Gastric carcinoma

P.A. Wright, G.T. Williams

The molecular study of gastric carcinoma has much to offer the medical
scientist, pathologist, clinician, and patient. Morphological
classifications are not completely satisfactory in predicting biological
and clinical behaviour. Molecular diagnostic markers are needed to
discriminate between inflammation and neoplasia in biopsy and cytology
specimens, to show invasion in those biopsies currently difficult to
interpret, to define disease subgroups with differing natural history or
response to various treatments, and to identify high risk groups for
surveillance. It is hoped that molecular analysis of the disease will
identify new therapeutic targets for more precise and effective
treatments with fewer side effects.

The genes underlying inherited susceptibility to gastric carcinoma have
not yet been identified, and the molecular mechanisms of cancer
promotion by environmental influences, such as diet and Helicobacter
pylori infection, await elucidation. Some progress has been made,
however, in identifying abnormalities of oncogenes, tumour suppressor
genes, and growth factors including their receptors and related gene
products. This will provide a structural framework for analysing
molecular mechanisms of disease development. Comparisons can now be made
between gastric carcinoma and other cancers (particularly colorectal
cancer), between diffuse and intestinal morphological types of gastric
carcinoma, and between early and advanced cases. Many of the problems
which have limited progress are now being resolved.

Which chromosomes are involved?

Early cytogenetic studies were done on malignant effusions from very
advanced cases so it was unsurprising that a large variety of
iiromosomal aberrations were found, differing from case to case. Jiore
recent studies of solid tumour material have also found parked
heterogeneity both between cases and within each case, but two studies,
have reported frequent breakpoints at 3p21, which is the site of
deletions in lung cancer and may involve a phosphatase tumour
suppressor. Rodriguez et al. found translocations and deletions of
llpl3-15 (often abnormal in other solid tumours) in four of five cases,
two involving reciprocal translocations of 3p21. They observed the same
rearrangement in adenocar-cinomas of the lower oesophagus, supporting
the view that these cancers are similar in pathogenesis to gastric
carcinomas, as they usually arise in the metaplastic glandular mucosa of
Barrett's oesophagus. A cytogenetic analysis of nine cases of Barrett's
metaplasia, however, did not find chromosomal abnormalities at either
the llpl3-15 or 3p21 sites.

Chromosomal losses important in the pathogenesis of cancers can be
studied by using DNA polymorphism marker techniques to identify
separately maternal and paternal alleles (comparable with HLA typing).
Patients are informative if their constitutional DNA (from peripheral
blood, or hair roots) shows two patternsone for each allele. Allele
loss is indicated if one of these disappears in tumour tissue, and may
signify loss of a tumour suppressor gene. Studies on gastric carcinomas
were relatively few and analysed only small numbers of tumours, most
finding a low level of deletion. The high proportion of stromal cells in
many gastric cancers, particularly the diffuse type, may have caused a
dilution of tumour DNA and consequent failure to detect allele loss. A
recent study by Sano et al. selected cases comprising over 50% of tumour
cells microscopically assessed on immediately adjacent tissue, and found
a much higher rate of allele loss. Deletion of 5q was found only in well
differentiated cases and not in undifferentiated tumours. The 5q sites
correspond with the sites of the familial adenomatous polyposis (APC)
gene, also deleted in some spontaneous colon carcinomas, and the
adjacent mutated in colon carcinoma (MCC) gene. In contrast, both types
of gastric tumour showed a 60% rate of allele loss on 17p at the site of
the p53 gene. Extrapolating from this, we would then expect to find
mutations of the p53 gene in both cancer types, but 5q gene mutations
only in differentiated carcinomas. Paradoxically, Horii et al. have
found APC gene mutations in only diffuse or poorly differentiated types,
out not in differentiated intestinal types of gastric carcinoma, in a
study of 44 tumours. It will be interesting to see if this curious
paradox is clarified by other studies in progress.

Allele loss has also been reported in the 18q chromosomal site of the
deleted in colon cancer (DCC) gene in a series of intestinal type
gastric carcinomas (diffuse cancers were not examined). 18q deletions
were found in over 60% of informative cases including many early
intramucosal cases, compared to 17p losses (p53 gene) found at a lower
rate and in more advanced cases. Thus the overall pattern of allele loss
in gastric cancers shows many similarities with colon cancer, with
losses at 5q, 18q, and 17p, with 17p loss associated with progression.

Ras oncogenes and proteins

The ras protein p21 is encoded by three ras genes, Ha-ras, K-ras, and
N-ras, and is thought normally to have a role in signal transduction,
proliferation and differentiation. Many tumours have single base
mutations or increased expression of one or more of these genes and
amplification of K-ras has also been reported. Studies of Chinese
populations have shown H-ras mutations in about a third of gastric
carcinomas, associated with presence of distant metastases and short
survival, but other studies of different populations have failed to find
a significant rate of ras mutation. Differences could be caused by the
methodology, or different carcinogenic mechanisms possibly related to
diet or pathogen prevalence, since different carcinogens have been shown
to target different sites on the genome. Studies of ras expression show
raised p21 protein in tumours, using immunohisto-chemistry, Western
blotting, radioimmunoassay, and in situ hybridisation. There have been a
variety of ras antibodies available, however, and many studies do not
identify which antibodies were used, making comparison of conflicting
results difficult. Most report increased immunostaining in intestinal
type gastric carcinomas compared with the diffuse type and in advanced
compared with early cancers. Correlation of ras immunostaining with
depth of invasion, metastases, and worse prognosis has been reported in
a large survey of 171 cases, but not in smaller studies.

Ras overexpression, however, does not appear to be specific for
carcinoma. Ras immunocytochemistry has been evaluated for diagnostic use
on cytology brush smears. While all cancers were positive, so was the
one peptic ulcer, and this technique obviously cannot be used to
discriminate neoplasia from reactive atypia. High levels of ras have
been reported in dysplasia, intestinal metaplasia, regenerating
epithelium adjacent to peptic ulceration, and even in normal mucosa
adjacent to neoplastic lesions. Cerniak et al. observed immunopositivity
in mucosa adjacent to diffuse but not intestinal types of carcinoma, and
also in dysplasia and intestinal metaplasia adjacent to intestinal type
carcinomas. They argued that this supported the view that intestinal
carcinomas arise from intestinal metaplasia or dysplasia, while diffuse
cancers arise de novo from morphologically normal epithelium, with
increased ras expression occurring at an early stage of carcinogenesis
in both. It is not possible, however, to show a temporal sequence from
samples of the same resection specimen. Over expression of ras in the
adjacent mucosa may be in response to a trophic stimulus from the
tumour, and thus may follow rather than precede tumour development. An
interesting study by Yama-moto et al. on 174 gastric carcinomas showed
increased coexpression of ras and TGFa (transforming growth factor a),
correlating with stage, grade, depth of invasion, presence of lymphatic
metastases, and worse prognosis. This raises the possibility that an
important action of ras on gastric cells may be to upregulate production
of TGFa, and differences in ras expression between tumours may reflect
different proliferation rates ralher than different patterns of

P53 tumour suppressor gene

P53 mutations are the most common defect found in human cancer, and
several studies have found a high proportion of abnormalities in gastric
carcinoma by direct DNA analysis and by immunohistochemistry to detect
the mutant protein which is usually not degraded as rapidly as the
normal p53 protein and therefore accumulates in the cell. Abnormalities
have been found in both intestinal and diffuse cases, more commonly in
metastases tnan primary cancers. Advanced cancers have a higher rate of
p53 abnormality than early cancers, and p53 immunopositivity is
associated with poorer survival. One study found that all cases with
mutant p53 were aneuploid, and that no diploid tumour had a p53
mutation, which supports the view that the role of normal p53 is to
prevent cells with damaged DNA from replicating.  P53 abnormalities are
not always late events in gastric carcinogenesis as immunopositivity has
been shown in early cancers, and in dysplasia. Some mutations, such as
deletions causing a frame shift, result in a truncated protein not
detected by immunohisto-chernistry, and certain mutations are only
weakly stabilising. Conversely immunopositivity may occur in the absence
of mutation, possibly caused by increased transcription in rapidly
proliferating cells, inactivation of a factor required for degradation
or another mechanism of non-mutational stabilisation. It is currently
unclear which analysis immunopositivity or DNA mutation gives the more
useful information about biological and clinical behaviour. Further
evaluation of p53 as a prognostic and diagnostic marker for use on
biopsies and cytology specimens is merited.

Growth factors, receptors and related genes

Abnormalities of several growth factor/receptor systems have been found
in gastric carcinoma. Those from the fibroblast growth factor system
seem to be more associated with diffuse carcinomas, whereas intestinal
carcinomas tend to have a higher rate of abnormality of the genes
related to epidermal growth factor and their receptors. The K-sam
oncogene was originally isolated from a cell line from a diffuse
carcinoma, and comprises a rearranged fibro-blast growth factor receptor
gene. (It is also related to another oncogene, ret, found in thyroid
papillary carcinomas). Amplification has been reported in 10 of 48
undifferentiated (diffuse) carcinomas compared with none of 35 well
differentiated (intestinal) carcinomas, thus adding to the evidence that
the two main morphological types of gastric carcinoma have a different
molecular pathogenesis. With the ligands of the fibroblast growth factor
group, the evidence is less clear. The hst-I oncogene encodes fibroblast
growth factor 4, and is often coamplified with the int-2 oncogene
encoding fibroblast growth factor 3, but there are conflicting reports
as to how often they are coamplified in gastric carcinoma. Basic
fibroblast growth factor 2 over expression has been observed in gastric
cancer, more frequently in scirrhous tumours. Thus there is some
evidence that the epithelial stromal interactions observed in the
scirrhous diffuse cancers are at least partially mediated through the
fibroblast growth factor system. Intestinal tumours tend to show a
higher frequency of overexpression of receptors of the epidermal growth
factor system than diffuse or undifferentiated tumours. This has mainly
been shown with the receptors EGFR, ERBB2, and ERBB3, and is sometimes
but not always associated with amplification of the receptor gene. The
situation with the numerous ligands is less clear, and raised levels of
TGFa and epidermal growth factor have been shown in both intestinal and
diffuse carcinomas, although some studies show a higher frequency of
increased expression in tumours of intestinal type. While the more
advanced tumours in most studies have higher levels of expression of
both receptors and ligands than early cancers, there are conflicting
reports as to their prognostic value. Over expression has also been
found in non-neoplastic tissue.

These studies of growth factors and their receptors are of great
interest regardless of any prognostic predictive power. It is believed
that herein may lie a mechanism of cancer promotion by Helicobacter
pylori. There are complex interactions with the trefoil peptidesa group
ofpeptides including pS2 and hSP associated with cells involved in
gastrointestinal inflammatory repair, which may be of significance in
neoplasia as well as inflammation. Both the fibroblast growth factor and
epidermal growth factor systems offer therapeutic targets for inhibition
of tumour growth, and for development of receptor-specific targeting

Differentiated gastric carcinomas: a paradox?

Most human cancers with well differentiated and poorly differentiated
subtypes show a pattern with a loss of differentiation paralleled by
increased aggressiveness, and a higher rate of genetic abnormalities
such as aneuploidy and allele loss. In early gastric cancer (Tl),
however, the raised tubulopapillary Pen A subtype is morphologically
well differentiated but has the worst prognosis, and the highest rate of
aneuploidy, of any early gastric cancer while early gastric carcinomas
of diffuse type are almost always cured by surgery. Advanced intestinal
carcinomas have been shown to have a higher rate of allele loss than
diffuse and poorly differentiated carcinomas at several chromosomal
sites, and are more frequently sneuploid. Yet diffuse cancers have a
worse prognosis when they become advanced. Some of these paradoxical
observations may be the result of artefactual underrepresentation of
tumour cells in poorly differentiated and diffuse cases in some of the
studies, and the inconsistency of histological subtyping between the
various researchers. Nevertheless gastric cancer appears to be an
intriguing model for studying the interactions between differentiation
proliferation, and carcinogenesis in epithelial cells.

Clearly knowledge of the molecular pathogenesis of gastric carcinoma is
at an early stage. Differences have been found between intestinal type
and diffuse carcinomas, and between early and advanced cancers. Some
studies have shown correlations with prognosis. Abnormalities have been
found in dysplasia and intestinal metaplasia in intestinal type cancers,
and in adjacent mucosa in diffuse cancers. At present we cannot
recommend direct application of molecular techniques for routine
diagnostic use, although some areas such as p53 immunohistochemistry
merit further clinical evaluation. We confidently predict, however, an
exciting future for the exploration of this complex and fascinating

1 Williams GT. Early gastric cancer. In: Filipe MI, Jass JR, eds.
Gastric carcinoma. Edinburgh: Churchill Livingstone, 1986: 172-96.

2 Isaacson PG. Biopsy appearances easily mistaken for malignancy in
gastro-intestinal endoscopy. Hktoputhology 1982; 6: 377-89.

3 Sandberg AA. Solid tumours and metastatic cancertumours of the
alimentary tract. In: Sandberg AA, ud. The chromosomes in human cancer
and leukaemia. New York: Elsevier, 1980:468-84.

4 Ochi H, Douglass HO, Sandberg AA. Cytogenetic studies in primary
gastric cancer. Cancer Genet Cytogenet 1986; 22: 295-307.

5 Rodriguez E, Rao PH, Ladanyi M, et al. llpl3-15 is a specific region
of chromosome rearrangement in gastric and esophageal adenocarcinomas.
Cancer Res 1990; 50: 6410-16.

6 Garewal HS, Sampliner R, Liu Y, Trent JM. Chromosomal rearrangements
in Barren's oesophagus: a premalignant lesion of esophageal
adenocarcinoma. Cancer Genet Cvtogenel 1989; 42:281-96.

7 Sano T, Tsujino T, Yoshida K, el al. Frequent loss of heterozygosity
on chromosomes lq)5q, and 17p in human gastric carcinomas. Cancer Res
1991; 51: 2926-31.

8 Horii A, Nakatsuru S, Miyoshi Y, el al. The APC gene responsible for
Familial Adenoma-tous Polyposis is mutated in human gastric cancer.
Cancer Res 1992; 52: 3231-3.

9 Uchino S, Tsuda H, Noguchi M, el al. Frequent loss of heterozygosity
at the DCC locus in gastric cancer. Cancer Res 1992; 52: 3099-102.

10 Deng G, Lui X, Wang J. Correlation of mutations ofoncogene c-Ha-ras
at codon 12 with metastases and survival of gastric cancer patients.
Oncogene Res 1991; 6: 33-8.

11 Fujida K, Ohuchi N, Yao T, et al. Frequent overexpression, but not
activation by point mutation ofras genes in primary human gastric
cancers. Gastroenterology 1987; 93: 1339-45.

12 JiangW,Kahn SM,GuillemJG,Lu SH,WeinsteinlB. Rapid detection of
rasoncogenes in human tumours: applications to colon, oesophageal and
gastric cancer. Oncogene 1989; 4: 923-8.

13 Victor T, Du Toil R, Jordaan AM, Bester AJ, van Helden PU. No
evidence for point mutations in codons 12,13 and 61 of the ras gene in a
high incidence area for esophageal and gastric cancers. Cancer Res 1990;
50: 4911-14.

14 Yoshida K, Hamatani K, Koide H, et al. Preparation of anti-ras Mr
21,000 protein monoclonal antibodies and immunohistochemical analyses on
expression of ras genes in human stomach and thyroid cancers. Cancer Res
1988; 48: 5503-9.

15 Tahara E, Yasui W, Taniyama K, et al. Ha-ras oncogene product in
human gastric carcinoma: correlation with invasiveness, metastasis, or
prognosis, fpn f Cancer Res 1986; 77:517-22.


16 de Blasi F, del Sal G, Hand Pll. Evidence of enhancement of the ras
oncogene protein product (p21) in a spectruin of human tumours. BrJ
Cancer 1989; 43: 431-5.

17 Ohuchi N, Hand PH, Merio G, et al. Enhanced expression of c-Ha-ras
p21 in human stomach adenocarcinomas defined by immunoassays using
monoclonal antibodies and in situ hybridisation. Cancer Res 1987; 47:

18 Yoshida K, Hamatani K, Koide H, et al. Analysis ofras gene expression
in stomach cancer by anti-ras p21 monoclonal antibodies. Cancer Detect
Prev 1988; 12: 369-76.

19 Czerniak B, Herz F, Gorczyca W, Koss L. Expression ofras oncogene p21
protein in early gastric carcinoma and adjacent gastric epithelia.
Cancer 1989; 64: 1467-73.

20 Czerniak B, Herz F, Koss LG, Schlom J. Ras oncogene p21 as a tumour
marker in the cytodiagnosis of gastric and coionic carcinomas. Cancer
1987; 60: 2432-6.

21 Yamamolo T, Hattori T, Tahara E. Interaction between transforming
growth factor alpha and c-Ha-ras p21 in progression of human gastric
carcinuma. Pathol Res Praet 1988; 183: 663-9.

22 Tamura G, Kihana T, Nomura K, el al. Detection of frequent p53
mutations in primary gastric cancer by cell sorting and polymerase chain
reaction single-strand conformation analysis. Cancer Res 1991; 51:

23 Martin HM, Filipe MI, Morris RW, et al. P53 expression and prognosis
in gastric carcinoma, 1m J Cancer 1992; 50: 859-62.

24 Kim JH, Takahashi T, Chiba I, et al. Occurrence of p53 abnormalities
in gastric carcinoma tumours and cell lines, f Natl Cancer Insi 1991;
83: 938^3.

25 Yamada Y, Yoshida T, Hayashi J, et al. P53 gene mutations in gastric
cancer metastases and in gastric cancer cell lines derived from
metastases. Cancer Res 1991; 51: 5800-5.

26 Kastan MB, Onyekwere 0, Sidransky D, Vogelstein B, Craig RW.
Participation of p53 protein in the cellular response to DNA damage.
Cancer Res 1991; 51: 6304II.

27 Brito MJ, Filipe MI, I-ane D, Morris R. P53 overexpression in early
gastric carcinoma and precancerous lesions. J Pathol 1992; 167: 95A.

28 Wynford-Thomas D. P53 in tumour pathology: can we trust
immunocytochemistry? 3 Pathol 1992; 166: 331-2.

29 Nakatani H) Sakamoto H, Yoshida T, ef al. Isolation of an amplified
DNA sequence in stomach cancer. Jpn J Cancer Res 1990; 81: 707-10.

30 Yoshida MC, Wada M, Satoh H, et al. Human HSTI (HSTFI) gene maps to
chromosome band llql3 and co-amplifies with the INT2 gene in human
cancer. Proc Nail Acad Sci USA 1988; 85:4861-4.

31 Tsuda T) Tahara E, Kajiyarna G, et al. High incidence of
co-amplification ofhst-I and int-2 genes in human esophageal carcinomas.
Cancer Res 1989; 49: 5505-8.

32 Tamimoto H, Yoshida K, Yokozaki H, el al. Expression of basic
fibroblast growth factor in human gastric carcinomas. Virchows Arch [B]
cell Pathol 1991; 61: 263-7.

33 Lemoine NR) Jain S, Silvestre F, et al. Amplification and
overexpression of the EGF receptor and c-erbB2 protooncogenes in human
stomach cancer. BrJ Cancer 1991; 64: 79-83.

34 Faick VG, Gullick WJ. c-erbB2 oncogene product staining in gastric
adenocarcinoma. An immunohistochemical study.. f Pathol 1989; 159:

35 Lemoine NR) Prigent SA) Hughes CM) et al. Expression of c-erbB-3
protein in normal human adult and letal tissues. J Pathol 1992; 167:

36 Muller W, Borchard F. Expression of transforming growth factor alpha
in gastric carcinoma and normal gastric mucosa cells. Cancer 1992; 69:

37 Hirayama D) Takahiro F) Satonaka K) et al. Immunohistochemical study
of epidermal growth factor and transforming growth factor p in the
penetrating type of early gastric cancer. Hum Pathol 1992; 23: 681-5.

38 Nasirn MM) Thomas DM) Alison MR) Filipe MI. Transforming growth
factor a expression in normal gastric mucosa, intestinal metaplasia)
dysplasia, and gastric carcinomaan immunohistochemical study.
Histopathology 1992; 20: 339-43.

39 Yasui W, Hata J, Yokozaki H) et al. Interaction between epidermal
growth factor and its receptor in progression of human gastric
carcinoma. Int f Cancer 1988; 41: 211-17.

40 Wright NA, Poulsom R, Stamp GWH, et al. Epidermal growth factor
(EGF/URO) induces expression of regulatory peptides in damaged human
gastrointestinal tissues. J Pathol 1990; 162:279-84.

41 Theisinger B, Welter C, Seitz G) et al. Expression of the breast
cancer associated gene pS2 and the pancreatic spasmolytic polypeptide
gene (hSP) in diffuse type of stomach carcinoma. F.urJ Cancer 1991; 27:

42 Wright PA) Lemoine NR. The molecular basis of gastric cancer. In:
Lemoine NR) ed. Cancer: a molecular approach. Oxford: Blackwell
Scientific 1993 (in press).