Expert Review of Molecular Diagnostics

ISSN: 1473-7159 (Print) 1744-8352 (Online) Journal homepage: http://www.tandfonline.com/loi/iero20

Molecular classification of gastric cancer

Christoph Röcken

To cite this article: Christoph Röcken (2017) Molecular classification of gastric cancer, Expert
Review of Molecular Diagnostics, 17:3, 293-301, DOI: 10.1080/14737159.2017.1286985

To link to this article: http://dx.doi.org/10.1080/14737159.2017.1286985

Accepted author version posted online: 24

Jan 2017.
Published online: 06 Feb 2017.

Submit your article to this journal

Article views: 183

View related articles

View Crossmark data

Citing articles: 2 View citing articles

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=iero20

Download by: [La Trobe University] Date: 01 October 2017, At: 13:01
 EXPERT REVIEW OF MOLECULAR DIAGNOSTICS, 2017
VOL. 17, NO. 3, 293–301
http://dx.doi.org/10.1080/14737159.2017.1286985

REVIEW

Molecular classification of gastric cancer

Christoph Röcken
Department of Pathology, Christian-Albrechts-University, Kiel, Germany

ABSTRACT ARTICLE HISTORY

Introduction: Gastric cancer is among the most common cancers worldwide. Despite declining inci- Received 27 November 2016
dences, the prognosis remains dismal in Western countries and is better in Asian countries with national Accepted 23 January 2017
cancer screening programs. Complete endoscopic or surgical resection of the primary tumor with or KEYWORDS
without lymphadenectomy offers the only chance of cure in the early stage of the disease. Survival of Gastric cancer; molecular
more locally advanced gastric cancers was improved by the introduction of perioperative, adjuvant and classification; histology;
palliative chemotherapy. However, the identification and usage of novel predictive and diagnostic Epstein-Barr virus;
targets is urgently needed. microsatellite instability;
Areas covered: Recent comprehensive molecular profiling of gastric cancer proposed four molecular HER2
subtypes, i.e. Epstein-Barr virus-associated, microsatellite instable, chromosomal instable and genomi-
Downloaded by [La Trobe University] at 13:01 01 October 2017

cally stable carcinomas. The new molecular classification will spur clinical trials exploring novel targeted
therapeutics. This review summarizes recent advancements of the molecular classification, and based
on that, putative pitfalls for the development of tissue-based companion diagnostics, i.e. prevalence of
actionable targets and therapeutic efficacy, tumor heterogeneity and tumor evolution, impact of
ethnicity on gastric cancer biology, and standards of care in the East and West.
Expert commentary: The overall low prevalence of actionable targets and tumor heterogeneity are the
two main obstacles of precision medicine for gastric cancer.

1. Introduction sporadic GC: a chronic atrophic gastritis leads to intestinal

metaplasia, dysplasia, and finally the occurrence of GC [2]. In
Gastric cancer (GC) is the fifth most common cancer world-
his initial model, a diet rich in salt, and medication were
wide with an estimated 952,000 new cases (7% of total cancer
considered as driving forces of chronic gastritis. Later on,
incidence) and 723,000 deaths in 2012 [1]. The incidence and
after Barry J. Marshall and J. Robin Warren had identified H.
prevalence of GC vary greatly geographically. The highest
pylori as a cause of chronic bacterial (type B) gastritis, Correa
rates are observed in East Asia, Central Europe, and Eastern
modified his model and added H. pylori as causative agent of a
Europe accounting collectively for 87% of all new GC cases in
chronic inflammation [3]. Since then ample epidemiological,
the world. Significant lower rates are found in Africa and North
animal experimental, and therapeutic evidence support the
America [1]. Worldwide, the standardized incidence rates of
contention that chronic H. pylori-induced gastritis leads to
noncardia GC have been declining steadily over the last 30–
GC formation. In 1994, H. pylori was classified as group 1
50 years. To the contrary, cancer of the proximal stomach/
human carcinogen by the International Agency on Cancer.
cardia and esophago-gastric junction has been stable or even
Chronic colonization of the stomach mucosa by H. pylori may
increased in its incidence. Patient prognosis remains dismal in
lead to different phenotypes of chronic gastritis. The antrum-
countries without GC screening programs. GC is a disease of
predominant form is associated with ulcers of the stomach
the elderly often reaching the peak incidence in the seventh
and duodenal mucosa without an increased risk of cancer
and eighth decade of life. Five-year survival rate then ranges
development. A corpus-predominant gastritis carries a sub-
between 30–35% for men and women. Particularly, in low-
stantial risk for the development of GC through mucosal
incidence regions the disease is often diagnosed at an
atrophy, hypacidia, and intestinal and pseudopyloric metapla-
advanced stage limiting the therapeutic options. Prognosis
sia [4]. The indolent form of H. pylori-associated gastritis pre-
was improved in countries with high incidence rates and
sents neither with ulcers nor with an increased risk of GC
national cancer screening programs, e.g. Japan and Korea,
formation. Several bacterial virulence factors for colonization,
leading to diagnosis of GC in earlier disease stages.
persistence, ulcer, and cancer risk have been identified in
recent years, e.g. CagA, DupA, IceA1, NapA, oipA, and VacA
2. Etiology [5]. These are modulated by host factors, e.g. polymorphisms
in genes coding for pro- and anti-inflammatory cytokines, as
GC may occur sporadically, familial, or as a hereditary disease.
well as environmental and socioeconomic cues. The individual
In 1988, Correa described a model for the carcinogenesis of

CONTACT Christoph Röcken christoph.roecken@uksh.de Department of Pathology, Christian-Albrechts-University, Arnold-Heller-Str. 3, Haus 14, D-24105
Kiel, Germany
© 2017 Informa UK Limited, trading as Taylor & Francis Group
 294 C. RÖCKEN
disease manifestation and cancer risk is a result of a complex
interaction between host and microbiotic traits [6].
The familial forms of GC may share risk factors, such as H.
pylori-infection, nutritional habits, and gene polymorphisms,
e.g. in pro- and anti-inflammatory cytokine genes.
Approximately 5–10% of the patients with GC carry germ
line mutations and are truly hereditary forms [7,8]. A total of
30–40% of the cases with a diffuse-type hereditary GC harbor
mutations or copy number changes in the CDH1 gene, which
codes for E-cadherin. More than 50 different mutations of the
CDH1 gene have been described hitherto. The mutations are
found across the entire gene. α-E-catenin (CTNNA1) was also
linked to diffuse type hereditary GC and germline mutations,
including nonsense and frameshift mutations, were found in
three families [9,10]. Other genes possibly associated with
hereditary GC are insulin receptor (INSR), F-Box-protein 24
(FBXO24), mitogen-activated protein kinase kinase kinase 6
(MAP3K6), protease serine 1 (PRSS1), and DOT1-like histone
Downloaded by [La Trobe University] at 13:01 01 October 2017
H3K79 methyltransferase (DOT1L), and require verification
[10,11]. Recently, Worthley et al. described the first kindred
with an intestinal type hereditary GC syndrome associated
with proximal polyposis [12]. The underlying mutation was
found in the APC Promoter 1B [13]. GC can also be part of
other hereditary cancer syndromes such as familial adenoma-
tous polyposis coli (APC), Lynch syndrome (hMLH1, hMLH2),
Cowden syndrome (PTEN), juvenile polyposis, Li-Fraumeni syn-
drome (TP53), MUTYH-associated adenomatous polyposis
(MUTYH), Peutz-Jeghers syndrome (STK11), and hereditary
breast and ovarian cancer (BRCA1/2) [14].
3. Histology of GC
The World Health Organization (WHO) published in 2010 a
new classification system for malignant tumors of the gastro-
intestinal tract [15]. Several modifications were introduced for
GC. The adenocarcinomas were subclassified into tubular,
papillary, mucinous, poorly cohesive (including the signet
ring cell variant), and mixed type. The carcinoma with lym-
phoid stroma (medullary carcinoma) and the hepatoid adeno-
carcinoma were added. While no recommendation was given
for ancient classification schemes [15], classification according
to Laurén is still widely used in many clinical trials and GC
research publications [16]. According to Laurén, GC is classi-
fied into intestinal, diffuse, mixed, and unclassified [16]. The
vast majority (>90%) of GCs are adenocarcinomas. Recent
comprehensive molecular characterization studies were car-
ried out with gastric adenocarcinomas and utilized the
Laurén classification [17]. Thus, the morpho-molecular classifi-
cation of GC matches only rudimentary with the current WHO
classification [18].
4. Molecular subtypes of GC
The genetic complexity of GC has been recently shown in an
integrative genomic analysis including array-based somatic
copy number analysis, whole-exome sequencing, array-based
DNA methylation profiling, messenger RNA sequencing,
microRNA sequencing, and reverse-phase protein array analy-
sis [18]. A molecular classification was proposed, which
categorizes four subtypes, i.e. microsatellite instable (MSI),
Epstein-Barr-virus (EBV)-positive, chromosomal instable, and
genomically stable GCs [17,18]. These studies for the first
time build a bridge between distinct molecular subtypes of
GC and some phenotypes. It may form the basis for the
development of novel therapeutic strategies [17,18]. The chro-
mosomal instable type often shows an intestinal phenotype
according to Laurén, frequently harbors mutations in the P53-
tumor suppressor gene as well as activating mutations in
genes coding for receptor tyrosine kinases. The genomically
stable GC often is of a diffuse type, with mutations occurring
in CDH1 and RHOA. They also show alterations in cell adhesion
molecules. The EBV-associated variant and microsatellite
instable GCs (MSI-GCs) usually present with an unusual phe-
notype [18,19].
Based on the molecular classification of GC one might raise
the question as which genotype is associated with which
phenotype (=morpho-molecular classification of GC).
4.1. Microsatellite instable gastric cancer (MSI-GC)
MSI-GCs are hypermutated tumors carrying a CpG-island
methylator phenotype (CIMP). They usually demonstrate silen-
cing of MLH1, which encodes a DNA-mismatch repair protein.
In the literature, the prevalence of MSI-GC ranges from 0% to
44.5% [20]. MSI-GCs are significantly more prevalent in elderly
patients, the distal stomach and are associated with a signifi-
cantly lower number of lymph node metastases and a signifi-
cantly better overall and tumor-specific survival. MSI-GC lack
BRAF-mutations commonly found in sporadic MSI colorectal
cancer [18]. In fact, BRAF-mutations are virtually absent in
GC [21].
In our own GC cohort we found 34 (7.5%) patients with
MSI-GC [20]. A total of 21 (6.2%) patients with MSI-GC showed
unusual histological features. The tumors consisted predomi-
nantly of highly pleomorphic tumor cells with large vesicular
nuclei in a trabecular, nested, microalveolar, or solid growth
pattern. Tumor cell size was variable, sometimes exhibiting
tumor cells that had lymphocytoid or blastoid appearance.
An abundant tumor-associated inflammatory stroma consist-
ing of either polymorphs and/or lymphocytes was frequently
observed in these cases, usually with little or no desmoplastic
stroma. Thirteen patients demonstrated several of these his-
tological features, whereas eight had only few discriminating
features mainly consisting of pushing margins and/or a lym-
phocytes or polymorphs. In three cases the inflammatory
component was the only unusual feature [20]. Interestingly,
the tumors of 13 (3.8%) patients showed no distinctive histo-
logical features and were indistinguishable from any of the
microsatellite stable GCs. We found no pure poorly cohesive,
signet ring cell carcinomas in the MSH-GC group even though
signet ring cells can be part of a mixed pattern in MSI-GC.
The accurate classification of MSI-GC may become clinically
relevant for two reasons: (1) MSI-GCs may not require any stan-
dard adjuvant (radio-)chemotherapy in a curative setting [20]. (2)
MSI-GCs express the immune checkpoint molecules PD-L1 and
PD-1 and may be considered suitable for the treatment with
immune checkpoint inhibitors in the palliative setting [18,22].
The classification and identification of MSI-GCs can be done

with antibodies directed against MLH1, PMS2, MSH2, and MSH6
as well as usage of the mononucleotide markers BAT-25, BAT-26,
NR-21, NR-24, and NR-27 [20]. However, there is currently no
international consensus on the diagnosis of MSI-GCs and our
group adapted the procedure utilized for colorectal cancer [20].
Tumor heterogeneity is an issue in MSI-GCs. We identified
a single case which showed a biphasic MSH2-expression
pattern. Staining of the whole tissue slides unexpectedly
demonstrated a loss of MSH2 in the major part (between
77% and 95%) of the tumor whereas a small part (min. 5%,
max. 23%) expressed MSH2. Subsequent molecular biologi-
cal studies of the microdissected tumor areas confirmed
microsatellite stability in the MSH2-positive and MSI in the
MSH2-negative area. Furthermore, analyses of the lymph
node metastases confirmed that only the microsatellite
stable tumor subclone had metastasized. Interindividual
and intratumoral heterogeneity of MSI-GCs may significantly
impact on clinical decision-making. Histological phenotype
Downloaded by [La Trobe University] at 13:01 01 October 2017
may raise suspicion in the majority of cases. However, one
third of the tumors fail to exhibit any specific phenotype
and MSI-testing may need a more general approach. The
vast majority of MSI-GC demonstrates loss of MLH1-expres-
sion and immunostaining for MLH1 may be considered as
the most cost-effective approach to screen for MSI [23].
However, it has to be kept in mind that other members of
the DNA-repair machinery can be lost and that in patients
with lymph node metastases, testing the primary tumor and
lymph nodes metastases might help to unravel intratumoral
heterogeneity.
4.2. Epstein-bar-virus-associated gastric cancer
(EBVaGC)
EBVaGC is another molecular subtype of GC [17,18]. EBVaGC
also show a CIMP-phenotype, frequently harbor PIK3CA- and
ARID1A-mutations and dysregulation of immune cell signal-
ing. EBVaGC are characterized by male predominance, occur-
rence primarily in the proximal and postgastrectomy
remnant stomach, multiplicity, and association with lympho-
cytic infiltration [24]. The number of lymph node metastases
was negatively associated with EBV status [25], and in Asian
populations EBVaGc may exert a better prognosis compared
with EBV-negative GCs [26]. EBVaGC and MSI-GC are
mutually exclusive [20]. The phenotype of EBVaGC is variable
with tubular and intestinal differentiation. It may exert an
undifferentiated phenotype (lymphoepithelioma-like or
medullary). Any GC with an unusual phenotype should be
forwarded to EBV-testing using EBER-in situ-hybridization,
which gives a nuclear signal. Immunostaining is less sensi-
tive, because GC is commonly of latency type 1 and hence
immunonegative for EBNA2, LMP1, and ZEBRA [24,27]. Given
the wide morphological spectrum of EBVaGC, testing should
be done at leisure [28]. In addition, all EBVaGCs harbor
promotor hypermethylation of CDKN2A (p16INK4A) and immu-
nostaining of p16 may serve as a putative surrogate marker
for EBVaGC [18].
Similar to MSI-GC, EBVaGCs significantly more commonly
express PD-L1 and PD-1 and might thus be sensitive to therapy
with immune checkpoint inhibitors [22]. Thus recognition and
EXPERT REVIEW OF MOLECULAR DIAGNOSTICS 295
diagnosis of EBVaGC may become clinically relevant in the near
future.
4.3. Chromosomal instable GCs
The chromosomal instable GCs often show an intestinal
phenotype according to Laurén, frequently harbor mutations
in the P53-tumor suppressor gene and activating mutations
in genes coding for receptor tyrosine kinases [18].
Approximately 37% of all GCs harbor amplifications of
genes coding for receptor tyrosine kinases such as epider-
mal growth factor receptor (EGFR), FGFR2, HER2, and MET
[19,29]. EGFR-amplified GCs present with a male preponder-
ance and affect more commonly the distal stomach [28].
HER2-amplified GCs are equally distributed among men
and women, and tend to be more prevalent in the proximal
stomach. HER2-amplification does not correlate with patient
prognosis [30,31]. MET-amplified GCs are found more com-
monly in men and proximal GCs. They have significantly
more lymph node and distant metastases, and are charac-
terized by a very poor prognosis with a median overall
survival of less than 6 months [32,33].
FGFR2 is another receptor tyrosine kinase, which is ampli-
fied in approximately 7–10% of GCs [19,33]. Up to now, no
association was found between FGFR2 amplification and
patient gender, anatomical site, histology, or TNM classifica-
tion [19,33]. However, there has been some evidence that
FGFR2-amplified GCs may exert poor survival outcome [19].
Amplification of FGFR2 is not restricted to chromosomal
instable GCs and can also be observed in genomically stable
GCs. This might explain why no association was found
between tumor type and FGFR2 status [33].
There is currently no uniform approach for the classification
of chromosomal instable GCs. Immunostaining of p53 and
detection of receptor tyrosine kinase expression and amplifi-
cation have been proposed recently as surrogate markers
[23,28]. However, in view of the observations made with
FGFR2, other markers might be needed.
4.4. Genomically stable GCs
Genomically stable GCs are supposed to have primarily a
diffuse histological phenotype according to Laurén, harbor
CDH1- and RHOA mutations as well as rearrangements
between CLDN18 and ARHGAP26 or ARHGAP6 [18,34].
Claudin 18 is a component of the tight junctions and its
down regulation was associated with impaired survival [35].
Elevated expression of cell adhesion pathways, including the
B1/B3 integrins, syndecan-1-mediated signaling, and angio-
genesis-related pathways are also associated with the geno-
mically stable genotype. However, our own validation
studies demonstrated that RHOA mutations also occur in
intestinal type GCs [36] and the integrins αvβ3 and αvβ5
are expressed significantly more often in the intestinal phe-
notype [37]. The expression of αvβ5 on stroma cells was
found to be an independent prognostic factor of intestinal-
type GC [37]. Thus, while comprehensive molecular charac-
terization of GC has generated highly informative data, vali-
dation studies are needed to make the findings more robust.
 296 C. RÖCKEN

The morpho-molecular classification of GC
cumbersome.
5. Translational aspects
Molecular classification raises hopes not only for an improved
morpho-molecular classification of GC but also for novel treat-
ment options (see Table 1). This applies for the prognostica-
tion of patient outcome and for the biomarker-based
prediction of therapeutic response (companion diagnostics).
With regard to prognosis, there has been some evidence in
colorectal cancers that stage II MSI colorectal cancers may not
benefit from therapy with 5-fluoruracil [38,39]. This may also
apply to GC. It is a disease of the elderly, often presenting with
co-morbidities, and avoiding perioperative/neoadjuvant or
adjuvant oncological treatment might be an option, worth to
consider prospectively in MSI-GC.
With regard to the other three molecular subtypes, i.e.
Downloaded by [La Trobe University] at 13:01 01 October 2017
remains interindividual variability of mutational load [40]. Only P53-
mutations are highly prevalent in GC. The vast majority of
mutations is in the range of <10%, which supports the con-
tention that GC is a complex disease. A low prevalence of
actionable targets complicates clinical trials. Pharmaceutical
companies consider issues of financing and re-imbursement.
Principle investigators need to consider patient numbers for
statistical analyses. Basket trials with upfront molecular testing
can help to overcome these limitations. However, they may
need to focus on GC rather than being part of diverse other
solid organ tumors under investigation, and each actionable
target requires verification for each individual tumor type. The
EXPAND study, which tested capecitabine and cisplatin with or
without cetuximab for patients with previously untreated
advanced GC, nicely illustrated the obstacles: adding cetuxi-
mab to capecitabine–cisplatin as first-line treatment did not
result in any benefit [41]. Cetuximab is a fully humanized
antibody directed against the EGFR and provides significant

EBVaGC, chromosomal instable, and genomically stable GCs, survival benefit to patients with advanced colorectal cancer
currently there is not sufficient evidence that they might be with wild-type KRAS status [42]. As in comparison with color-
suitable for risk stratification [18]. However, single genetic ectal cancer, KRAS mutations are rare in GC while they do
alterations have been associated with poor patient outcome, express EGFR. Therefore, it was somewhat reasonable to test
e.g. MET-amplification, and might outperform grouping efficacy and safety of cetuximab in KRAS-wildtype GCs.
according to one of the three other aforementioned molecular However, cetuximab failed to show any benefit and the rela-
subtypes. tive success in colorectal cancer could not be recapitulated in
With regard to predictive biomarkers, recent in-depth GC [41]. These data provide evidence that the presence of an
molecular analyses of GC unraveled many actionable targets actionable target per se does not guarantee therapeutic effi-
and raise hope that these offer new chances for targeted cacy in a given tumor/patient.
therapy of GC [17,18] (see Table 1). However, a few issues
merit consideration. 5.2. Tumor heterogeneity and tumor evolution
Our own recent validation studies showed that the different
5.1. Low prevalence of actionable targets and molecular subtypes overlap. GC may harbor microsatellite
therapeutic efficacy stable and instable areas adjacent to each other [20]. RHOA
mutations are found in intestinal type GCs [36]. In addition,
The subgrouping into four molecular subtypes cannot mask
the amplification of HER2 and MET may occur in the same
the problem of an overall low prevalence of the individual
primary tumor albeit in different areas [32]. These observa-
genetic alterations. GC belongs to the group of cancers with a
tions highlight a specific issue of GC, i.e. intratumoral hetero-
high frequency of somatic mutations as well as a substantial
geneity. Approximately 30% of all GCs are heterogeneous
and may compromise morpho-molecular classification of GC
Table 1. Comprehensive molecular characterization of gastric cancer classified and hence selection of an appropriate targeted therapy.
four molecular subtypes, which harbor several interesting actionable targets and Tumor heterogeneity is becoming one of the main obstacles
treatment options.
of cancer treatment in the era of precision medicine.
Molecular subtype Target Treatment/drug
Sequencing of multiple samples of primary clear cell renal
MSI No adjuvant chemotherapy?
MSI/EBV-associated PD-L1/PD-1 Nivolumab, Pembrolizumab, cell carcinoma and its metastatic sites revealed the complex-
Atezolizumab, ity of intraindividual cancer genetics, and allowed reconstruc-
Durvalumab, tion of its evolutionary history [43,44]. Genetically distinct
Avelumab,
Ipilimumab subclones were found in the primary tumor and distant
EBV-associated PIK3CA/AKT/mTOR PX-866, NVP-BYL719, metastases. Cancer is now viewed as a highly dynamic evolu-
MLN1117, NVP-BEZ236, tionary disease [45], with continuing mutations favoring the
Temsirolimus, Everolimus,
Ridaforolimus emergence of new (sub)clones with distinct biological prop-
Chromosomal instable EGFR Cetuximab, Panitumumab erties already in precursor lesions [46]. Subclones may show
HER2 Trastuzumab, Pertuzumab different patterns of interaction; they may compete, overtake
MET Onartuzumab, Rilotumumab,
Crizotinib, Foretinib, other clones, parasitize, or peacefully coexist. A ‘trunk and
Tivantinib branch’ model has been proposed for tumor evolution [47].
VEGFA Bevacizumab Common events, found in every subclone of the tumor
VEGFR2 Ramucirumab
Genomically stable Integrins Cilengetide region, are represented in the trunk. Heterogeneous somatic
EpCam Catumaxumab events occurring in a limited number of (or even a single)
MSI: microsatellite instable; EBV: Epstein-Barr-virus subclone(s) represent the branches of the tree [47]. While
 EXPERT REVIEW OF MOLECULAR DIAGNOSTICS 297

targeting alterations in the trunk might be most promising as

they occur in all tumor cells, many actionable targets exclu-
sively occur in branches. This subclonal heterogeneity of
actionable targets poses problems in diagnosis (tissue sam-
pling error) and therapy (emergence of resistance).
Treatment and eradication of a subclone may provide a
growth advantage for a competing non-responsive subclone,
which finally may lead to patient death. This is probably why
the efficacy of several targeted therapies was limited in time
and also somewhat unpredictable [47]. Intratumoral molecu-
lar heterogeneity appears to be inherent to tumor evolution
and has also consequences for sampling procedures. We
have studied the impact of tissue sampling on Her2/neu
testing in GC. Her2/neu has been introduced as a predictive
biomarker for treatment of GC with trastuzumab [48].
Amplification of genes encoding receptor-tyrosine kinases
usually occurs in genomically instable GCs [18], which is
probably why in GC Her2/neu expression might be intrinsi-
Downloaded by [La Trobe University] at 13:01 01 October 2017

cally heterogeneous (Figure 1). We assessed Her2/neu status

(according to the GC scoring system [49]) using a tissue
micro array approach, in which five tissue cores per paraffin
‘donor’ block served as surrogate for a biopsy procedure, and
compared the results with those obtained on whole tissue
sections cut from the same paraffin block. On the TMA cores
we obtained a ‘false-negative’ rate of 24% and a ‘false-posi-
tive’ rate of 3% [31]. Silva et al. [33] made similar observa-
tions when they explored gene copy number status of
receptor tyrosine kinase and downstream signaling genes in
primary GC and matched lymph node metastases. A total of
68% of their 103 patients with results from primary GC and
matched lymph node metastasis showed a conversion of the
copy number status: they found a significant intraindividual
difference for KRAS, MYC, CCNE, HER2, and FGFR2 [33].
Heterogeneous expression of predictive biomarkers poses a
major challenge in clinical trials but this issue is often neglected.
The phenomenon is often first explored in retrospective bio-
marker studies, once a treatment modality has been formally
approved by a regulatory body. As an example, many studies
on Her2/neu expression were carried out after approval of
trastuzumab for the treatment of GC (for a review see [50]).
These issues should be addressed prior to implementation of a
trial and whether (expression of) the molecular target of the
treatment is homogeneous (‘trunk-alteration’) or heteroge-
Figure 1. Morpho-molecular classification of gastric cancer. A molecular
neous (‘branch-alteration’) should be explored in advance and classification was proposed, which categorizes four subtypes, i.e. microsatel-
translated into suitable biopsy procedures [51]. However, many lite instable (a, b), Epstein-Barr-virus-positive (c, d), chromosomal instable (d-
studies address treatment in later stages of cancer with pallia- g), and genomically stable gastric cancers (h). This microsatellite instable
gastric cancer has a solid growth pattern (a) and a loss of MLH1-expression
tive intent and analyses rely on biopsy samples only, as patients (b). MLH1-immunopositive tumor-infiltrating lymphocytes serve as internal
may not be eligible for tumor resection. However, an essential positive control (b). Epstein-Barr-virus-positive cancers (c) often yield a
requirement for any trial is the exploration of tumor heteroge- lymphocytic infiltration (lymphoepithelioma-like) and nuclear staining after
EBER-in situ hybridization (d). Chromosomal instable gastric cancers often
neity before the trial takes off, to avoid sampling errors and have an intestinal phenotype according to Laurén (e) and activating muta-
favor representative tissue biopsies [51]. In this respect it might tions in genes coding for receptor tyrosine kinases, such as Her2/neu (f).
not be surprising that results of recent studies communicated However, Her2/neu-overexpression (f) and HER2-gene amplification (g) can
be heterogeneous, with positive tumor areas (arrow heads) and negative
during the annual meeting of the American Society of Clinical areas (arrows) within the same primary tumor. Genomically stable gastric
Oncologists failed to show a benefit of MET inhibitors in GC. In cancers are often of a diffuse phenotype according to Laurén (h).
both studies, drug efficacy was tested along with a potential Hematoxylin and eosin (a, c, e, h); anti-MLH1-immunostaining (b); EBER-in
situ hybridization (d); Her2/neu-immunostaining (d); HER2-in situ hybridiza-
companion diagnostic [52,53]. In fact, apart from HER2, no other tion (g). Original magnifications 400-fold.
 298 C. RÖCKEN
companion diagnostic reached clinical application in GC, so far.
Ramucirumab, which targets VEGF-receptor 2, does not require
companion diagnostics.
Based on the observations made for Her2/neu, an interdis-
ciplinary group of German experts has taken on the challenges
of Her2/neu testing in GC and recommended that at least five
tumor-bearing biopsies should be obtained from different
sites of the primary tumor [54]. Similarly, the College of
American Pathologists, the American Society of Clinical
Pathology and the American Society of Clinical Oncology
recently published recommendations for HER2 testing [55].
Testing of multiple biopsy fragments (from a primary or meta-
static site), or from the resected primary tumor, is preferred.
For biopsy specimens, current recommendations state that,
when possible, a minimum of five biopsy specimens, and
optimally six to eight, should be obtained to account for
intratumoral heterogeneity and to provide sufficient tumor
specimens for diagnosis and biomarker testing [55].
Downloaded by [La Trobe University] at 13:01 01 October 2017
5.3. Impact of ethnicity on GC biology
Several studies carried out on the U.S. patient populations
have shown that ethnicity impacts on patient survival prob-
ably reflecting a difference in GC tumor biology [56–63]. The
recent comprehensive molecular characterization also
explored the putative impact of ethnicity. The authors did
not observe any systematic differences in distribution of sub-
types between patients of East Asian and Western origin and
they also were unable to identify a strong biologic difference
between tumors of East Asian origin compared with other
tumors [18]. However, patient numbers were low and further
analysis with larger sample cohorts will be required to better
delineate differences that may exist between GCs originating
from different regions of the world and in patients of different
ethnic backgrounds [18]. This may become particularly impor-
tant, when treatment with checkpoint inhibitors is to be
explored. GCs in Asian and non-Asian patient populations
exhibit distinct tumor immunity signatures related to T-cell
function [64]. This may have an effect on the development
of companion diagnostics and therapeutic efficacy [22].
5.4. Standards of care for the treatment of GC
Standards of care vary in different regions of the world,
which may also effect molecular classification and compa-
nion diagnostics of GC. Perioperative and adjuvant chemo
(radio-)therapy of GC has become standard of care in
Western countries [65,66], while in Eastern Asia, adjuvant
chemotherapy is standard of care [67]. In general, adjuvant
chemotherapy offers the chance for an in-depth exploitation
of the resected primary tumor and its corresponding lymph
node metastases prior to (radio-)chemotherapy. Intratumoral
heterogeneity can be explored systematically on a patient-
by-patient basis. Neoadjuvant/perioperative chemotherapy
carries the risk to compromise molecular classification in
two ways. At first, therapy-induced regression may prohibit
accurate molecular classification by reducing overall tumor
burden. Companion diagnostic may need biopsy sampling
of the recurrent tumor in a palliative setting. Second, ther-
apy affects tumor biology. Companion diagnostics explored
in untreated tumors may necessitate validation in neoadju-
vantly treated tumors due to subclone eradiation and sub-
clone selection. This may become particularly relevant for
the use of immune checkpoint inhibitors and their cognate
companion diagnostics.
6. Expert commentary
The recent comprehensive molecular characterization has
advanced our understanding of the complexity and diversity
of GC. Trastuzumab (Her2/neu) [48] and ramucirumab (VEGF-
receptor 2) have been approved as targeted therapy of GC.
Administration of trastuzumab only applies to a small sub-
group of patients with overexpression of Her2/neu and/or
amplification of HER2 [40]. No other companion diagnostic
has reached clinics. Neither EGFR nor MET had been validated
as suitable targets for precision medicine of GC. However, as
shown in Table 1, there are a substantial number of drugs
available, which are directed against a variety of dysregulated
molecules of the four molecular subtypes. The overall low
prevalence of many actionable targets and tumor heteroge-
neity of GC are probably the two main obstacles. Sampling
errors leading to false negative and false positive test results is
a specific risk in GC. Patients with advanced GC may be
considered ineligible for gastrectomy and biopsy samples
might be the only tissue specimen available for genetic testing
and companion diagnostics. The histological phenotype may
not be representative in a biopsy specimen as well as the
diverse genetic alterations described earlier. Recently, expert
panels suggested that at least five tumor-bearing biopsies
should be studied in GC in order to enable adequate HER2-
testing. This may also apply to other putative biomarkers of
GC. Thus, future studies also need to consider the risk of
misinterpretation, and lack of response as well as tumor pro-
gression may require repeat tissue sampling in order to clas-
sify GC correctly. In our own cohort the evidence of tumor
heterogeneity in GC was overwhelming (even including the
Ki67-status [68]) and we discontinued the use of tissue micro
arrays for biomarker analyses.
7. Five-year view
In the following years, further efforts will be spent to improve
(perioperative, adjuvant, and palliative) oncological treatment
of GC. These will include checkpoint inhibitors and they
should explore the necessity of chemotherapy in MSI-GC.
Validation studies hopefully unravel the impact of ethnicity
on the molecular classification of GC and should also spend
major attention on tumor heterogeneity. Ultimately, the
upcoming years should show whether the four-tire molecular
classification stands the test of time or needs revision/
amendment.
Key issues
● Gastric cancer is among the most common cancers world-
wide with a dismal prognosis. It may occur sporadically,

familial or as a hereditary disease. Nutritional habits and
colonization of the stomach by H. pylori are major causes of
gastric cancer.
● Histological phenotypes of gastric cancer according to the
World Health Organization-criteria are the tubular, papillary,
mucinous, poorly cohesive, mixed type, medullary and
hepatoid carcinoma. According to Laurén, gastric cancer is
classified into intestinal, diffuse, mixed and unclassified.
● A recent molecular classification proposed four subtypes,
i.e. microsatellite instable, Epstein-Barr-virus-positive, chro-
mosomal instable, and genomically stable gastric cancers.
● Microsatellite instable and Epstein-Barr-virus associated
gastric cancers significantly more commonly express PD-
L1 and PD-1 and might thus be sensitive to therapy with
immune checkpoint inhibitors.
● Chromosomal instable gastric cancers frequently harbor
P53-mutations and activating mutations in genes coding
for receptor tyrosine kinases, e.g. EGFR, FGFR2, HER2,
Downloaded by [La Trobe University] at 13:01 01 October 2017
and MET.
● Genomically stable cancers are characterized by CDH1- and
RHOA mutations as well as rearrangements between
CLDN18 and ARHGAP26 or ARHGAP6.
● Overall low prevalence of mutations in actionable targets,
intra- and interindividual tumor heterogeneity, as well as
different standards of stomach cancer treatment in the East
and West are obstacles for the development of companion
diagnostics.
● Implementation of check point inhibitors in gastric cancer
care needs consideration of ethnicity: Asian and non-Asian
patient populations exhibit distinct tumor immunity
signatures.
Funding
This paper was not funded.
Declaration of interest
The author has no relevant affiliations or financial involvement with any
organization or entity with a financial interest in or financial conflict with
the subject matter or materials discussed in the manuscript. This includes
employment, consultancies, honoraria, stock ownership or options, expert
testimony, grants or patents received or pending, or royalties.
References
Papers of special note have been highlighted as either of interest (•) or of
considerable interest (••) to readers.
1. Carneiro F. Stomach cancer. In: Stewart BWWCP, editor. World
cancer report 2014. Lyon: IARC; 2014. p. 383–391.
2. Correa P. A human model of gastric carcinogenesis. Cancer Res.
1988;48(13):3554–3560.
• This publication describes a etiopathogentic model of gastric
cancer development, commonly known as the Correa-model.
3. Correa P. Human gastric carcinogenesis: a multistep and multi-
factorial process–First American Cancer Society award lecture on
cancer epidemiology and prevention. Cancer Res. 1992;52
(24):6735–6740.
EXPERT REVIEW OF MOLECULAR DIAGNOSTICS 299
4. Uemura N, Okamoto S, Yamamoto S, et al. Helicobacter pylori
infection and the development of gastric cancer. N Engl J Med.
2001;345(11):784–789.
5. Malfertheiner P, Link A, Selgrad M. Helicobacter pylori: perspectives
and time trends. Nat Rev Gastroenterol Hepatol. 2014;11(10):628–
638.
6. Amieva M, Peek RM Jr. Pathobiology of Helicobacter pylori-induced
gastric cancer. Gastroenterology. 2016;150(1):64–78.
7. Oliveira C, Pinheiro H, Figueiredo J, et al. Familial gastric cancer:
genetic susceptibility, pathology, and implications for manage-
ment. Lancet Oncol. 2015;16(2):e60–e70.
8. Yamada H, Sakamato H, Sugimura H. Sexy small copy numbers in
hereditary gastric carcinogenesis. J Gastroint Dig Syst. 2014;4:4.
9. Majewski IJ, Kluijt I, Cats A, et al. An α-E-catenin (CTNNA1) mutation
in hereditary diffuse gastric cancer. J Pathol. 2013;229(4):621–629.
10. Petrovchich I, Ford JM. Genetic predisposition to gastric cancer.
Semin Oncol. 2016;43(5):554–559.
11. Donner I, Kiviluoto T, Ristimäki A, et al. Exome sequencing reveals
three novel candidate predisposition genes for diffuse gastric can-
cer. Fam Cancer. 2015;14(2):241–246.
12. Worthley DL, Phillips KD, Wayte N, et al. Gastric adenocarcinoma
and proximal polyposis of the stomach (GAPPS): a new autosomal
dominant syndrome. Gut. 2012;61(5):774–779.
13. Li J, Woods SL, Healey S, et al. Point mutations in exon 1B of APC
reveal gastric adenocarcinoma and proximal polyposis of the sto-
mach as a familial adenomatous polyposis variant. Am J Hum
Genet. 2016;98(5):830–842.
14. van der Post RS, Vogelaar IP, Carneiro F, et al. Hereditary diffuse
gastric cancer: updated clinical guidelines with an emphasis on
germline CDH1 mutation carriers. J Med Genet. 2015;52(6):361–
374.
15. Lauwers GY, Carneiro F, Graham DY, et al. Tumours of the stomach.
In: Bosman FT, Carneiro F, Hruban RH, et al., editors. WHO classifi-
cation of tumours of the digestive system. Lyon: International
Agency for Research on Cancer; 2010. p. 48–80.
16. Lauren P. The two histologic main types of gastric carcinoma:
diffuse and so-called intestinal-type carcinoma. Acta Pathol
Microbiol Scand. 1965;64:31–49.
17. Wang K, Yuen ST, Xu J, et al. Whole-genome sequencing and
comprehensive molecular profiling identify new driver mutations
in gastric cancer. Nat Genet. 2014;46(6):573–582.
•• This and the following publication are seminal papers on the
genetics of gastric cancer using whole-genome sequencing
and comprehensive molecular profiling.
18. Cancer Genome Atlas Research . Comprehensive molecular char-
acterization of gastric adenocarcinoma. Nature. 2014;513
(7517):202–209.
•• This seminal publication describes four molecular subtypes of
gastric cancer based on a comprehensive molecular character-
ization of 295 gastric cancer patients.
19. Deng N, Goh LK, Wang H, et al. A comprehensive survey of geno-
mic alterations in gastric cancer reveals systematic patterns of
molecular exclusivity and co-occurrence among distinct therapeu-
tic targets. Gut. 2012;61(5):673–684.
• This study applied single nucelotide poylmorphism arrays in a
panel of 193 gastric cancer patients and 40 gastric cancer cell
lines and found 22 recurrent genomic alterations including
several actionable targets.
20. Mathiak M, Warneke VS, Behrens HM, et al. Clinicopathologic char-
acteristics of microsatellite instable gastric carcinomas revisited:
urgent need for standardization. Appl Immunohistochem Mol
Morphol. 2017;25:12–24.
21. Warneke V, Behrens H-M, Haag J, et al. Prognostic and putative
predictive biomarkers of gastric cancer for personalized medicine.
Diagn Mol Pathol. 2013;22(3):127–137.
22. Böger C, Behrens H-M, Mathiak M, et al. PD-L1 is an independent
prognostic predictor in gastric cancer of Western patients.
Oncotarget. 2016;7(17):24269–24283.
 300 C. RÖCKEN
23. Gonzalez RS, Messing S, Tu X, et al. Immunohistochemistry as a
surrogate for molecular subtyping of gastric adenocarcinoma. Hum
Pathol. 2016;56:16–21.
24. Fukayama M, Ushiku T. Epstein-Barr virus-associated gastric carci-
noma. Pathol Res Pract. 2011;207(9):529–537.
25. Tokunaga M, Land CE. Epstein-Barr virus involvement in gastric
cancer: biomarker for lymph node metastasis. Cancer Epidemiol
Biomarkers Prev. 1998;7(5):449–450.
26. Liu X, Liu J, Qiu H, et al. Prognostic significance of Epstein-Barr virus
infection in gastric cancer: a meta-analysis. BMC Cancer.
2015;15:782.
27. Tokunaga M, Land CE, Uemura Y, et al. Epstein-Barr virus in gastric
carcinoma. Am J Pathol. 1993;143(5):1250–1254.
28. Park CK, Park JS, Kim HS, et al. Receptor tyrosine kinase amplified
gastric cancer: clinicopathologic characteristics and proposed
screening algorithm. Oncotarget. 2016. DOI:10.18632/
oncotarget.12291
29. Kiyose S, Nagura K, Tao H, et al. Detection of kinase amplifications
in gastric cancer archives using fluorescence in situ hybridization.
Pathol Int. 2012;62(7):477–484.
30. Shen G-S, Zhao J-D, Zhao J-H, et al. Association of HER2 status with
prognosis in gastric cancer patients undergoing R0 resection: a
Downloaded by [La Trobe University] at 13:01 01 October 2017
large-scale multicenter study in China. World J Gastroenterol.
2016;22(23):5406–5414.
31. Warneke VS, Behrens H-M, Böger C, et al. Her2/neu testing in
gastric cancer: evaluating the risk of sampling errors. Ann Oncol.
2013;24(3):725–733.
32. Metzger M-L, Behrens H-M, Böger C, et al. MET in gastric cancer -
discarding a 10% cutoff rule. Histopathology. 2016;68(2):241–253.
33. Silva ANS, Coffa J, Menon V, et al. Frequent coamplification of
receptor tyrosine kinase and downstream signaling genes in
Japanese primary gastric cancer and conversion in matched
lymph node metastasis. Ann Surg. 2016;1. DOI:10.1097/
SLA.0000000000002042
34. Kakiuchi M, Nishizawa T, Ueda H, et al. Recurrent gain-of-function
mutations of RHOA in diffuse-type gastric carcinoma. Nat Genet.
2014;46(6):583–587.
35. Oshima T, Shan J, Okugawa T, et al. Down-regulation of claudin-18
is associated with the proliferative and invasive potential of gastric
cancer at the invasive front. PLoS One. 2013;8(9):e74757.
36. Röcken C, Behrens H-M, Böger C, et al. Clinicopathological charac-
teristics of RHOA mutations in a Central European gastric cancer
cohort. J Clin Pathol. 2016;69(1):70–75.
37. Böger C, Warneke VS, Behrens HM, et al. Integrins αβ3 and αβ5 as
prognostic, diagnostic, and therapeutic targets in gastric cancer.
Gastric Cancer. 2015;18(4):784–795.
38. Sargent DJ, Marsoni S, Monges G, et al. Defective mismatch repair
as a predictive marker for lack of efficacy of fluorouracil-based
adjuvant therapy in colon cancer. J Clin Oncol. 2010;28(20):3219–
3226.
39. Sinicrope FA, Foster NR, Thibodeau SN, et al. DNA mismatch repair
status and colon cancer recurrence and survival in clinical trials of
5-fluorouracil-based adjuvant therapy. J Natl Cancer Inst. 2011;103
(11):863–875.
40. Lawrence MS, Stojanov P, Polak P, et al. Mutational heterogeneity
in cancer and the search for new cancer-associated genes. Nature.
2013;499(7457):214–218.
41. Lordick F, Kang Y-K, Chung H-C, et al. Capecitabine and cisplatin
with or without cetuximab for patients with previously untreated
advanced gastric cancer (EXPAND): a randomised, open-label
phase 3 trial. Lancet Oncol. 2013;14(6):490–499.
42. De Roock W, Piessevaux H, De Schutter J, et al. KRAS wild-type state
predicts survival and is associated to early radiological response in
metastatic colorectal cancer treated with cetuximab. Ann Oncol.
2008;19(3):508–515.
43. Gerlinger M, Rowan AJ, Horswell S, et al. Intratumor heterogeneity
and branched evolution revealed by multiregion sequencing. N
Engl J Med. 2012;366(10):883–892.
44. Gerlinger M, Horswell S, Larkin J, et al. Genomic architecture and
evolution of clear cell renal cell carcinomas defined by multiregion
sequencing. Nat Genet. 2014;46(3):225–233.
45. Yates LR, Campbell PJ. Evolution of the cancer genome. Nat Rev
Genet. 2012;13(11):795–806.
46. Humphries A, Cereser B, Gay LJ, et al. Lineage tracing reveals
multipotent stem cells maintain human adenomas and the pattern
of clonal expansion in tumor evolution. Proc Natl Acad Sci USA.
2013;110(27):E2490–E2499.
47. Swanton C. Intratumor heterogeneity: evolution through space and
time. Cancer Res. 2012;72(19):4875–4882.
48. Bang Y-J, Van CE, Feyereislova A, et al. Trastuzumab in combination
with chemotherapy versus chemotherapy alone for treatment of
HER2-positive advanced gastric or gastro-oesophageal junction
cancer (ToGA): a phase 3, open-label, randomised controlled trial.
Lancet. 2010;376(9742):687–697.
•• This seminal study introduced targeted therapy for gastric
cancer.
49. Rüschoff J, Dietel M, Baretton G, et al. HER2 diagnostics in gastric
cancer-guideline validation and development of standardized
immunohistochemical testing. Virchows Arch. 2010;457(3):299–307.
50. Lordick F, Röcken C. The identification of predictive factors for
perioperative chemotherapy in esophago-gastric cancer. Ann
Oncol. 2013;24(5):1135–1138.
51. Röcken C. Quality assurance in clinical trials-the role of pathology.
Virchows Arch. 2016;468(1):83–92.
52. Cunningham D, Tebbutt NC, Davidenko I, et al. Phase 3, rando-
mized, double-blind, multicenter, placebo-controlled trial of rilotu-
mumab plus epirubicin, cisplatin, and capecitabine as first-line
therapy in patients with advanced MET-positive gastric or gastro-
esophageal junction cancer: RILOMET-1 study. J Clin Oncol. 2015;33
(Suppl.):abstr4000.
53. Shah MA, Bang YJ, Lordick F, et al. METGastric: a phase III study of
onartuzumab plus mFOLFOX6 in patients with metastatic HER2-
negative (HER2-) and MET-positive (MET+) adenocarcinoma of the
stomach or gastroesophageal junction (GEC). J Clin Oncol. 2015;33
(Suppl.):abstr4012.
54. Baretton G, Dietel M, Gaiser T, et al. HER2 testing in gastric cancer:
results of a meeting of German experts. Pathologe. 2016;37(4):361–
366.
55. Bartley AN, Washington MK, Ventura CB, et al. HER2 testing and
clinical decision making in gastroesophageal adenocarcinoma:
guideline from the College of American Pathologists, American
Society for Clinical Pathology, and American Society of Clinical
Oncology. Arch Pathol Lab Med. 2016;140:1345–1363.
DOI:10.5858/arpa.2016-0331-CP
56. Al-Refaie WB, Tseng JF, Gay G, et al. The impact of ethnicity on the
presentation and prognosis of patients with gastric adenocarci-
noma. Results from the National Cancer Data Base. Cancer.
2008;113(3):461–469.
57. Byfield SAD, Earle CC, Ayanian JZ, et al. Treatment and outcomes of
gastric cancer among United States-born and foreign-born Asians
and Pacific Islanders. Cancer. 2009;115(19):4595–4605.
58. Kim J, Sun C-L, Mailey B, et al. Race and ethnicity correlate with
survival in patients with gastric adenocarcinoma. Ann Oncol.
2010;21(1):152–160.
59. Kunz PL, Gubens M, Fisher GA, et al. Long-term survivors of gastric
cancer: a California population-based study. J Clin Oncol. 2012;30
(28):3507–3515.
60. Nelson R, Ko EB, Arrington A, et al. Race and correlations between
lymph node number and survival for patients with gastric cancer. J
Gastrointest Surg. 2013;17(3):471–481.
61. Röcken C, Behrens H-M. Validating the prognostic and discriminat-
ing value of the TNM-classification for gastric cancer - a critical
appraisal. Eur J Cancer. 2015;51(5):577–586.
62. Theuer CP, Kurosaki T, Ziogas A, et al. Asian patients with gastric
carcinoma in the United States exhibit unique clinical features and

superior overall and cancer specific survival rates. Cancer. 2000;89
(9):1883–1892.
63. Theuer CP. Asian gastric cancer patients at a southern California
comprehensive cancer center are diagnosed with less advanced
disease and have superior stage-stratified survival. Am Surg.
2000;66(9):821–826.
64. Lin SJ, Gagnon-Bartsch JA, Tan IB, et al. Signatures of tumour
immunity distinguish Asian and non-Asian gastric adenocarcino-
mas. Gut. 2015;64(11):1721–1731.
65. Cunningham D, Allum WH, Stenning SP, et al. Perioperative che-
motherapy versus surgery alone for resectable gastroesophageal
cancer. N Engl J Med. 2006;355(1):11–20.
Downloaded by [La Trobe University] at 13:01 01 October 2017
EXPERT REVIEW OF MOLECULAR DIAGNOSTICS 301
• The MAGIC-trial introduced perioperative chemotherapy in
Europe.
66. Macdonald JS, Smalley SR, Benedetti J, et al. Chemoradiotherapy
after surgery compared with surgery alone for adenocarcinoma of
the stomach or gastroesophageal junction. N Engl J Med. 2001;345
(10):725–730.
• This study introduced adjuvant chemoradiotherapy in the U.S.
67. Japanese Gastric Cancer . Japanese gastric cancer treatment guide-
lines 2014 (ver. 4). Gastric Cancer. 2017;20(1):1–19.
68. Böger C, Behrens HM, Röcken C. Ki67 – An unsuitable marker of
gastric cancer prognosis unmasks intratumoral heterogeneity. J
Surg Oncol. 2015;113(1):46–54.