD4, while Next Generation Sequencing identified 61% [35].

There may still be cryptic mutations in these genes causing JPS, undetectable to any current gene sequencing technologies.

Gene-specific phenotype associationsAge of diagnosis

DCV negative JPS cases have a significantly lower mean age of diagnosis compared to those carrying either a BMPR1a or SMAD4 DCV (13.1 years vs 21.4 years, p = 0.05) [3]. MacFarland et al. found the median age of diagnosis in a selected paediatric population without an identified DCV was 5 years compared to 18 years in those with an identified DCV (p < 0.001) [14]. The mean age of diagnosis was similar and not statistically significant for BMPR1a compared to SMAD4 DCV carriers (24.5 years and 28.0 years, respectively) [4, 14, 21].

Location of polyps

The location of polyps in BMPR1a DCV carriers are predominantly colorectal [15]. Gastric polyps are significantly less common inBMPR1a DCV carriers than SMAD4 DCV carriers (8% vs 73%, p < 0.001 [21], and 13% vs 39%, p = 0.001 [4]). Similarly, family history of upper GI polyps is significantly lower in BMPR1a DCV carriers compared to SMAD4 DCV carriers (10% vs 85–86%, p < 0.01) [3]. In a European retrospective study, polyps in the small intestine were even less common in BMPR1a DCV carriers (3.2%) compared to SMAD4 DCV carriers (15.7%) [4].

Number of polyps

Colonic polyps are seen in both carriers of BMPR1a and SMAD4 DCVs [21], however SMAD4 DCV carriers have higher colorectal polyp numbers than BMPR1a DCV carriers [15].

BMPR1a DCV carriers do not display massive gastric polyposis (> 100 gastric polyps) [17], with 86% of BMPR1a DCV carriers having < 5 gastric polyps, whereas 17% of SMAD4 DCV carriers had > 100 gastric polyps (p = 0.0001) [4].

There was no significant difference in those required to have a colectomy for colorectal polyp control between BMPR1a and SMAD4 DCV carriers. However there was a significant difference in need for colectomy between those with an identified DCV compared to those without an identified DCV (33% vs 3.1%, p = 0.03) [14]. There were no BMPR1a DCV carriers who required gastrectomy for gastric polyp control or gastric cancer [14].

Type of polyps

While juvenile polyps are the most common type of polyp seen in JPS, other polyp types have been reported, including hyperplastic, adenomatous and pseudo-polyps [13, 21,22,23]. This mixed polyposis is seen in BMPR1a DCV carriers but not in SMAD4 DCV carriers, who typically present with homogenous juvenile polyposis [4].

Deletion of PTEN/BMPR1a

In addition to large deletions of BMPR1a being identified in patients with JPS, cases of JPI have been observed in patients with contiguous deletion of PTEN and BMPR1a, which often presents in the first two years of life with severe GI bleeding, diarrhoea, exudative enteropathy and rectal prolapse [4, 5, 18, 36,37,38]. However, there is variability in the phenotype of reported cases, with some presenting later in childhood [39, 40], and some also being associated with dysmorphic features, developmental delay and macrocephaly [5, 41].

Patients with a large deletion of PTEN and BMPR1a are diagnosed with JPS at a significantly younger age than those who possess a DCV in BMPR1a alone (1.5 years vs 23 years, p < 0.001) [4].

PTEN DCVs cause the PTEN Hamartomatous Tumour Syndromes (PHTS) such as Cowden Syndrome and Bannayan-Riley-Ruvalcaba syndrome, which often present with skin and GI hamartomas, macrocephaly, intellectual disability and developmental delay. The variability in phenotype seen in those with contiguous deletion of both PTEN and BMPR1a may be explained by heterogeneous and mixed phenotypes of PHTS and JPS.

Additionally, the young age of presentation in contiguous deletion of PTEN and BMPR1a suggests a synergistic effect of the contiguous gene deletion [5].

Extra-intestinal manifestations

Extra-intestinal manifestations are not commonly seen in BMPR1a DCV carriers. Importantly, those with JPS caused by BMPR1a DCVs do not display any features of HHT (epistaxis, telangiectasia and arteriovenous malformations) as seen in SMAD4 DCV carriers [4].

Congenital heart defects have been reported in BMPR1a DCV carriers [4, 10, 42, 43]. In a cohort of patients with JPS, congenital heart defects were observed more in BMPR1a than SMAD4 DCV carriers (9.1% vs 4.2%), however due to limited sample size, this finding was not statistically significant [4].

Other extra-intestinal manifestations seen in BMPR1a DCV carriers include facial dysmorphism, macrocephaly, short stature and delayed puberty. Interestingly, these extra-intestinal features are also seen in PHTS, and may have actually been caused by undetected DCVs in PTEN rather than BMPR1a, due to use of older sequencing techniques.

Malignancy

JPS is a precancerous condition, with a mean age of diagnosis of 43.9 years and a cumulative lifetime risk at 70 years of age of GI malignancy of 38.7% [1] and for any malignancy of 86.2% [2]. The rates of malignancy reported in cohorts of JPS patients in the literature varies from 11 to 22% [2, 4, 12,13,14,15]. Differences in the years of observations between studies accounts for some of the variability in malignancy rates.

In JPS malignant transformation has been thought to occur from permanent mechanical insults, inflammation and repair, following a dysplasia-carcinoma sequence, with cancer arising on a background of generalised mucosal instability, as seen in other hereditary precancerous GI polyposis syndromes such as Familial Adenomatous Polyposis [13, 44].

GI malignancy seen in JPS includes both gastric and colorectal cancers, with colorectal cancer being more common (62% vs 21%) [4] and developing at a younger age than gastric cancer [13]. In a cohort of JPS patients, colorectal cancer was more common than gastric cancer in BMPR1a DCV carriers (88% vs 0%) [4]. Indeed, gastric cancer has not been reported in BMPR1a DCV carriers [4, 21].

A gene-phenotype association in terms of cancer risk in JPS has been identified, with cancer being less frequently observed in BMPR1a DCV carriers than SMAD4 DCV carriers (8.5% vs 20.5%) [4]. A phenotype associated cancer risk has also been identified, with cumulative risk of malignancy being significantly lower in patients with JPC, than those with GJP (58.7% vs 77.6%, p = 0.005) [2].

Genotype–phenotype correlation

Those with DCVs in BMPR1a who present at a young age (under 10 years), with high numbers of colorectal polyps (> 50) or who develop colorectal cancer are presumed to carry a DCV which causes a more penetrant phenotype of JPS (see Fig. 3 and 4). Only some DCVs identified from the literature included the accompanying phenotype, compromising the interpretation of a genotype–phenotype correlation from the reported DCVs and penetrance for the phenotype (see Additional File 3).

Fig. 3

Age of Diagnosis and Number of Colorectal Polyps in patients possessing Large Deletions, Missense, Nonsense or Frameshift DCVs in BMPR1a

Fig. 4

Frequency of DCV type across functional domains of BMPR1a, and patients with Colorectal Cancer

Amongst patients with large deletions, none were reported to have a high number of colorectal polyps. There were however two patients possessing a large deletion of the entire signal peptide of the BMPR1a gene who were diagnosed at a young age (two and eight) with a moderate number of colorectal polyps (10–50) [21].

There were two patients diagnosed at a young age with high numbers of colorectal polyps with nonsense DCVs in the Intracellular Domain; one with the genotype c.1010C > G (p.Ser337Ter), was diagnosed at age six with > 80 colorectal polyps [17, 21], and the other with c.1081C > T (p.Arg361Ter), who was diagnosed with > 150 colorectal polyps, but at a slightly later age of 14 [21].

There were four patients diagnosed at a young age with high numbers of colorectal polyps possessing frameshift DCVs. Two had a duplication in the MH1 Domain: one with a c.351dup (p.Leu118Alaf