Most individuals with Familial Adenomatous Polyposis (FAP) harbor mutations in the APC gene on chromosome 5q21. They are at an increased risk of brain tumors, including cerebellar medulloblastoma, when compared with the general population (Brain Tumor Polyposis—BTP Type 2). Genotype-phenotype correlations between APC gene mutations and central nervous system (CNS) tumors have, thus far not been successful. Herein the authors have pooled their registry experience in BTP type 2 with the published reports.
The authors analyzed their established hereditary CRC Registry for brain tumors in FAP pedigrees (56 families, 213 individuals), pooled their patients with BTP and known APC mutations with those reported thus far elsewhere, and compared the resulting mutation distribution of FAP-BTP with the mutation distribution for APC mutations in the US.
Twenty-eight patients from 24 families were accrued, the most common brain tumor in BTP was medulloblastoma (60%) predominantly in females (12:5) under the age of 20 (mean age 14.7 SD 9.2). Other histologic subtypes included astrocytoma and ependymoma. Analysis of the pooled APC mutation data by Chi-square test of association shows an odds ratio of 3.7 (P < .005) for all brain tumor subtypes and 13.1 (P < .001) for medulloblastoma in patients harboring segment 2 APC mutation (codons 679–1224) compared to nonsegment 2 mutation.
Hereditary syndromes predisposing to cancer including colorectal polyposis syndromes are rare, but may be associated with a markedly increased risk of developing both colorectal and extraintestinal malignancy. The impact of aggressive surveillance and, when applicable, prophylactic surgery in these patients has been borne out with improved morbidity and mortality.1
Since its first description by Turcot et al in 1959,2 the association of brain tumors and colorectal polyposis and cancer has been refined. This was initially based on the pattern of inheritance and colorectal polyp burden,3 and further elucidated by the associated gene mutations in the brain tumor-polyposis (BTP) Syndrome Type1 (Hereditary Non-Polyposis Colon Cancer [HNPCC]; MLH1, MSH2 mutation associated) and Type 2 (FAP; adenomatous polyposis coli [APC] gene mutation associated).4
Familial adenomatous polyposis (FAP) is a dominantly inherited predisposition to colorectal cancer occurring usually by the fourth decade of life, and diverse extracolonic and extraintestinal tumors. FAP is, in most cases, related to mutations in the APC gene on chromosome 5q21. Mutations within specific loci on the gene have been correlated with differences in the clinical course or phenotype of FAP.
In BTP Type 2 the colorectal adenoma burden is typical for FAP. However, brain tumors frequently predate the development of colorectal cancer (CRC). The central nervous system (CNS) tumors associated with FAP include medulloblastoma, astrocytomas,5 and less frequently ependymoma, pinealoblastoma,6 and ganglioglioma.7 In families with FAP the relative risk of developing brain tumors in general and specifically medulloblastoma is increased 7 and 92 times, respectively, when compared with that of the general population.5 Hamilton et al5 and subsequent investigators noted clustering of brain tumors in FAP families. Thus far, however, no group of investigators has been able to correlate the development of brain tumors with either a specific or a spectrum of APC mutations, akin to the observation of certain clinical characteristics of FAP correlating with mutational hotspots in the APC gene.
In this study we performed a retrospective analysis on our own hereditary gastrointestinal cancer gene registry, specifically analyzing the occurrence of brain tumors within families with FAP and, wherever possible, correlating the histologic subtype of brain tumor with the APC mutation. We then performed a pooled data analysis on APC mutations in FAP families including the BTP Type 2 from our registry and those previously reported in the existing medical literature.
MATERIALS AND METHODS
We analyzed all of our accrued hereditary colon cancer pedigrees (1965–2005) for the clinical diagnosis of FAP (Gardner syndrome, APC, and attenuated FAP included) and for nonmetastatic brain tumors in the extended family. Families were included when they had a history of numerous, early (before age 50) colorectal adenomas, with or without APC mutation testing. Affected individuals were included if they were reported as having been diagnosed with a primary brain cancer and they, or a blood relative, were diagnosed with FAP based on clinical presentation and APC mutation testing. Affected individuals were included when they were further characterized based on the age at presentation and histologic subtype of the brain tumor. When not reported in the clinical chart, permission to review additional, including hospital medical records, was requested from the patient or next of kin.
A PubMed search using MeSH terms Gardner syndrome, Adenomatous Polyposis Coli, and brain neoplasms or CNS neoplasms or CNS tumors was performed. The results were filtered to include reference to the APC gene. Demographic, tumor histology, and APC mutation results were pooled with the identified pedigree from our registry. Data analysis included a chi-square test on the resulting contingency table (GraphPad In-Stat, San Diego, CA). APC mutation frequency defined as segmental involvement in affected families was compared with the APC mutation distribution in the North American population of patients with FAP as reported elsewhere8 (Table 1).
Table 1. Segmental Definition of APC Gene Codons
APC indicates adenomatous polyposis coli.
Brain Tumors in Patients With FAP: Single-Center Analysis
Since its inception, our registry has included 56 families with FAP, with 213 individuals considered at risk of FAP. In this group, 3 (1%) developed primary CNS tumors (Fig. 1). In 1 family (Fig. 1A) harboring an APC mutation involving segment 2, 1 of the siblings had undergone colectomy at 18 years of age but was also diagnosed with cerebellar medulloblastoma at 19 years of age; no other CNS tumors were reported from the family. In the other 2 FAP families harboring individuals diagnosed with brain tumors, FAP mutations in segment 1; codon 141 (Fig. 1C) and 159 (Fig. 1B) were detected. Primary brain tumors were reported in the medical records recovered, although no further details on tumor histology could be identified despite multiple attempts.
Brain Tumors in Patients With FAP: Pooled Analysis
APC gene mutation was reported in 28 individuals (17F : 9M) with CNS tumors belonging to 24 FAP families (Table 2). The mean age at presentation for CNS involvement was 18.5 (SD 12.5) years and the most commonly encountered histologic subtype was medulloblastoma (17; 60%, mean age at diagnosis, 14.7 years, SD 9.2), other subtypes were astrocytoma (4; 14%), and ependymoma (3; 10%). The most frequently reported APC mutation involved codon 1061 in 3 unrelated individuals, 2 diagnosed with medulloblastoma. APC gene segment 2 (codon 679–1224) mutations were present in 16 (57%) individuals and 13 (54%) families. Comparison of pooled APC mutation data between the North American population with FAP and the reported FAP families with CNS tumors by chi-square test of association showed an odds ratio of 3.74 (95% confidence interval [CI]: 1.71–8.17, 2-sided P value = .001) for all brain tumor subtypes and 13.1 (95% CI : 3.69–46.50, P < .0001) for medulloblastoma in patients with segment 2 APC mutation compared with nonsegment 2 mutation.
Table 2. Pooled APC Mutation Analysis in Patients With FAP and CNS Tumors
Method of testing
Age at CNS Lesion, y
APC indicates adenomatous polyposis coli; FAP, familial adenomatous polyposis; CNS, central nervous system; PTT, protein truncation test; PCR, polymerase chain reaction; SSCP, single-strand conformation polymorphism analysis; CSGE, conformation sensitive gel electrophoresis; CHRPE, congenital hypertrophy of the retinal pigment epithelium; N/A, not available.
Brain tumors and early colon cancer occur both in FAP and HNPCC. This is the first study to identify, within the population with FAP, a subpopulation with a markedly increased risk of brain tumors in general and medulloblastoma specifically. Our analysis suggests that FAP patients harboring the APC gene mutation between codons 697–1224 are at a 3-fold increased risk of brain tumors in general, and 13-fold increased risk of medulloblastoma specifically, when compared with FAP patients with other APC gene mutations. This observation has implications ranging from genetic counseling for individuals with FAP as well as potentially impacting the risk-benefit assessment for surveillance for brain tumors within this subpopulation.
The limitations of our study include the small number of affected individuals within our individual cohort having both a brain tumor and FAP. Despite the staggering higher risk of brain tumors in FAP compared with the general population, it is still a relatively rare occurrence (1 in 3735 patient-years);5 this has beset earlier reports on the APC mutation correlates of BTP-2, necessitating a pooled analysis to develop statistically sound observations. Our study is also limited by the increased likelihood of referral of atypical, or more aggressive variants of the disease to centers dedicated to the recognition and management of hereditary cancer syndromes. In that sense, our comparison population, the APC mutations detectable in the North American FAP population, is likely a better reflection of APC mutations in the FAP population than in that followed through dedicated registries. An additional assumption included in our statistical analysis was that the sensitivity of different testing methods was comparable between the 2 groups.
The incidence of medulloblastoma in patients with FAP appears to be highest in the first 2 decades of life and was validated by our analysis. Although brain tumors are a recognized feature of FAP, to date, however, there have been no formal recommendations on routine imaging or surveillance for CNS lesions in this population. This is despite evidence that diagnosis and treatment at an early stage entails a significantly improved prognosis.18 In our analysis of pooled cases, twice as many patients were female than male (F:M ratio 2.4:1); this is in contrast to most reported series of medulloblastoma in the general population showing a clear preponderance in males.19
The concept of genotype-phenotype correlations in FAP is far from novel. Our observation of a milder, attenuated form of FAP (attenuated FAP: AFAP) with later development of colorectal cancer was followed by correlation of this clinical phenotype with mutations at the extreme 3′ and 5′ end of the APC gene.20 Similar correlations have been made for aggressive upper intestinal and various extraintestinal characteristics of FAP.8 Mutations within segment 2 of the APC gene overlap with mutations associated with an increased risk of hepatoblastoma21 and pigmented retinal lesions, namely, congenital hypertrophy of the retinal pigment epithelium (CHRPE). Although poorly understood, genotype-phenotype correlations appear to be mediated by a dominant-negative effect of mutated APC gene products.22
Although associated with the same mutation spectrum, CHRPE was not consistently present in our or other reported patients with BTP (7/22); furthermore, when present CHRPE was not segregating to those individuals harboring a mutation in segment 2 of the APC gene.
The APC gene encodes a 2843-amino acid product with several functional domains. Its main role appears to be the down-regulation of beta-catenin as part of the wnt-signaling pathway. The part of the gene corresponding to segment 2: codons 697–1224, translates into part of the Armadillo region, involved in the stabilization and motility of the cellular actin cytoskeleton, and several 15-amino acid repeats that appear to be beta-catenin binding but not degradation sites. Activation of the wnt-signaling pathway and intranuclear localization of beta-catenin appear to be present in a subset of patients with medulloblastoma23 that have a better prognosis.24
In conclusion, we have described the impact of BTP-2 in the context of a tertiary referral center and cancer registry. Our 56 families with FAP included 3 with BTP-2, which is comparable to the experience reported from other registries and reflects the overall increased risk of brain tumors in FAP when compared with the general population. Our analysis of pooled data on APC mutations associated with BTP-2 suggests that brain tumors in general, and cerebellar medulloblastoma specifically, are far more likely to develop in individuals harboring APC mutation between codons 679–1224. This in turn impacts the cost-effectiveness of potential CNS tumor surveillance in this specific subset of patients with FAP and has important ramifications in the counseling offered to these patients. Our group recommends heightened clinical suspicion and more ready referral to neuroradiologic imaging in individuals with FAP with segment 2 APC gene mutation.
We thank Kristy Livesay for her invaluable editorial and secretarial assistance.