’Original article Adenocarcinomas at different positions at the gastro-oesophageal junction show distinct association with gastritis and gastric preneoplastic conditions Jan Bornscheina, Andrea Dingwertha, Michael Selgrada, Marino Veneritoa, Patrick Stuebsb, Katrin Frauenschlaegerc, Achilleas Achilleosd, Albert Roessnerc and Peter Malfertheinera Objective Adenocarcinomas at the gastro-oesophageal junction (GOJ) are currently stratified by tumour location. This retrospective study examines the association of preneoplastic conditions and inflammation of the gastric mucosa with GOJ cancer at different locations and compares them with nonjunctional gastric cancers. Patients and methods A total of 520 patients with junctional and nonjunctional gastric cancer were assessed for the presence and degree of intestinal metaplasia, glandular atrophy and inflammation in the stomach. Histopathological data were complete for 428 patients (68.9% men, median age 67.7 years), including 172 patients with GOJ cancer (GOJ1: 1–5 cm proximal to the junction, GOJ2: ‘true’ junctional, GOJ3: 2–5 cm distal to the junction). Gastric inflammation and preneoplastic conditions were scored according to the updated Sydney classification and further stratified into respective operative link on gastritis assessment (OLGA) and operative link on gastritis assessment on intestinal metaplasia (OLGIM) stages. Results The prevalence and degree of gastric atrophy and intestinal metaplasia were significantly lower in GOJ1 than GOJ3 (P < 0.01). Preneoplastic conditions in the stomach were similar in GOJ3 compared with nonjunctional gastric cancer. GOJ1 were almost exclusively (98.4%) of the intestinal type, whereas GOJ2 and GOJ3 were the diffuse type in 22.6 and 22.4% of the patients (P < 0.001). Of all patients, only 8.5 and 12.7% presented with stage III/IV according to OLGA and OLGIM, respectively. However, data for OLGA and OLGIM staging were only available in 61.2 and 67.9% of patients, respectively. Conclusion GOJ1 are less likely to be associated with gastric pathology compared with GOJ3 or nonjunctional gastric cancer. OLGA or OLGIM staging in patients with advanced gastro-oesophageal cancer seems to be of limited value. Eur J Gastroenterol Hepatol 27:492–500 Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Introduction
The International Gastric Cancer Association (IGCA) and the International Society for Diseases of the Esophagus (ISDE) defined adenocarcinoma of the gastro-oesophageal junction (GOJ) as those located within 5 cm proximal or distal of the proximal end of the gastric folds [1]. In contrast, the seventh edition of the UICC TNM classification focuses on the simplification of therapeutic algorithms and defines all tumours that are located even within 5 cm distal European Journal of Gastroenterology & Hepatology 2015, 27:492–500 Keywords: cardiac cancer, gastric atrophy, gastric cancer, gastrooesophageal junction, intestinal metaplasia, OLGA, OLGIM a
Department of Gastroenterology, Hepatology and Infectious Diseases, Department of General, Visceral and Vascular Surgery, cInstitute of Pathology, Otto-von-Guericke University, Magdeburg, Germany and dCancer Research UK Cambridge Institute, University of Cambridge, Cambridge, UK b
Correspondence to Peter Malfertheiner, MD, Department of Gastroenterology, Hepatology and Infectious Diseases, Otto-von-Guericke University, Leipziger Str. 44, 39104 Magdeburg, Germany Tel: + 49 391 671 3100; fax: + 49 391 671 3105; e-mail: [email protected] Received 14 October 2014 Accepted 7 January 2015 Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal's website (www.eurojgh.com).
from the GOJ as oesophageal cancer as long as the tumour straddles the junction [2]. The TNM system thus suggests that all adenocarcinomas at the GOJ should be treated as a single (oesophageal) entity. This will potentially include proximal gastric cancers and distal oesophageal adenocarcinomas. Distal GOJ cancers have previously been reported to be less differentiated, and to show more aggressive behaviour, more often lymph node involvement, and a shorter overall survival compared with proximal GOJ cancers [3]. Furthermore, these cancers seem to develop by different biological mechanisms, with proximal GOJ cancers being more associated with gastrooesophageal reflux disease, Barrett’s oesophagus and obesity, whereas distal GOJ cancers are linked to Helicobacter pylori infection and preneoplastic conditions in the stomach [4]. Systems in current use to distinguish the origin of GOJ cancers are either based on the anatomical location of the main tumour mass [such as the TNM classification or the stratification proposed by Siewert and Stein (see the Materials and methods section)] or on paraclinical surrogates such as serum pepsinogens and anti-H. pylori immunoglobulin G [5,6]. Although epithelial dysplasia is the only ‘true’ premalignant lesion of the gastric mucosa, changes related to chronic inflammation such as intestinal metaplasia (IM) or glandular atrophy of the gastric mucosa are considered as
0954-691X Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
DOI: 10.1097/MEG.0000000000000299
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of the article is prohibited.
492
Gastro-oesophageal cancer and gastritis Bornschein et al.
preneoplastic conditions. The gold standard for identification of preneoplastic changes of the gastric mucosa is the histopathological assessment of endoscopic biopsies. The updated Sydney classification of gastritis remains the most widely used system for characterization of the status of the gastric mucosa for both research and clinical purposes [7]. New classification systems have been developed to enable a more consistent stratification of the individual risk of gastric cancer development. The OLGA and the operative link on gastritis assessment on intestinal metaplasia (OLGIM) staging systems for gastric premalignant lesions aim to allow simple stratification requiring the same biopsy protocol recommended by the Sydney system. The abbreviation OLGA stands for ‘Operative Link on Gastritis Assessment’ and is based on the grading of glandular atrophy, pooling the degree of atrophy in the antrum and corpus into a simple scoring system ranging from ‘0’ to ‘IV’ [8]. Further progression towards gastric cancer is expected in patients who present initially with stages III and IV [9]. The OLGIM system classifies only the degree of IM instead of all atrophic changes with or without metaplastic transformation. It was introduced as an alternative to OLGA as the interobserver agreement was shown to be better for the assessment of IM compared with atrophy [10]. OLGA stages correlate well with serum pepsinogens, which are used as surrogate markers for gastric atrophy; OLGA also correlated with IM and dysplastic changes [11–13] and the modified OLGIM system seems to provide complementary information [14,15]. It is still under debate as to which of these approaches is superior for the assessment of gastric cancer risk [10,14,16]. Although these scores were established to assess the risk of cancer associated with preneoplastic gastric conditions, recent studies have used OLGA or OLGIM scores in patients with a diagnosis of gastric cancer [17,18]. The aim of our study was to assess the association of GOJ tumours in their strict allocation to gastric inflammation and related preneoplastic conditions of the gastric mucosa. The secondary aim was to evaluate the clinical value of OLGA/OLGIM staging in patients with gastric cancer. Materials and methods Data collection
This is a retrospective analysis of histopathological changes of the gastric mucosa in patients with gastric and GOJ cancer. Tumour-specific data (Laurén type, location of the tumour, TNM stage, grading) were identified and the updated Sydney classification was applied to the nonmalignant gastric mucosa. The H. pylori infection status was recorded, when available. The electronic patient documentation system of the Department of Gastroenterology, Hepatology and Infectious Diseases at the University Hospital, Magdeburg, Germany, was used to retrieve patients who had been diagnosed with or treated for gastric adenocarcinoma, distal oesophageal adenocarcinoma or GOJ cancer between January 2000 and February 2013. A total of 707 patients fulfilled these criteria; 428 were included for the final analysis, 172 of whom had tumours at the GOJ. The reasons for exclusion of patients are shown in Fig. 1. About half of the patients in the final study cohort underwent surgical (or endoscopic) resection of the tumour. Patients who underwent resection showed a higher proportion of antrum cancers, less advanced local invasion
www.eurojgh.com
493
(T-stage), less lymph node involvement and less often distant metastases (Supplementary Table 1, Supplemental digital content 1, http://links.lww.com/EJGH/A8). The study was in line with the requirements of the local ethics authorities. Assessment of tumour location
Tumour location was assessed using either endoscopy or endoscopic ultrasound report, or the postsurgical pathological report, in the case of gastrectomy. GOJ cancers were classified according to the classification proposed by Siewert and Stein [1]. GOJ1 were considered as ‘distal oesophageal’ with the main tumour mass within 1–5 cm proximal to the GOJ, GOJ2 as ‘true junctional’ cancers within 1 cm proximal and 2 cm distal to the junction, and GOJ3 as ‘proximal gastric’ within 2–5 cm distal from the GOJ. For comparison with nonjunctional gastric cancer, gastric adenocarcinomas located in the corpus were also stratified into ‘proximal’ and ‘distal’ tumours, as described previously [19]; carcinomas with the main tumour mass in the fundus and in the proximal third of the corpus were classified as ‘proximal’ and tumours located in the distal two-thirds of the corpus were classified as ‘distal’. Histopathological assessment
Histopathological changes of the gastric mucosa were assessed according to the updated Sydney system [7]. The degree of IM and mucosal atrophy were graded 0–3 according to the predefined scoring system, with grade 0 in case of no alterations and grade 3 in case of severe changes – that is, almost complete replacement of the gastric mucosa. The activity of inflammation was graded according to the infiltration of the mucosa with neutrophil granulocytes and the chronicity of inflammation determined by the degree of infiltration with lymphocytes. Antrum and corpus were scored separately. In cases where the original report did not provide adequate information for scoring, specimens were re-evaluated by an expert pathologist. In the case of tertiary referrals who underwent gastrectomy at our hospital, but did not have local diagnostic gastroscopy data, histology was assessed on the surgical resection specimen or on pretreatment endoscopic biopsies. OLGA and OLGIM stages have been stratified as described previously [9,10,20]. However, for 38.8% of the patients, data on atrophy in the antrum or the corpus or both were not available, and data were missing on IM in 32.1% of the patients (Fig. 2). For the final analysis, only cases with either complete OLGA or OLGIM staging were included. Barrett’s oesophagus has been assessed in endoscopy reports or in the final pathology report after gastrectomy, where available. H. pylori status was evaluated according to any diagnostic reports that were available from (a) histology, (b) serology, (c) 13C-urea breath test, (d) stool antigen test or (e) a rapid urease test on biopsies. In case of any positive finding, the patient was considered currently or previously (serology) positive. Patients were considered negative with a negative result in serology or with negative results in at least two of the other modalities. This approach was adopted by consent of the authors.
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of the article is prohibited.
494
May 2015 • Volume 27 • Number 5
European Journal of Gastroenterology & Hepatology
Total number assessed:
N = 707
Gastro-oesophageal adenocarcinoma:
n = 643
Double entries or other entry than AC (including undifferentiated) n = 64 Unknown Laurén (e.g. preoperated) n = 107
n = 536
Complete Laurén classification: (mixed and mucinous type also excluded)
No data on tumour location n = 16 n = 520
Complete Laurén and location:
Incomplete histopathological data (IM and/or atrophy) n = 92
n = 428
Complete histopathological data:
Nonjunctional cancer n = 256 n = 172
Tumor location GOJ:
Fig. 1. Generation of the study population. Reason for exclusion of datasets from the final assessment. AC, adenocarcinoma; GOJ, gastro-oesophageal junction; IM, intestinal metaplasia.
IM GOJ
Corpus
Missing 19%
Incomplete 12%
Incomplete 16% Complete 66%
Complete 69%
GOJ
Complete 69%
Corpus
Antrum
Missing 23%
Missing 25%
Missing 24%
Incomplete 12%
Missing 19%
Missing 18%
Incomplete 12%
Atrophy
Antrum
Complete 64%
Incomplete 16%
Complete 59%
Incomplete 19%
Complete 58%
Fig. 2. Availability of histopathological data for the assessment of OLGA and OLGIM stages. Cases for which complete histopathological data on the prevalence and degree of IM and atrophy were available for staging according to either OLGA or OLGIM are shown. For about a third of cases, the data were either completely missing or incomplete for the antrum or the corpus. GOJ, gastro-oesophageal junction; IM, intestinal metaplasia; OLGA, Operative Link on Gastritis Assessment.
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of the article is prohibited.
Gastro-oesophageal cancer and gastritis Bornschein et al.
www.eurojgh.com
Statistical analysis
Group comparison of categorical data was performed using Fisher’s exact test. The parametric t-test was applied to compare age between the groups. The Kruskal–Wallis test was applied for a nonparametric comparison of Sydney scaled inflammatory scores and the Mann–Whitney U-test for post-hoc analyses in case of positive results. Spearman’s rank correlation test was used for correlation analyses. A two-sided significance level was defined as P-value less than 0.05, applying further Bonferroni’s correction in case of multiple comparisons. Data were analysed using IBM SPSS Statistics 20.0 (IBM SPSS Inc., Chicago, Illinois, USA). Results Demographic and tumour-specific data for gastrooesophageal junction cancers of different locations
Table 1 shows the demographics and tumour-specific data of the patients with GOJ tumours (n = 172), stratified by the Siewert classification. GOJ1 were almost exclusively of the intestinal type, whereas about a quarter of GOJ2 and GOJ3 showed a diffuse or a mixed phenotype. Distally located tumours were of more advanced stage and less differentiated compared with GOJ1. Data on lymph node involvement and distant metastases are incomplete because of the retrospective nature of the study. Demographics and tumour-specific data of the broader cohort (GOJ cancer: n = 211) are shown in Supplementary Table 2 (Supplemental digital content 1, http://links.lww.com/EJGH/A8). The distribution of the factors analysed was similar to that of the final study cohort. Preneoplastic and inflammatory changes of the gastric mucosa in patients with gastro-oesophageal junction cancer
The prevalence of both IM and glandular atrophy of the gastric mucosa was higher for GOJ3 compared with
495
GOJ1, with GOJ2 showing intermediate values (P < 0.001, P = 0.011, respectively; Fig. 3). Also, the degree of both IM and glandular atrophy was more severe for distally located adenocarcinomas at the GOJ (Fig. 3a and b). Similarly, activity and chronicity of gastric inflammation were more severe in GOJ3 (Fig. 4c and d). The degree of IM correlated strongly with the degree of atrophy both in the antrum and in the corpus (r = 0.362–0.659, P < 0.001; Supplementary Table 3, Supplemental digital content 1, http://links.lww.com/EJGH/A8). The same held true for the degree of inflammatory activity and chronicity, respectively (r = 0.384–0.529, P < 0.001; Supplementary Table 3, Supplemental digital content 1, http://links.lww.com/EJGH/A8). In addition, chronicity of inflammation in the corpus seemed to be an indicator of metaplastic or atrophic changes as it correlated with the both the maximal degree of IM (r = 0.335, P < 0.001) and atrophy (r = 0.289, P < 0.001). IM in the stomach was significantly detected more often in patients with less differentiated GOJ cancers (G1/2: 28.6% vs. G3/4: 45.7%, P = 0.033). However, there was no difference in the presence of IM in the stomach between cancers of intestinal or diffuse type (34.9 vs. 34.6%, P = 1.000). Gastric atrophy was only detected in patients with intestinal-type cancers (22.8 vs. 0.0%, P = 0.005). Comparison with nonjunctional gastric cancer
Table 2 shows the clinicopathological characteristics of carcinomas located in the antrum and corpus compared with GOJ tumours. Supplementary Table 4 (Supplemental digital content 1, http://links.lww.com/EJGH/A8) shows these data for the extended study cohort (n = 520). However, as discussed above, the distribution of the factors addressed has been similar to the final study cohort (Supplementary Table 5, Supplemental digital content 1, http://links.lww.com/ EJGH/A8); thus, we refer here only to the final cohort.
Table 1. Demographic and tumour-specific data of patients with gastro-oesophageal junction cancer n (%) GOJ1 (n = 61) GOJ2 (n = 53) GOJ3 (n = 58) Age (median, IQR) (years) Sex (male) Laurén (intestinal type) T-stagea (n = 147) T1 T2 T3 T4 Lymph node involvement (N-positive)a (n = 136) Distant metastases (M1)a (n = 64) Grading/differentiationa (n = 168) Well Moderate Poor
66.0 60.9–73.9 53 (86.9) 60 (98.4)
69.6 62.5–74.9 45 (84.9) 41 (77.4)
68.1 57.8–75.3 45 (77.6) 45 (77.6)
18 13 17 4 23
7 13 18 5 29
4 16 28 4 42
(34.6) (25.0) (32.7) (7.7) (53.5)
(16.3) (30.2) (41.9) (11.6) (70.7)
(7.7) (30.8) (53.8) (7.5) (80.8)
P-value all groups
P-value GOJ1 vs. GOJ2
P-value GOJ1 vs. GOJ3
P-value GOJ2 vs. GOJ3
(0.737)
–
–
–
(0.376) < 0.001
– 0.001
– < 0.001
– (1.000)
(19.7) (28.6) (42.9) (8.8) (69.1)
0.036
(0.247)
0.006
(0.462)
0.017
(0.120)
0.007
(0.328)
Total (N = 172) 67.8 60.8–75.0 143 (83.1) 146 (84.9) 29 42 63 13 94
9 (32.1)
7 (46.7)
12 (57.1)
28 (43.8)
0.209
–
14 (24.6) 26 (45.6) 17 (29.8)
7 (13.2) 26 (49.1) 20 (37.7)
4 (6.9) 21 (36.2) 33 (56.9)
25 (14.9) 73 (43.5) 70 (41.7)
0.015
(0.310)
Group comparison was performed using Fisher’s exact test. Percentages are given per available cases. Post-hoc P-values are given if comparison of all groups has been significant; Bonferroni’s correction for multiple comparisons was applied. GOJ, gastro-oesophageal junction; IQR, interquartile range. a Data on these features were not available for all cases. Significance for P < 0.05.
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of the article is prohibited.
–
0.003
–
(0.124)
496
May 2015 • Volume 27 • Number 5
European Journal of Gastroenterology & Hepatology
80% 70%
IM Atrophy
60% 50% 40% 30% 20% 10% 0%
GOJ1
GOJ2
GOJ3
Proximal corpus
Distal corpus
Antrum
Fig. 3. Prevalence of atrophy and intestinal metaplasia (IM) in the stomach. The prevalence for both IM (P < 0.001) and glandular atrophy (P = 0.011) was significantly higher in tumours located distally at the junction, showing a pattern similar to nonjunctional gastric cancer. GOJ, gastro-oesophageal junction.
With respect to the degree of preneoplastic and inflammatory changes of the gastric mucosa, all conditions (IM, atrophy, activity and chronicity of inflammation) were more severe for nonjunctional tumours compared with GOJ cancers (P < 0.001, data not shown). In the posthoc analysis, the prevalence and degree of IM were significantly lower for GOJ1 and GOJ2 compared with nonjunctional cancers of different locations (P < 0.01; Fig. 3). This was not the case for GOJ3, which showed comparable prevalence to nonjunctional cancers and a lower degree of IM in the corpus only compared with cancers in the distal corpus (P = 0.005) and of IM in the antrum compared with antrum cancers (P = 0.002). Also,
(a) 100.0% 90.0% 80.0% 70.0% 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0%
None
Antrum
Corpus
Mild
Antrum Corpus
GOJ1 (c) 100.0% 90.0% 80.0% 70.0% 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0%
GOJ2
None
Antrum
Corpus
GOJ1
Moderate
Mild
Moderate
Antrum
Corpus
GOJ2
OLGA and OLGIM stages for patients with gastrooesophageal junction cancer and nonjunctional gastric cancer
Here, we included patients with both complete and incomplete histopathology records for the entire cohort with gastric and junctional adenocarcinomas (n = 520;
(b)
Severe
Antrum
the prevalence of atrophy was significantly lower for GOJ1 compared with nonjunctional cancers (P < 0.01) as well as the severity of corpus atrophy (P < 0.001). Atrophy in the antrum was lower only compared with antral cancers (P < 0.001). GOJ2 showed a similar pattern of gastric atrophy compared with proximal corpus cancers, but the prevalence was lower compared with more distal gastric cancers (P < 0.01). The degree of atrophy in the corpus was lower compared with distal corpus carcinomas (P = 0.001) and the degree of antral atrophy was lower compared with antral cancers (P = 0.003). For GOJ3, there was no difference in the prevalence or the degree of gastric atrophy compared with nonjunctional cancers. Activity of gastric inflammation was similar for GOJ1 compared with proximal corpus cancers, but significantly lower compared with cancers in the distal corpus and the antrum (P ≤ 0.001). Inflammatory activity was not different between nonjunctional cancers and GOJ2 as well as GOJ3. Chronicity of inflammation was lower in GOJ1 relative to nonjunctional cancers (P < 0.001). GOJ2 and GOJ3 showed less inflammatory chronicity in the corpus compared with distal corpus cancers (P = 0.002), but similar chronicity in the antrum and corpus compared with nonjunctional cancers of other locations.
Corpus
100.0% 90.0% 80.0% 70.0% 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0%
Antrum
Corpus
(d)
Severe
Corpus
GOJ3
Mild
Antrum
GOJ1
GOJ3
Antrum
None
100.0% 90.0% 80.0% 70.0% 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0%
Corpus
GOJ1
Severe
Corpus
Antrum
GOJ2
None
Antrum
Moderate
Mild
Moderate
Severe
Antrum Corpus
Antrum
GOJ2
Corpus
GOJ3
Corpus
GOJ3
Fig. 4. Distribution of the degree of preneoplastic and inflammatory changes in patients with GOJ cancer. The severity of preneoplastic conditions and inflammatory changes of the gastric mucosa in patients with GOJ cancer graded according to the updated Sydney classification is shown, with 0: no changes, 1: mild, 2: moderate, 3: severe changes (a) gastric IM, (b) gastric glandular atrophy, (c) activity of gastric inflammation by infiltration with neutrophils, (d) chronicity of gastric inflammation by infiltration with lymphocytes. GOJ, gastro-oesophageal junction; IM, intestinal metaplasia.
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of the article is prohibited.
Gastro-oesophageal cancer and gastritis Bornschein et al.
www.eurojgh.com
497
Table 2. Demographic and tumour-specific data of patients with gastric and junctional adenocarcinomas n (%) P-value all groups
P-value GOJ vs. corpus
P-value GOJ vs. antrum
P-value corpus vs. antrum
(0.092)
–
–
–
< 0.001 < 0.001
< 0.001 < 0.001
< 0.001 < 0.001
(0.894) (0.599)
(21.6) (27.5) (41.1) (9.8) (69.6)
(0.062)
–
–
–
0.026
(0.190)
(0.144)
0.009
–
–
–
GOJ (n = 172)
Corpus (n = 167)
Antrum (n = 89)
Total (N = 428)
67.8 60.8–75.0 143 (83.1) 146 (84.9)
65.8 54.6–75.0 100 (59.9) 85 (50.9)
69.0 60.7–76.0 52 (58.4) 49 (55.1)
67.7 57.2–75.1 295 (68.9) 280 (65.4)
21 22 17 9 35
73 93 139 33 211
Age (median, IQR) (years) Sex (male) Laurén (intestinal type) T-stagea (n = 338) T1 T2 T3 T4 Lymph node involvement (Npositive)a (n = 303) Distant metastases (M1)a (n = 142) Grading/differentiationa (n = 418) Well Moderate Poor
29 42 63 13 94
(19.7) (28.6) (42.9) (8.8) (69.1)
23 29 59 11 82
(18.9) (23.8) (48.4) (9.0) (77.4)
(30.4) (31.9) (24.6) (13.0) (57.4)
28 (43.8)
25 (50.0)
11 (43.8)
64 (45.1)
(0.651)
25 (14.9) 73 (43.5) 70 (41.7)
15 (9.3) 47 (29.0) 100 (61.7)
6 (6.8) 25 (28.4) 57 (64.8)
46 (11.0) 145 (34.7) 227 (54.3)
0.001
0.001
0.002
(0.836)
Group comparison was performed using Fisher’s exact test. Percentages are given per available cases. Post-hoc P-values are given if comparison of all groups has been significant. GOJ, gastro-oesophageal junction; IQR, interquartile range. a Data on these features were not available for all cases. Significance for P < 0.05.
Fig. 1; Supplementary Table 4, Supplemental digital content 1, http://links.lww.com/EJGH/A8). OLGA staging was possible in 318 (61.2%) and OLGIM staging in 353 (67.9%) patients. In the remaining cases, histopathological scoring from either the antrum or the corpus was incomplete (Fig. 3). For GOJ tumours (n = 211), OLGA and OLGIM staging was possible in 64.5 and 68.7%, respectively. The data availability on OLGA and OLGIM was lowest for GOJ1 (61.2 and 63.5%, respectively) compared with GOJ2 (69.8 and 73.0%, respectively) and GOJ3 (63.5 and 71.4%, respectively). Considering all junctional and nonjunctional cancers, location itself was not an influencing factor on data availability for OLGA and OLGIM. However, Laurén type (P = 0.024 and 0.024, respectively), T-stage (P = 0.046 and 0.015, respectively) and N stage (P = 0.009 and 0.001, respectively) did influence data availability, with a lower proportion of tumours of the diffuse type and those with advanced T-stage and nodal involvement showing complete data. The distribution of OLGA and OLGIM stages in our study population is shown in Table 3. In our cohort, 8.5 and 12.7% of the patients presented with stages III and IV according to OLGA and OLGIM, respectively. Nonjunctional cancers showed higher stages of both OLGA and OLGIM stages than GOJ cancers (P < 0.001 and < 0.001, respectively), and OLGA stages were higher in intestinal-type tumours (P = 0.007). Otherwise, there was no correlation of the distribution of OLGA or OLGIM stages with clinicopathological characteristics in our cohort. Discussion
In our study, GOJ3 showed a strong association with the degree of IM and atrophy of the background gastric mucosa, which was not the case for GOJ1 cancers. This suggests a different aetiology for adenocarcinomas at this location. It has been postulated previously that junctional cancers are either
associated with H. pylori-induced changes of the gastric mucosa or with a history of gastro-oesophageal reflux disease (GORD) [5,6]. The Irish FINBAR study investigated the association of H. pylori infection and gastric atrophy with GORD-induced alterations of the distal oesophagus, namely, oesophagitis, Barrett’s oesophagus and oesophageal adenocarcinoma. There was an inverse relationship between H. pylori infection and gastric atrophy with oesophageal adenocarcinomas that could not be confirmed for true junctional cancers [21]. In a recent Japanese study, Barrett’s associated cancers at the GOJ showed significantly less gastric atrophy relative to tumours without evidence of adjacent Barrett’s metaplasia [22]. The distinct association of gastric atrophy with cancer from different locations at the junction was also confirmed by a recent meta-analysis [23]. Overall, inflammatory changes in the gastric mucosa are less severe in patients with GOJ cancer than nonjunctional gastric cancer [24]. However, although this difference was distinct for GOJ1, GOJ3 showed a pattern that is comparable with nonjunctional gastric cancer. Thus, it is fair to infer that GOJ3 are related to gastric inflammation, whereas GOJ1 are not. Once more, GOJ2 have an intermediate position, showing less severe preneoplastic conditions, but gastric inflammation comparable to adenocarcinomas in the distal corpus and the antrum. The positive correlation of the degree of inflammation in the corpus with preneoplastic changes in the stomach further suggests that corpus-dominant inflammation could indicate a higher risk for the development of gastric cancer (including GOJ3) as reported previously [25–27]. Similar data have been reported for IM, where not only the degree but also the distribution of IM defines the risk for the development of gastric cancer [16,28,29]. In a Japanese study on patients who developed gastric cancer after H. pylori eradication, IM in the corpus at baseline was the only independent predictor for progression in multivariate analysis [30].
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of the article is prohibited.
498
May 2015 • Volume 27 • Number 5
European Journal of Gastroenterology & Hepatology
Table 3. Distribution of OLGA and OLGIM stages for patients with gastric and junctional adenocarcinomas n (%) GOJ (n = 172) OLGA (n = 318) Stage 0 107 Stage 1 11 Stage 2 11 Stage 3 6 Stage 4 1 a OLGIM (n = 353) Stage 0 98 Stage 1 25 Stage 2 14 Stage 3 7 Stage 4 1
Corpus (n = 167)
Antrum (n = 89)
Total (N = 428)
P-value all groups
P-value GOJ vs. corpus
P-value GOJ vs. antrum
P-value corpus vs. antrum
a
(78.7) (8.1) (8.1) (4.4) (0.7)
74 15 17 11 2
(62.2) (12.6) (14.3) (9.2) (1.7)
28 12 16 6 1
(44.4) (19.0) (25.4) (9.5) (1.6)
209 38 44 23 4
(65.7) (11.9) (13.8) (7.2) (1.3)
0.001
(0.057)
< 0.001
(0.109)
(67.6) (17.2) (9.7) (4.8) (0.7)
43 40 25 18 7
(32.3) (30.1) (18.8) (13.5) (5.3)
18 23 22 9 3
(24.0) (30.7) (29.3) (12.0) (4.0)
159 88 61 34 11
(45.0) (24.9) (17.3) (9.6) (3.1)
< 0.001
< 0.001
Introduction
The International Gastric Cancer Association (IGCA) and the International Society for Diseases of the Esophagus (ISDE) defined adenocarcinoma of the gastro-oesophageal junction (GOJ) as those located within 5 cm proximal or distal of the proximal end of the gastric folds [1]. In contrast, the seventh edition of the UICC TNM classification focuses on the simplification of therapeutic algorithms and defines all tumours that are located even within 5 cm distal European Journal of Gastroenterology & Hepatology 2015, 27:492–500 Keywords: cardiac cancer, gastric atrophy, gastric cancer, gastrooesophageal junction, intestinal metaplasia, OLGA, OLGIM a
Department of Gastroenterology, Hepatology and Infectious Diseases, Department of General, Visceral and Vascular Surgery, cInstitute of Pathology, Otto-von-Guericke University, Magdeburg, Germany and dCancer Research UK Cambridge Institute, University of Cambridge, Cambridge, UK b
Correspondence to Peter Malfertheiner, MD, Department of Gastroenterology, Hepatology and Infectious Diseases, Otto-von-Guericke University, Leipziger Str. 44, 39104 Magdeburg, Germany Tel: + 49 391 671 3100; fax: + 49 391 671 3105; e-mail: [email protected] Received 14 October 2014 Accepted 7 January 2015 Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal's website (www.eurojgh.com).
from the GOJ as oesophageal cancer as long as the tumour straddles the junction [2]. The TNM system thus suggests that all adenocarcinomas at the GOJ should be treated as a single (oesophageal) entity. This will potentially include proximal gastric cancers and distal oesophageal adenocarcinomas. Distal GOJ cancers have previously been reported to be less differentiated, and to show more aggressive behaviour, more often lymph node involvement, and a shorter overall survival compared with proximal GOJ cancers [3]. Furthermore, these cancers seem to develop by different biological mechanisms, with proximal GOJ cancers being more associated with gastrooesophageal reflux disease, Barrett’s oesophagus and obesity, whereas distal GOJ cancers are linked to Helicobacter pylori infection and preneoplastic conditions in the stomach [4]. Systems in current use to distinguish the origin of GOJ cancers are either based on the anatomical location of the main tumour mass [such as the TNM classification or the stratification proposed by Siewert and Stein (see the Materials and methods section)] or on paraclinical surrogates such as serum pepsinogens and anti-H. pylori immunoglobulin G [5,6]. Although epithelial dysplasia is the only ‘true’ premalignant lesion of the gastric mucosa, changes related to chronic inflammation such as intestinal metaplasia (IM) or glandular atrophy of the gastric mucosa are considered as
0954-691X Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
DOI: 10.1097/MEG.0000000000000299
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of the article is prohibited.
492
Gastro-oesophageal cancer and gastritis Bornschein et al.
preneoplastic conditions. The gold standard for identification of preneoplastic changes of the gastric mucosa is the histopathological assessment of endoscopic biopsies. The updated Sydney classification of gastritis remains the most widely used system for characterization of the status of the gastric mucosa for both research and clinical purposes [7]. New classification systems have been developed to enable a more consistent stratification of the individual risk of gastric cancer development. The OLGA and the operative link on gastritis assessment on intestinal metaplasia (OLGIM) staging systems for gastric premalignant lesions aim to allow simple stratification requiring the same biopsy protocol recommended by the Sydney system. The abbreviation OLGA stands for ‘Operative Link on Gastritis Assessment’ and is based on the grading of glandular atrophy, pooling the degree of atrophy in the antrum and corpus into a simple scoring system ranging from ‘0’ to ‘IV’ [8]. Further progression towards gastric cancer is expected in patients who present initially with stages III and IV [9]. The OLGIM system classifies only the degree of IM instead of all atrophic changes with or without metaplastic transformation. It was introduced as an alternative to OLGA as the interobserver agreement was shown to be better for the assessment of IM compared with atrophy [10]. OLGA stages correlate well with serum pepsinogens, which are used as surrogate markers for gastric atrophy; OLGA also correlated with IM and dysplastic changes [11–13] and the modified OLGIM system seems to provide complementary information [14,15]. It is still under debate as to which of these approaches is superior for the assessment of gastric cancer risk [10,14,16]. Although these scores were established to assess the risk of cancer associated with preneoplastic gastric conditions, recent studies have used OLGA or OLGIM scores in patients with a diagnosis of gastric cancer [17,18]. The aim of our study was to assess the association of GOJ tumours in their strict allocation to gastric inflammation and related preneoplastic conditions of the gastric mucosa. The secondary aim was to evaluate the clinical value of OLGA/OLGIM staging in patients with gastric cancer. Materials and methods Data collection
This is a retrospective analysis of histopathological changes of the gastric mucosa in patients with gastric and GOJ cancer. Tumour-specific data (Laurén type, location of the tumour, TNM stage, grading) were identified and the updated Sydney classification was applied to the nonmalignant gastric mucosa. The H. pylori infection status was recorded, when available. The electronic patient documentation system of the Department of Gastroenterology, Hepatology and Infectious Diseases at the University Hospital, Magdeburg, Germany, was used to retrieve patients who had been diagnosed with or treated for gastric adenocarcinoma, distal oesophageal adenocarcinoma or GOJ cancer between January 2000 and February 2013. A total of 707 patients fulfilled these criteria; 428 were included for the final analysis, 172 of whom had tumours at the GOJ. The reasons for exclusion of patients are shown in Fig. 1. About half of the patients in the final study cohort underwent surgical (or endoscopic) resection of the tumour. Patients who underwent resection showed a higher proportion of antrum cancers, less advanced local invasion
www.eurojgh.com
493
(T-stage), less lymph node involvement and less often distant metastases (Supplementary Table 1, Supplemental digital content 1, http://links.lww.com/EJGH/A8). The study was in line with the requirements of the local ethics authorities. Assessment of tumour location
Tumour location was assessed using either endoscopy or endoscopic ultrasound report, or the postsurgical pathological report, in the case of gastrectomy. GOJ cancers were classified according to the classification proposed by Siewert and Stein [1]. GOJ1 were considered as ‘distal oesophageal’ with the main tumour mass within 1–5 cm proximal to the GOJ, GOJ2 as ‘true junctional’ cancers within 1 cm proximal and 2 cm distal to the junction, and GOJ3 as ‘proximal gastric’ within 2–5 cm distal from the GOJ. For comparison with nonjunctional gastric cancer, gastric adenocarcinomas located in the corpus were also stratified into ‘proximal’ and ‘distal’ tumours, as described previously [19]; carcinomas with the main tumour mass in the fundus and in the proximal third of the corpus were classified as ‘proximal’ and tumours located in the distal two-thirds of the corpus were classified as ‘distal’. Histopathological assessment
Histopathological changes of the gastric mucosa were assessed according to the updated Sydney system [7]. The degree of IM and mucosal atrophy were graded 0–3 according to the predefined scoring system, with grade 0 in case of no alterations and grade 3 in case of severe changes – that is, almost complete replacement of the gastric mucosa. The activity of inflammation was graded according to the infiltration of the mucosa with neutrophil granulocytes and the chronicity of inflammation determined by the degree of infiltration with lymphocytes. Antrum and corpus were scored separately. In cases where the original report did not provide adequate information for scoring, specimens were re-evaluated by an expert pathologist. In the case of tertiary referrals who underwent gastrectomy at our hospital, but did not have local diagnostic gastroscopy data, histology was assessed on the surgical resection specimen or on pretreatment endoscopic biopsies. OLGA and OLGIM stages have been stratified as described previously [9,10,20]. However, for 38.8% of the patients, data on atrophy in the antrum or the corpus or both were not available, and data were missing on IM in 32.1% of the patients (Fig. 2). For the final analysis, only cases with either complete OLGA or OLGIM staging were included. Barrett’s oesophagus has been assessed in endoscopy reports or in the final pathology report after gastrectomy, where available. H. pylori status was evaluated according to any diagnostic reports that were available from (a) histology, (b) serology, (c) 13C-urea breath test, (d) stool antigen test or (e) a rapid urease test on biopsies. In case of any positive finding, the patient was considered currently or previously (serology) positive. Patients were considered negative with a negative result in serology or with negative results in at least two of the other modalities. This approach was adopted by consent of the authors.
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of the article is prohibited.
494
May 2015 • Volume 27 • Number 5
European Journal of Gastroenterology & Hepatology
Total number assessed:
N = 707
Gastro-oesophageal adenocarcinoma:
n = 643
Double entries or other entry than AC (including undifferentiated) n = 64 Unknown Laurén (e.g. preoperated) n = 107
n = 536
Complete Laurén classification: (mixed and mucinous type also excluded)
No data on tumour location n = 16 n = 520
Complete Laurén and location:
Incomplete histopathological data (IM and/or atrophy) n = 92
n = 428
Complete histopathological data:
Nonjunctional cancer n = 256 n = 172
Tumor location GOJ:
Fig. 1. Generation of the study population. Reason for exclusion of datasets from the final assessment. AC, adenocarcinoma; GOJ, gastro-oesophageal junction; IM, intestinal metaplasia.
IM GOJ
Corpus
Missing 19%
Incomplete 12%
Incomplete 16% Complete 66%
Complete 69%
GOJ
Complete 69%
Corpus
Antrum
Missing 23%
Missing 25%
Missing 24%
Incomplete 12%
Missing 19%
Missing 18%
Incomplete 12%
Atrophy
Antrum
Complete 64%
Incomplete 16%
Complete 59%
Incomplete 19%
Complete 58%
Fig. 2. Availability of histopathological data for the assessment of OLGA and OLGIM stages. Cases for which complete histopathological data on the prevalence and degree of IM and atrophy were available for staging according to either OLGA or OLGIM are shown. For about a third of cases, the data were either completely missing or incomplete for the antrum or the corpus. GOJ, gastro-oesophageal junction; IM, intestinal metaplasia; OLGA, Operative Link on Gastritis Assessment.
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of the article is prohibited.
Gastro-oesophageal cancer and gastritis Bornschein et al.
www.eurojgh.com
Statistical analysis
Group comparison of categorical data was performed using Fisher’s exact test. The parametric t-test was applied to compare age between the groups. The Kruskal–Wallis test was applied for a nonparametric comparison of Sydney scaled inflammatory scores and the Mann–Whitney U-test for post-hoc analyses in case of positive results. Spearman’s rank correlation test was used for correlation analyses. A two-sided significance level was defined as P-value less than 0.05, applying further Bonferroni’s correction in case of multiple comparisons. Data were analysed using IBM SPSS Statistics 20.0 (IBM SPSS Inc., Chicago, Illinois, USA). Results Demographic and tumour-specific data for gastrooesophageal junction cancers of different locations
Table 1 shows the demographics and tumour-specific data of the patients with GOJ tumours (n = 172), stratified by the Siewert classification. GOJ1 were almost exclusively of the intestinal type, whereas about a quarter of GOJ2 and GOJ3 showed a diffuse or a mixed phenotype. Distally located tumours were of more advanced stage and less differentiated compared with GOJ1. Data on lymph node involvement and distant metastases are incomplete because of the retrospective nature of the study. Demographics and tumour-specific data of the broader cohort (GOJ cancer: n = 211) are shown in Supplementary Table 2 (Supplemental digital content 1, http://links.lww.com/EJGH/A8). The distribution of the factors analysed was similar to that of the final study cohort. Preneoplastic and inflammatory changes of the gastric mucosa in patients with gastro-oesophageal junction cancer
The prevalence of both IM and glandular atrophy of the gastric mucosa was higher for GOJ3 compared with
495
GOJ1, with GOJ2 showing intermediate values (P < 0.001, P = 0.011, respectively; Fig. 3). Also, the degree of both IM and glandular atrophy was more severe for distally located adenocarcinomas at the GOJ (Fig. 3a and b). Similarly, activity and chronicity of gastric inflammation were more severe in GOJ3 (Fig. 4c and d). The degree of IM correlated strongly with the degree of atrophy both in the antrum and in the corpus (r = 0.362–0.659, P < 0.001; Supplementary Table 3, Supplemental digital content 1, http://links.lww.com/EJGH/A8). The same held true for the degree of inflammatory activity and chronicity, respectively (r = 0.384–0.529, P < 0.001; Supplementary Table 3, Supplemental digital content 1, http://links.lww.com/EJGH/A8). In addition, chronicity of inflammation in the corpus seemed to be an indicator of metaplastic or atrophic changes as it correlated with the both the maximal degree of IM (r = 0.335, P < 0.001) and atrophy (r = 0.289, P < 0.001). IM in the stomach was significantly detected more often in patients with less differentiated GOJ cancers (G1/2: 28.6% vs. G3/4: 45.7%, P = 0.033). However, there was no difference in the presence of IM in the stomach between cancers of intestinal or diffuse type (34.9 vs. 34.6%, P = 1.000). Gastric atrophy was only detected in patients with intestinal-type cancers (22.8 vs. 0.0%, P = 0.005). Comparison with nonjunctional gastric cancer
Table 2 shows the clinicopathological characteristics of carcinomas located in the antrum and corpus compared with GOJ tumours. Supplementary Table 4 (Supplemental digital content 1, http://links.lww.com/EJGH/A8) shows these data for the extended study cohort (n = 520). However, as discussed above, the distribution of the factors addressed has been similar to the final study cohort (Supplementary Table 5, Supplemental digital content 1, http://links.lww.com/ EJGH/A8); thus, we refer here only to the final cohort.
Table 1. Demographic and tumour-specific data of patients with gastro-oesophageal junction cancer n (%) GOJ1 (n = 61) GOJ2 (n = 53) GOJ3 (n = 58) Age (median, IQR) (years) Sex (male) Laurén (intestinal type) T-stagea (n = 147) T1 T2 T3 T4 Lymph node involvement (N-positive)a (n = 136) Distant metastases (M1)a (n = 64) Grading/differentiationa (n = 168) Well Moderate Poor
66.0 60.9–73.9 53 (86.9) 60 (98.4)
69.6 62.5–74.9 45 (84.9) 41 (77.4)
68.1 57.8–75.3 45 (77.6) 45 (77.6)
18 13 17 4 23
7 13 18 5 29
4 16 28 4 42
(34.6) (25.0) (32.7) (7.7) (53.5)
(16.3) (30.2) (41.9) (11.6) (70.7)
(7.7) (30.8) (53.8) (7.5) (80.8)
P-value all groups
P-value GOJ1 vs. GOJ2
P-value GOJ1 vs. GOJ3
P-value GOJ2 vs. GOJ3
(0.737)
–
–
–
(0.376) < 0.001
– 0.001
– < 0.001
– (1.000)
(19.7) (28.6) (42.9) (8.8) (69.1)
0.036
(0.247)
0.006
(0.462)
0.017
(0.120)
0.007
(0.328)
Total (N = 172) 67.8 60.8–75.0 143 (83.1) 146 (84.9) 29 42 63 13 94
9 (32.1)
7 (46.7)
12 (57.1)
28 (43.8)
0.209
–
14 (24.6) 26 (45.6) 17 (29.8)
7 (13.2) 26 (49.1) 20 (37.7)
4 (6.9) 21 (36.2) 33 (56.9)
25 (14.9) 73 (43.5) 70 (41.7)
0.015
(0.310)
Group comparison was performed using Fisher’s exact test. Percentages are given per available cases. Post-hoc P-values are given if comparison of all groups has been significant; Bonferroni’s correction for multiple comparisons was applied. GOJ, gastro-oesophageal junction; IQR, interquartile range. a Data on these features were not available for all cases. Significance for P < 0.05.
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of the article is prohibited.
–
0.003
–
(0.124)
496
May 2015 • Volume 27 • Number 5
European Journal of Gastroenterology & Hepatology
80% 70%
IM Atrophy
60% 50% 40% 30% 20% 10% 0%
GOJ1
GOJ2
GOJ3
Proximal corpus
Distal corpus
Antrum
Fig. 3. Prevalence of atrophy and intestinal metaplasia (IM) in the stomach. The prevalence for both IM (P < 0.001) and glandular atrophy (P = 0.011) was significantly higher in tumours located distally at the junction, showing a pattern similar to nonjunctional gastric cancer. GOJ, gastro-oesophageal junction.
With respect to the degree of preneoplastic and inflammatory changes of the gastric mucosa, all conditions (IM, atrophy, activity and chronicity of inflammation) were more severe for nonjunctional tumours compared with GOJ cancers (P < 0.001, data not shown). In the posthoc analysis, the prevalence and degree of IM were significantly lower for GOJ1 and GOJ2 compared with nonjunctional cancers of different locations (P < 0.01; Fig. 3). This was not the case for GOJ3, which showed comparable prevalence to nonjunctional cancers and a lower degree of IM in the corpus only compared with cancers in the distal corpus (P = 0.005) and of IM in the antrum compared with antrum cancers (P = 0.002). Also,
(a) 100.0% 90.0% 80.0% 70.0% 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0%
None
Antrum
Corpus
Mild
Antrum Corpus
GOJ1 (c) 100.0% 90.0% 80.0% 70.0% 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0%
GOJ2
None
Antrum
Corpus
GOJ1
Moderate
Mild
Moderate
Antrum
Corpus
GOJ2
OLGA and OLGIM stages for patients with gastrooesophageal junction cancer and nonjunctional gastric cancer
Here, we included patients with both complete and incomplete histopathology records for the entire cohort with gastric and junctional adenocarcinomas (n = 520;
(b)
Severe
Antrum
the prevalence of atrophy was significantly lower for GOJ1 compared with nonjunctional cancers (P < 0.01) as well as the severity of corpus atrophy (P < 0.001). Atrophy in the antrum was lower only compared with antral cancers (P < 0.001). GOJ2 showed a similar pattern of gastric atrophy compared with proximal corpus cancers, but the prevalence was lower compared with more distal gastric cancers (P < 0.01). The degree of atrophy in the corpus was lower compared with distal corpus carcinomas (P = 0.001) and the degree of antral atrophy was lower compared with antral cancers (P = 0.003). For GOJ3, there was no difference in the prevalence or the degree of gastric atrophy compared with nonjunctional cancers. Activity of gastric inflammation was similar for GOJ1 compared with proximal corpus cancers, but significantly lower compared with cancers in the distal corpus and the antrum (P ≤ 0.001). Inflammatory activity was not different between nonjunctional cancers and GOJ2 as well as GOJ3. Chronicity of inflammation was lower in GOJ1 relative to nonjunctional cancers (P < 0.001). GOJ2 and GOJ3 showed less inflammatory chronicity in the corpus compared with distal corpus cancers (P = 0.002), but similar chronicity in the antrum and corpus compared with nonjunctional cancers of other locations.
Corpus
100.0% 90.0% 80.0% 70.0% 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0%
Antrum
Corpus
(d)
Severe
Corpus
GOJ3
Mild
Antrum
GOJ1
GOJ3
Antrum
None
100.0% 90.0% 80.0% 70.0% 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0%
Corpus
GOJ1
Severe
Corpus
Antrum
GOJ2
None
Antrum
Moderate
Mild
Moderate
Severe
Antrum Corpus
Antrum
GOJ2
Corpus
GOJ3
Corpus
GOJ3
Fig. 4. Distribution of the degree of preneoplastic and inflammatory changes in patients with GOJ cancer. The severity of preneoplastic conditions and inflammatory changes of the gastric mucosa in patients with GOJ cancer graded according to the updated Sydney classification is shown, with 0: no changes, 1: mild, 2: moderate, 3: severe changes (a) gastric IM, (b) gastric glandular atrophy, (c) activity of gastric inflammation by infiltration with neutrophils, (d) chronicity of gastric inflammation by infiltration with lymphocytes. GOJ, gastro-oesophageal junction; IM, intestinal metaplasia.
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of the article is prohibited.
Gastro-oesophageal cancer and gastritis Bornschein et al.
www.eurojgh.com
497
Table 2. Demographic and tumour-specific data of patients with gastric and junctional adenocarcinomas n (%) P-value all groups
P-value GOJ vs. corpus
P-value GOJ vs. antrum
P-value corpus vs. antrum
(0.092)
–
–
–
< 0.001 < 0.001
< 0.001 < 0.001
< 0.001 < 0.001
(0.894) (0.599)
(21.6) (27.5) (41.1) (9.8) (69.6)
(0.062)
–
–
–
0.026
(0.190)
(0.144)
0.009
–
–
–
GOJ (n = 172)
Corpus (n = 167)
Antrum (n = 89)
Total (N = 428)
67.8 60.8–75.0 143 (83.1) 146 (84.9)
65.8 54.6–75.0 100 (59.9) 85 (50.9)
69.0 60.7–76.0 52 (58.4) 49 (55.1)
67.7 57.2–75.1 295 (68.9) 280 (65.4)
21 22 17 9 35
73 93 139 33 211
Age (median, IQR) (years) Sex (male) Laurén (intestinal type) T-stagea (n = 338) T1 T2 T3 T4 Lymph node involvement (Npositive)a (n = 303) Distant metastases (M1)a (n = 142) Grading/differentiationa (n = 418) Well Moderate Poor
29 42 63 13 94
(19.7) (28.6) (42.9) (8.8) (69.1)
23 29 59 11 82
(18.9) (23.8) (48.4) (9.0) (77.4)
(30.4) (31.9) (24.6) (13.0) (57.4)
28 (43.8)
25 (50.0)
11 (43.8)
64 (45.1)
(0.651)
25 (14.9) 73 (43.5) 70 (41.7)
15 (9.3) 47 (29.0) 100 (61.7)
6 (6.8) 25 (28.4) 57 (64.8)
46 (11.0) 145 (34.7) 227 (54.3)
0.001
0.001
0.002
(0.836)
Group comparison was performed using Fisher’s exact test. Percentages are given per available cases. Post-hoc P-values are given if comparison of all groups has been significant. GOJ, gastro-oesophageal junction; IQR, interquartile range. a Data on these features were not available for all cases. Significance for P < 0.05.
Fig. 1; Supplementary Table 4, Supplemental digital content 1, http://links.lww.com/EJGH/A8). OLGA staging was possible in 318 (61.2%) and OLGIM staging in 353 (67.9%) patients. In the remaining cases, histopathological scoring from either the antrum or the corpus was incomplete (Fig. 3). For GOJ tumours (n = 211), OLGA and OLGIM staging was possible in 64.5 and 68.7%, respectively. The data availability on OLGA and OLGIM was lowest for GOJ1 (61.2 and 63.5%, respectively) compared with GOJ2 (69.8 and 73.0%, respectively) and GOJ3 (63.5 and 71.4%, respectively). Considering all junctional and nonjunctional cancers, location itself was not an influencing factor on data availability for OLGA and OLGIM. However, Laurén type (P = 0.024 and 0.024, respectively), T-stage (P = 0.046 and 0.015, respectively) and N stage (P = 0.009 and 0.001, respectively) did influence data availability, with a lower proportion of tumours of the diffuse type and those with advanced T-stage and nodal involvement showing complete data. The distribution of OLGA and OLGIM stages in our study population is shown in Table 3. In our cohort, 8.5 and 12.7% of the patients presented with stages III and IV according to OLGA and OLGIM, respectively. Nonjunctional cancers showed higher stages of both OLGA and OLGIM stages than GOJ cancers (P < 0.001 and < 0.001, respectively), and OLGA stages were higher in intestinal-type tumours (P = 0.007). Otherwise, there was no correlation of the distribution of OLGA or OLGIM stages with clinicopathological characteristics in our cohort. Discussion
In our study, GOJ3 showed a strong association with the degree of IM and atrophy of the background gastric mucosa, which was not the case for GOJ1 cancers. This suggests a different aetiology for adenocarcinomas at this location. It has been postulated previously that junctional cancers are either
associated with H. pylori-induced changes of the gastric mucosa or with a history of gastro-oesophageal reflux disease (GORD) [5,6]. The Irish FINBAR study investigated the association of H. pylori infection and gastric atrophy with GORD-induced alterations of the distal oesophagus, namely, oesophagitis, Barrett’s oesophagus and oesophageal adenocarcinoma. There was an inverse relationship between H. pylori infection and gastric atrophy with oesophageal adenocarcinomas that could not be confirmed for true junctional cancers [21]. In a recent Japanese study, Barrett’s associated cancers at the GOJ showed significantly less gastric atrophy relative to tumours without evidence of adjacent Barrett’s metaplasia [22]. The distinct association of gastric atrophy with cancer from different locations at the junction was also confirmed by a recent meta-analysis [23]. Overall, inflammatory changes in the gastric mucosa are less severe in patients with GOJ cancer than nonjunctional gastric cancer [24]. However, although this difference was distinct for GOJ1, GOJ3 showed a pattern that is comparable with nonjunctional gastric cancer. Thus, it is fair to infer that GOJ3 are related to gastric inflammation, whereas GOJ1 are not. Once more, GOJ2 have an intermediate position, showing less severe preneoplastic conditions, but gastric inflammation comparable to adenocarcinomas in the distal corpus and the antrum. The positive correlation of the degree of inflammation in the corpus with preneoplastic changes in the stomach further suggests that corpus-dominant inflammation could indicate a higher risk for the development of gastric cancer (including GOJ3) as reported previously [25–27]. Similar data have been reported for IM, where not only the degree but also the distribution of IM defines the risk for the development of gastric cancer [16,28,29]. In a Japanese study on patients who developed gastric cancer after H. pylori eradication, IM in the corpus at baseline was the only independent predictor for progression in multivariate analysis [30].
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of the article is prohibited.
498
May 2015 • Volume 27 • Number 5
European Journal of Gastroenterology & Hepatology
Table 3. Distribution of OLGA and OLGIM stages for patients with gastric and junctional adenocarcinomas n (%) GOJ (n = 172) OLGA (n = 318) Stage 0 107 Stage 1 11 Stage 2 11 Stage 3 6 Stage 4 1 a OLGIM (n = 353) Stage 0 98 Stage 1 25 Stage 2 14 Stage 3 7 Stage 4 1
Corpus (n = 167)
Antrum (n = 89)
Total (N = 428)
P-value all groups
P-value GOJ vs. corpus
P-value GOJ vs. antrum
P-value corpus vs. antrum
a
(78.7) (8.1) (8.1) (4.4) (0.7)
74 15 17 11 2
(62.2) (12.6) (14.3) (9.2) (1.7)
28 12 16 6 1
(44.4) (19.0) (25.4) (9.5) (1.6)
209 38 44 23 4
(65.7) (11.9) (13.8) (7.2) (1.3)
0.001
(0.057)
< 0.001
(0.109)
(67.6) (17.2) (9.7) (4.8) (0.7)
43 40 25 18 7
(32.3) (30.1) (18.8) (13.5) (5.3)
18 23 22 9 3
(24.0) (30.7) (29.3) (12.0) (4.0)
159 88 61 34 11
(45.0) (24.9) (17.3) (9.6) (3.1)
< 0.001
< 0.001
