Introduction

JPS is a rare autosomal dominant syndrome affecting one per 100 000 births, where 50 – 75% of affected patients have a positive family history [19, 26]. Hamartomatous polyps occur throughout the GIT, increasing cumulative lifetime risk of GI cancer by 9 – 50%, which is decreased through increased surveillance [65]. Such “juvenile” polyps (JP) are histologically described as having dense stroma with inflammatory infiltrate with mucus-filled cystic glands in the lamina propria (Figs. 1, 2, and 3).

Fig. 1
figure 1

Macroscopic appearance of JPs. A Multiple pedunculated polyps with smooth surfaces post-bowel resection in a JPS patient. B JP from a patient with JPS, noted for its smooth surface [14]

Full size image
Fig. 2
figure 2

Structure of the SMAD4 gene, involving MH1 domain, linker domain and MH2 domain (self-made)

Full size image
Fig. 3
figure 3

Illustration showing SMAD4’s (blue oval) involvement in the TGF-β and BMP pathways [39]

Full size image

Clinical diagnosis of JPS is confirmed if any one of the following 3 criteria are met, given there is absence of syndromic extra-intestinal features that define other hamartomatous polyposis syndromes:

  1. 1.

     > 5 colonic hamartomatous polyps at one time or recurrent

  2. 2.

    Any number of hamartomatous polyps in a patient with family history of JPS

  3. 3.

    Extracolonic hamartomatous polyps (e.g., stomach, small bowel)

It is estimated that 40–60% of JPS cases are caused by DCVs in SMAD4 or BMPR1A genes, mostly consisting of missense, nonsense, deletions, and small insertions, together with large genomic deletions [4]. Germline SMAD4 DCVs have been observed in 20–30% JPS cases [3], similarly to BMPR1A, located at chromosome 10q23.2. PTEN DCVs have been sequenced in approximately 5% of JPS patients, near BMPR1A at chromosome 10q23.3, though confounds with Cowden syndrome. Finally, ENG DCVs have recently been associated with JPS, located at chromosome 9q34.1 with known associations to HHT [48, 77]. For example, Howe and colleagues [45] found ENG DCVs in 13/31 JPS patients without DCVs in SMAD4 or BMPR1A, whereas 2/11 JPS patients were ENG+ in Sweet and colleagues’ study [102]. Currently, > 50% of JPS cases have had causative genes described, all involved in the TGF-β signalling pathway, modulating colonic epithelial growth [38, 47].

The SMAD4 gene is located on chromosome 18q21.1 with 55 000 base pairs encompassing 11 exons, encoding a 551-amino acid protein. Functional domains include the MH1 domain, involved in DNA binding, a domain linking the two MH domains, and the MH2 domain which is involved in homodimerisation, heterodimerisation, and transcriptional activation and nuclear location of SMAD4 [48, 75].

Its protein product acts as an intracellular mediator to TGF-β and downstream bone morphogenetic protein (BMP) signalling, having a major role in colonic epithelial growth [21, 48]. Following activation of TGF-β, members of the SMAD family form a complex with SMAD4 via its COOH terminus. Mutations disrupting this complex formation result in loss of TGF-β signalling, thus partially explaining why most germline mutations of SMAD4 map to this domain [92]. Physiologically, these complexes are then transported to the nucleus, signalling growth inhibition. It is hypothesised that heterozygous loss-of-function (LOF) SMAD4 mutations may prevent formation of these complexes, thus promoting growth, resulting in cellular proliferation and subsequently neoplasia [39]. Potentially, DCVs in the linker region are less prominent, given this region is often deleted during alternative splicing.

Bosman’s hamartoma-adenoma-carcinoma sequence theory [13] postulates a “landscaper” defect, where SMAD4 mutations disrupt epithelial architecture, differentiation, and proliferation via altering the microenvironment. This pathway begins with JP formation, adenomatous change, dysplasia, then finally carcinoma [54, 55].

Current knowledge of genotype–phenotype correlations between SMAD4 variants and JPS is that individuals are more inclined to have upper gastrointestinal (UGI) polyposis and higher gastric cancer risk, as compared to BMPR1A [91]. HHT is exclusively seen within SMAD4 + JPS patients [8]. Thus, current treatment involves regular surveillance via routine endoscopy with polypectomy, together with monitoring and treatment of HHT manifestations [65].

Further detailed genotype–phenotype correlations between SMAD4 DCVs and JPS subtypes are yet to be fully described in this field of research. Better characterisation of these associations will help modify clinical diagnosis, screening, surveillance, and management of SMAD4 + JPS patients. In addition, this research will also aid pathogenicity classification for SMAD4 variants where phenotypic manifestations are incorporated into modified American College of Medical Genetics (ACMG) criteria, allowing, for example, segregation analyses to assist in classification.

This narrative review’s overarching research question is thus, in SMAD4 variant carriers, what are existing genotype–phenotype correlations relating to sites and types of variants within the gene, particularly focusing on phenotypes of JPS? Further from this, what implications will these correlations have on clinical management and for gene specific modifications to ACMG criteria?

Methods

A literature search was performed on July 27, 2021, via three online search databases: Ovid MEDLINE, Embase Classic + Embase, and PubMed. Search strategy utilised key words surrounding the research question: SMAD4, JPS, and gene association studies, via MeSH on Demand.

Regarding JPS, key words included juvenile polyposis, intestinal polyposis, JPS and hamartomatous polyposis syndrome. Aliases of SMAD4 gene included SMAD family member 4, MADH4, DPC4, JIP, MYHRS, mothers against decapentaplegic homolog 4, deletion target in pancreatic carcinoma 4, MAD homolog 4, deleted in pancreatic carcinoma locus 4 and HSMAD4. To capture keywords related to genotype–phenotype association studies, words included genotype, gene, genome, phenotype, DNA, mutation, chromosome, variant, and variations of gene associated studies. Boolean operators including AND and OR were utilised, adjacency operators, as well as word truncation to enable different forms of words to be searched for simultaneously (Appendix A).

Specific inclusion criteria included English and human studies, retrospective and prospective gene studies, unique case reports and reviews which discuss JPS, SMAD4 DCVs and/or genotype–phenotype correlations. Exclusion criteria included any non-English and animal studies, non-significant case reports, and studies that did not mention juvenile polyposis syndrome nor SMAD4.

Results

From this search, 829 studies were identified from three databases (Ovid MEDLINE = 251, Embase Classic + Embase = 386, PubMed = 192). 396 of these were duplicates, 30 were animal studies not excluded from search strategy, and 82 were considered irrelevant as they did not involve SMAD4, JPS or its causative genes, amounting to 321 studies. Full text screening thus isolated 110 papers, including narrative reviews, retrospective and prospective gene studies, together with pertinent case reports relevant to this review, pertaining to SMAD4 and its genotype–phenotype correlations to JPS and relevant conditions (Fig. 4). Results are tabulated in Fig. 5, Tables 1 and 2, and are further elucidated in the discussion.

Fig. 4
figure 4

PRISMA Diagram presenting the review process for this narrative review (self-made)

Full size image
Fig. 5
figure 5

Sites and types of SMAD4 DCVs, protein change and phenotypes, self-made via Microsoft Visio. All 291 variants were collated from the literature. In the top section, SMAD4 gene structure is represented via its exons demarcated by nucleotide numbers. DCVs causing JPS phenotypes have been depicted with lines extending below the diagram in brown (JPS), green (massive gastric polyposis in JPS) and red (GI cancer in JPS). Above the illustrations represent extra-gastrointestinal phenotypes, including HHT (dark red), aortopathy (pink), IBD (purple), Myhre syndrome (dark blue), JIA (light blue), hypertrophic osteoarthropathy (yellow), Ménétrier’s disease (grey) and concurrent GI cancer boxed in bright red. In the bottom section, large deletions and chromosomal translocations are portrayed via solid lines, surrounding the SMAD4 gene, where JPS phenotypes are below, and extra-GI phenotypes are above. c.1245_1248delCAGA (p.Asp415Glufs*20), a highly prevalent DCV, is represented by asterisks (*). Variants of uncertain significance (VUS) are marked with adjacent stars, sourced from JPS registries and case reports, with pathogenicity reviewed by ClinVar

Full size image
Table 1 List of gene studies, including title, sample size, study design, results, and limitations
Full size table
Table 2 Tabulated form of SMAD4 variants and associated phenotypes, where families are bolded
Full size table

Discussion

Genotype–phenotype correlations to JPS

Variant hotspot

In SMAD4 + JPS patients, the majority of germline DCVs are in the MH2 domain. Up to 80% of DCVs are located between exon 8 and 11, allowing complex formation and translocation to the nucleus in the TGF-β pathway [74]. Small deletions in this location have caused serious cases of JPS with colonic and gastric juvenile polyposis, particularly c.1245_1248del [34, 48, 86, 91]. In another study, 40% (10/25) harboured a genetic alteration at codon 361 in exon 8 [56].

Variant types

Most common variant types among SMAD4 DCVs are missense, deletions and small deletions, resulting in frameshift and premature stop codons [27]. Less common are nonsense, insertions, duplications and intronic mutations [20, 23, 93]. In a study by Jones and colleagues [56], 22/25 patients had missense mutations, 2/25 had frameshift mutations and 1/25 had a nonsense mutation, whereas Aretz and colleagues [4] reported 5/17 nonsense, 6/17 frameshift and 6/17 missense mutations, all found to be pathogenic variants, apart from one VUS (missense mutation c.425_426A > G). Chromosomal translocations are rare causes of JPS, historically requiring chromosomal analysis, as described in a JPS-HHT patient [1] with exons 6–11 deleted. Another case report involved a balanced translocation causing loss of the entire SMAD4 gene in a JPS-HHT patient, with dysmorphic features, intellectual disability, developmental delay, and corpus callosum agenesis [79].

Histologic phenotype

JPs with SMAD4 DCVs tend to be more epithelial with high crypt-to-stroma ratio, as compared to BMPR1A, with a tendance to be more stromal with a lower ratio. Despite this, dysplasia was equally common in JPS polyps with either mutation [104]. Polyp phenotype is variable, ranging from sessile to pedunculated, with adenomatous, hyperplastic, and inflammatory polyps described in the literature, especially in BMPR1A + JPS [69, 77].

Extracolonic polyposis

Patients with SMAD4 DCVs, especially in the linker and MH2 domains, tend to develop and have a family history of UGI polyps, including the small bowel and stomach [34, 91]. SMAD4 DCVs are associated with higher gastric polyp numbers, massive gastric polyposis, and thus, partial or total gastrectomy and gastric cancer [4, 5, 10, 28, 72]. In most cases, SMAD4 DCVs have been detected in patients with both UGI and LGI polyps, whereas polyps are restricted to LGI and anal canal for BMPR1A [100].

Aggressive phenotype and variant correlation

Patients with SMAD4 DCVs can develop a more aggressive GI phenotype, with polyps associated with low-grade adenoma, high grade adenocarcinoma, upper GI location, and presence of malformed vessels within the stroma [41]. This is especially the case in patients with DCVs in exons 8–11, especially c.1245_1248del and c.1421delC, involving massive gastric polyposis and GI cancer [33, 34, 55, 80, 89].

Genotype–phenotype correlations to cancer

Lifetime risk for development of GI cancers in JPS families in different studies range from 9 to 50%, attenuated by improved surveillance and polypectomies over time. Overall, most SMAD4 + JPS patients with GI cancer had DCVs in the MH2 region [36, 77]. They have a higher incidence of GI cancer than those with BMPR1A. In a study by Aytac and colleagues [6], following regular surveillance and appropriate polypectomies, 4/27 individuals with SMAD4 DCVs developed cancer, in comparison to 0/8 of BMPR1A + JPS patients. In another study by Blatter and colleagues [10], incidence of cancer was also higher in SMAD4 carriers, with 20.5% of patients with GI cancer (26/127), compared to 8.4% (8/94) in BMPR1A carriers (p = 0.015).

Gastric cancer

As aforementioned, gastric polyposis is more common in SMAD4 carriers with JPS, with gastric cancer risk occurring up to 30% in those with SMAD4 DCVs. 7/17 JPS patients with SMAD4 variants had gastric cancer in a study by Aretz and colleagues [4], compared to 0/13 for BMPR1A carriers. In Blatter and colleagues’ study [10], 7/127 SMAD4 carriers had gastric cancer, and 0/94 in BMPR1A.

Colorectal cancer

Colorectal cancer occurs at a similar incidence in both causative genes of JPS, where 15/127 (11.8%) had CRC in SMAD4 carriers, compared to 7/94 (7.4%) in BMPR1A carriers [10]. In another study by Schwenter and colleagues [93], 3/14 (21.4%) SMAD4 + JP-HHT patients developed early onset CRC.

Somatic studies

From somatic studies, SMAD4 is not seen to be a driver gene for GI cancer, though 16% of primary colorectal tumours have alterations in SMAD4, and 6% in SMAD2. SMAD4 follows APC mutation and precedes TP53 in CRC development [25]. Loss of SMAD4 expression is associated with worse overall survival in patients with CRC, given associations with metastasis and advanced disease [73].

In 30% of pancreatic cancers, SMAD4 is deleted following inactivation of K-ras, increasing TGF-β expression and creating an environment for tumour progression [74]. It is postulated that SMAD4 mutations do not initiate tumour formation, as germline mutations are not associated with pancreatic tumours, but instead promote metastases via LOH and intragenic mutations [73].

Genotype–phenotype correlations to genetically related allelic disorders of SMAD4

HHT & JPS-HHT syndrome

HHT is an autosomal dominant disorder affecting 1 in 5000 to 10 000 individuals, leading to vascular dysplasia with facial and peripheral telangiectasias, together with arteriovenous malformations (AVM) of lung, central nervous system, and GIT. In JPS-HHT, patients share symptoms of JPS and the full range of HHT features [53].

80–85% HHT patients have DCVs in ENG, ACVRL1 or SMAD4, where the former two encode for endothelial receptors of the TGF-β family, necessary to maintain vascular integrity and angiogenesis [42, 93]. SMAD4 DCVs account for < 2% of HHT patients [50]. It is hypothesised that its genetic loss disrupts the balance regulating vascular remodelling and angiogenesis, as well as communication between TGF-β and BMP signalling pathways, as SMAD4 is common to both [35]. Frequency of pulmonary AVMs and gastric involvement were higher amongst SMAD4 + JPS-HHT patients, than those not due to SMAD4 +  [55, 101]. Such DCVs are mostly found in SMAD4’s MH2 region [38], where up to 80% of SMAD4 + JPS are accompanied by HHT. Particularly prevalent DCVs include c.1228_1229delCA, c.1245_1248del, and missense variants in exon 8 [74, 75, 80].

Myhre syndrome (MS)

SMAD4 DCVs are solely responsible for MS, a rare developmental disorder with < 100 cases reported. It is characterised by dysmorphic features, joint limitation, muscular pseudohypertrophy, intellectual disability and deafness. DCVs include de novo missense mutations around codon 496–500 in exon 11 (Figs. 6 and 7).

Fig. 6
figure 6

Diagram of variant location in the SMAD4 gene causing MS, mostly around the Ile500 residue in the MH2 domain [67]

Full size image
Fig. 7
figure 7

Diagram depicting SMAD4’s involvement in the TGF-β pathway, as shared with TGFB11/2 and FBN1, the signalling pathway to transcribe profibrotic genes [2]

Full size image

All reported MS cases have occurred independently of JPS [67], with associations to neoplasia. In a recent study, 6/61 MS patients exhibited neoplasia, including endometrial (3/6) and brain tumours (3/6). Given LOF mutations in SMAD4 cause JPS, it is hypothesised that gain-of-function mutations observed in MS may contribute to neoplasia [70].

Other conditions observed in carriers of SMAD4 variants

Cardiac pathologies

Cardiac pathologies have been reported in SMAD4 DCV carriers both independently and in conjunction with JPS-HHT. Manifestations include aortic root dilatation, aneurysm, aortic dissection, and mitral valve dysfunction, including regurgitation and prolapse.

This has been reported in SMAD4 + JPS-HHT patients with variants in the MH2 region, particularly c.1245_1248del and c.1333C > T [2, 16, 103, 106, 107]. Heald and colleagues [43] observed cardiac pathologies in 6/16 SMAD4 + JPS-HHT patients, while Wain and colleagues [106] reported 7/34 JPS patients had connective tissue defects including enlarged aortic root, aortic and mitral valve insufficiency and aortic dissection. Thus, these cardiac pathologies have been postulated to be part of SMAD4-induced HHT manifestations.

Notably, without JPS-HHT, hereditary thoracic aortic disease was described in patients with rare MH1 domain missense mutations. In Duan and colleagues’ case report [29], two family members exhibited different phenotypes, one with ascending aortic dissection, and the other with aortic aneurysm and bicuspid aortic valve. Two unrelated patients both had early onset type A aortic dissection. The proposed pathway is that SMAD4’s involvement in TGF-β signalling is shared with TGFBR1/2 and FBN1, genes involved in connective tissue disorders, where SMAD4 is a transcriptional regulator and tumour suppressor [2].

Juvenile idiopathic arthritis (JIA) & Hypertrophic osteoarthropathy (HOA)

In SMAD4 + JPS-HHT patients, JIA has been described in carriers of MH2 domain missense mutations, particularly c.1052A > T in exon 8. Along with colonic JPs with HHT features, patients had upper and lower limb joint swelling, erythema, and digital clubbing [9, 61].

HOA has also been observed in SMAD4 + JPS families, marked by digital clubbing and extensive new bone formation, in MH2 domain DCVs, particularly c.1236C > G [7, 64]. In addition, digital clubbing has been examined in many SMAD4 + JPS-HHT patients, potentially as a manifestation of pulmonary AVMs and right-to-left cardiac shunts [9, 35, 36, 50, 59, 61]. It is postulated that SMAD4 mediates intracellular signals of TGF-β and BMP, found at high levels in bone and cartilage, potentially having a role in bone formation, thus explaining HOA and JIA.

Ménétrier’s disease (MD)

MD has been diagnosed concurrently in SMAD4 + JPS patients with gastric polyposis, marked by giant mucosal folds in gastric fundus and body, with diminished acid secretory capacity and protein losing state causing hypoalbuminemia. All MD cases were caused by the SMAD4 variant, c.1245_1248del. In one family, there were 5 MD cases, 3 JPS cases and 1 case of JPS-MD, and two other studies reported familial JPS-MD cases [82, 83]. Mechanistically it is proposed that TGF-alpha overexpression leads to TGF-β pathway inactivation, promoting cell proliferation, where MD could be a manifestation of gastric polyposis in JPS, or be confounding given their similar pathology.

Diagnosis and genetic testing

SMAD4 + JPS patients mostly have generalised and colonic juvenile polyposis, together with JPS-HHT syndrome in some patients. JPS subtypes include:

  • 1. Infantile JP (< 2 years old): rare in SMAD4 + JPS patients, but more common in large deletions involving BMPR1A and PTEN, this is the most severe form of JPS with poor prognosis given aggressive polyp formation [49]

  • 2. Generalised JP: JPs throughout the GIT

  • 3. Colonic JP or JP coli: JPs exclusively in the colon, common to BMPR1A carriers.

  • The latter two phenotypes, generalised and colonic JP, are common to SMAD4, and are caused by DCVs throughout the gene. Malignancy mainly occurs in those with DCVs in the MH2 region, whereas non-malignant polyposis can occur anywhere.

  • 4. JPS-HHT syndrome: Exclusive to SMAD4 + patients, JPS-HHT patients have features of both JPS and HHT, caused by DCVs in MH2 region with few exceptions [35, 36].

Once diagnosed with JPS, genetic testing of SMAD4 and BMPR1A germline mutations for probands should occur, in combination with familial genetic counselling [106]. Molecular genetic testing approaches can include BMPR1A and SMAD4 concurrent testing, including multiplex-ligation dependent probe amplification for single or partial gene deletions. Also, serial-gene testing in patients with suspected JPS-HHT can occur, via sequence analysis and gene-targeted duplication or deletion analysis for SMAD4. Contemporarily, multigene panels with BMPR1A, SMAD4, PTEN and other genes, exome and genome sequencing, and chromosomal analysis for translocations are utilised [68, 88].

In addition, all SMAD4 DCV carriers should be screened following JP and HHT protocols, further elucidated in management.

Implications on management

Surveillance for asymptomatic SMAD4 or BMPR1A DCV carriers, or at-risk family members with no variant detected, are distinct between LGI and UGI tracts. In general, careful surveillance should occur for SMAD4 DCVs in the MH2 domain.

LGI management

Asymptomatic LGI surveillance follows conventional endoscopic monitoring, involving 3-yearly full blood examination and colonoscopy from 12–15 years old if no abnormalities are found, or commence screening earlier if symptomatic [24, 109]. Otherwise, if polyps are found, annual screening and endoscopic polyp resection would occur until polyp free [48]. Others suggest patients should be screened annually or biennially regardless, until 70 years old [30]. If colonic polyps are unable to be monitored, controlled or demonstrate malignant potential, this warrants consideration of total abdominal colectomy with ileo-rectal anastomosis or proctocolectomy with or without pouch reconstruction [32, 96]. There are no randomised controlled trials of surveillance to provide a strong evidence base for surveillance and its frequency.

UGI management

In terms of UGI surveillance, there are competing thoughts, but are tailored towards known genotype–phenotype correlations. Howe and colleagues [48] suggest UGI endoscopy should take place concomitantly with colonoscopy, in conjunction with biliary and/or pancreatic duct brushings in the context of abnormal liver function tests or elevated amylase. Dunlop [30] advises one or two-yearly UGI endoscopies with colonoscopy from 25 years old. Sayed and colleagues [91] differentiate screening between SMAD4 + JPS patients, who should receive it one to three-yearly, while BMPR1A + or DCV negative patients should be screened five-yearly. Similarly differentiating management based on causative genes, Monahan and colleagues [76] have suggested UGI endoscopic surveillance from 18 years old in SMAD4 + JPS patients, and from 25 years old for BMPR1A + patients, at a frequency of one to three-yearly. Nonetheless, given current data, no UGI surveillance for BMPR1A carriers could be justified due to lack of UGI pathology reported, especially cancer. Additionally, Ménétrier’s disease could be considered during UGI endoscopy. If polyps are detected, UGI endoscopy would be repeated annually with appropriate resection, though complete or partial gastrectomy may be warranted in cases of massive gastric polyposis, dysplasia and gastric cancer, as seen in SMAD4 + JPS patients [40, 66, 97, 98].

Extra-intestinal manifestations

SMAD4 + JPS patients should be evaluated for HHT within 6 months of diagnosis, examining for manifestations such as telangiectasia, AVM and digital clubbing. Together with complete blood count, annual targeted clinical examination should occur to monitor for HHT and cardiac pathologies, including full facial observation, peripheral examination to assess for clubbing and joint swelling as per JIA, and cardiorespiratory examination. If HHT is confirmed, screening would thus include 2 yearly bubble contrast echocardiography and pulse oximetry for pulmonary AVMs, followed by CT pulmonary angiogram if abnormal, and a single MRI brain to exclude brain AVMs [32, 35,36,37, 51].

Limitations

Overall, limitations of most studies were that they were retrospective, with limited patient numbers given JPS’ rarity, and had incomplete screening of all findings of interest. As a result, low numbers often precluded statistically significant observations. In some cases, there was overrepresentation of SMAD4 + patients given recruitment methods and did not compare phenotypes with other causative genes. As such, larger scale follow-up studies of JPS patients should occur both retrospectively and prospectively to assess genotype–phenotype correlations, with complete screening of all potentially associated syndromes and conditions.

Conclusion

In conclusion, truncating, missense and nonsense mutations around the MH2 region of SMAD4 are most prevalent and hence more likely to be pathogenic. In SMAD4 + JPS patients, given association with extracolonic polyposis and higher risks of GI cancers, endoscopic screening should occur from 12–15 years at a 3-yearly frequency, especially for patients with DCVs in the MH2 region. With associated genetically related allelic disorders like HHT, cardiac pathologies, HOA and potentially JIA, symptoms should be monitored for these via regular targeted clinical examination. Where HHT is suspected, further investigations should include 2-yearly bubble echocardiogram and a single brain MRI.

This review may help modify clinical diagnosis, screening, surveillance, and management of SMAD4 + JPS patients, as well as aid development of gene specific modifications to the ACMG/AMG criteria for pathogenicity assessment of SMAD4, thus supporting the work of the planned SMAD4 InSiGHT ClinGen Variant Curation Expert Panel.