A 12-year-old boy with no documented familial history of cancer presented weakness, malaise, abdominal pain, dysphagia and loss of 14 kg before referral. Upper digestive endoscopy detected a 3.5 cm-ulcerated lesion in the cardia. Surgical exploration revealed an inoperable gastric tumor with adjacent lymph-node involvement, histopathologically proven liver metastases and probable diaphragm, esophageal and celiac axis infiltration. A gastrostomy was performed and the patient received chemotherapy without significant response. The patient died with disseminated disease after seven months.
Methods
The patient was referred to the Department of Abdominal Surgery of Hospital AC Camargo, São Paulo, Brazil, and biopsies were obtained during diagnostic endoscopy and surgery. The biopsied samples included tissues fixed in 10% buffered formalin and embedded in paraffin (FFPE, Formalin-Fixed, Paraffin-Embedded) and a frozen liver metastasis sample. DNA was extracted from areas of gastric tissue sections defined by a pathologist as normal, metaplastic, dysplatic or cancerous. Blood from relatives was obtained after genetic counseling and informed consent. DNA was extracted from FFPE tissues and white blood cells using standard protocols. TP53 mutations were detected by direct sequencing of the entire coding region of the gene using an automated procedure and protocols as described elsewhere [10]. Immunohistochemistry was performed using the following antibodies for p53 (DO-7, DakoCytomation, Glostrup, Denmark), E-cadherin (36, BD Transduction, San Jose, CA), MLH1 (G168-728, BD Biosciences Pharmigen, San Jose, CA), MSH2 (G219-1129, BD Biosciences Pharmigen, San Jose, CA), PMS2 (A16-4, BD Biosciences Pharmigen, San Jose, CA) and MSH6 (44/MSH6, BD Biosciences Pharmigen, San Jose, CA) expression.
Genetic and Molecular Pathology findings
Endoscopic biopsies included intestinal metaplasia in the cardia region (S1) and high-grade dysplasia/in situ gastric carcinoma (S2). Surgical biopsies included liver metastasis (S3M), well-differentiated gastric adenocarcinoma (S3G) and a histopathologically normal celiac lymph node (S4L). TNM classification was T4N2M1. Despite early-onset, tumor histology did not support a case of Hereditary Diffuse Gastric Cancer (HDGC). Immunohistochemistry showed E-cadherin membrane expression in tumor cells. No alterations were found in MLH1, MSH2, PMS2 and MSH6 expression, ruling against an extra-colonic manifestation of Lynch Syndrome (data not shown). Figure 1 shows immunostaining for p53 in metaplasia (Figure 1A) and in gastric adenocarcinoma (Figure 1B). In the latter, all cancer cells showed strong positivity in the nucleus as well as weak cytoplasmic staining. In metaplasia, p53 was detected in the nucleus of cells in small, defined areas with intestinal differentiation (as shown by the presence of goblet cells) whereas areas of non-intestinal metaplasia were negative for p53 expression. Sequencing of TP53 exons 2 to 11 identified wild-type sequence (CGC) in S1 (metaplasia) while S2, S3G and S3M showed a mutation at codon 337 (c.1010G > A, p.R337H) (Figure 2A). The celiac lymph node (S4L) was negative for p.R337H (data not shown).
Presence of the founder haplotype associated with p.R337H was assessed by analysis of SNP 28 (rs9894946) in the TP53 gene with the T allele indicating the founder haplotype [11]. This allele was detected in the patient's samples containing p.R337H (S2, S3G and S3M) but not in those with wild-type sequence (Figure 2B). Allele-specific PCR demonstrated that p.R337H was located on the haplotype carrying SNP28, identical to the founder haplotype (data not shown). Presence of p.R337H was confirmed by Restriction Length Fragment Polymorphism (RFLP) using HhaI, which cuts within codon 337, in S2, S3G and S3M but not in S1 or S4L (data not shown). Genomic DNA was obtained from peripheral blood of the father, mother and of one brother. The mutant haplotype was constitutive (in heterozygosity) in the father's genomic DNA (Figure 2), whereas mother and brother were wild-type for TP53. DNA from peripheral blood cells of the patient was not available.
Discussion
This case is one of the earliest cases of gastric cancer ever reported. Gastric adenocarcinoma is extremely uncommon under 21 years of age [12]. Data from Brazilian cancer registries do not show any evidence of occurrence of gastric adenocarcinoma in children or teenagers [13]. This patient was the youngest case of gastric cancer over a period of 18 years in the pathology archives of Hospital AC Camargo, one of the largest cancer centers in Brazil. The patient did not present any evidence of familial history of cancer and had no report of personal history of other disease including cancer. In previous studies, we observed that the TP53 p.R337H mutation is present in the germline of about 1:300 individuals (0.3%) in Southern Brazil as the consequence of a founder effect predisposing to early cancer [8]. However, the penetrance of this mutation is incomplete. In families with clinical definitions of LFS/LFL, about 25 to 30% of p.R337H carriers remain apparently cancer-free over their lifetime (compared to a penetrance of over 90% in carriers of germline TP53 mutations occurring at "hotspot" codons). Furthermore, the penetrance at age 30 is about 15%, compared to 50% in carriers of mutations at "hotspot" codons. Because of this incomplete penetrance, a germline p.R337H mutation may be present in individuals without specific familial cancer history, in particular in the absence of extensive pedigree data. These considerations led us to test for the presence of p.R337H in retrospective, pathology archived tissues of this very young case.
The TP53 mutation p.R337H, borne by a haplotype carrying the T allele of rs9894946 (SNP28) was detected in tumor and metastasis but not in non-cancer tissues of this 12-year-old patient. This haplotype corresponds to the documented founder allele detected in many Brazilian LFS/LFL families and is present in only 2% of Brazilians. On the other hand, p.R337H has never been documented as a somatic mutation in any cancer at the IARC TP53 database (http://www-p53.iarc.fr). Therefore, the probability that the mutation may arise as somatic mutation on the allele carrying SNP28 independently of the founder mutation is extremely low. Moreover, neither the p.R337H mutation nor the T allele of rs9804946 was detected in non-tumor tissues of the patient, although both variants were detected in cancer tissues. The presence of the founder haplotype in tumor, but not in non-tumor tissue of the patient is therefore highly puzzling. The analysis was repeated multiple times by different technicians and using different DNA extracts, ruling out a technical artifact. On the other hand, the hypothesis of a selective loss of heterozygosity for p.R337H both in metaplasia and in the non-involved celiac lymph node seems extremely unlikely. Consistent with a germline origin, the mutant haplotype was found in the father's DNA, but not in the mother and a brother. This observation indicates that the mutant haplotype detected in the patient's tumor is of paternal origin.
The mechanisms that have led to the presence of this mutant haplotype only in cancer cells of the patients are obscure. The results are consistent with the hypothesis that the patient may have harbored two populations of cells, one containing the paternal TP53 wild-type allele, and another one containing the paternal TP53 mutant allele, the latter present at very low levels, undetectable by sequencing in most tissues. The maternal, wild-type allele appears to be identical in both cell populations. This situation would be consistent with genetic microchimerism. Microchimerism refers to harboring a small number of cells or DNA from a genetically different individual, often due to maternal-fetal or feto-fetal trafficking during pregnancy [14–16]. In this patient, we suggest that microchimerism may have been caused by feto-fetal cell trafficking between dizygotic twin fetuses, one of whom had the paternal mutant haplotype and underwent very early developmental failure. It should be noted that the loss of a dizygotic fetus has not been documented in the mother's history of pregnancy, making this hypothesis unverifiable.