Portal de Periódicos da CAPES

Gastric ulcer and malignancy – is there a need for follow‐up endoscopy?

2004; Wiley; Volume: 19; Issue: 9 Linguagem: Inglês

10.1111/j.1365-2036.2004.01933.x

1365-2036

John A. Todd, Cathy Richards, Adam D. Dixon, Richard Robinson,

Gastrointestinal disorders and treatments

The incidence of gastric adenocarcinoma has declined over the last 30 years. At the same time there has been an improvement in 5-year survival rates. However, only 10% of patients with gastric cancer will still be alive 5 years after diagnosis.1 Gastric cancer can present as an exophytic lesion, diffuse infiltration of the gastric mucosa or even as a gastric ulcer, which can mimic a benign gastric ulcer. The majority of gastric cancers are diagnosed in an advanced state when curative treatment is no longer possible or in patients whose co morbidity does not allow curative treatment to be attempted. This has lead to an emphasis on detecting early gastric cancers. It is recommended that all patients with gastric ulcers should undergo repeat gastroscopy to assess ulcer healing and exclude malignancy, even if the initial endoscopic appearance is benign and the biopsies show no evidence of malignancy. This has been re-inforced in the latest British Society of Gastroenterology guidelines.2 We sought to investigate if repeat endoscopy was worthwhile in patients with gastric ulcers and to establish the yield of follow-up endoscopy in diagnosing gastric cancer. Glenfield General Hospital is a teaching hospital in Leicester. Approximately 2500 upper gastrointestinal (GI) endoscopies are performed per year in the endoscopy unit. Since October 1997 all endoscopic procedures have been recorded on an Endoscopic Database (v2.21; Unisoft Computers Ltd, Enfield, Middlesex, UK). Using this database, all endoscopically diagnosed gastric ulcers and suspected gastric cancers between 1 January 1998 and 30 June 2002 were identified. The endoscopists were either consultant gastroenterologists/surgeons or trainee gastroenterologists/surgeons. The Leicestershire Pathology Service analyses histological specimens from all of Leicestershire. Using the Leicestershire Pathology database, all patients in Leicestershire who had a diagnosis of gastric adenocarcinoma, gastric lymphoma, gastric dysplasia (low or high grade) between 1 January 1998 and 30 June 2003 were identified. The results from the two databases were combined and analysed in order to: Establish the accuracy of the index endoscopy in detecting gastric cancer. Determine the number of gastric cancers detected at follow-up endoscopy that had not been suspected at index endoscopy. From 1 January 1998 to 30 June 2002, 11 679 upper GI endoscopies were performed in the endoscopy unit. One hundred gastric carcinomas were diagnosed, of which six were lymphomas. Three hundred and forty-eight benign gastric ulcers were diagnosed. Macroscopically the endoscopist suspected 108 patients to have malignant pathology. Of these, eighty-one were subsequently found to have confirmed carcinoma. Seventy-nine of these cases of gastric cancer were confirmed at the time of the first endoscopy by histology. Of the two remaining patients, one was confirmed on histology 23 days later and the other 112 days later, after repeat endoscopies. The endoscopist diagnosed benign gastric ulcers in 340 patients. Of these, 15 were found to have adenocarcinoma on biopsies taken at the index endoscopy. Five patients had dysplasia on the initial biopsies in this group, three had high-grade dysplasia, which on later biopsy turned out to be adenocarcinoma. Two patients with low-grade dysplasia underwent repeat endoscopy, which demonstrated that the ulcers had healed with no evidence of dysplasia on histology. Two hundred and twelve patients who had no suspicion of malignancy either macroscopically or microscopically at index endoscopy underwent follow-up endoscopy at Glenfield. One patient initially thought to have a benign ulcer on the basis of macroscopic and microscopic appearance, had high-grade dysplasia on repeat gastroscopy 66 days later. The patient subsequently underwent a Bilroth I gastrectomy and histology from the resection specimen confirmed high-grade dysplasia. The patient is alive and well with no signs of recurrence 5 years later. Based on the endoscopist report alone, the sensitivity for detecting a malignant lesion was 81% and specificity was 91.7%. On the basis of histology alone, the sensitivity for detecting a malignant lesion was 97%, with a specificity of 99.4%. The average number of biopsies taken was four. Using the endoscopist report and histology for the index endoscopy, the sensitivity was 99% and the specificity was 91.1%. Repeat endoscopy and biopsy improved the sensitivity from 99 to 100%. During this study period 100 gastric malignancies were diagnosed. Ninety-nine of these were either diagnosed or suspected on the basis of endoscopic appearance and/or histology at the index endoscopy. Only one malignancy was found at follow-up endoscopy that had not been suspected either on the basis of endoscopic appearance or histology at the first endoscopy. This study showed that for a single endoscopy using the combined approach of endoscopic appearance and histology, the sensitivity for detection of gastric malignancy was 99% and the specificity was 91.1%. This is similar to other published studies. Six studies have looked at the ability to detect gastric cancer on first endoscopy through a combination of endoscopic appearance and histology.3–8 Out of a total of 592 gastric malignancies, 567 were detected on first endoscopy, giving an overall sensitivity of 95.8%. Nine studies have looked at the diagnosis of gastric cancer after a first endoscopy with a macroscopically benign looking gastric ulcer (GU) and no evidence of malignancy on histology.3–5, 7, 9–13The main follow-up method in these studies was endoscopy, often combined with clinical and surgical follow-up. Out of a total 2341 patients with gastric ulcers who were considered benign after index endoscopy (normal macroscopic appearance and no histological evidence of malignancy), only 20 patients during follow-up were found to have a gastric malignancy. Therefore 117 repeat endoscopies had to be performed to detect one case of undiagnosed gastric cancer. Of the 348 patients who have been diagnosed as having benign gastric ulcers after index endoscopy and follow-up for a minimum of 1 year using the Leicestershire Pathology database, only 211 had a follow-up endoscopy at Glenfield hospital. The reasons for this are not clear. Potential reasons include nonsteroidal induced gastric ulcers, frailty of individual patients, comorbidity and patient preference. They may have had repeat endoscopy at another hospital in Leicestershire. However if they had developed gastric cancer within the time frames of this study, this would have been detected using the Leicestershire Pathology database. There is the possibility that patients may have chosen to go to another hospital outside the area covered by the Leicestershire pathology service, leading to us not recognizing missed cancers at index endoscopy. This is unlikely given the catchment area of Glenfield hospital. The optimum approach to detect gastric malignancy at the index endoscopy, is the combined approach of endoscopic appearance along with histology. The correct diagnosis at first endoscopy depends on several factors: (i) the endoscopist suspecting that an ulcer is malignant; and (ii) the endoscopist's skill in taking sufficient biopsies from the appropriate sites around the lesion even when malignancy is not suspected. It should be noted that the size of a gastric ulcer cannot be relied upon to exclude cancer. Large gastric ulcers are more likely to be malignant, however gastric ulcers that are <1 cm in size can still be malignant.14 Biopsies should also be taken from ulcer scars as some malignant gastric ulcers can undergo partial healing.6 The number of biopsies is also important. Sancho-Poch et al. calculated on the basis of 174 cases of gastric cancer, a minimum of eight biopsy specimens should be taken to guarantee at least one fragment, would be positive for malignancy.15 There is a concern that if routine endoscopy is not performed for gastric ulcers then early gastric carcinoma may be missed. Eckardt et al. looked at the survival curves of those who had complied with endoscopic follow-up for gastric ulcers and those who did not.11 At the end of 5 years the survival curves were similar with a 5-year survival of 84% in those who had complied with follow-up and 81% in those who did not comply with follow-up. The causes of death were similar in each group. Podolsky et al. in their study described six patients whose diagnosis of gastric carcinoma was not made at first endoscopy.6 Two of the six patients had early gastric cancer, i.e. disease limited to the mucosa and sub mucosa and no lymph node involvement. They point out that patients whose gastric cancer presents as a benign ulcer looking lesion have a better long-term outlook than those with a malignant appearing lesion, however they base this on very small numbers. In this study, the case of malignant gastric cancer that was missed at the index endoscopy was high-grade dysplasia that was successfully resected. The numbers of patients diagnosed with previously unsuspected gastric malignancy on second upper GI endoscopy are small. From our study over 200 repeat upper GI endoscopies would have to be performed to detect one case of gastric cancer, which may not be curable. Reviewing the literature, approximately 120 repeat upper GI endoscopies are required to detect one case of unsuspected gastric malignancy. We suggest that repeat endoscopy for gastric ulcers should only be performed if the endoscopist suspects a malignancy or there is dysplasia on the first set of biopsies and the patient is fit for surgery.