Article Text
PDF
Abstract
Aims α-Fetoprotein (AFP)-producing gastric carcinoma (AFPGC) is one of the most aggressive GC subtypes. Frequent expression of human epidermal growth factor receptor 2 (HER2) has previously been reported in hepatoid adenocarcinoma (HAC), a major histological subtype of AFPGC originating from common-type GC (CGC). However, HER2 expression levels in other AFPGC histological subtypes are unknown. In this study, we analysed HER2 expression in GCs with primitive phenotypes in addition to HAC.
Methods HER2 expression was evaluated in representative complete sections from 16 HACs, 19 GCs with enteroblastic differentiation (GCEDs) and 334 GCs of other histological types as controls. The Ruschoff/Hofmann method was used to score HER2 immunohistochemistry. Samples with a HER2 score of 2+ were further assessed using fluorescence in situ hybridisation. Oncofetal protein (OFP) expression in HAC and GCED was tested via immunohistochemical staining for AFP, glypican 3 and Sal-like protein 4.
Results Thirty of 35 HAC/GCED cases comprised more than two histological patterns. The HER2 positivity rates of each histological component in the HACs/GCEDs were 25.0% for HAC (n=16), 34.6% for GCED (n=26) and 48.1% for CGC (n=27), which were higher than those of the control group (13.8%). Additionally, the majority of CGC components in HACs/GCEDs were positive for OFP (88.9%).
Conclusions HER2 is frequently overexpressed not only in HAC but also in GCED and CGC components of HACs/GCEDs, which suggests an association between HER2 and OFP expression. Moreover, our findings suggest that HER2-positive CGC has a higher risk of progression to HAC/GCED than HER2-negative GC.
- gastric
- gastric cancer
- gastric pathology
- gut pathology
Statistics from Altmetric.com
Introduction
Gastric carcinoma (GC), the fifth most common cancer worldwide, is highly prevalent in eastern Asia1 and is responsible for 50 000 cancer-related mortalities annually in Japan.2 Trastuzumab (Herceptin), which is a recombinant humanised monoclonal antibody against human epidermal growth factor receptor 2 (HER2), was recently shown to be an effective treatment in patients with advanced HER2-positive GC in the large-scale phase III international ToGA clinical trial.3 Since then, trastuzumab-containing chemotherapy regimens have become a treatment option for patients with HER2-positive advanced GC.
The ToGA trial revealed a HER2-positivity rate of 22.1% in GC without evident geographical differences.4 However, HER2 status differs significantly among various histological types of GC. The HER2 positivity rate is significantly higher in intestinal type GCs (range 21.5%–32.7%) than in diffuse type GCs (range 2.9%–11.7%).4–8
Data pertaining to HER2 status in rare histological variants of GC are limited.9–11 Giuffre et al previously reported a high HER2 positivity rate (42.9%) in hepatoid adenocarcinoma (HAC), which is the major histological type of α-fetoprotein (AFP)-producing gastric carcinoma (AFPGC) that accounts for approximately 3% of total GCs.12 13 Horita et al also reported a positive relationship between HER2 expression and AFP production in GC, wherein 40% of 15 AFP-positive patients were HER2 positive; however, the histomorphologic relevance of this observation was not well described.14
The association between AFP and HAC implies a close relationship between oncofetal protein (OFP) expression and GC histology.15–20 Other than HAC, histological types such as GC with enteroblastic differentiation (GCED),5 19 yolk sac tumor-like carcinoma (YST),16 and intestinal type GC are also known to express AFP and/or other OFPs such as glypican 3 (GPC3) and Sal-like protein 4 (SALL4).18 20 However, HER2 expression levels in OFP-positive GCs other than HAC are still unknown.
In this study, we compared the HER2 expression levels among HAC, GCED and other histological types of GC to further investigate the HER2 status of GCs with a primitive phenotype.
Methods
Case selection
This retrospective study was approved by the appropriate Institutional Ethics Committee (No. 1993; 29 March 2017). Sixteen HACs and 19 GCEDs that were surgically resected from 35 patients between 2004 and 2017 were retrieved from the respective pathology archives of Wakayama Medical University and Japanese Red Cross Wakayama Medical Centre. Three hundred and thirty-four primary GCs of other histological types that were surgically resected from 333 patients at Wakayama Medical University between 2011 and 2016 were included as controls. Clinicopathological parameters including age, sex, serum AFP level, main tumour location, macroscopic tumour type and status of nodal and distant metastasis were obtained from each patient’s pathology report and/or medical records.
Histological evaluation
All resected specimens were formalin fixed, sectioned and subjected to conventional hematoxylin and eosin staining. Histological classification was based on the WHO and Lauren’s classification; the immunohistochemistry (IHC) slides from selected cases were scored by at least two of the authors.21 Intraepithelial high-grade neoplasia/dysplasia as per the WHO classification was diagnosed as intramucosal tubular carcinoma. Tumours composed of columnar neoplastic cells with clear cytoplasm growing in a tubular and/or papillary pattern and that were associated with the expression of at least one OFP (ie, AFP, GPC3 and/or SALL4) were designated GCEDs.15 19 In cases where HAC and GCED coexisted, the tumours were diagnosed as HAC. The percentages of each of the different histological patterns of HAC/GCED in a representative glass slide section were recorded. All cases were staged in accordance with the Union for International Cancer Control TNM classification (eighth edition).22
HER2 testing
Complete paraffin sections of all 369 tumour specimens were subjected to HER2 expression analysis. Initially, a representative full tissue section was evaluated using IHC for HER2 (clone SV2-61γ; Nichirei, Tokyo, Japan) using an automated immunostainer (Histostainer48A, Nichirei, Tokyo, Japan).
The Ruschoff/Hofmann method was used to score HER2 IHC staining: 0, no reactivity or membranous reactivity in <10% of tumour cells; 1+: faint/barely perceptible membranous reactivity in ≥10% of tumour cells; 2+: weak-to-moderate, complete, basolateral, or lateral membranous reactivity in ≥10% of tumour cells; and 3+: strong, complete, basolateral, or lateral membranous reactivity in ≥10% of tumour cells.23 In cases of HAC/GCED, HER2 status was assessed as a whole and for each histological component individually.
Cases with HER2 scores of 2+ were further tested by fluorescence in situ hybridization (FISH) using the Histra HER2 FISH kit (Jokoh, Tokyo, Japan) as per the manufacturer’s instructions.24 For each specimen, the total number of HER2 +and chromosome enumeration probe 17 (CEP17) signals were counted in 20 tumour cell nuclei and those with HER2/CEP17 ratios ≥2.0 were defined as HER2 positive.
IHC for AFP, GPC3 and SALL4
In HAC and GCED, immunostaining for AFP (polyclonal; DAKO, Glostrup, Denmark), GPC3 (clone 1G12; Nichirei, Tokyo, Japan) and SALL4 (clone 6E3; Abnova Corporation, Taipei, Taiwan) were performed using an automated immunostainer (Bond III; Leica Microsystems, Wetzlar, Germany or Ventana XT System Benchmark, Ventana Medical Systems, Tucson, AZ, USA). In the control group, all HER2-positive GCs and 15 randomly selected HER2-negative intestinal type GCs were also immunostained for AFP, GPC3 and SALL4 in the same manner. IHC was performed on samples sectioned from the same paraffin blocks as those on which the HER2 test was performed.
IHC evaluation of AFP was performed by assessing cytoplasmic staining in tumour cells, while GPC3 staining in both the membrane and cytoplasm was evaluated. Only nuclear staining of SALL4 was assessed. For GPC3 and SALL4, a positive status was defined as >5% staining of the tumour sample; a positive AFP status was defined as >1% staining of the tumour section.19 In each case, the results of IHC in different histological components were evaluated individually.
Statistical analysis
All statistical analyses were conducted using GraphPad Prism (GraphPad Software, San Diego, CA, USA) or JMP Pro 13 Statistical Discovery Software (SAS Institute, Cary, NC, USA). Comparisons between groups of categorical variables were performed using Fisher’s exact test and P values <0.05 were considered statistically significant.
Results
HAC and GCED
Clinical and pathological data of 16 patients with HAC and 19 patients with GCED are shown in table 1. Immunohistochemically, 5 of 16 HAC samples (31.3%) had a HER2 score of 3+ while 11 (68.8%) had a score of 0. Among 19 patients with GCED, 5 (26.3%) had samples with a HER2 score of 3+, 3 (15.8%) had a score of 2+, 1 (5.3%) had a score of 1+ and 10 (52.6%) had a score of 0. All three GCED cases with a HER2 IHC score of 2+ were HER2 positive according to FISH. Hence, 5 of 16 cases of HAC (31.3%) and 8 of 19 cases of GCED (42.1%) were HER2 positive overall (figure 1A–D). In total, 13 of 35 cases of HAC/GCED (37.1%) were determined to be HER2 positive. As for AFP positivity, 11 of 30 cases of AFP-positive HAC/GCED (36.7%) were HER2 positive.
Expression of human epidermal growth factor receptor 2 (HER2) in hepatoid adenocarcinoma (HAC), gastric carcinoma with enteroblastic differentiation (GCED) and yolk sac tumor-like gastric carcinoma (YST). A sample from patient 9, who had a histology of HAC, is composed of hepatocyte-like large polygonal eosinophilic cells with an inset showing α-fetoprotein (AFP) staining (A) and HER2 overexpression (score 3+) (B). A sample from patient 33 shows a GCED histology composed of columnar cells with clear cell cytoplasm growing in a tubular pattern with the inset showing glypican 3 (GPC3) staining (C) and HER2 overexpression in both conventional intestinal type (right area) and GCED components (left area), with the inset showing a high-power view of GCED component (D). A sample from patient 34 shows YST histology that is characterised by a papillary fibrovascular structure reminiscent of a Schiller-Duval body with the inset showing SALL4 staining (E) and HER2 overexpression with the inset showing a high-power view (F).
Clinical and pathological features of 16 patients with hepatoid adenocarcinoma (patients 1–16) and 19 patients with gastric carcinoma with enteroblastic differentiation (patients 17–35)
Thirty of 35 HAC/GCED tumour samples (85.7%) were composed of at least two histological subtypes. Consequently, 16 HACs, 26 GCEDs, 2 YSTs and 27 common type GC (CGC) components were evaluated in the 35 HAC/GCED samples. The 27 CGC components consisted of 22 intestinal type, four mixed type and one diffuse type GC. Results of HER2, AFP, GPC3 and SALL4 expression within different histological patterns in each patient sample are shown in table 2. In two cases (patients 11 and 34), YST components showed the same IHC profiles as the coexisting HAC and/or GCED components (figure 1E and F). Among 35 cases of GCED/HAC, 27 comprised CGC components, of which 26 were of intestinal and mixed types. The majority of CGC components were positive for SALL4 (88.9%). In one each of a HAC and GCED case that were HER2 positive (from patients 7 and 28, respectively), HER2 overexpression was observed only in the intestinal type component and not in the HAC or GCED components (figure 2A and B). Additionally, in one case each of a HAC and GCED that were HER2 negative (from patients 3 and 32, respectively), a HER2 score of 3+ was observed in 10% of the intramucosal mixed type and intestinal type GC components, respectively, accounting for <10% of the total tumour (figure 2C and D). As a result, the HER2 positivity rates of HAC and GCED components were 25.0% and 34.6%, respectively. Regarding CGC components associated with HAC/GCED, 13 of 27 samples (48.1%) showed HER2 overexpression and all 13 tumour areas with HER2 overexpression were histologically intestinal type GC. Consequently, the HER2 positivity rate increased in order of HAC, GCED and CGC components.
The human epidermal growth factor receptor 2 (HER2) positivity rate of common gastric carcinoma components was significantly higher than that of the intestinal type gastric carcinoma components in the control group (48.1% vs 24.8%, respectively, P<0.05). A sample from patient 7 shows hepatoid adenocarcinoma (HAC; lower area) with intestinal type gastric carcinoma (upper area) with the inset showing GPC3 staining in the HAC component (A). HER2 overexpression is limited in the intestinal type component (B). A sample from patient 3 shows HAC (upper area) admixed with mixed type gastric carcinoma (lower area) with the inset showing GPC3 staining in the HAC component (C). HER2 overexpression is limited in the mucosal intestinal type component; the inset shows a high-power view of the HER2 stained area (D).
Immunohistochemical profiles of the 16 cases of HAC and 19 cases of GCED
Control group
Among the 334 cases, 35 (10.5%) had a HER2 score of 3+, 29 (8.7%) had a score of 2+, 34 (10.2%) had a score of 1+ and 236 (70.7%) had a score of 0. Among the 29 HER2 score 2+ samples, 11 (37.9%) were HER2 positive according to FISH. In total, HER2 was determined to be positive in 46 cases (13.8%). A comparison of the clinicopathological characteristics between HER2-positive and HER2-negative cases is shown in table 3. HER2 positivity was significantly and positively correlated with advanced age, male sex, tubular and papillary histology (WHO classification), intestinal type histology in Laurens’ classification and liver metastasis (P<0.05).
Clinicopathological characteristics and HER2 positivity in the control group consisting of non-hepatoid and non-enteroblastic gastric carcinomas
Comparison of HER2 positivity rates between the histological subtypes
HER2 positivity rates among HACs (31.3%) and GCEDs (42.1%) were higher than those of the control group (13.8%) (P=0.0667 and P=0.0035, respectively). Although not statistically significant, the HER2 positivity rate of the HAC component was higher than that of the control group (25.0% vs 13.8%, P=0.2615) and similar to that of the intestinal type GC in the control group (24.8%). The 26 GCED components in 35 HAC/GCED samples showed a significantly higher HER2 positivity rate than those of the control group (34.6% vs 13.8%, respectively, P=0.0091), whereas their HER2 positivity rate was not significantly different than that in intestinal type GC of the control group (34.6% vs 24.8%, respectively, P=0.3358). However, the HER2 positivity rate of 27 CGC components in 35 samples with HAC/GCED was significantly higher than that of the intestinal type GC of the control group (48.1% vs 24.8%, respectively, P=0.0197).
AFP, GPC3 and SALL4 expression in control group
Within the control group, the positivity rates of AFP, GPC3 and SALL4 in 46 HER2-positive GC samples were higher than those in 15 representative HER2-negative intestinal type GCs (AFP positivity rate: 21.7% versus 6.7%, P=0.2646; GPC3 positivity rate: 34.8% versus 20.0%, P=0.3507; and SALL4 positivity rate: 63.0% versus 33.3%, P=0.0714, respectively). Additionally, the AFP, GPC3 and SALL4 positivity rates in these 46 HER2-positive control group GCs were significantly higher than those reported in previous studies (AFP: 2.6%–3.0%; GPC3: 17.0%–18.6%; and SALL4: 10.9%–32.3%; P<0.05).17 20 25 26
Discussion
We investigated HER2 expression in 16 HAC, 26 GCED and 27 CGC components in 35 HAC/GCED patient samples and found that all these components had higher rates of HER2 overexpression than the control group. Moreover, HER2-positive CGCs in the control group revealed higher OFP positivity rates than HER2-negative CGCs, implying an association between HER2 and OFP expression in GC. Ours is the largest study that investigated HER2 expression in GCs with a primitive phenotype to date.
The highest HER2 scores were observed in the intestinal type CGC components of HAC/GCED, which were significantly higher than those of intestinal type GCs in the control group. HAC/GCED is believed to develop from intestinal type GC;18 27 most of our HAC/GCED cases were admixed with intestinal type GC, supporting this theory. Our findings indicate that HER2-positive intestinal type GCs may have a higher risk for progression to HAC/GCED than HER2-negative intestinal type GCs.
GC with a primitive enterocyte phenotype characterised by AFP, GPC3, SALL4 and claudin-6 expression is usually classified as having ‘chromosomal instability (CIN)’ as per the molecular classification of The Cancer Genome Atlas and is associated with poor patient prognosis.20 28 It is noteworthy that most HER2-amplified GCs also belong to the CIN subgroup, indicating an overlap between HER2-positive and OFP-positive GC.28 While at least two studies have shown an association between HER2 and AFP expression in GC,11 14 a causal relationship between HER2 and OFP in GCs has not yet been established.
While evidence of the molecular interaction between HER2 and OFP in GC is presently lacking, several studies suggest an association between HER2 and stem cell regulatory genes such as NANOG, OCT3/4 and SOX2 in breast carcinomas.29 30 These stem cell-regulating genes are known to form a regulatory circuit with SALL4 to maintain embryonic stem cell pluripotency and abundant evidence suggests a strong biological similarity between embryonic stem cells and cancer cells.25 31 32 Our study further supports a role for HER2 in primitive cancer cell proliferation and we speculate that HER2 may be one of the upstream factors regulating OFP expression in GC. Additional investigations are required to clarify whether the association between HER2 and OFP expression observed in our study is correlative or causative.
The association between HER2 expression and a primitive phenotype has only been reported in breast carcinoma to date. Notably, HER2 and SALL4 were separately reported to be elevated in urothelial carcinoma.25 26 33 Additionally, HER2- and OFP-positive carcinomas are separately reported in foregut-derived organs other than the stomach, such as the duodenum and lungs.34–38 Future cross-sectional studies may yet confirm a relationship between HER2 and OFP expression in carcinomas.
Heterogeneity is a characteristic of HER2 expression in GC.9 24 39–45 Among 15 HACs/GCEDs that comprised HER2-overexpressing areas, discordance in HER2 status between the histological components was identified in four cases (26.7%) with limited HER2 expression in intestinal type GCs. Consequently, the HER2 positivity rates in each histological component among 35 cases of HAC/GCED decreased in order of CGC (highest), GCED and HAC (lowest). Even though HER2 overexpression is frequently observed in HACs/GCEDs, its expression appears to diminish during high-grade transformation in some tumours. Since HAC/GCED is known to be associated with distant organ metastasis more frequently than CGC,12 13 17–19 lower HER2 positivity rates in HAC/GCED may cause heterogeneity in HER2 status between primary and metastatic sites. Currently, there are limited data on treatment outcomes of trastuzumab-containing chemotherapy for HER2-positive HAC/GCED, but primary-metastatic conversion of the HER2 status that might occur in such cases can possibly lead to limited trastuzumab efficacy and worse prognosis.39–45
In conclusion, HER2 is frequently overexpressed in GCs with a primitive phenotype, indicating that a possible association may exist between HER2 and OFP expression. The high HER2 positivity rate observed in the CGC components of HACs/GCEDs suggests that HER2-positive CGCs have a higher risk of progression to HAC/GCED than HER2-negative GCs.
Take home messages
Human epidermal growth factor receptor 2 (HER2) is frequently overexpressed in gastric carcinomas with a primitive phenotype, such as hepatoid adenocarcinoma, gastric carcinoma with enteroblastic differentiation and intestinal type gastric carcinoma with oncofetal protein (OFP) expression.
Moreover, the OFP positivity rate is high in HER2-positive conventional gastric carcinoma, which implies an association between HER2 and OFP expression in gastric carcinoma.
HER2-positive intestinal type gastric carcinoma has a higher risk of progression to hepatoid carcinoma and gastric carcinoma with enteroblastic differentiation than HER2-negative gastric carcinoma.
Abstract translation
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.Acknowledgments
The authors wish to thank Saeka Ikoma, Sayuri Sato, and Maki Higashide for their technical assistance.
References
Footnotes
Handling editor Runjan Chetty.
Competing interests None declared.
Ethics approval Wakayama Medical University.
Provenance and peer review Not commissioned; externally peer reviewed.