Combination of Paris and Vienna Classifications may Optimize Follow-Up of Gastric Epithelial Neoplasia Patients

Background

Gastric cancer is a malignancy associated with considerable morbidity and mortality worldwide. The prognosis of advanced gastric cancer is significantly worse compared with early gastric cancer (EGC) [1]; therefore, how to detect and treat EGC is an important issue around the world. Since the inflammation-atrophy-metaplasia-dysplasia-carcinoma sequence was raised as a human model of gastric carcinogenesis [2], dysplasia, which is the last step of the development process for precancerous lesions, has been the major topic of discussion between endoscopists and pathologists.

Dysplasia was precisely defined by an international IBD-dysplasia morphology study group as lesions showing “unequivocal, non-invasive (confined within the basement membrane), neoplastic transformation of the epithelium excluding all reactive changes” and was then applied extensively to the whole gastrointestinal tract [3]. Dysplasia mainly depends upon the recognition of morphological features such as cytological and architectural changes in routinely processed hematoxylin and eosin (H&E)-stained sections [4]. Because large discrepancies existed between Western and Japanese pathologists, a series of meetings were held in Padova and Vienna for developing common terminologies for gastrointestinal epithelial neoplasia (GEN) [5–8]. Recently, Korean pathologists have made an effort to achieve an improved diagnostic consensus for GEN [9]. The revised Vienna classification categorizes biopsy-proven dysplasias into 5 groups and gives recommendations for each group [6]. The cut-off for choosing between endoscopic resection or follow-up for low-grade neoplasia (LGN) was not clearly established, while the choice of management for higher or lower grades was generally accepted.

In contrast to the pathological perspective of the Vienna classification of GEN, the Paris classification is an endoscopic/macroscopic standard for GEN [10]. In the outstanding clinical research performed by Japanese experts, the medical term “type 0” was defined for the endoscopic gross appearance of any tumor resembling early carcinoma, including adenoma/dysplasia and advanced carcinoma [11]. Not all type 0 lesions are malignant; for example, a type 0–IIa lesion can be a juvenile or proliferative polyp, and a type 0–III lesion can be a benign ulcer [10].

The above indicates that diagnosis of a precancerous lesion cannot be accomplished by pathology or endoscopy alone. We presumed that combining the 2 procedures may be a more optimal method during our follow-up study.

Material and Methods

PATIENTS:

Our prospective study was conducted in the Second Affiliated Hospital of Zhejiang University School of Medicine. Patients who had undergone gastroscopy and biopsy-proven GEN were included between January 2003 and September 2010. Exclusion criteria included patients with a past history of cancer, those who have been using NSAIDs medications, those who have previously undergone surgical or endoscopic treatments, and those with severe concomitant illnesses (e.g., cardiac, respiratory, hepatic, or renal insufficiency). A total of 410 consecutive patients were enrolled in our prospective study. A total of 170 patients with a follow-up of more than 12 months (at least 3 examinations, mean 28 months, range 12–80 months) were finally documented. There were 101 male patients (mean age 56 years, range 29–81 years) and 69 female patients (mean age 56 years, range 33–82 years). Written informed consent was obtained from all of the above patients (see detailed flowchart in Figure 1).

ENDOSCOPIC ASSESSMENT:

The clinical study was carried out with conventional single-channel endoscopes (Olympus GIF-XQ 240 and GIF-XQ 260; Olympus Optical Co., Ltd, Tokyo, Japan).

The macroscopic description of GEN was classified into the following 3 types: type 0 (including type 0–I, IIa, IIb, and type 0–IIc, IIa+IIc), no visible lesions (including patients without specific lesions but with background lesions such as erosion, atrophy, and superficial gastritis), and the ulcer type (including type 0–III; before we knew the pathologic results, we used “ulcer” instead of “type 0–III”). Type 0 lesions were divided into depressed and non-depressed categories.

The stomach was divided into 5 sections: cardia, fundus, body, antrum, and pylorus. The cross-sectional circumference was divided into 4 equal sections: greater curvature, lesser curvature, anterior wall, and posterior wall [12].

The detailed information (e.g., size and place) should be added after the macroscopic description.

Biopsies were taken from visible lesions ranging 2–5, and biopsies were randomly taken from 2 sites with less curvature in the gastric antrum if there were no visible lesions. Biopsies were taken from the same place during follow-up.

:

To minimize intra-procedural variation, all forceps biopsies were evaluated by the same experienced pathologist. All biopsy samples were fixed in 10% formalin and embedded in paraffin. Sections were cut into 4-μm thickness for H&E staining. Endoscopic mucosal resection (EMR), endoscopic submucosal dissection (ESD) or polypectomy specimens were washed in normal saline, fixed in 10% formalin, sectioned serially at 2-mm intervals, and entirely embedded in paraffin. Surgical specimens were fixed in 10% formalin, after recording lesion location, number, size and macroscopic type, and each lesion was removed, embedded in paraffin, and sectioned along the longitudinal axis for H&E staining. Lymph nodes around the stomach were processed in the same way, and the number of metastases was recorded.

Helicobacter pylori infection was defined as positive if a breath test, an H. pylori culture, or a histological evaluation according to the updated Sydney system [13] was positive. Breath tests for H. pylori were performed 1 month later after withdrawal of PPIs, bismuth, and any antibiotics in case of a false-negative result. The biopsy specimens for H. pylori tests were taken from the lesser curvature of the antrum.

MEDICAL TREATMENTS:

Most patients mainly had problems such as abdominal distension, acid regurgitation, and dyspepsia. Proton pump inhibitors (PPIs) and mucosal-protective drugs were prescribed to patients to relieve their symptoms. If the H. pylori test was positive, patients received H. pylori eradication therapy for 2 weeks until a negative result was obtained.

ENDOSCOPIC AND SURGICAL TREATMENTS:

Endoscopic resection included ESD or EMR, with patients under intravenous sedation. With the application of marking dots around the lesion and necessary hemostatic treatment such as endoscopic clipping or thermocoagulation, every lesion was completely removed with negative margins and no complications.

EMR was performed with a snare and the Endocut mode (50 W) of an electrosurgical generator (PSD-60; Olympus), after a saline solution containing 0.005 mg/mL epinephrine was injected into the submucosa beneath the lesion.

ESD was performed with an insulated tip knife (Olympus). After a 10% glycerin solution that contained 0.005 mg/mL epinephrine was injected into the submucosa, a circumferential incision of the mucosa was then made outside the marking dots. The whole tissue, including some submucosa connective tissue, was dissected en bloc from the muscle layer [14].

Patients with HGN, or LGA but suspicious for potential malignancy received surgical gastrectomy due to incomplete endoscopic resection. The endoscopist marked the lesion locations with forceps or methylene blue staining.

FOLLOW-UP REGIMEN:

According to the Vienna classification (Revised, 2002) [7], biopsy-proven LGA or dysplasia belongs to category 3 (low-grade GEN), and high-grade adenoma or dysplasia belongs to category 4 (high-grade GEN).

Previous reports strongly suggested that high-grade adenoma/dysplasia (HGD, category 4 in the Vienna classification) is highly predictive of invasive carcinoma (category 5), which either coexists or appears thereafter. Therefore, complete endoscopic or surgical resection is strongly recommended for patients with HGD, regardless of the macroscopic type [2,6,7].

We focused on the patients who have biopsy-proved dysplasia during the enrollment process. Patients with Paris classification (which may indicate dysplasia in pathology) were paid special attention. We followed up this group of patients every half year. Patients were enrolled into our group as soon as dysplasia was found. Patients without dysplasia in 2-year follow-up were not traced thereafter.

Gastric adenoma is a benign tumor of the glandular epithelium with varying degrees of cellular atypia and has papillary and/or tubular structures [15]. Gastric adenomas tend to develop into invasive adenocarcinoma [13–16]. Tumor size and grading of dysplasia have been indicated as prognostic factors of gastric adenomas [17], and the definite malignant potential of adenomas is indicative for endoscopic resection even if the adenomas are low-grade.

While the recommendations for follow-up or endoscopic resection for category 3 (LGA/LGD) listed by the Vienna classification (Revised, 2002) seem very reasonable for managing LGN, endoscopists and gastroenterologists are still confused about how to proceed. No consensus or guideline for the detailed management of biopsy-proven LGN has yet been reached. Based on the present follow-up, we subdivided patients of this group into several categories with different recommendations. For patients who had no visible lesions, follow-up alone was recommended. For type 0 lesions, endoscopic resection was recommended for depressed lesions, and close follow-up was suitable for non-depressed lesions.

Ulcer lesions should be carefully examined. Biopsy-proven HGN surrounding an ulcer lesion demonstrate potential malignancy and should be removed immediately. While LGD is not very indicative, it could indicate gastric cancer or inflammatory changes. Therefore, close follow-up should be initiated, and re-biopsy should be performed until healing takes place or higher-level neoplasia is observed. This regimen is summarized in Table 1.

Results

EFFICIENCY OF OUR FOLLOW-UP AND TREATMENT REGIMEN:

There were totally 4 patients who had type 0 lesions without dysplasia which were not traced after 2-year follow-up, and 4 patients who had no dysplasia initially and were found to have dysplasia in follow-up were thus finally enrolled into our study.

In total, there were 29 patients of gastric mucosal neoplasia who received endoscopic or surgical resections in our follow-up study, including 24 patients at enrollment (Table 3). The indications for interventional therapy were: HGN, LGA, depressed type 0 lesions with LGD, and other lesions with potential for bleeding.

By comparing the biopsy and post-resection diagnoses, the misdiagnosis rate was 62.07% (18/29), and the under-diagnosis rate was 51.72% (15/29), with 38.9% (7/18) in the LGN group and 72.7% (8/11) in the HGN group, respectively (Table 4).

Discussion

Dysplasia was initially proposed by an international IBD-dysplasia morphology study group as lesions showing “unequivocal, non-invasive (confined within the basement membrane), neoplastic transformation of the epithelium excluding all reactive changes”. However, the presence of dysplasia was confusing. Diagnosis of dysplasia mainly depends on the recognition of morphological features such as cytological and architectural changes in routinely processed hematoxylin and eosin-stained sections, and our pathologists could not clearly differentiate these changes. Consequently, most endoscopists or gastroenterologists will be confused about dysplasia, and a definite diagnosis of a neoplastic lesion cannot be determined; thus, resection cannot be confidently recommended. Based on the consensus reached at the international pathological conferences held in Padova and Vienna, biopsy-proven HGN became a definite indication for resection, while LGD/LGN was still confusing. Previous studies showed that most cases of LGN disappeared due to over-diagnosing regenerative (i.e., inflammatory) lesions as neoplastic lesions [18]. In fact, we determined that most “dysplasia” specimens in pathological reports were randomly biopsied, and no visible lesions were observed in the endoscopic examination. There were 132/170 (77.6%) patients with LGN but no visible lesions; of the 132 cases, 111 (84.1%) regressed and 18 (13.6%) persisted, which agreed with previous studies. Depending on the recognition of outcomes in this group of neoplasia, we can presume that patients with neoplasia and no visible lesions may display reactive or regenerative inflammatory changes and can be reversed with or without medications.

The Paris classification of superficial neoplastic lesions assisted us greatly in recognizing “real neoplastic dysplasia”. Since the macroscopic classification of early gastric carcinomas was established by Japanese experts, the prefix “type 0” (e.g., 0–I, 0–IIa) has been widely used to distinguish early and superficial carcinomas from advanced carcinomas and is becoming the general practice in detecting endoscopic gross appearances of any tumor resembling early carcinoma, including adenoma/dysplasia and advanced carcinoma [11]. To extend the useful methodology of detecting precursor lesions endoscopically, Japanese experts convinced other participants to reach a consensus regarding the Paris endoscopic classification of superficial neoplastic lesions in 2002. “Superficial” neoplasia includes neoplastic lesions with no invasion in the lamina propria and carcinoma with invasion of the lamina propria and a depth of penetration limited to the mucosa (stomach and esophagus) or the submucosa (large bowel). Almost every superficial neoplasia has a “superficial” morphology, but not every lesion of superficial morphology suggests neoplasias (e.g., hyperplastic polyps in the GI tract have little or no potential for transformation into neoplastic lesions but always have a protruding morphology). It is reasonable that polypoid precursors play a minor role in the development of advanced gastric cancer, while depressed non-polypoid lesions (type 0–IIc) are the usual precursors [10]. Cho et al. [19] analyzed 236 LGD lesions treated with endoscopic resection and showed that depressed morphology, surface erythema, or a size of 1 cm or greater were risk factors for HGD/carcinoma; thus, endoscopic resection is recommended even with biopsy-proven LGD. Based on the above, we showed that the Paris classification is an efficient and accurate way to screen neoplastic and non-neoplastic lesions, and depressed appearance can be a specific indication for endoscopic or surgical resection. In our study, 4 patients who had type 0 lesions without dysplasia were not traced after 2-year follow-up, and 4 patients who had no dysplasia initially and found to have dysplasia thus were finally enrolled into our study. Eighteen patients with type 0 lesions and LGN were divided into 2 groups: 9 patients with non-depressed lesions and 9 with depressed lesions; clinical management revealed that 6/9 of the latter group were progressing to cancer (Table 3).

Gastric adenomas are benign epithelial neoplastic tumors with a glandular organization that are characterized by localized proliferation of adenomatous epithelium with tubular and/or papillary structures [15]. Kamiya et al. reviewed 74 patients in the 1980s and observed gastric adenomas that underwent malignant changes with gradual transformation from moderate to severe dysplasia [16]. Kasuga et al. recently retrospectively reviewed 231 gastric adenomas by comparing forceps biopsy and post-resection diagnoses and recommended en bloc resection even when forceps biopsy indicated LGA [14]. In our study, we subdivided LGN into LGA and LGD groups to treat them. Seven patients with biopsy-proven LGA were resected quickly (except for 1 who failed in ESD), including 1 patient with high-grade adenoma and 6 with LGAs.

Gastric ulcers have been frequently noted for their probability of malignancy. There are 2 hypotheses regarding malignant ulcers: first, long-time chronic ulcers lead to malignant changes, and second, the carcinoma itself presented with ulcer appearance [20]. There is no doubt that biopsy-proven HGN should be treated immediately, while in the case of biopsy-proven LGN, we can re-biopsy the lesion if suspicious for carcinoma or perform close follow-up within 1 month. In our study, 2 patients with LGN progressed to HGN within 1 month, which means that misdiagnosis took place the first time but the proper diagnosis was later discovered through close follow-up. There were 4 patients who developed multiple ulcers with LGN after the initial examinations, which may not indicate gastric cancer; however, close follow-up was required to exclude gastric lymphoma.

From our follow-up chart of 146 patients, except for 129 non-visible-lesion patients fluctuating between “LGD” and “normal”, 3 patients progressed to LGA or HGN ranging from 1 to 7 years, and 1 patient with non-depressed lesions progressed to LGA but finally had the lesions resected. Although the regression ratio was very high in the LGN group, a few did progress to HGN even after 6 years, which would be a possible sign of gastric cancer. We recommend using a follow-up strategy (we recommend at least once a year) for all cases of this group. For patients who have visible lesions (e.g., non-depressed type 0 lesions) with LGN, we recommend that the first follow-up should be initiated in 3–6 months to reduce the possibility of misdiagnosing gastric adenomas. Special attention should be paid to ulcer lesions with LGN, as we mentioned above. The follow-up interval should be strictly controlled for 1 month until definite endoscopic mucosal healing occurs.

In total, 29 patients received en bloc resection. All patients had 1 of the following indications: biopsy-proven HGN, biopsy-proven LGA, biopsy-proven LGD with depressed type 0 appearance, or potential for bleeding. By comparing the forceps biopsy and post-resection diagnoses, we determined that the misdiagnosis rate was 62.07% (18/29), and the under-diagnosis rate was 51.72% (15/29), with 38.9% (7/18) in the LGN group and 72.7% (8/11) in the HGN group. Endoscopic forceps biopsy (EFB) may cause errors in histological diagnosis and thus cause significant discrepancy between EFB and en bloc resection or surgical diagnosis [21]. Forceps biopsy errors are the leading cause of the high misdiagnosis rate, excluding variance among endoscopists and pathologists. In follow-up using the Paris classification and endoscopic techniques (EMR or ESD), with increasing attention to the probability of biopsy misdiagnosis, the problems seem to be solved to a certain extend; however, overtreatment can occur.

The limitations of the present study include the following points. First, we enrolled 410 patients in the initial examination; however, due to poor patient compliance, only 170 patients could complete our follow-up. Second, medications for H. pylori eradication were prescribed in our follow-up group, and the efficiency of medications for reversing low-grade gastric neoplasia have not been well evaluated; thus, we could not exclude the possibility that medications influenced the regression of gastric neoplasia. We plan to continue our study by obtaining results from patients on a larger scale.

Conclusions

Diagnosis of GEN by endoscopy or pathology alone was shown be insufficiently accurate. Additionally, the combination of the Vienna and Paris classifications could be helpful for follow-up and timely treatment of suspicious precancerous lesions and for reducing the misdiagnosis rate of EGC, especially in countries where, unlike Japan, the early detection and treatment rates for gastric cancer are low and advanced equipment (such as narrow-band imaging, NBI, or magnifying endoscopy) or detection experience are lacking.