Patients and methods

Patients with JPS were identified from sources in the UK, Australia, Israel, USA, Japan, and Korea. The diagnostic criteria for JPS used in this study were in accordance with the proposal of Ko and colleagues30 that affected individuals have either five or more juvenile polyps throughout the gastrointestinal tract or any number of juvenile polyps and a family history of JPS. All affected individuals had more than one typical juvenile polyp as confirmed by histology. The following clinical data were also obtained using a standardised proforma: family history; occurrence of benign and malignant tumours; dysmorphic features; arteriovenous malformations and other cardiovascular anomalies; pigmentation; and other notable clinical features including those of CS, BRRS, and GS. Where possible, material from JPS polyps and other tumours was obtained for histological review and confirmation of diagnosis.

Constitutive DNA from individuals was extracted from EDTA blood samples using a standard sucrose lysis method. Published oligonucleotide and reaction conditions were used to amplify specifically each exon of theDPC4, PTEN, andPTCH genes.20 24 31 The search for germline mutations in DPC4 was performed using conformation specific gel electrophoresis (CSGE). Two independent workers using either CSGE or single strand conformational gel polymorphism (SSCP) undertook screening for mutations inPTEN. A combination of CSGE and SSCP was used to screen the PTCH gene. All samples with band shifts were sequenced in duplicate and in forward and reverse orientations, after reamplification of the appropriate exon from genomic DNA in the polymerase chain reaction, using the ABI Ready Reaction Dye Terminator Cycle Sequencing kit and the 377 Prism sequencer.

Results

Fifty six patients with JPS were studied; 47 were ascertained from 15 families and nine apparently sporadic patients had no known relatives affected with JPS, although in three cases there was a strong family history of colorectal cancer. Table 1 details the clinical characteristics of all patients. Mean age at presentation of juvenile polyposis was 23 (SD 16) years (range 1–65). All affected individuals from whom histological material was available had juvenile polyps of a typical type.

The extent of polyposis within the gastrointestinal tract varied considerably between families in terms of both the numbers and sites of polyp formation. Some of the families showed marked evidence of intrafamilial differences in disease expression (table 1) even allowing for factors such as patient age and screening by colonoscopy. Furthermore, in a number of patients there was documented evidence of adenomatous intestinal polyposis although none had the atypical juvenile polyps found in HMPS.

The increased risk of gastrointestinal cancer reported in association with JPS was clearly present in our patients. Fifteen (27%) had developed some type of gastrointestinal cancer by the age of 65. Of these, six were small bowel carcinomas, two gastric cancers, and seven colorectal cancers. Based on cancer incidence rates in the UK,32 this equates to about a 16-fold increased risk of gastrointestinal malignancy in JPS.

None of the patients had prototypical dermatological (e.g. tricholemmomas) or skeletal phenotypic features indicative of CS, BRRS, or GS. There was some overlap between the clinical features of JPS patients and these other syndromes—three patients had macrocephaly for example—but no specific features to suggest misdiagnosis. Furthermore, none of the patients had a past history of thyroid cancer and only one had developed breast cancer. Three patients had hyper- or hypothyroidism but no tissue diagnosis suggestive of CS. The increased risk of cancer in these JPS patients appeared to be confined to the gastrointestinal tract in contradistinction to CS and GS. Arteriovenous and other cardiovascular anomalies were present in three patients (5%). The reported association between JPS and these anomalies may therefore be genuine.

No JPS patient had a germline PTEN mutation. Control samples with known PTEN mutations showed aberrant SSCP bands and previously reported polymorphisms were detected. PTEN mutations are typically found in 80% of CS and 50% of BRRS cases which adhere to operational diagnostic criteria established by the International Cowden Consortium.33 Also, germlinePTCH mutations were not detected in our JPS cases. PTCH mutations are typically detectable in 39% of GS patients.31

Five germline DPC4 mutations were identified (table 2). Three of these were deletions ranging in size from two to 11 base pairs in exons 1, 4, and 11. One of the mutations was a single base substitution creating a stop codon in exon 10. The fifth mutation was a missense mutation in exon 8. This variant has been reported previously.24 Our series therefore confirms that mutations in DPC4 are associated with JPS but the mutations we detected did not involve the trimerisation domain of theDPC4 protein. There were no distinguishing clinicopathological features in those patients harbouringDPC4 mutations.

Discussion

Several syndromes—JPS, CS, BRRS, GS, and HMPS—include juvenile polyps in their phenotype. These diseases can have subtle manifestations and some of their clinical features are not unique. Hence diagnosis is not straightforward. However, it is important to distinguish between these syndromes as the types of cancer associated with each appear to be quite different. A more reliable and objective means of differentiating syndromes than reliance solely on clinical features is clearly desirable, especially in the case of JPS the diagnosis of which is made in part by exclusion. Determining the molecular basis of each of these syndromes clearly offers the best method of establishing a diagnosis.

Until recently the molecular basis of the hamartomatosis polyposis syndromes was unknown. Early somatic data suggested the existence of a putative locus for JPS at 10q22–24, encompassingPTEN.17 This observation led several researchers to examine PTEN as a candidate for JPS. No linkage to 10q22 or mutation inPTEN was detected in either the JPS families or isolated cases studied by Marsh and colleagues.20However, both Olschwang and colleagues21 and Lynch and colleagues18 reported germlinePTEN mutations in patients with multiple juvenile polyps, suggesting that PTEN was the cause of some JPS cases. Eng has recently questioned their conclusion.33 In the study by Olschwang and colleagues,21 the adult male patient (G116) also had a laryngeal carcinoma and a heterogeneous thyroid nodule, which is highly suggestive of CS. Furthermore, although both children in the report (G796 and G710) did not have features indicative of CS or BRRS, they were less than 15 years of age. As the penetrance of CS is only 10% by this age34 it is possible that both children will develop features of CS later in life. Similarly, the affected members of the family reported by Lynch and colleagues18 had major clinical features consistent with a diagnosis of CS.

Our mutational analysis of JPS patients is the most comprehensive to date. The findings concur with previous work suggesting that the probability of detecting a constitutive mutation inPTEN is not high if a patient's clinical phenotype does not adhere to the International Cowden Consortium operational diagnostic criteria for CS33 or to the specific clinical features of BRRS.

PTCH is a good candidate for JPS but has not previously been tested for germline mutations in JPS patients. We have excluded this gene as a cause of juvenile polyps outside the setting of GS. Although PTCH mutations may directly lead to gastrointestinal hamartoma formation it remains possible that reports of juvenile polyps in GS result from a chance association or from contiguous deletion of PTCH and at least one other gene.

Recent work has identified germline mutations in theDPC4 (SMAD4) gene on 18q21.1 as a cause of some cases of JPS.23 In our study 21% of JPS could be ascribed to germline mutations inDPC4 (five of 24; comprising 15 families and nine sporadic cases tested). This provides further support for the role of mutations in this gene as a cause of JPS. Failure to detect mutations in DPC4 in the other patients is unlikely to reflect problems in the screening method alone: it is possible that some mutations may be in the UTRs, introns, or promoter region of the gene but we have screened all exons and splice sites and have found that CSGE can detect all small insertions, deletions, and 90% of single base substitutions under such conditions.

DPC4/SMAD4 is a pivitol component of the transforming growth factor β signal transduction pathway.35 Through hetero-oligomer formation by interaction with SMAD1 (and possibly 5) andSMAD2 (and possibly 3),SMAD4 mediates apoptotic and growth inhibition responses. Hamartoma formation in JPS presumably results from disruption of the transforming growth factor β signal transduction pathway.35 In our study mutations inSMAD4 were detected in exons 1, 4, 8, 10, and 11. Pathogenic mutations previously reported were in exons 5, 8, and 9.23 A wide range of mutations can clearly affect cellular responses to transforming growth factor β leading to hamartoma formation. Whether specific genotype-phenotype relationships exist await further studies.

We conclude that the clinical features of Cowden syndrome, Bannayan-Riley-Ruvalcaba syndrome, and Gorlin syndrome can be used to distinguish patients who are likely to carry germline mutations inPTEN and PTCH and who do not, therefore, have a diagnosis of JPS. The rare syndrome HMPS can probably be distinguished from JPS on histological grounds, and there is no evidence for linkage of JPS to the HMPS locus (although this cannot be confirmed until the HMPS gene is identified). Combining our findings and those reported by Howe and colleagues23suggests that up to 40% of JPS cases might be caused by germline mutations in DPC4 although these patients cannot be distinguished from the majority of JPS patients who harbour germline mutations in uncharacterised genes. Current evidence does not suggest any common mode of action of PTEN, PTCH, and DPC4. Therefore, there are few clues as to the nature of the uncharacterised genes for JPS as it appears that several different cellular defects are associated with hamartomatous juvenile polyps of identical appearance.