HER2 overexpression in gastric and esophageal adenocarcinoma: a clinicopathological and diagnostics analysis
Scientific Master Thesis T. Koopman, 1720317 Dr. A.L.T. Imholz, internist-oncologist Deventer Hospital, Internal Medicine 01-11-2012 to 24-03-2013 In collaboration with M.M. Smits, M. Louwen, and dr. M. Hage, pathologists
2
Abstract
Background: The HER2/neu (Human Epidermal growth factor Receptor 2/neu) oncogene
shows overexpression in 15-30% of gastric and esophageal adenocarcinoma. Trastuzumab
was recently shown to provide a survival benefit in HER2 positive patients with advanced
gastric and gastro-esophageal junction cancer. The need is increasing for diagnostic centers
and pathologists capable of gastro-esophageal HER2 diagnostics. A standardized modified
scoring system was recently introduced for gastro-esophageal HER2 scoring. Primary tumor
classification of gastric or esophageal cancer changed, with the introduction of the 7th TNM
(tumor, node, metastasis) edition of Classification of Malignant Tumours of the UICC (Union
of International Cancer Control).
Patients and methods: Tumor samples of 323 patients diagnosed with invasive gastric or
esophaeal adenocarcinoma between 2004 and 2011 in the Deventer Hospital were collected.
HER2 status was determined using immunohistochemistry (IHC) followed by chromogenic in
situ hybridization (CISH) in samples scored equivocal or HER2 positive. IHC 3+ or IHC
2+/CISH positive tumors were considered HER2 positive. Association between
clinicopathological variables and HER2 positivity was analyzed. A comparison was made
between the 6th and 7th TNM edition on classifying gastric or esophageal primary tumors.
Inter-observer variability on IHC scoring using the currently standard modified HER2 scoring
system was determined among three pathologists and the scoring system was validated.
Additionally, HER2 positivity and survival rates were analyzed in a subgroup of 46 patients
with advanced disease receiving the currently recommended palliative triplet chemotherapy
consisting of a fluoropyrimide, a platinum compound and an anthracyclin, without
trastuzumab.
Results: We found HER2 positive disease in 50 of 323 patients (15.6%). In univariate and
multivariate logistic models, HER2 positivity rates were significantly higher in esophageal
primary tumors (esophageal 25% vs. gastric 7.4%) and in intestinal histological tumor type
(intestinal 24.3% vs. diffuse or mixed 11.5%). Using the 7th TNM edition, many tumors
previously classified as gastric cancer are now classified as esophageal cancer (30.5% of all
tumors in this study). Inter-observer agreement on IHC scoring was good (κ = 0.78) with
higher levels of disagreement found in diffuse or mixed tumor type and in weak to moderate
stained IHC samples (2+). The HER2 scoring system was found to be an excellent method to
establish HER2 status. In the triplet chemotherapy subgroup, no significant difference in
survival rates was found between HER2 positive and HER2 negative patients.
Conclusions: HER2 positivity occurs in 15.6% of invasive gastro-esophageal
adenocarcinoma in Western patients, with positivity rates varying depending on primary
tumor location and histological tumor type. The currently used standardized HER2 scoring
system is an excellent, clinically applicable method to establish HER2 status in appropriately
trained pathologists. Further studies are needed on HER2 status in advanced disease patients
receiving the recommended palliative triplet chemotherapy.
3
Samenvatting
Achtergrond: Het HER2/neu (Human Epidermal growth factor Receptor 2/neu) oncogen
vertoont overexpressie in 15-30% van de adenocarcinomen van de maag en slokdarm. Recent
is gebleken dat trastuzumab leidt tot een betere overleving in HER2 positieve patiënten met
vergevorderde kanker van de maag of gastro-oesofageale overgang. Er is een stijgende
behoefte aan diagnostische centra en pathologen die bekwaam zijn in de HER2 diagnostiek bij
maag- en slokdarmkanker. Sinds kort wordt een gestandaardiseerd scoringssysteem gebruikt
voor de HER2 bepaling in de maag en slokdarm. Classificatie van de primaire tumor bij
maag- en slokdarmkanker is veranderd met de introductie van de 7de editie van de TNM
(tumor, node, metastasis) Classification of Malignant Tumours van de UICC (Union of
International Cancer Control).
Patiënten en methoden: Van 323 patiënten met de diagnose invasief adenocarcinoom van de
maag of slokdarm in de periode 2004 tot en met 2011 in het Deventer Ziekenhuis werd op
tumormateriaal de HER2 status bepaald door middel van immunohistochemie (IHC) gevolgd
door chromogene in situ hybridisatie (CISH) bij monsters met een onduidelijke of positieve
HER2 score. Tumoren met een IHC 3+ score of IHC 2+/CISH positieve test werden als HER2
positief beschouwd. Het verband tussen verschillende clinicopathologische variabelen en
HER2 positiviteit werd geanalyseerd. De classificatie van primaire tumoren van de maag of
slokdarm volgens de 6de en 7de TNM editie werd vergeleken. Inter-observer variabiliteit bij
het scoren van IHC met het huidige gestandaardiseerde aangepaste HER2 scoringssysteem
werd bepaald bij drie pathologen en het scoringssysteem werd gevalideerd. Daarnaast werden
HER2 positiviteit en overlevingskansen geanalyseerd in een subgroep van 46 patiënten met
vergevorderde kanker die het huidige aanbevolen palliatieve triplet chemotherapie schema
krijgen bestaande uit een fluoropyrimide, een platinumderivaat en een anthracycline, zonder
trastuzumab.
Resultaten: In deze studie werd HER2 positiviteit gevonden in 50 van de 323 patiënten
(15.6%). In univariate en multivariate logistische modellen kwam HER2 positiviteit
significant vaker voor bij primaire slokdarmtumoren (slokdarm 25% vs. maag 7.4%) en bij
het intestinale histologische tumor type (intestinaal 24.3% vs. diffuus of gemengd 11.5%). Als
gebruik gemaakt wordt van de 7de TNM editie worden veel tumoren die eerder als
maagkanker geclassificeerd werden nu geclassificeerd als slokdarmkanker (30.5% van alle
tumoren in deze studie). De inter-observer overeenstemming bij het scoren van IHC was goed
(κ = 0.78) en verschil van interpretatie betrof merendeels diffuse of gemengde tumor types en
zwak tot matig aankleurende IHC monsters (2+). Het HER2 scoringssysteem bleek een
uitstekende methode om de HER2 status te bepalen. In de triplet chemotherapie subgroep
werd geen significant verschil gevonden in overlevingskansen tussen HER2 positieve en
HER2 negatieve patiënten.
Conclusies: HER2 positiviteit komt voor in 15.6% van invasieve adenocarcinomen van de
maag en slokdarm in Westerse patiënten, met wisselende positiviteit afhankelijk van de
primaire tumorlocatie en het histologische tumor type. Het huidige gebruikte HER2
scoringssysteem is een uitstekende, klinisch toepasbare methode om HER2 status te bepalen
door adequaat geschoolde pathologen. Er moet verder onderzoek gedaan worden op het
gebied van HER2 status in patiënten met vergevorderde ziekte die het aanbevolen palliatieve
triplet chemotherapie schema krijgen.
4
Contents
Abstract
Samenvatting
Introduction
Epidemiology
Anatomy
Pathology
Diagnosis
Survival
Treatment
HER2: molecular aspects
HER2: clinical impact
HER2: treatment modalities
HER2: diagnostics
Objectives and hypothesis
Methods
Patients
Triplet chemotherapy subgroup
Pathology review
Immunohistochemistry
In situ hybridization
Statistical analysis
Results
HER2 positivity
Inter-observer variability in IHC scoring
Clinicopathological characteristics
Triplet chemotherapy subgroup
Discussion
Tumor location
HER2 overexpression: primary tumor location
HER2 overexpression: histological tumor type
Inter-observer variability
Inter-observer variability: primary tumor location and histological tumor type
HER2 diagnostics: scoring system validation
Triplet chemotherapy: HER2 status and survival rates
Conclusions
Acknowledgements
References
Appendix: histology images
Page
2
3
5
5
5
6
6
6
6
7
7
8
8
8
10
10
12
13
13
15
17
19
19
20
20
23
26
26
27
27
28
29
29
31
33
34
35
38
5
Introduction
Epidemiology
Gastric cancer is, with approximately 750.000 deaths every year, the second leading cause of
cancer death worldwide. Additionally, esophageal cancer accounts for over 400.000 deaths
yearly. Every year, approximately 1.000.000 new cases of gastric cancer and 500.000 new
cases of esophageal cancer present worldwide, making the upper gastrointestinal tract tumors
together the second most commonly diagnosed type of cancer.1,2
Men are affected about twice
as often as women. Globally, the incidence of gastric cancer per 100.000 varies depending on
continent and sex, from 3.4 in North American females to 26.9 in Asian males. Esophageal
cancer shows a similar epidemiological distribution, with these numbers being 1.3 and 15.3
per 100.000, respectively.3
Risk factors for development of both gastric and esophageal cancer are low socioeconomic
status, tobacco smoking, alcohol consumption and a diet high in salty or preserved (pickled or
smoked) foods and low in fresh fruit and vegetables. Additionally, Helicobacter pylori
infection is a major risk factor for developing gastric cancer. Obesity and gastro-esophageal
reflux causing Barrett’s esophagus are additional major risk factors for esophageal cancer,
cancer of the gastro-esophageal junction and gastric cancer of the cardia.3-5
Over the last decades, incidence of non-cardia gastric cancers has declined while incidence of
cardiac gastric cancer, gastro-esophageal junction cancer and cancer of the esophagus is
rapidly increasing. This shift is believed to be caused by an increased prevalence of gastro-
esophageal reflux and obesity, along with a decline of gastric Helicobacter pylori
infection.3,5,6
In countries with high gastric cancer incidence such as Japan, endoscopic screening programs
are used to screen high risk groups, improving survival rates. However, evidence for
worldwide screening is insufficient.3,4
Anatomy
In cancer staging, the esophagus is divided in the cervical esophagus, the intrathoracic
esophagus and the gastro-esophageal junction. The intrathoracic esophagus is divided in an
upper, mid, and lower thoracic portion. The lower thoracic part of the esophagus is about 8
centimeters in length and is defined as the distal half of the esophagus between the tracheal
bifurcation and the gastro-esophageal junction. The gastro-esophageal junction is situated
approximately 40 centimeters from the upper incisor teeth.7
The stomach is divided (from proximal to distal) in the cardia, fundus, corpus, antrum and
pylorus.7,8
The cardia is the portion of the stomach just below the gastro-esophageal junction.
In the 6th TNM (tumor, node, metastasis) edition of Classification of Malignant Tumours of
the UICC (Union of International Cancer Control), both cardia- and gastro-esophageal
junction tumors were classified and staged as gastric cancer.9 Since the recent 7th TNM
edition, in use since 2010, gastro-esophageal junction tumors are staged as esophageal
cancer. Moreover, a cardia tumor is now classified as gastro-esophageal junction and thus
staged as esophageal, when its epicenter is within 5 centimeters of the gastro-esophageal
junction and the tumor grows into the esophagus.7,8
Carcinomas in the cardia are
histologically and morphologically similar to carcinomas of the esophagus. Moreover,
adenocarcinomas of the cardia show more similarity to esophageal adenocarcinomas than to
lower gastric adenocarcinomas, both in ontogenetic patterns (age and sex distribution) as
pathomorphologically.10
This explains the changes made in the TNM classification system.
When compared to the 6th TNM edition, the 7th TNM edition for esophageal adenocarcinoma
was found to result in a better prognostic stratification of overall survival in patients with
resectable esophageal and gastro-esophageal junction tumors.11,12
In advanced disease,
6
survival patterns of esophageal and gastro-esophageal junction cancer are similar to that of
advanced gastric cancer, regardless of the TNM version used to stage the tumor.13
Pathology
Adenocarcinoma is the predominant type of gastric cancer, comprising 95% of all malignant
tumors.14
In esophageal cancer or cancer of the gastro-esophageal junction, adenocarcinomas
comprise 60-80% of all cases, a markedly increased frequency compared to earlier decades.
Most of these adenocarcinomas develop in premalignant Barrett’s esophagus.5
Most other
esophageal tumors are squamous cell carcinoma.3
Histologically, gastric carcinoma is classified using Laurén's classification, distinguishing
between the intestinal and diffuse tumor type.15
If characteristics of both types are expressed,
the carcinoma is classified as mixed type. The intestinal tumor type is characterized by
expansive growth and metastasis to the liver, while the diffuse tumor type grows infiltrative
and metastasizes predominantly peritoneal. The diffuse tumor type is usually less well
differentiated than the intestinal type, and is associated with worse prognosis.16,17
Esophageal
adenocarcinomas predominantly show histological intestinal type characteristics. This is
thought to be related to the intestinal metaplasia caused by esophageal reflux.18
Diagnosis
Gastric- and esophageal cancer do not present with typical symptoms. Early stages usually do
not cause symptoms, while advanced stages may present with symptoms such as epigastric
pain, obstruction, aversion to meat (horror carnis), bleeding, perforation and weight loss.
Diagnosis is made using endoscopic biopsy samples. Staging is done using the UICC TNM
classification system, of which the above mentioned 7th edition is in use since 2010. Clinical
T status and in lesser extent N status can be assessed using endoscopic ultrasound (EUS). CT
scans are used to assess N and M status, and can differentiate between advanced T stages.
PET scans, visualizing enhanced metabolic activity, are useful to identify active tumor and
metastatic lesions. Laparoscopy is most sensitive for peritoneal metastases.4
Pathological
TNM status is assessed after surgical resection. Additionally, biopsies taken of suspected
metastatic lesions are be used to determine pathological M status.
Survival
Because the early stages predominantly proceed asymptomatic, upper gastrointestinal tract
tumors often present in advanced stages. Because of this, survival is poor. Even though
survival rates are improving, overall 5-year survival of gastric cancer is still lower than 30%
and 5-year survival of esophageal adenocarcinoma is lower than 20%.3,5,19
Advanced gastric
cancer, defined as metastatic or inoperable locally advanced cancer, has a poor prognosis even
if treated with chemotherapy. Median survival of these patients ranges from 7.5 to 12 months,
and median survival in patients receiving best supportive care is 3 to 5 months.20,21
Advanced
esophageal and gastro-esophageal junction cancer survival patterns are similar to that of
advanced gastric cancer.13
Treatment
Early gastric and lower esophageal cancer is treated curatively with surgery. Small, local
tumors are removed by endoscopic mucosal resection (EMR) or submucosal dissection.
Larger resectable tumors, with or without locoregional lymph node metastases, are operated
mainly using open surgery. During surgery, depending on the location, size and presence of
metastases in lymph nodes, the stomach and/or esophagus is partially or completely resected.4
Locally advanced tumors are tumors which have grown too far into the adjacent structures or
which are metastasized to locoregional lymph nodes in such a way that the surgeon is unable
7
to resect them. Metastasized tumors are disseminated to distant lymph nodes or other organs.
Locally advanced tumors and metastasized tumors together are defined as advanced disease.
For advanced gastric and lower esophageal cancer, the main treatment is systemic
chemotherapy. Though survival remains poor, it is associated with marked improvements of
overall survival. A regimen consisting of a fluoropyrimide, a platinum compound and an
anthracyclin achieves the best survival results.21,22
In Europe, this triplet chemotherapy
treatment is the standard recommended palliative chemotherapy regimen in advanced gastric
carcinoma.23
Advanced distal esophageal and gastro-esophageal junction adenocarcinomas
are treated with the same drug regimen, though chemoradiation involving other drug
combinations is common as well.
HER2: molecular aspects
New treatment options are emerging in targeted therapies, the most promising target being the
Human Epidermal growth factor Receptor 2 (HER2/ ErbB2). Located on chromosome 17,
HER2 is part of the epidermal growth factor receptor (EGFR) family also known as the ErbB
protein family, consisting of four structurally related receptor tyrosine kinases. Receptor
tyrosine kinases are high-affinity cell surface receptors which are key regulators of
fundamental cellular processes such as cell proliferation, differentiation, metabolism,
adhesion, migration and apoptosis. 24,25
The EGFR family consists of: 1) HER1/ErbB1, named the EGFR, 2) HER2/ErbB2, also
called HER2/neu as the receptor in rodents is called neu, 3) HER3/ErbB3 and 4)
HER4/ErbB4. These receptors are activated by ligand binding. EGFR for instance binds with
growth factors such as epidermal growth factor (EGF) and transforming growth factor alpha
(TGF-α). For HER2, no specific ligand has been recognized yet. Activation of these receptors
activate intracellular signaling pathways, which promote cell proliferation and oppose
apoptosis. EGFR family overexpression is therefore associated with many human cancers.
The HER2/neu oncogene is well known from its presence and clinical significance in breast
cancer, but has been demonstrated in adenocarcinomas of other organs as well, such as colon,
lung and gynaecological tumors.24,25
HER2 overexpression, also called HER2 positivity, in
adenocarcinomas of the gastric and esophageal cancer was first described over two decades
ago, and has received a gradual increase of attention in the past decade.26,27
HER2: clinical impact
Studies up to this date on HER2 in gastro-esophageal cancer studied either gastric cancer or
esophageal cancer, but often not the two together.
In gastric cancer, HER2 positivity occurs in 15-30% of adenocarcinomas.27-30
HER2 in
esophageal adenocarcinoma is less extensively studied, but many studies report similar HER2
positivity rates.31-36
Histologically, HER2 is more common in Laurén's intestinal tumor type
of adenocarcinoma than in diffuse tumor type (15-35% vs. <10%).15,28-30,37,38
Reports on impact of HER2 positivity on survival rates in gastric cancer are conflicting. Many
studies find an association between HER2 positivity and lower survival rates, especially in
operable gastric cancer.20,28,30
In studies on advanced gastric cancer, HER2 positive cancer is
has been associated with both higher and lower survival rates.39-41
In more recent studies
however, on both operable and advanced gastric and gastro-esophageal junction cancer
patients, HER2 positivity is not independently associated with a difference in overall
survival.27,42
In esophageal adenocarcinoma, conclusions on HER2 positivity and survival rates are
conflicting as well.31-33
8
HER2: treatment modalities
In breast cancer, anti-HER2 therapy with trastuzumab (Herceptin®), is associated with
improved prognosis in all stages and is standard of care.43-46
Trastuzumab is a monoclonal
antibody that specifically targets the HER2 receptor, preventing tyrosine kinase activation,
decreasing HER2 receptor signaling and inducing antibody-dependent cellular cytotoxicity.47
Besides breast cancer, gastric cancer and cancer of the gastro-esophageal junction are the only
cancer locations where anti-HER2 therapy is currently validated. A recent study found that
the use of trastuzumab in HER2-positive advanced gastric cancer or gastro-oesophageal
junction cancer was associated with a significant improvement of median survival. This study,
called the ‘ToGA’ study, found an increase in median survival of 11.1 to 13.8 months in
patients treated with trastuzumab.37
As such, HER2 diagnostics and anti-HER2 therapy
appears to be a promising part of palliative gastro-esophageal cancer treatment.
HER2: diagnostics
HER2 status in gastric and esophageal adenocarcinoma is determined using a modified
scoring system defined by Hofmann et al. in 2008.38
This scoring system is also used in the
ToGA trial. Recently (2010), Rüschoff et al. established additional guidelines to enhance the
scoring system.48
The scoring system differs from the scoring system on HER2 in breast
cancer, because of a high level of tumor heterogeneity and because of irregular membrane
staining in gastro-esophageal tissue caused by the secretory nature of the epithelium. Thus, if
the breast cancer scoring system would be applied on gastro-esophageal cancer, many tumors
would receive a false-negative score.38
The rate of HER2 overexpression is visualized using immunohistochemistry (IHC) with a
HER2 marker and is scored 0 or 1+ (negative), 2+ (equivocal) or 3+ (positive) according to
the visible quantitative HER2 positive tissue. In samples scored 2+ (equivocal), IHC testing is
followed by in situ hybridization (ISH) which uses chromosome labeling to visualize gene
amplification. IHC scoring is done by an individual pathologist, is prone to inter-observer
differences, as has been shown in breast cancer49,50
and gastric cancer.48,51,52
Because of the
higher level of tumor heterogeneity and irregular membrane staining in gastro-esophageal
tumors, IHC scoring is harder than in breast cancer, potentially giving rise to higher
discordance between observers than in breast cancer. 48
Objectives and hypothesis
The HER2/neu receptor is a promising target for therapy in gastric and esophageal
adenocarcinoma. In the National Comprehensive Cancer Network (NCCN) guidelines used in
the United States, HER2 assessment and treatment is recommended for advanced gastric- and
esophageal adenocarcinoma.53,54
Assessment and treatment of HER2 status in clinical practice
in Europe is however not yet part of standard practice. In the Deventer Hospital, a peripheral
hospital in the Netherlands where this study is conducted, HER2 status in gastric and
esophageal cancer was rarely determined before this study.
In this retrospective study, we determined HER2 positivity rate in gastric and esophageal
adenocarcinoma in the Deventer Hospital in the Netherlands. Data on HER2 overexpression
in esophageal cancer is scarce. To the best of our knowledge, this study is the first where a
direct comparison on HER2 overexpression between gastric and esophageal cancer, including
gastro-esophageal junction cancer, was performed. As HER2 overexpression rates in studies
on gastric and esophageal cancer showed similar HER2 positivity rates27-36
, we expected that
HER2 overexpression in both locations was similar in our study.
Additionally, we performed a direct comparison of tumor location classification according to
the 7th and 6th TNM editions, as studies before 2010 used the 6th edition. As a consequence
9
we expected that many tumors previously classified as gastric tumors using the 6th TNM
edition, were classified as esophageal tumors using the 7th TNM edition.
As mentioned before, IHC scoring of HER2 cancer is prone to inter-observer differences.
Studies about inter-observer variability in HER2 testing in gastro-esophageal cancer are
scarce; in most of the studies IHC scoring was only investigated in gastric and gastro-
esophageal junction cancer,48,51,52
not in esophageal cancer. Moreover, these studies were
carried out among different laboratories51
or were executed with pathologists already familiar
with the HER2 testing protocols.48
Our study provides new data on inter-observer variability
in IHC scoring on HER2 in gastro-esophageal adenocarcinoma among newly trained
pathologists in a single laboratory. Additionally we analyzed inter-observer variability in IHC
scoring on HER2 among different primary tumor locations and histological tumor types,
which was not done in the previous studies. The aim was to demonstrate that appropriate
education and training of pathologists on the modified HER2 scoring system currently used in
gastric and esophageal cancer when identifying HER2 status, is feasible and will result in
similar inter-observer variability to the other studies conducted.
Anthracyclins are known to improve survival by several months, if added to the two drug
regimen used in the ToGA study.21
Reports on the influence of HER2 status on survival rates
in advanced gastro-esophageal cancer are conflicting39-41
. A recent study on advanced gastric
cancer found that HER2 was not an independent predictor of prognosis. However, patients in
this study received a variety of palliative chemotherapy regimens, all without anthracyclins.27
The association between HER2 status and survival rates in advanced gastro-esophageal cancer
patients receiving the recommended palliative triplet chemotherapy treatment has not been
studied individually. We selected a subgroup of patients with advanced stage gastric or
esophageal adenocarcinoma receiving triplet chemotherapy treatment. In this subgroup, we
compared survival rates of HER2 positive patients versus HER2 negative patients. We
expected that the HER2 positive patients had better survival rates than the HER2 negative
patients.
10
Methods
This study is a retrospective analysis, consisting of two main parts. Firstly, a describing
clinicopathological analysis of the total study population, which includes an analysis of inter-
observer variability in the scoring of IHC on HER2. Secondly, a subgroup receiving palliative
chemotherapy treatment will be divided in two cohorts, depending on HER2 status, of whom
survival rates will be retrospectively analyzed.
Patients
Our study includes all patients diagnosed with histologically confirmed gastric or esophageal
(including gastro-esophageal junction) adenocarcinoma, diagnosed in the Deventer Hospital
(Netherlands) from January 2004 to December 2011, or recidivating in this period from early
stage cancer diagnosed before this period.
The Dutch Integral Cancer Centre
(Integraal Kankercentrum Nederland,
IKNL) was consulted to identify all
patients with gastric or esophageal cancer
in the Deventer Hospital in the specified
period, which resulted in 433 patients.
From the records of the Deventer Hospital,
five additional patients with recidivating
cancer were identified, adding to a total of
438 patients. Patients of whom
pathological material was not present in the
archives of the Deventer Hospital were
excluded from the study, leaving 387
patients. Pathological reports were
assessed for tumor type and 333 patients
were found to have adenocarcinoma,
eligible for IHC staining on HER2. The
patient selection procedure is shown in figure 1.
Clinical and pathological characteristics including age, sex, pathological specimen type,
primary tumor location, tumor stage, disease status (advanced versus not advanced),
chemotherapy regimen (if given), were collected retrospectively from the Deventer Hospital
patient records. Patients were added to the database anonymously, so that approval from the
Medical Ethical Testing Committee was not necessary.
Primary tumor location definition, according to the 7th TNM edition of Classification of
Malignant Tumours, is displayed in table 1 and figure 2. The primary tumor location was
based on tumor location as described in endoscopy reports available in the medical records.
Tumor stage (I through IV), according to the 7th TNM edition, was based on endoscopic
ultrasound (EUS) reports, radiological examination reports including computed tomography
(CT) scans and positron emission tomography (PET) scan or pathological reports of
histologically confirmed metastatic locations or of tissue obtained in surgical specimen. If
data on tumor stage were not available in the Deventer Hospital medical records, IKNL data
were used to determine tumor stage, if available. Advanced disease was defined as metastatic
disease or technically inoperable locally advanced disease. Distinction between advanced or
not advanced disease status among patients was made using tumor stage data, surgical
operation reports and medical reports and letters by attending physicians.
Figure 1 Patient selection procedure HER2=human epidermal growth factor receptor 2.
11
Table 1 Definition of primary tumor location*
Location Definition
Distal esophagus Located in the distal esophagus without growth into the gastro-esophageal junction.
Gastro-esophageal junction
1) Located directly in the GEJ 2) Located in the distal esophagus with growth into the GEJ 3) Located in the cardia with growth into the esophagus and the epicenter of the tumor being within 5 cm of the GEJ
Stomach (cardia) Located in the gastric cardia without growth into the gastro-esophageal junction
Stomach (non-cardia) Located anywhere in the stomach but the cardia
*According to TNM 7 edition guidelines. GEJ =gastro-esophageal junction.
Figure 2 Definition of primary tumor location (according to TNM 7 edition guidelines)
Distal esophagus
Gastro-esophageal junction
Cardia Non-cardia
Esophageal Gastric
12
Triplet chemotherapy subgroup
A subgroup of patients who received palliative
triplet chemotherapy were analyzed separately.
These patients were included if they were
older than 18 years of age at time of diagnosis
and had histologically confirmed, advanced
stage (defined as technically inoperable locally
advanced, recurrent, or metastatic)
adenocarcinoma of the stomach or esophagus
including the gastro-esophageal junction;
Eastern Cooperative Oncology Group (ECOG)
performance status 0-2; and measurable or
non-measurable disease. As such, of the total
study population of 333 patients, 174 patients
were classified as advanced stage, of which 73
received palliative chemotherapy. Patients
who received at least one cycle of a three-drug
regimen containing a fluoropyrimide, a
platinum compound and an anthracyclin were
selected, amounting to a subgroup of 49
patients. The patient selection procedure is
shown in figure 3.
Triplet chemotherapy was given as either an
intravenous epirubicin 60mg/m2 and cisplatin
50mg/m2 at day one with oral capecitabine
1000mg/m2 twice daily for 14 days (ECC) or as intravenous epirubicin 60mg/m
2 and
oxaliplatin 130mg/m2 at day one with oral capecitabine (Xeloda®) 625mg/m
2 twice daily for
21 days (EOX). Both the ECC and EOX cycles are repeated every three weeks.
In addition to the data on clinicopathological characteristics already collected for all HER2
tested patients in this study (age, sex, pathological specimen type, primary tumor location,
tumor stage and disease status), medical records of the triplet chemotherapy subgroup patients
were assessed for clinical and/or pathological TNM stage (according to the 7th TNM edition),
location and number of metastatic lesions (if present), date of advanced stage diagnosis,
chemotherapy period and the number of received cycles, date of progression (if available) and
date of death or date last known alive. The date of advanced stage diagnosis was defined as
the date advanced stage at diagnosis was histologically confirmed, or as the date recurrent
disease was confirmed histologically or by CT imaging in patients with known early stage
cancer. Pathological or, if not available, clinical TNM stage was determined, depending on
information available in the medical records. Pathological TNM stage was based on
pathological reports of tissue obtained by surgical treatment (T and N stage) and
histologically confirmed metastatic locations (M stage). Clinical TNM stage was based on
endoscopic ultrasound (EUS) reports (T and N stage) or radiological examination reports
including computed tomography (CT) scans and positron emission tomography (PET) scans
(T, N and M stage). If radiological reports did not provide information on TNM stage, the CT
scan at diagnosis was reviewed with a radiologist (RvD or AF) to determine missing TNM
stage values.
According to the HER2 status established in the first part of the study, triplet chemotherapy
subgroup patients were divided in two groups; a HER2 positive group and a HER2 negative
group. These groups were compared for the primary objective progressive-free survival (PFS)
and secondary objective overall survival (OS).
Figure 3 Subgroup patient selection procedure HER2=human epidermal growth factor receptor 2.
13
Progression free survival was defined as the time from starting chemotherapy until
progression was shown by pathological confirmation of new metastatic lesions or by CT
imaging, using the RECIST 1.1 criteria.55
Additionally, death was accounted as progression
when progression was not clinically evaluated and established. In one patient a follow-up CT
was not executed, but progression was identified using endoscopy, showing significant tumor
growth in comparison to earlier endoscopic examinations. Overall survival was defined as
time from starting chemotherapy until death from any cause. Patients without an event
(progression or death in PFS or death in OS) at the time this study was conducted, were
censored at the date that they were last known to be alive without progression.
Pathology review
Formalin-fixed paraffin-embedded tumor samples from 333 patients, being biopsies of the
primary tumor; surgical resection material; or biopsies of metastatic lesions, were retrieved
from the Deventer Hospital pathology department archives. If a surgical resection specimen
was available, this was taken instead of an available primary tumor biopsy specimen. If
primary tumor biopsy or resection material was not present, a metastatic lesion biopsy was
taken if present. The tumor samples were evaluated for HER2 protein expression by
immunohistochemistry (IHC) by three pathologists, followed by gene amplification using
chromogenic in situ hybridization (CISH) if the consensus IHC score was 2+, 3+ or
inconclusive. Corresponding hematoxylin-eosin slides were reviewed to assess sample
adequacy. In all tumor samples, tumor type according to Laurén's classification (intestinal,
diffuse or mixed) was established by a consensus between the study pathologists. All IHC and
CISH tests were performed and evaluated in the Deventer Hospital (Netherlands).
Immunohistochemistry
IHC staining for HER2 is scored visually, ranging from 0 (no expression) to 3+ (high
expression). Our study used the modified scoring system for gastro-esophageal tumors
established by Hofmann et al. in 2008,38
with additional guidelines on this scoring system
provided by Rüschoff et al. in 2010.48,56
The IHC scoring protocol is displayed in table 2. In
comparison to breast cancer, this HER2 IHC scoring system differs in two points. Firstly,
unlike in breast cancer where membranous staining must be complete, incomplete
membranous staining can be considered as positive staining in gastro-esophageal tumors.
Secondly, in breast cancer both biopsies and resection specimens have to contain >10% (and
in case of a 3+ sample, > 30%) stained tumor cells in order to be considered as positive
staining, while in gastric and esophageal cancer this criterion is only restricted to resection
specimens. In gastric and esophageal cancer biopsies, a number of at least five clustered
stained tumor cells is used as criterion instead of the >10% (or >30%) criterion. These five
clustered cells must have membranous staining that is complete, basolateral or only lateral
between cell-cell contacts.48,56,57
IHC images are displayed in images A through E in the
appendix.
When a sample is scored IHC 0 or IHC 1+ the sample is pronounced HER2 negative and
when a sample is scored IHC 3+, it is pronounced HER2 positive. In samples scored IHC 2+,
IHC testing is followed by an in situ hybridization technique. The IHC 2+ sample is then only
considered positive if the ISH test shows HER2 gene amplification. HER2 diagnostic
algorithm is summarized in figure 4.
14
Table 2 IHC scoring for HER2 status in gastric- and esophageal adenocarcinoma
Score Reactivity* HER2 status
0 No reactivity or no membranous reactivity (visible at 40x)
Negative
1+ Faint or barely visible membranous reactivity (visible at 40x)
Negative
2+ Weak to moderate complete, basolateral or lateral membranous reactivity (visible at 10-20x)
Equivocal, should be followed by ISH
3+ Strong complete, basolateral or lateral membranous reactivity (visible at 2,5-5x)
Positive
* There must be sufficient well-preserved tumor tissue, staining must be membranous, and there must be a cluster of >5 stained tumor cells in biopsies or staining in >10% of tumor cells in resection specimens. HER2=human epidermal growth factor 2. IHC=immunohistochemistry. ISH=in situ hybridization.
Figure 4 HER2 diagnostics algorithm in gastro-esophageal adenocarcinoma HER2=human epidermal growth factor 2. IHC=immunohistochemistry. ISH=in situ hybridization.
New slides were prepared for IHC immunostaining on HER2 on all 333 retrieved tumor
samples. Paraffin sections were cut at 4 µm using the Leica RM2255 rotary microtome (Leica
Microsystems BV) and mounted on glass slides with validated control samples also cut at 4
µm. These slides were deparaffinized in the Tissue-Tek® Prisma®/Film® Automated Slide
Stainer and Coverslipper (Sakura Finetek Europe BV). Immunostaining for HER2 was
performed using the Ventana BenchMark® XT immunostainer (Ventana Medical Systems,
Inc.), following the manufacturers protocol. Slides were heated at 75° Celsius and
deparaffinized in the BenchMark® XT with EZ Prep solution (Ventana Medical Systems,
Inc.), followed by heat-induced antigen retrieval using CC1 (Cell Conditioning 1, pH 9,
Ventana Medical Systems, Inc.) at 95° Celsius during 30 minutes. The Ventana iViewTM
DAB detection kit (Ventana Medical Systems, Inc.) was added, combined with the
PATHWAY® HER2/neu antibody (Ventana Medical Systems, Inc.), to achieve HER2
immunostaining. The slides were counterstained with hematoxylin. To ensure heterogeneity,
all samples were stained in full batches of 30 samples in the BenchMark® XT.
15
The PATHWAY® HER2/neu antibody (Ventana Medical Systems, Inc.) contains the 4B5
antibody. This 4B5 antibody is a monoclonal rabbit antibody, which binds to the HER2
receptor on cells, and was found to be an excellent antibody to detect HER2 status in breast
cancer.58
The 4B5 antibody is validated for use in gastro-esophageal cancer as well, showing
equal or even better results for HER2 detection in comparison to the other frequently used
monoclonal rabbit antibody HercepTestTM
(Dako Denmark A/S).48
Both antibodies are FDA
(U.S. Food and Drug Administration) approved. The 4B5 antibody is used in combination
with the Ventana iViewTM
DAB detection kit (Ventana Medical Systems, Inc.), which
contains an indirect biotine streptavidine system that binds with the 4B5 antibody bound to
the HER2 receptor, resulting in a dark brown color visible under bright field microscope.
Three pathologists in the Deventer Hospital (MMS, ML and MH), previously unfamiliar with
HER2 scoring in gastric and esophageal adenocarcinoma, were educated on the modified
scoring system. This was achieved using the HER2 testing internet site by F. Hoffmann-La
Roche Ltd.59
, which offers expert instructions, lectures and exercises on IHC scoring of HER2
in gastro-esophageal tumors, based on the scoring system by Hofmann et al. Additionally, all
three pathologists were provided the additional guidelines established by Rüschoff et al. A
random sample of ten IHC HER2 stained slides providing different HER2 amplification levels
was taken from the total batch of 333 stained samples, and was evaluated by the three
pathologists simultaneously. Doubtful cases were then re-assessed by a pathologist
specialized in HER2 scoring in gastric- and esophageal tumors, situated in the Academic
Medical Centre (AMC) in Amsterdam (MvdV), to clarify to the pathologists in the Deventer
Hospital how to score these cases.
The three pathologists were given identical scoring lists, in which they could score IHC
staining (0, 1+, 2+ or 3+) and tumor type (diffuse, intestinal or mixed type) according to
Laurén's classification.15
Moreover, they were given the opportunity to mark a slide
recognized from the random sample used in the simultaneous evaluation. Of all tumor
samples, hematoxylin-eosin slides were retrieved from the archives and presented to the
pathologists simultaneously with the HER2 stained IHC slides, to assess sample adequacy.
Slides were evaluated using a Zeiss Axioskop® 2 plus microscope (MMS and ML) or Zeiss
Axioskop® 20 microscope (MH) equipped with 2.5x, 5x, 10x, 20x, 40x and 63x dry objective
lenses (Carl Zeiss Microscopy, LLC).
In our study, three individual pathologists determined IHC status. If at least two pathologists
determined a 3+ score, the sample was treated as IHC 3+ and classified HER2 positive.
Additionally, CISH was performed on these HER2 positive samples, to verify gene
amplification. If all pathologists scored the sample 0 or 1+, the sample status was classified
HER2 negative. CISH was not performed on these samples. If at least one pathologist
determined a 2+ score, except in cases already pronounced HER2 positive as the other two
pathologists both scored 3+, the sample was treated as IHC 2+ and followed by CISH. If one
or two pathologists scored the sample as HER2 positive (IHC 3+) while the other one or two
gave a HER2 negative score (IHC 0 or 1+), with none of the pathologists giving a IHC 2+
score, HER2 status was pronounced inconclusive and IHC was followed by CISH to establish
true HER2 status.
In situ hybridization
In situ hybridization (ISH) can be performed using different techniques, such as fluorescence
in situ hybridization (FISH), silver-enhanced in situ hybridization (SISH) or chromogenic in
situ hybridization (CISH). The FISH technique uses fluorescent probes which bind to specific
DNA sequences on chromosomes and is a test with high sensitivity and specificity. In breast
cancer HER2 diagnostics, FISH testing was shown to be highly specific and more accurate
than IHC testing, though IHC was nearly as accurate as FISH.50,60
The SISH technique
16
achieves HER2 gene labeling with silver signals, and was shown to be a practical alternative
for FISH in breast cancer HER2 testing.61
The CISH technique uses conventional peroxidase
reactions to visualize gene amplification. CISH is also shown to be a accurate and practical
alternative for FISH to detect HER2 gene amplification, with similar results in both breast
cancer and gastric cancer samples.62-64
In contrast to FISH, both SISH and CISH allow for
morphological and amplification evaluation simultaneously and can be performed with a
bright field microscope instead of a fluorescence microscope. CISH and SISH are less labor
intensive and allow longer storage of stained slides.61-63
In situ hybridization scoring protocols
on HER2 in gastro-esophageal cancer are identical to scoring protocols on HER2 in breast
cancer.38,62,63
In our study, the CISH technique was used for HER2 chromosome labeling.
CISH testing was performed using the ZytoDot® SPEC HER2 Probe kit (Zytovision GmbH).
New slides were prepared from the formalin-fixed paraffin-embedded tumor samples used in
IHC. Paraffin sections were cut at 5 µm Leica RM2255 rotary microtome (Leica
Microsystems BV) and mounted on glass slides with validated control samples also cut at 5
µm. These slides were heated in a 60° Celsius stove overnight, and deparaffinized in the
Tissue-Tek® Prisma®/Film® Automated Slide Stainer and Coverslipper (Sakura Finetek
Europe BV). The slides were then incubated in pretreatment buffer in a temperature-
controlled water basin at 100° Celsius for 15 minutes, using ZytoDot® CISH Heat
Pretreatment Solution EDTA (Zytovision GmbH), and immediately afterwards washed in
distilled water at room temperature during two minutes for three times. Enzymatic digestion
was achieved by applying ES1 Pepsin Solution (Zytovision GmbH) for 90 seconds. The slides
were then again washed in distilled water during two minutes for three times and dehydrated
with graded ethanols in five steps, every step lasting two minutes. The slides were allowed to
dry for at least 20 minutes. The digoxigenin-labeled HER2 probe (ZytoDot® SPEC HER2
Probe) was then added to the sections by applying 8 to 16 µl probe mixture per slide,
depending on the amount of tissue, and covered with coverslips. The slides were denatured
for 5 minutes at 95° Celsius and hybridized at 37° Celsius in the Dako hybridizer (Dako
Denmark A/S) overnight. After hybridization, coverslips were removed and slides were
washed in a phosphate buffer saline (PBS)/0.05% Tween 20 solution at room temperature.
The slides were then washed with saline-sodium citrate buffer (SSC) and Tween 20, washing
10 minutes in a 2x SSC/0.05% Tween 20 solution at 42° Celsius followed by washing 10
minutes in a 0.1x SSC/0.05% Tween 20 solution at 60° Celsius. This was followed by two
washes for five minutes in PBS/0.05% Tween 20 solution at room temperature. To block
peroxidase activity and unspecific staining, the slides were put in a 3% hydrogen
peroxidase/methanol solution (prepared with 30% hydrogen peroxidase and 100% methanol)
for 10 minutes, followed by application of BS1 Blocking Solution (Zytovision GmbH) during
10 minutes. HER2 probe detection was achieved by sequential incubations with a mouse anti-
digoxigenin antibody (AB1 Mouse-Anti DIG, Zytovision GmbH) during 30 minutes and an
antimouse-peroxidase polymer (AB2 Anti-Mouse-HRP-Polymer, Zytovision GmbH) during
30 minutes. Slides were washed with PBS/0.05% Tween 20 solution during two minutes for
three times after every step of the detection process. Chromogenic visualization was achieved
by putting the slides in a 3,3’-diaminobenzidine (DAB) solution during at least 8 minutes.
Slides were lightly counterstained with hematoxylin and were dehydrated and embedded in
the Tissue-Tek® Prisma®/Film® Automated Slide Stainer and Coverslipper (Sakura Finetek
Europe BV).
CISH sections of samples which received an IHC 2+ score by at least one pathologist were
reviewed by the pathologist who gave this 2+ score. CISH sections of samples where IHC
scoring was inconclusive were scored by one of the study pathologists (MMS), to which the
sections were presented without information on what IHC scores the sample had received and
by whom. For CISH evaluation, corresponding HER2 IHC slides and hematoxylin-eosin
17
slides were provided. CISH evaluation was performed using a Zeiss Axioskop® 2 plus
microscope (MMS and ML) or Zeiss Axioskop® 20 microscope (MH) equipped with 2.5x,
5x, 10x, 20x, 40x and dry objective lenses (Carl Zeiss Microscopy, LLC). Non-amplified,
unaltered gene copy number was defined as one to five signal dots per nucleus, showing
either diploid (two signal dots) or polysomia (three to five signal dots). Amplification was
defined as six to 10 signal dots per nucleus or presence of small gene copy clusters (low
amplification) or more than 10 signal dots per nucleus or presence of large gene copy clusters
(high amplification), in more than 50% of cancer cells in a tissue area selected for
enumeration of at least 20 cells. This CISH scoring protocol on HER2 in gastro-esophageal
cancer is identical to the scoring protocol on HER2 in breast cancer.38,62,63
A CISH test is
pronounced HER2 positive when either low- or high amplification is present, and HER2
negative when the signal is diploid or shows polysomia. If no staining was present, CISH was
repeated on the sample to rule out staining process errors. If the repeated test showed no
staining, the CISH was classified as 'no signal' and not assessable. No staining is most likely
due to poor tissue fixation failure, known to be the most common source of error in HER2
testing.65
CISH images are displayed in images F through I in the appendix.
Statistical analysis
Kappa statistics was used for analysis of inter-observer variability. Because of the ordinal
IHC scoring scale of 0 through 3, weighed kappa statistics were applied. In weighed kappa
statistics, a linear or quadratic model can be applied. Because of the clinical significance of
larger distances between scores in the HER2 IHC scoring scale, quadratic kappa weights were
applied in this study. These weighed kappa statistics were conducted between pairs of two out
of three observers. To establish the kappa value of inter-observer variability between all three
pathologists, the simple mean weighed kappa value averaged over the three pairs of raters was
determined. It was shown that in trials using ordinal scales where the sample size is much
larger compared to the number of raters, as is the case in this study, inter-observer variability
for multiple raters can be determined using this method.66
The kappa statistic can be
interpreted as the agreement between multiple observers adjusted for chance agreement. A
kappa value of 1 indicates perfect agreement, whereas a value of 0 indicates no agreement
better than chance. The kappa value cannot be used to determine statistical significance.
Though not universally accepted, a widely used categorization of agreement by kappa score
suggested by Landis and Koch is as follows: <0.2 'slight', 0.21 to 0.40 'fair', 0.41 to 0.60
'moderate', 0.61 to 0.80 'substantial', 0.81 to 1.00 'almost perfect' agreement.67
Univariate and multivariate relationship between HER2 positivity rate and clinicopathological
variables was determined using a binary logistic regression model. Risk factors included in
the model were sex, age below versus above 65 years, advanced versus not advanced disease
status, the pathological specimen sample used for HER2 testing (primary tumor versus
metastatic lesion and biopsy versus surgical specimen), esophageal versus gastric primary
tumor location and intestinal versus diffuse or mixed histological tumor type. Patients with
missing values in one of these variables were excluded from multivariate analysis.
Comparison of primary tumor location and tumor type was performed using the Chi-square
test.
In the triplet chemotherapy subgroup, the primary objective was to determine progression free
survival (PFS) and the secondary objective was to determine overall survival (OS). In this
subgroup, comparison of PFS and OS of the HER2 positive group in comparison to the HER2
negative group was performed using the Kaplan-Meier method and univariate statistical
significance was analyzed using the log-rank test. Multivariate statistical comparison was
performed using a Cox proportional hazards regression model including sex, age below versus
above 65 years, intestinal versus diffuse or mixed histological tumor type, esophageal versus
18
gastric primary tumor location, T stage, N stage and M stage (metastasized versus locally
advanced). Patients with missing values in one of these variables were excluded from
multivariate analysis. T stage was dichotomized as T stage with growth only within the organ
versus growth through outermost organ connective tissue and beyond (growth through serosa
(T4a) or growth in adjacent structures (T4b) in gastric cancer, or growth through advantitia
(T3) or growth in adjacent structures (T4) in esophageal cancer). N stage was dichotomized as
less than 7 positive lymph nodes (N0 through N2) versus 7 or more positive nodes (N3,
including N3a or N3b in gastric cancer).
Inter-observer variability was statistically analyzed using Microsoft® Excel® for Windows
2007 and the statistical program ‘R’ for Windows version 2.15.2. Kappa statistics were
calculated using package ‘irr’ of program ‘R’ for Windows. All other statistical analyses were
performed using Microsoft® Excel® for Windows 2007 and IBM® SPSS® Statistics for
Windows version 20.0.0. All p values were two-sided, with a value of p < 0.05 being
considered clinically significant.
19
Results
HER2 positivity
Of the 333 samples eligible for HER2 diagnostics, five samples could not be assessed due to
lack of remaining tumor tissue in the formalin-fixed paraffin-embedded tumor samples. In
these patients, no other tumor sample was available and patients were excluded from the rest
of study. In five samples, the tumor turned out to be non-invasive (i.e. carcinoma in situ or
high grade dysplasia) and these samples were excluded from analysis. None of the 10 slides
used in the simultaneous evaluation during the IHC teaching process were recognized by any
of the pathologists during IHC scoring of the tumor samples.
50 of 323 patients (15.5%) tested HER2 positive. Study population HER2 diagnostics and
results are summarized in figure 5. In the 323 samples scored by the pathologists, 182 samples
received a unanimous HER2 negative score, being scored 0 or 1+ by all three pathologists
(IHC negative). A total of 42 samples received a consensus IHC 3+ score and were classified
as HER2 positive (IHC positive). Of these 42 samples, 34 received a unanimous 3+ score,
while 8 were scored 3+ by two and 2+ by one pathologist. CISH showed amplification in 39
of the 42 HER2 positive samples (92.9%), two were non-amplified (diploid) CISH and one
could not be assessed due to insufficient tumor tissue (lost after IHC testing). The two
samples with non-amplified CISH were among the 8 samples which were scored 3+ by two
and 2+ by one pathologist.
In the remaining group of 99 samples without IHC negative or IHC positive score which were
followed by CISH, 8 samples showed HER2 amplification and were classified as HER2
positive, while 89 samples showed no amplification and were classified as HER2 negative
and two samples showed 'no signal' in repeated CISH tests. Of the 99 samples on which CISH
was performed without HER2 positive IHC score, 95 samples were treated as IHC 2+ when at
least one of the pathologists scored the sample 2+ (without the other two pathologists scoring
the sample 3+). CISH showed amplification in 7 of these samples. The other four samples on
which CISH was performed had received an inconclusive IHC score by the three pathologists
as one or two gave the sample a HER2 negative score (0 or 1+) while the other one or two
scored the sample HER2 positive (3+). One of these four inconclusive samples showed
amplification in CISH and the other three showed no amplification.
Figure 5 HER2 diagnostics in this study HER2=human epidermal growth factor 2. IHC=immunohistochemistry. CISH=chromogenic in situ hybridization.
20
Inter-observer variability in IHC scoring
In three of all 323 IHC scored samples, one of the three pathologists did not give a IHC score
because of lacking tumor tissue amount or quality in the sample, while the other two
pathologists did score the sample. These values were excluded for statistical inter-observer
agreement analysis among all three pathologists.
A HER2 deviation score of ≤1 was found in 89.4% of IHC samples scored by all three
pathologists (286 of 320 samples). Perfect agreement among all three pathologists was found
in 52.2% of these examined samples. Score deviation between two observers is displayed in
table 3. Calculation of quadratic weighed kappa values showed a 'substantial' level of
agreement between observer 1 versus observer 2 (κ = 0.757) and observer 2 versus observer 3
(κ = 0.744). Observer 1 and observer 3 achieved an 'almost perfect' agreement (κ = 0.839).
The mean weighed kappa value averaged between all three pathologists was κ = 0.780, a
'substantial' level of inter-observer agreement.
When both IHC 0 and IHC 1+ were classified as negative, IHC 2+ as equivocal and IHC 3+
as positive, perfect agreement among all three pathologists rose to 76.3% of cases, while
kappa value remained stable at κ = 0.787. Inter-observer agreement on IHC 3+ scoring was
'almost perfect' (κ = 0.854, 95.0% perfect agreement among all three pathologists).
By tumor type, inter-observer agreement in IHC scoring on HER2 was markedly higher in
intestinal tumor type than in diffuse or mixed tumor type. Agreement among the intestinal
tumor type showed 'almost perfect' agreement with a mean squared weighed kappa value of κ
= 0.815 among all three pathologists (52.9% perfect agreement among all three pathologists),
while agreement among diffuse or mixed tumor types was 'moderate' agreement with a mean
squared weighed kappa value of only κ = 0.566 among all three pathologists (40,9% perfect
agreement among all three pathologists).
By tumor location, inter-observer agreement in IHC scoring on HER2 was similar in
esophageal primary tumor location and gastric primary tumor location, with mean squared
weighed kappa value of respectively κ = 0,792 (51.0% perfect agreement among all three
pathologists) and κ = 0.719 (45.4% perfect agreement among all three pathologists). Table 3 Inter-observer agreement in IHC scoring on HER2 by pairs
Pair of raters κ value* Score deviation (n) 0 1 2 3 total
1 vs. 2 0.757 209 90 18 3 320 1 vs. 3 0.839 229 82 9 0 320 2 vs. 3 0.744 202 99 20 2 323
*Calculated using squared weighed kappa statistics. IHC=immunohistochemistry. HER2=human epidermal growth factor receptor 2.
Clinicopathological characteristics
Patient characteristics and HER2 positivity are displayed in table 4. Of the 323 samples on
which IHC and CISH (if indicated) was performed, HER2 status in two samples could not be
established, as IHC was scored 2+ and repeated CISH tests showed no signal. These two
samples were excluded from clinicopathological characteristic and HER2 positivity analysis.
Of the 321 remaining evaluated samples of gastric or esophageal adenocarcinoma, patients
were predominantly male (68.5%) and older than 65 years of age (64.5%). Many patients
presented with advanced disease (52.3%) compared to not advanced disease (40.2%). Of 168
patients who presented with advanced disease, most patients (139 patients, 82.7%) presented
with stage IV metastasized disease compared to (technically inoperable) locally advanced
stage III disease (29 patients, 17.3%). Of 129 patients who presented with not advanced
21
disease, 35 (27.1%) had stage I disease, 37 (28.7%) had stage II disease, 55 (42.6%) had stage
III disease and in 2 patients (1.6%), tumor stage was unknown.
The majority of samples were taken from primary tumors (98.8%) as only four samples came
from metastatic lesions. These primary tumor samples were predominantly biopsy specimens
obtained by endoscopy (71.0%). All metastatic lesion samples were biopsies. Esophageal
primary tumor (47.4%) was nearly as common as gastric primary tumor (50.5%). Primary
tumors located in the esophagus were mostly classified as gastro-esophageal junction tumors
(64.5% of esophageal primary tumors). Tumor type according to Laurén's classification was
predominantly intestinal type (53.9%), followed by diffuse (25.5%) and mixed type (20.6%).
The intestinal tumor type was more common in esophageal primary tumors than in gastric
primary tumors, with 98 out of 152 cases (64.5%) compared to 74 out of 162 cases (45.7%), a
strongly significant (p=0.001) difference.
HER2 positivity was found in 50 of 321 patients (15.6%). HER2 positivity rate was similar in
males and females (17.8% versus 10.9%), age below or above 65 years of age (13.2% versus
16.9%) and in advanced or not advanced disease status (17.2% versus 14.0%). Both univariate
and multivariate analysis showed no statistical difference in these comparisons. Biopsy
specimens showed statistically significant higher HER2 positivity rate in comparison to
surgical specimens (18.0% versus 9.0%, p=0.048), but this was not significant in multivariate
analysis (p=0.492).
By location, 76% (38 samples) of HER2 positive tumors were primary esophageal tumors and
24% (12 samples) were primary gastric tumors, a difference statistically strongly significant
in both univariate and multivariate analysis (both p<0.001). Of the HER2 positive primary
esophageal tumors, 55% (21 samples) were classified as GEJ tumors. As such, GEJ tumors
were HER2 positive in 21.4% of cases, while distal esophageal tumors were HER2 positive in
31.5% of cases. In 25 of 54 primary distal esophageal tumors, the primary tumor was located
in a Barrett's esophagus. Of these, 11 (44%) showed HER2 positivity.
By tumor type, HER2 positive tumors were predominantly of the intestinal type (84%),
instead of diffuse (6%) or mixed type (10%). The intestinal tumor type showed a statistically
strongly significant higher rate of HER2 positivity than diffuse or mixed type in univariate
(p<0.001) and multivariate (p=0.001) analysis, with HER2 positivity among 24.3% of all
intestinal type tumors compared to 3.7% in diffuse type tumors and 7.8% in mixed type
tumors.
22
23
Triplet chemotherapy subgroup
Of 49 patients initially selected for the subgroup analysis, tumor material was absent in two
patients. In one patient, HER2 status could not be established due to IHC 2+ score and
repeated 'no signal' CISH. These three patients were excluded from subgroup analysis (figure
6).
Figure 6 Chemotherapy subgroup HER2 status HER2=human epidermal growth factor receptor 2.
Of the remaining 46 patients, all of whom received triplet chemotherapy treatment, seven
tested HER2 positive and 39 HER2 negative. This 15.2% HER2 positivity rate is in line with
the 15.6% HER2 positivity rate found in the total study group. Clinicopathological
characteristics of patients receiving triplet chemotherapy are displayed in table 5.
All patients received a triplet chemotherapy regimen of a fluoropyrimide, a platinum
compound and an anthracyclin as ECC (32.6%) or as EOX (67.4%). Patients received 1 to 8
cycles of chemotherapy (median 4 cycles), depending on tumor response, side effects and
patient condition.
A total of 42 patients (91.3%) showed progression (clinical progression or death) after starting
chemotherapy. A total of 40 patients (87.0%) had died at the time this study was conducted.
Of the seven HER2 positive patients, three were still alive of whom two without progression
at the time this study was conducted.
24
Table 5 Triplet chemo therapy population characteristics (n = 46)
Characteristics Total, n (%)
HER2 positive HER2 negative
39 7
(84.8) (15.2)
Male Female
32 14
(69.6) (30.4)
Age at diagnosis <65 ≥65
25 21
(54.3) (45.7)
Specimen used for HER2 tests Primary tumor biopsy Primary tumor resection Metastatic lesion
37
8 1
(80.4) (17.4) (2.2)
Tumor type* Diffuse Intestinal Mixed
16 22
8
(34.8) (47.8) (17.4)
Primary tumor location†
Esophagus Distal esophagus Gastro-esophageal junction Stomach Cardia Non-cardia Unknown
16
5 11 26
8 18
4
(34.8) (56.5) (8.7)
Chemotherapy regimen ECC EOX
15 31
(32.6) (67.4)
*Tumor type according to Laurén's classification. †
Primary tumor location according to TNM7 guidelines. HER2=human epidermal growth factor 2. ECC=epirubicin, cisplatin and capecitabine. EOX=epirubicin, oxaliplatin and capecitabine (Xeloda®).
Total subgroup median PFS was 6.1 months (0.4-73.2 months) and median OS was 7.7
months (range 0.5-73.2 months). Kaplan-Meier curves on PFS and OS in HER2 positive
versus HER2 negative patients are displayed in figure 7. HER2 positive patients had higher
PFS than HER2 positive patients (hazard ratio 3.08). Median PFS was 4.4 months (range 0.5-
73.2 months) versus 6.1 months (range 0.4-15.3 months). This difference was not significant
in univariate analysis (log-rank test p = 0.129). In multivariate analysis, HER2 status was not
an independent predictor of PFS (the hazard ratio of 3.08 had a 95% CI of 0.83-11.4,
p=0.093). None of other variables included in multivariate analysis showed statistical
significance (table 6).
HER2 positive patients had higher OS compared to HER2 negative patients (hazard ratio
2.07). Median OS was 8.5 months (range 0.5-73.2 months) versus 7.5 months (range 1.0-22.1
months), a not significant difference in univariate analysis (log-rank test p = 0.160). In
multivariate analysis, HER2 status was not an independent predictor of OS (the hazard ratio
of 2.07 had a 95% CI of 0.55-7.83 , p=0.282). None of the other variables included in
multivariate analysis showed statistical significance (table 7).
25
Figure 7 Kaplan-Meier curves for progression free survival (PFS) and overall survival (OS) for patients with HER2-positive (n=7) and HER2 negative (n=39) disease. HER2=human epidermal growth factor receptor 2.
Table 6 Multivariate survival analysis*
Characteristics Progression free survival Overall survival
Hazard ratio (95% CI) p Hazard ratio (95% CI) p
HER2 status HER2 positive vs HER2 negative 3.08 (0.83-11.4) 0.093 2.07 (0.55-7.83) 0.282
Sex Male vs female 0.88 (0.36-2.15) 0.775 0.75 (0.30-1.89) 0.543
Age <65 years vs ≥65 years 1.40 (0.62-3.13) 0.417 1.48 (0.65-3.34) 0.349
Tumor type** Diffuse or mixed vs intestinal 0.92 (0.43-1.95) 0.817 0.77 (0.34-1.76) 0.534
Primary tumor location† Esophageal vs gastric 1.02 (0.39-2.65) 0.917 1.31 (0.47-3.61) 0.607
T stage No breakthrough vs breakthrough 1.26 (0.61-2.58) 0.537 1.24 (0.58-2.66) 0.577
N stage <7 nodes vs ≥ 7 nodes 2.02 (0.88-4.61) 0.096 1.48 (0.62-3.57) 0.380
M stage Locally advanced vs metastasized
2.32 (0.71-7.58) 0.163 2.31 (0.59-9.07) 0.229
*Cox proportional hazard model, cases with missing values (n = 3) were excluded.
**Tumor type according to Laurén's classification. †
Primary tumor location according to TNM7 guidelines. 95% CI=95% confidence interval. HER2=human epidermal growth factor receptor 2.
26
Discussion
Recently, targeted anti-HER2 therapy with trastuzumab was found to offer a new treatment
modality in advanced gastric and gastro-esophageal junction cancer.37
As adenocarcinomas of
the distal esophagus show similar etiological and pathomorphological characteristics, these
tumors too are believed to benefit from anti-HER2 therapy.3-5,7-10
The need is increasing for
appropriate gastro-esophageal HER2 diagnostic protocols, performed by trained pathologists
in capable diagnostic centrums. Unlike other studies, our study performs a direct comparison
between gastric and esophageal cancer, including gastro-esophageal junction cancer, is
performed. Moreover, we performed a direct comparison of tumor location classification
according to the 7th and 6th TNM editions, as studies before 2010 used the 6th edition. Our
study provides new data on inter-observer variability in IHC scoring of HER2 in gastro-
esophageal adenocarcinoma among newly trained pathologists in a single laboratory, and
compares this inter-observer variability among different primary tumor locations or
histological tumor types. Furthermore, we investigate the validity of the modified HER2
scoring system currently used in gastric and esophageal cancer for identifying HER2 status.
Finally, we analyze survival rates of HER2 positive versus HER2 negative patients in a
subgroup of advanced gastro-esophageal cancer patients receiving the recommended triplet
chemotherapy treatment.
One should keep in mind that our study was conducted retrospectively, and
clinicopathological characteristics such as tumor location and disease status are based on
medical reports made before this study. This is especially important in primary tumor location
determination, as these reports were made before the existence of the 7th TNM edition.
Tumor location
In 2010, the 7th TNM edition of Classification of Malignant Tumours was introduced.7 A
major change in this classification protocol in comparison to the previous 6th edition was that
gastro-esophageal junction (GEJ) tumors were no longer classified as gastric cancer, but
instead as esophageal cancer. Moreover, gastric cardia tumors within 5 cm of the esophagus
and reaching in the esophagus were now classified as GEJ tumors and thus treated as
esophageal tumors. Classifying GEJ and gastric cardia tumors as esophageal cancer would
seem a logical alteration, as these adenocarcinomas show similar etiological, epidemiological
and pathomorphological patterns, in contrast to non-cardia gastric tumors3-5,7-10
. Additionally,
because of these similarities, one could even argue to include all gastric cardia tumors in
esophageal cancer staging instead of only those with growth into the esophagus, as stated by
the current 7th TNM classification system. On the other hand, it is known that differentiation
between GEJ, cardia and gastric non-cardia tumor location by endoscopy varies with different
investigators.8
Most studies on gastro-esophageal cancer up to 2010, including the ToGA study, used the 6th
TNM edition. In our study, using the 7th TNM edition, we found a similar distribution of
gastric primary tumors (162 patients, 50.5%) and esophageal tumors (152 patients, 47.4%).
However, of the esophageal tumors, 98 (30.5% of all tumors) were GEJ tumors, including
gastric cardia tumors within 5 cm of the esophagus and reaching in the esophagus. These
tumors would have been classified as gastric tumors in the 6th TNM edition, in which the
primary tumor location would primarily have been gastric instead of esophageal (81.0% vs.
16.8%). This illustrates that a large portion of studies on gastric cancer using the 6th TNM
edition will have included a marked number of tumors that would be staged as esophageal
tumors when using the 7th TNM edition. Moreover, the N category has undergone major
changes in esophageal tumor staging, as the 7th TNM edition applies an N0 to N3 scale
instead of an N0 to N1 scale as in the 6th TNM edition, along with more lymphatic nodes
27
being accounted as regional instead of metastatic nodes.7,9
Studies using the 7th TNM edition
can therefore not be readily compared to studies using the 6th TNM edition. However,
differences in clinical outcome between the use of 6th versus 7th TNM editions are mostly
found in early stage, resectable tumors.11,12
In advanced disease, survival patterns of
esophageal and GEJ cancer are similar to that of advanced gastric cancer, regardless which
version of the TNM classification is used to establish cancer stage.13
HER2 overexpression: primary tumor location
The HER2 overexpression rate in our study, comprising roughly half gastric and half
esophageal adenocarcinoma according to the 7th TNM edition, was 15.6% (50 out of 321
patients). This is similar compared to the 15-30% overexpression found in most other studies
on both gastric and esophageal adenocarcinoma.27-36
Of the 50 tumors showing HER2
overexpression, primary tumor location was classified as gastric in 12 patients (7.4% of all
gastric tumors) and as esophageal in 38 patients (24.8% of all esophageal tumors). As the
number of both primary tumor locations was similar in our study, this shows HER2 positivity
is more common in esophageal cancer, which is in contrast to our hypothesis that HER2
positivity would be similar in both primary tumor locations.
Many other studies on gastro-esophageal cancer included either only gastric cancer (with GEJ
cancer) or esophageal cancer, not both tumor locations together. In studies including both
gastric and GEJ tumors, a higher HER2 positivity rate was found in GEJ tumor location than
in gastric tumor location, with differences of 24-33% versus 16-21% respectively.27-29
Additionally, higher HER2 positivity rates were found in proximal gastric cancer (including
the GEJ) than in distal gastric cancer.52,68
Another study, on esophageal and GEJ
adenocarcinoma, found a GEJ primary tumor HER2 positivity rate of 28% versus 15% in
esophageal primary tumors.33
As GEJ tumors would now be classified as esophageal tumors,
studies on gastric cancer including GEJ tumors before the use of the 7th TNM edition would
most likely show lower HER2 positivity rates if these GEJ tumors were not included. In
hindsight, a tendency towards higher HER2 positivity rate in proximal gastric tumors which
are now classified as esophageal tumors is thus apparent in these studies. On the other hand,
distal esophageal primary tumors (without growth in the GEJ) in our study showed 31.5%
HER2 positivity, which is higher than the 15-29% HER2 positivity rate found in studies on
distal esophageal adenocarcinoma alone.31-36
This could be related to underlying Barrett's
esophagus in the primary distal esophageal tumors included in our study, which was
frequently reported (25 of 54 distal esophageal tumors). HER2 positivity in these cases was
44%. Indeed, other studies have found HER2 overexpression to be associated with tumor
progression from Barrett's esophagus and with adenocarcinomas in Barrett's esophagus.32,69,70
The higher tendency towards esophageal tumor HER2 positivity found in our study is thus
partly explained by the classification of GEJ primary tumors as esophageal since the 7th TNM
edition, and partly by a higher HER2 positivity rate in distal esophageal tumors in underlying
Barrett's esophagus.
HER2 overexpression: histological tumor type
Many studies have shown that HER2 overexpression is histologically more common in
Laurén's intestinal type of adenocarcinoma than in the diffuse type (15-35% vs. <10%).15,28-
30,37,38 In our study we found similar results, with HER2 positivity in 24.3% of all intestinal
type tumors, 3.7% of all diffuse type tumors and 7.8% of all mixed type tumors. As such, the
vast majority of HER2 positive tumors in our study were of the intestinal type as this tumor
type comprised 42 of all 50 (84%) HER2 positive samples.
It is known that esophageal adenocarcinomas predominantly show histological intestinal type
characteristics. This is thought to be related to the intestinal metaplasia caused by esophageal
28
reflux.18
As adenocarcinoma of the gastro-esophageal junction is associated with the same
etiological pattern, the intestinal tumor type would be expected to be the predominant tumor
type in this group as well. Indeed, we found that the intestinal tumor type was significantly
more common in esophageal primary tumors (including GEJ tumors) than in gastric primary
tumors (p=0.001). As such, the higher HER2 positivity rates in esophageal adenocarcinoma in
comparison to gastric adenocarcinoma could be related to a higher prevalence of intestinal
tumor type. This would however not provide a full explanation, as our study showed that both
primary esophageal tumor location and intestinal tumor type are independently associated
with positive HER2 status in a multivariate analysis model including both factors. The reason
for the selective overexpression of HER2 in intestinal type adenocarcinoma is thought to be
complex, and no explanation has been found yet.30,71
Inter-observer variability
In our study, immunohistochemistry on HER2 was scored by three pathologists who, though
familiar with HER2 scoring in breast cancer, were unfamiliar with HER2 scoring in gastric
and esophageal adenocarcinoma. Prior to the study, they were educated on the IHC scoring
system defined by Hofmann et al. in 2008 with additional guidelines established by Rüschoff
et al. in 2010. This study by Rüschoff et al. contains the only available data on inter-observer
variability in IHC scoring on HER2 in gastro-esophageal adenocardinoma, along with an
Australian study (the GaTHER study; Gastric HER2 Testing Study) on inter-laboratory
agreement in IHC and ISH among nine laboratories by Fox et al.48,51
Warneke at al. is a third,
very recent study investigating inter-observer variability, but as this study has compared
HER2 positive and HER2 negative results of HER2 testing including CISH tests among two
observers instead of IHC scoring alone, it cannot be compared to our study results.52
In the study by Rüschoff et al., inter-observer variability was determined in IHC scoring of
HER2 in gastric adenocarcinoma among six pathologists. As they used their established
guidelines in the study itself, the same modified scoring system and guidelines on HER2
scoring was used as in our study. The six pathologists participating were, in contrast to our
pathologists, thoroughly familiar with the scoring system. Additionally, they held a two-day
meeting on 30 tumor samples to reach consensus on pitfalls in the scoring system. They
found a HER2 deviation score of ≤1 between pathologists in 95.6% of 547 scored samples.
This was 89.4% of 321 samples in our study, which suggests that properly educated
pathologists previously unfamiliar with the IHC scoring of HER2 in gastro-esophageal
adenocarcinoma, can achieve results that are similar to pathologists that are experienced and
well-trained on this area. The kappa value found by Rüschoff et al. was κ = 0.61, being
'moderate' agreement among the six pathologists. However, the kappa statistic used was the
unweighed kappa for multiple raters according to Conger.72
This method does not take into
account the nature of the ordinal scale of the IHC scores on HER2. The kappa value would
therefore most likely have been higher if a weighed kappa was applied, as in our study.
Among the three pathologists in our study an inter-observer agreement kappa value of κ =
0.78 was found using the mean squared weighed kappa value between pairs of pathologists. If
the unweighed kappa for multiple raters according to Conger would have been used instead,
as was the case in the study by Rüschoff et al., the kappa value of our study would have been
κ = 0.504. Though lower than the κ = 0.61 value found in the study by Rüschoff, this kappa
value would still be classified as 'moderate' agreement.67
Again, this shows that inter-observer
agreement among newly trained pathologists in our study was similar to inter-observer
agreement among pathologists experienced in IHC scoring of HER2 in gastric and esophageal
adenocarcinoma.
In the Australian GaTHER study on inter-laboratory agreement among nine laboratories,
scoring was performed by pathologists affiliated to the individual laboratories, thus inter-
29
observer variability influenced the outcome.51
This study reported relative inexperience of
participating pathologists with the HER2 scoring system in IHC on gastro-esophageal
tumors, as was the case in our study. They used the scoring system by Hofmann et al. both
with and without the additional guidelines by Rüschoff et al. Moreover, the study compared
the agreement between the scoring system with and without the additional guidelines, and
concluded a good agreement between the two scoring systems. To analyze inter-laboratory
agreement, Fox et al. also used unweighed kappa statistics instead of weighed statistics. On
IHC scoring inter-laboratory agreement, they found a mean unweighed kappa value of κ =
0.46. If these statistics were applied in our study, the mean unweighed kappa of κ = 0.505
shows a similar result.
In conclusion, pathologists previously unfamiliar with the modified IHC scoring system on
HER2 in gastric and esophageal adenocarcinoma can become properly competent with this
scoring system by receiving appropriate training along with familiarization with the additional
guidelines by Rüschoff et al.
Inter-observer variability: primary tumor location and histological tumor type
To the best of our knowledge, our study is the first in which inter-observer variability on IHC
scoring of HER2 in gastro-esophageal adenocarcinoma was studied in relation to tumor
location and tumor type. We found that there was similar agreement in primary esophageal
and gastric tumor samples, but agreement among tumor types varied greatly. IHC scoring on
HER2 in the intestinal tumor type gave rise to an 'almost perfect' agreement (κ = 0.815), while
agreement in scoring diffuse and mixed type gave rise to a 'moderate' agreement (κ = 0.566).
This shows that inter-observer agreement is markedly higher when scoring the intestinal type
tumor on HER2 in comparison to diffuse or mixed type tumors. This could be explained by
the nature of the diffuse type carcinoma, which is usually less well differentiated than the
intestinal type.16,17
Because of the poor differentiated growth of diffuse type carcinoma, cell-
cell contact is more difficult to observe than in the better differentiated and thus histologically
more ordered intestinal type carcinoma. As this cell-cell contact in membranous HER2
staining is mandatory in the minimum of five clustered stained tumor cells in IHC scoring,
these stained clusters are harder to identify and thus more difficult to score. Moreover, the
diffuse type carcinoma can be of signet ring cell type. These signet rings have a center
cytoplasmatic vacuole, pushing the cytoplasma more towards the membranes of the cell. As
such, cytoplasma and membranes are close together in signet ring cells, complicating visual
differentiation between cytoplasmatic staining (no HER2 overexpression) and membraneous
staining (possible HER2 overexpression) in IHC scoring.52
Histology images of both tumor
types and images of the IHC scoring difficulties in the diffuse tumor type are displayed in
images J through M in the appendix.
HER2 diagnostics: scoring system validation
Immunohistochemistry
Since only recently, HER2 status in gastric and esophageal adenocarcinoma is determined
using a modified HER2 scoring system, defined by Hofmann et al. (in 2008) and Rüschoff et
al. (in 2010). Studies before these years did not use a standardized HER2 scoring system, and
cannot be readily compared with studies using the modified scoring system, as many tumors
would have received false-negative score. Conflicting reports on HER2 overexpression and
survival before 2008 could lie in lack of this standardized scoring system.38,48
When using the
modified HER2 scoring system in clinical practice, IHC is used to differentiate between
'negative', 'positive' and 'equivocal' HER2 status and ISH tests are only used to determine
equivocal status.
30
In our study, 50 of 323 samples were found HER2 positive. Of these, a majority of 42
samples had been identified as HER2 positive in IHC testing. As such, the majority of HER2
positive samples were classified as positive in IHC testing. Of these IHC positive samples, 34
received unanimous IHC 3+ score. Eight samples were scored IHC 3+ by two and IHC 2+ by
one of the pathologists. As CISH would have followed this IHC 2+ score, HER2 positive
status would have been determined regardless of the sample not receiving a IHC 3+ score by
this pathologist.
These findings show that IHC testing with the modified HER2 scoring system is an
appropriate method of establishing a positive HER2 status in gastric and esophageal
adenocarcinoma.
We found improved inter-observer agreement if instead of comparing all scores, scores of
both IHC 0 and 1+ were classified as negative, IHC 2+ as equivocal and IHC+ as positive.
Additionally, inter-observer agreement was 'almost perfect' on IHC 3+ scoring. This suggests
that there is a high level of inter-observer agreement in IHC 0/1+ and IHC 3+ scores, and that
the highest level of disagreement is among the tumor samples scored 2+ by at least one
pathologist. In fact, of the 95 cases scored 2+ by at least one pathologists, in only 16 cases
(16.8%) perfect agreement on the 2+ score existed. Indeed, as the nature of the 2+ score is
'equivocal', disagreement among this score could be expected. However, because the IHC 2+
scored samples are followed by ISH to ensure the true HER2 status, many of the cases of
disagreement in this study were clarified by ISH testing. Moreover, in our data the majority of
the 95 cases scored 2+ by at least one pathologist were HER2 negative as they showed no
amplification in CISH testing (88 of 95 cases, 92.6%). The inter-observer disagreement
among the 2+ scored samples would therefore have little consequence for HER2 status in
clinical practice.
Summarized, the current modified IHC scoring system on HER2 in gastro-esophageal
adenocarcinoma is an excellent method to identify HER2 positive and HER2 negative status,
with high rates of inter-observer agreement. Weak to moderate HER staining in IHC, which
could be scored as equivocal IHC 2+, gives rise to higher levels of inter-observer
disagreement, but often appeared HER2 negative in CISH tests, thus has relatively low
clinical consequence.
In situ hybridization
ISH testing is an excellent method of detecting HER2 overexpression in breast cancer and
gastro-esophageal cancer. In comparison to IHC testing, ISH testing is less prone to loss of
antigenic immunoreactivity during fixation procedure leading to loss of sensitivity.73
Additionally, IHC gives 10% false results (false positive or false negative) in comparison to
FISH in breast cancer, and IHC scoring is prone to inter-observer variability.48-52
Considering
this, some people argue that in gastro-esophageal cancer, where tumor heterogeneity is higher
and membranous staining more irregular than in breast cancer, ISH testing should be the
standard objectivation method of HER2 status. There are however a number of remarks.
Firstly, ISH testing is a more costly method compared to IHC. Secondly, ISH testing is less
readily available in most laboratories. The third, clinically important, remark is that
concordance between IHC and ISH results in gastro-esophageal cancer is much lower than in
breast cancer, where IHC 0 or IHC 1+ samples are usually ISH negative. This was found in
the ToGA trial, where 22.4% of included patients were FISH positive while IHC score was 0
or 1+. This frequency was in fact similar to the frequency of IHC 2+/FISH positive patients
(27.2%). A recent study on resected early stage gastric and GEJ tumors confirmed this
finding.52
The ToGA trial found that adding trastuzumab to chemotherapy in HER2 positive
patients (defined as either IHC 3+, FISH positive, or both) results in significantly improved
median survival of 13.8 versus 11.1 months. However, in included patients who received
31
negative IHC score (IHC 0 or IHC 1+) with positive FISH, the study found no statistically
significant difference in survival, with median survival seemingly impaired 8.7 versus 10.0
months.37
This suggests a high level of non-responders among IHC negative/FISH positive
patients, which means HER2 overexpression suggested by only a positive FISH test but with
negative IHC result would have limited clinical value. This notion would also account for
CISH testing, as CISH shows similar results on HER2 gene amplification detection in
comparison to FISH in both breast cancer and gastric cancer.63,64
Performing only ISH testing
as the standard objectivation method for HER2 status in gastro-esophageal adenocarcinoma
would therefore not be advisable, as anti-HER2 therapy in IHC negative/FISH positive
patients is not indicated.
Most of the samples classified HER2 positive in IHC testing showed amplification in
complementary CISH testing (39 of 41 assessable samples, 95.1%), showing only two cases
of disconcordance as IHC positive/CISH negative patients (4.9%). This is similar to the 3% of
IHC positive/FISH negative patients found in de ToGA trial. Administering trastuzumab to
patients with this IHC positive/ISH negative disconcordance seemed to improve overall
survival in the ToGA trial, however not statistically significant as the number of these patients
was few (n=15).
IHC testing is an excellent method of establishing positive or negative HER2 status as
concluded above. Equivocal IHC status often shows no amplification in ISH testing, and IHC
negative status has no clinical consequence regardless of ISH amplification. Therefore, IHC
testing on HER2 in gastro-esophageal adenocarcinoma as a whole is an excellent method to
differentiate HER2 status before applying ISH. As such, and as expected, the currently used
standardized modified scoring system on HER2 scoring, only implementing ISH in equivocal
IHC HER2 staining, is an excellent clinically applicable protocol to establish HER2 status in
gastro-esophageal adenocarcinoma.
Triplet chemotherapy: HER2 status and survival rates
Reports of influence of HER2 status on survival rates in gastric and esophageal cancer are
conflicting in both early stage and advanced stage disease.20,27,28,30-33,39-42
Recently, the use of
trastuzumab in HER2-positive advanced gastric cancer or gastro-oesophageal junction cancer
was associated with a significant improvement of median survival, of 11.1 to 13.8 months in
the ToGA trial.37
Surprisingly, the ToGA trial did not use the standard recommended triplet
chemotherapy of a fluoropyrimide, a platinum compound and an anthracyclin.21,23
Instead,
they used a doublet chemotherapy with only a fluoropyrimide and a platinum compound,
without the anthracyclin. This might be because of known cardiotoxic side effects of both
anthacyclins and trastuzumab, but it was shown that both drugs can be safely administered
simultaneously in HER2 positive breast cancer patients.74
To the best of our knowledge, the
influence of HER2 status on survival of patients with advanced gastro-esophageal cancer
receiving triplet chemotherapy regimen has not been studied up to date. The triplet
chemotherapy regimen has been studied in early gastro-esophageal cancer in the
'TransMAGIC' study, which found that HER2 status was not an independent prognostic factor
in a group of 385 patients (of which 42 HER2 positive, 10.9%). However, this study
contained only early gastro-esophageal cancer patients which received the triplet
chemotherapy as neo-adjuvant and adjuvant chemotherapy surrounding curative resection,
and can therefore not be compared to studies on advanced gastro-esophageal cancer patients.75
Janjigian et al. found that HER2 status was not an independent predictor of PFS or OS in a
group of 381 patients (of which 78 HER2 positive, 20%) with advanced gastro-esophageal
cancer receiving palliative chemotherapy (none received trastuzumab). In this study, OS of
HER2 positive patients and HER2 negative patients was 13.9 and 11.4 months, respectively.
Although the studied population received a variety of different chemotherapy regimens, none
32
received the recommended triplet therapy and none of the chemotherapy regimens contained
an anthracyclin.27
In our study, we selected a subgroup of patients with advanced stage gastric or esophageal
adenocarcinoma that received palliative triplet chemotherapy treatment of a fluoropyrimide, a
platinum compound and an anthracyclin. In this subgroup, median PFS was 6.1 months (0.4-
73.2 months) and median OS was 7.7 months (range 0.5-73.2 months), which are lower than
the survival rates of 11-14 months reported in other studies on advanced gastric
carcinoma.27,37
This is most likely because of the prospective nature of these studies compared
to the retrospective nature of our study, in which patients often did not receive chemotherapy
treatment immediately after diagnosis. As such, the difference in survival rates could be
explained by this delay.
The subgroup comprised only 46 patients of whom seven showed HER2 positivity. We found
a seemingly improved prognosis in the HER2 positive patients compared to the HER2
negative patients in terms of PFS (hazard ratio 3.08) and OS (hazard ratio 2.07). This would
suggest HER2 positivity is associated with better survival rates, as we expected. The survival
differences were however not significant in univariate and multivariate statistical analysis.
The number of patients in the triplet chemotherapy subgroup was too small to prove statistical
significance, and the survival rates of HER2 positive patients were greatly influenced by one
patient who survived without progression for at least 73.2 months compared to a median OS
and PFS of 8.5 months among both HER2 positive and HER2 negative patients. Moreover,
there is weakness in the determination of clinicopathological characteristics of the subgroup
in our study, especially in TNM stage, as this was based on different reports in different
patients, varying from pathological reports to EUS reports and old CT/PET scan reports to
presently (for the study) revised reports. To justly establish influence of HER2 status on
survival rates in advanced gastro-esophageal cancer patients receiving triplet chemotherapy, a
prospective study should be conducted with a larger patient population, with careful and
uniform establishment and monitoring patient characteristics and of tumor characteristics and
progression.
33
Conclusions
HER2 positivity occurs in 15.6% of invasive gastro-esophageal adenocarcinoma in Western
patients. HER2 positivity rates in esophageal primary tumors, including gastro-esophageal
junction tumors, are higher than in gastric primary tumors (esophageal 25% vs. gastric 7.4%).
HER2 positivity rates are higher in the intestinal histological tumor type than in the diffuse or
mixed tumor type (intestinal 24.3% vs. diffuse or mixed 11.5%).
Using the 7th TNM edition Classification of Malignant Tumours of the UICC, many tumors
previously classified as gastric cancer are now classified as esophageal cancer (30.5% of all
tumors in this study).
Inter-observer agreement on IHC scoring among pathologists newly educated on the gastro-
esophageal HER2 scoring system was good (κ = 0.78), and in fact similar to inter-observer
agreement among pathologists thoroughly familiar with this HER2 scoring system. Inter-
observer agreement was similar in gastric or esophageal tumor samples. Inter-observer
agreement was better in tumors of the intestinal tumor type compared to tumors of the diffuse
or mixed tumor type.
The currently used standardized modified scoring system on HER2 scoring is an excellent,
clinically applicable protocol to establish HER2 status in gastro-esophageal adenocarcinoma
in appropriately trained pathologists.
Further studies are needed on influence of HER2 status on survival in advanced disease
patients receiving the recommended palliative triplet chemotherapy treatment of a
fluoropyrimide, a platinum compound and an anthracyclin.
34
Acknowledgements
In the end, many people were involved in the conduction and completion of this project and
my gratitude goes out to all of them.
I would like to express my deep gratitude to dr. A.L.T. Imholz for his patient guidance,
enthusiastic encouragement and all his help in making this project possible. My grateful
thanks are extended to M.M. Smits, for her enthusiasm, advice and assistance in the
pathological analysis. I am particularly grateful to her and the other pathologists dr. M. Hage
and M. Louwen, for scoring and analyzing numerous tumor samples for the database. Special
thanks go to prof. Mark van de Vijver (AMC), for his small but important part in the
education of the pathologists.
I would like to thank Cobi van Rijn and Erik Eisink for their help in this project and
familiarizing me with histopathological techniques. My great appreciation goes out to the
other staff of the department of Clinical Pathology for their help and patience during my time
in the laboratory.
Special thanks go out to dr. W. van der Zwet, for his help and guidance on the statistical
analysis.
Assistance provided by R. van Dijk and A. Fortuin of the Radiology department was greatly
appreciated.
I would like to thank the Dutch Integral Cancer Centre (Integraal Kankercentrum Nederland,
IKNL) for their assistance in the collection of my data.
I am grateful to Roche Pharma, Germany and Ventana Medical Systems, Inc, by whom this
study was partly supported.
Finally, I would like to thank my family and friends for their unconditional support and my
fellow students at the Deventer Hospital for their moral support and distraction during coffee
breaks.
35
References
1. Garcia M, Jemal A, Ward EM, et al. Global cancer facts and figures 2007. Atlanta, GA: American Cancer
Society, 2007.
2. American Cancer Society. Global cancer facts and figures 2nd edition. Atlanta, GA: American Cancer
Society, 2011.
3. Kamangar F, Dores GM, Anderson WF, et al. Patterns of cancer incidence, mortality, and prevalence across
five continents: defining priorities to reduce cancer disparities in different geographic regions of the world. J
Clin Oncol 2006; 24:2137-2150.
4. Hartgrink HH, Jansen EPM, van Grieken NCT, et al. Gastric cancer. Lancet 2009; 374:477-490.
5. Pera M, Manterola C, Vidal O, et al. Epidemiology of esophageal adenocarcinoma. J Surg Oncol 2005;
92:151-159.
6. Moelans CB, van Diest PJ, Milne ANA, et al. HER-2/neu testing and therapy in gastroesophageal
adenocarcinoma. Pathol Res Int 2011; doi:10.4061/2011/674182.
7. Sobin LH, Gospodarowicz M, Wittekind C, (editors), International Union Against Cancer (UICC). TNM
Classification of Malignant Tumours. 7th ed. Hoboken, NJ: Wiley-Blackwell; 2010.
8. Edge SB, Byrd DR, Compton CC, et al., eds. AJCC Cancer Staging Manual. 7th ed. New York, NY: Springer,
2010, p103-125.
9. Sobin LH, Wittekind C. TNM Classification of Malignant Tumours. 6th ed. Hoboken, New Jersey: John
Wiley & Sons; 2002.
10. Heidl G, Langhans P, Mellin W, et al. Adenocarcinomas of esophagus and cardia in comparison with gastric
carcinoma. J Cancer Res Clin Oncol 1993; 120:95-99.
11. Talsma K, van Hagen P, Grotenhuis BA, et al. Comparison of the 6th and 7th editions of the UICC-AJCC
TNM Classification for Esophageal Cancer. Ann Surg Oncol 2012; 19:2142-2148.
12. Gaur P, Hofstetter WL, Bekele BN, et al. Comparison between established and the worldwide esophageal
cancer collaboration staging systems. Ann Thorac Surg 2010; 89:1797-1804.
13. Chau I, Norman AR, Cunningham D, et al. The impact of primary tumour origins in patients with advanced
oesophageal, oesophago-gastric junction and gastric adenocarcinoma – individual patient data from 1775
patients in four randomised controlled trials. Ann Oncol 2009; 20:855-891.
14. Dicken BJ, Bigam DL, Cass C, et al. Gastric adenocarcinoma. Review and considerations for future
directions. Ann Surg 2005; 241:27-39.
15. Laurén P. The two histological main types of gastric carcinoma: diffuse and so-called intestinal-type
carcinoma. An attempt at a histo-clinical classification. Acta Pathol Microbiol Scand 1965; 64:31-49.
16. Adachi Y, Yasuda K, Inomata M, et al. Pathology and prognosis of gastric carcinoma: well versus poorly
diferentiated type. Cancer 2000; 89(7):1418-1424.
17. Yamashita K, Sakuramoto S, Katada N, et al. Diffuse type advanced gastric cancer showing dismal prognosis
is characterized by deeper invasion and emerging peritoneal cancer cell: the latest comparative study to intestinal
advanced gastric cancer. Hepato-gastroenterology 2009; 56:276-281.
18. Derakshan MH, Liptrot S, Paul J, et al. Oesophageal and gastric intestinal-type adenocarcinomas show the
same male predominance due to a 17 year delayed development in females. Gut 2009; 58:16-23.
19. Horner MJ, Ries LAG, Krapcho M, et al. SEER Cancer Statistics Review, 1975-2009. 2012.
http://seer.cancer.gov/csr/1975_2009/ (accessed December 12, 2012).
20. Chua TC, Merrett ND. Clinicopathological factors associated with HER2-positive gastric cancer and its
impact on survival outcomes – A systematic review. Int J Cancer 2012; 130:2845-2856.
21. Wagner AD, Grothe W, Haerting J, et al. Chemotherapy in advanced gastric cancer: a systematic review and
meta-analysis based on aggregate data. J Clin Oncol 2006; 24:2903-2909.
22. Wagner AD, Unverzagt S, Grothe W, et al. Chemotherapy for advanced gastric cancer. Cochrane Database
Syst Rev. 2010; (3):CD004064.
23. Oncoline. Maagcarcinoom: recidief en metastasen. oncoline.nl 2009.
24. Schlessinger J. Cell signaling by receptor tyrosine kinases. Cell 2000; 103:211-225.
25. Yoong J, Michael M, Leong T. Targeted therapies for gastric cancer, current status. Drugs 2011;
71(11):1367-1384.
26. Houldsworth J, Cordon-Cardo C, Ladanyi M, et al. Gene amplification in gastric and esophageal
adenocarcinomas. Cancer Res 1990; 50(19):6414-6422.
27. Janjigian YY, Werner D, Pauligk C, et al. Prognosis of metastatic gastric and gastroesophageal junction
cancer by HER2 status: a European and USA International collaborative analysis. Ann Oncol 2012; 23(10):2656-
2662.
28. Tanner M, Hollmen M, Junttila TT, et al. Amplification of HER-2 in gastric carcinoma: association with
Topiosomerase IIalpha gene amplification, intestinal type, poor prognosis and sensitivity to trastuzumab. Ann
Oncol 2005; 16:273-278.
36
29. Bang Y, H. Chung H, Xu J, et al. Pathological features of advanced gastric cancer (GC): relationship to
human epidermal growth factor receptor 2 (HER2) positivity in the global screening programme of the ToGA
trial. J Clin Oncol 2009; 27:15s (suppl; abstr 4556).
30. Gravalos C, Jimeno A. HER2 in gastric cancer: a new prognostic factor and a novel therapeutic target. Ann
Oncol 2008; 19:1523-1529.
31. Yoon HH, Shi Q, Sukov WR, et al. Adverse prognostic impact of intratumor heterogeneous HER2 gene
amplification in patients with esophageal adenocarcinoma. J Clin Oncol 2012; 30:3932-3938.
32. Yoon HH, Shi Q, Sukov WR, et al. Association of HER2/ErbB2 expression and gene amplification with
pathological features and prognosis in esophageal adenocarcinomas. Clin Cancer Res 2012; 18:546-554.
33. Phillips BE, Tubbs RR, Rice TW, et al. Clinicopathologic features and treatment outcomes of patients with
human epidermal growth factor receptor 2-positive adenocarcinoma of the esophagus and gastroesophageal
junction. Dis Esophagus 2012 Jun 7 doi: 10.1111/j.1442-2050.2012.01369.x.
34. Reichelt U, Duesedau P, Tsourlakis MC, et al. Frequent homogeneous HER-2 amplification in primary and
metastastic adenocarcinoma of the esophagus. Mod Pathol 2007; 20:120-129.
35. Langer R, Rauser S, Feith M, et al. Assessment of ErbB2 (Her2) in oesophageal adenocarcinomas: summary
of a revised immunohistochemical evaluation system, bright field double in situ hybridisation and fluorescence
in situ hybridisation. Mod Pathol 2011; 24:908-916.
36. Hu Y, Bandla S, Godfrey TE, et al. HER2 amplification, overexpression and score criteria in esophageal
adenocarcinoma. Mod Pathol 2011; 24:899-907.
37. Bang YJ, van Cutsem E, 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; 376:687-697.
38. Hofmann M, Stoss O, Shi D, et al. Assessment of a HER2 scoring system for gastric cancer: results from a
validation study. Histopathology 2008; 52:797–805.
39. Gómez-Martin C, Garralda E, José Echarri M, et al. HER2/neu testing for anti-HER2-based therapies in
patients with unresectable and/or metastatic gastric cancer. J Clin Pathol 2012; 65(8):751-757.
40. Jaehne J, Urmacher C, Thaler HT, et al. Expression of Her2/neu oncogene product p185 in correlation to
clinicopathological and prognostic factors of gastric carcinoma. J Cancer Res Clin Oncol 1992; 118:474-479.
41. Song Y, Huang J, Wang JW. Correlation of HER-2/neu gene amplification and protein expression with the
prognosis of advanced gastric cancer patients. Chin J Cancer 2010; 29(1):76-81.
42. Grabsch H, Sivakumar S, Gray S, et al. HER2 expression in gastric cancer: rare, heterogeneous and of no
prognostic value - conclusions from 924 cases of two independent series. Cellular Oncol 2010; 32:57-65.
43. Romond EH, Perez EA, Bryant J, et al. Trastuzumab plus adjuvant chemotherapy for operable HER2-
positive breast cancer. N Engl J Med 2005; 353(16):1673-1684.
44. Piccart-Gebhart MJ, Procter M, Leyland-Jones B, et al. Trastuzumab after adjuvant chemotherapy in HER2-
positive breast cancer. N Engl J Med 2005; 353(16):1659-1672.
45. Smith I, Procter M, Gelber RD, et al. 2-year follow-up of trastuzumab after adjuvant chemotherapy in HER2-
positive breast cancer: a randomised controlled trial. Lancet 2007; 369(9555):29-36.
46. Slamon DJ, Leyland-Jones B, Shak S, et al. Use of chemotherapy plus a monoclonal antibody against HER2
for metastatic breast cancer that overexpresses HER2. N Engl J Med 2001; 344(11):783-792.
47. Hudis CA. Trastuzumab – mechanism of action and use in clinical practice. N Eng J Med 2007; 357:39-51.
48. Rüschoff J, Dietel M, Baretton G, et al. HER2 diagnostics in gastric cancer - guideline validation and
development of standardized immunohistochemical testing. Virchows Arch 2010; 457:299-307.
49. Gancberg D, Jarvinen T, di Leo A, et al. Evaluation of HER-2/NEU protein expression in breast cancer by
immunohistochemistry: an interlaboratory study assessing the reproducibility of HER-2/NEU testing. Breast
Cancer Res Treat 2002; 74(2):113-120.
50. Jacobs TW, Gown AM, Yaziji H, et al. Comparison of fluorescence in situ hybridization and
immunohistochemistry for the evaluation of HER-2/neu in breast cancer. J Clin Oncol 1999; 17(7):1974-1982.
51. Fox SB, Kumarasinghe MP, Armes JE, et al. Gastric HER2 testing study (GaTHER): an evaluation of
gastric/gastroesophageal junction cancer testing accuracy in Australia. Am J Surg Pathol 2012; 36:577-582.
52. Warneke VS, Behrens HM, Böger C, et al. Her2/neu testing in gastric cancer: evaluating the risk of sampling
errors. Ann Oncol 2013; 24(3):725-733.
53. National Comprehensive Cancer Network, Inc. NCCN Clinical Practice Guidelines in Oncology: Gastric
Cancer. Version 2.2012.
54. National Comprehensive Cancer Network, Inc. NCCN Clinical Practice Guidelines in Oncology: Esophageal
and Esophagogastric Junction Cancers. Version 2.2012.
55. Eisenhauer EA, Therasse P, Boaerts J, et al. New response evaluation criteria in solid tumours: Revised
RECIST guideline (version 1.1). Eur J Cancer 2009; 45(2):228-247.
56. Rüschoff J, Hanna W, Bilous M, et al. HER2 testing in gastric cancer: a practical approach. Mod Pathol
2012; 25(5):637-650.
37
57. Rüschoff J, Nagelmeier I, Baretton G, et al. Her2-Diagnostik beim Magenkarzinom. Was ist anders im
Vergleich zum Mammakarzinom? Pathologe 2010; 31:208-217.
58. Powell WC, Hicks DG, Prescott N, et al. A new rabbit monoclonal antibody (4B5) for the
immunohistochemical (IHC) determination of the HER2 status in breast cancer: comparison with CB11,
fluorescence in situ hybridization (FISH) and interlaboratory reproducibility. Appl Immunohistochem Mol
Morphol 2007; 15(1):94-102.
59. HER2 testing: gastric cancer. her2testing.org 2013.
60. Press MF, Slamon DJ, Flom KJ, et al. Evaluation of HER-2/neu gene amplification and overexpression:
comparison of frequently used assay methods in a molecularly characterized cohort of breast cancer specimens. J
Clin Oncol 2002; 20(14):3095-3105.
61. Papouchado BG, Myles J, Lloyd RV, et al. Silver in situ hybridization (SISH) for determination of HER2
gene status in breast carcinoma. Am J Surg Pathol 2010; 34:767-776.
62. Yan B, Yau EX, Bte Omar SS, et al. A study of HER2 gene amplification and protein expression in gastric
cancer. J Clin Pathol 2010; 63:839–842.
63. Kiyose S, Igarashi H, Nagura K, et al. Chromogenic in situ hybridization (CISH) to detect HER2 gene
amplification in breast and gastric cancer: Comparison with immunohistochemistry (IHC) and fluorescence in
situ hybridization (FISH). Pathol Int 2012; 62(11):728-734.
64. Tanner T, Gancberg D, di Leo A, et al. Chromogenic in situ hybridization. A practical alternative for
fluorescence in situ hybridization to detect HER-2/neu oncogene amplification in archival breast cancer samples.
Am J Pathol 2000; 157(5):1467-1472.
65. Middleton LP, Price KM, Puig P, et al. Implementation of American Society of Clinical Oncology/College
of American Pathologists HER2 guideline recommendations in a tertiary care facility increases HER2
immunohistochemistry and fluorescence in situ hybridization concordance and decreases the number of
inconclusive cases. Arch Pathol Lab Med 2009; 133(5):775–780.
66. Sawa J, Morikawa T. Interrater reliability for multiple raters in clinical trials of ordinal scale. Drug
Information Journal 2007; 41(5):595-605.
67. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977;
33:159-174.
68. Yu GZ, Chen Y, Wang JJ. Overexpression of Grb2/HER2 signaling in Chinese gastric cancer: their
relationship with clinicopathological parameters and prognostic significance. J Cancer Res Clin Oncol 2009;
135:1331-1339.
69. Rossi E, Grisanti S, Villanacci V, et al. HER-2 overexpression/amplification in Barrett's oesophagus predicts
early transition from dysplasia to adenocarcinoma: a clinico-pathologic study. J Cell Mol Med 2009; 13(9
B):3826-3833.
70. Rossi E, Villanacci V, Bassotti G, et al. Her-2/neu in Barrett esophagus: a comparative study between
histology, immunohistochemistry, and fluorescence in situ hybridization. Diagn Mol Pathol 2006; 15(3):125-
130.
71. Park DI, Yun JW, Park JH, et al. HER-2/neu amplification is an independent prognostic factor in gastric
cancer. Dig Dis Sci 2006; 51:1371-1379.
72. Conger AJ. Integration and generalization of Kappa for multiple raters. Psychol Bull 1980; 88:322-328.
73. Slamon DJ, Godolphin W, Jones LA et al. Studies of the HER-2/neu proto-oncogene in human breast and
ovarian cancer. Science 1989; 244(4905):707-712.
74. Untch M, Muscholl M, Tjulandin S, et al. First-line trastuzumab plus epirubicin and cyclophosphamide
therapy in patients with human epidermal growth factor receptor 2-positive metastatic breast cancer: cardiac
safety and efficacy data from the herceptin, cyclophosphamide, and epirubicin (HERCULES) trial. J Clin Oncol
2010; 28(9):1473-1480.
75. Okines AFC, Thompson LC, Cunningham D, et al. Effect of HER2 on prognosis and benefit from peri-
operative chemotherapy in early oesophago-gastric adenocarcinoma in the MAGIC trial. Ann Oncol 2012;
doi:10.1093/annonc/mds622.
38
Appendix: histology images
Histology images 1 Immunohistochemistry images of HER2 in gastro-esophageal adenocarcinoma. Illustrated are IHC scores 0 (A), score 1+ (B), score 2+ (C) and score 3+ (D and E, at different magnifications).
A
E
E C
B
39
Histology images 2 Chromogenic in situ hybridization images of HER2 in gastro-esophageal adenocarcinoma. Illustrated are diploid staining (F), polysomia (G), low amplification (H) and high amplification (I). The high amplification image shows large gene copy clusters (I, arrowhead).
I H
F G
40
Histology images 3 Hematoxylin-eosin images of the intestinal tumor type (J) and the diffuse tumor type (K). Immunohistochemistry staining of HER2 in the diffuse type is more difficult to score, because clusters of stained tumor cells are harder to identify due to poor differentiation (L) or because of difficult visual differentiation between cytoplasmatic and membraneous staining in signet ring cells (M).
J K
M L