High Grade Dysplasia of the Gastric Mucosa

Gut, 1990,31,977-983 977

High grade dysplasia ofthe gastric mucosa: a marker

for gastric carcinoma Gut: first published as 10.1136/gut.31.9.977 on 1 September 1990. Downloaded from

M Lansdown, P Quirke, M F Dixon, A T R Axon, D Johnston

Abstract encountered in interpreting the changes in The natural history of gastric epithelial gastric biopsy specimens are similar to those dysplasia and its relation to gastric cancer are experienced with ulcerative colitis, and therefore ill defined. A consecutive series of 40 patients a system with only two grades of dysplasia has with an initial diagnosis of gastric epithelial much to commend it.' A potential advantage of dysplasia based on examination of endoscopic such a simple two grade classification is that the biopsies has been reviewed to determine the natural history and prognostic importance of clinical outcome and to evaluate a two tier each category can be identified more readily, histological grading system as a predictor of which in turn allows more precise guidelines for the risk of cancer. On review, only 20 ofthe 40 clinical management to be formulated. patients were considered to have true dys- In this study we sought to determine the fate of plasia: seven patients had low grade dysplasia all patients who were diagnosed as having gastric and 13 had high grade dysplasia. Of the 13 epithelial dysplasia over a five year period and to patients with high grade dysplasia, 11 (85%) evaluate a two tier histological grading system as were found to have gastric cancer within 15 a predictor of the risk of cancer. months. Of the 10 patients with high grade dysplasia who underwent gastrectomy, six were found to have early gastric cancer, three Methods had cancer invading into the muscularis During the five years 1983 to 1987 inclusive, propria, and none had lymph node metastases. gastric epithelial dysplasia was diagnosed in 40 High grade dysplasia is thus a marker ofgastric patients at the Leeds General Infirmary, and cancer. Moreover, the cancers associated with categorised by the reporting pathologist into high grade dysplasia are usually pathologically mild, moderate, or severe grades9 or, more favourable and curable. The finding, by an recently, into low grade or high grade dys- experienced pathologist, of high grade dysplasia."' All the original biopsy specimens and

plasia in two separate sets of endoscopic any subsequent pathological material were http://gut.bmj.com/ biopsies is therefore an indication for radical reviewed and reclassified as (1) no dysplasia, (2) surgical treatment, provided that the patient's low grade dysplasia, (3) high grade dysplasia, age and general condition permit such an and (4) carcinoma. Gastric cancer was classified approach.

Little is known about the natural history of on October 3, 2021 by guest. Protected copyright. epithelial dysplasia in the stomach. Areas of epithelial dysplasia can often be found adjacent to established gastric cancer.' It is tempting to propose that gastric cancer arises from dysplastic epithelium just as colonic cancer arises from adenomas.4 Follow up studies of patients with dysplasia of the gastric epithelium tend to support this hypothesis,5-8 though the rate of progression to cancer appears to be slow. A firm diagnosis ofdysplasia made on the basis ofendoscopic biopsies raises two main problems: (a) Has gastric cancer been missed because of University Departments sampling error? (b) If dysplasia is present, is the ofSurgery M Lansdown risk of progression to cancer so great that the D Johnston patient should be advised to undergo gastrectomy? and Pathology P Quirke The degree or severity of dysplasia is M F Dixon obviously important. Gastric epithelial dysplasia is graded by most pathologists into mild, and Department of Medical moderate, and severe categories. Cancer is diag- Gastroenterology, nosed only if neoplastic cells breach the base- General Infirmary, Leeds ment membrane of the epithelium and invade LS1 3EX the lamina propria. The use of terms such as AT R Axon 'intraepithelial cancer' and 'carcinoma in situ' Correspondence to: for lesions with the Mr M Lansdown. cytological appearances of Figure 1: Biopsy specimen categorised as low grade dysplasia, Accepted for publication cancer but without evidence of invasion of the showing minor stratification ofelongated, hyperchromatic 6 November 1989 lamina propria is discouraged.9 The difficulties nuclei, extending up to the surface ofthe mucosa (x 640). 978 Lansdown, Quirke, Dixon, Axon,Johnston Gut: first published as 10.1136/gut.31.9.977 on 1 September 1990. Downloaded from

IM ,W Figure 2: Biopsy specimen showing high grade dysplasia, in Figure 3: Biopsy specimenfrom a patient not included in this which the nuclei are much more pleomorphic than in Figure 1. series where the bizarre appearances were interpreted as Pseudostratification occupies thefull thickness ofthe intramucosal carcinoma, but in whom no cancer was detected epithelium and there are misplaced mucin vacuoles (x640). in a subsequent gastrectomy specimen. Biopsy specimens showing eitherfrank intramucosal carcinoma or bizarre 'suspicious' appearances have not been classified as gastric according to the revised TNM classification of dysplasia (x 640). malignant tumours. " gastric cancer and in high grade dysplasia are often identical. Carcinoma was diagnosed only if DIAGNOSIS OF DYSPLASIA neoplastic cells were seen to be invading the http://gut.bmj.com/ The cytological and architectural features of lamina propria. Multiple sections through dysplasia of the gastric mucosa are well des- biopsy material were examined to reduce the cribed.'9I2-4 Diagnosis of low grade (Fig 1) or likelihood of invasion being missed. Cases in high grade dysplasia (Fig 2) depended on the which the original endoscopic biopsy material higher nucleus:cytoplasm ratio, higher mitotic showed bizarre cytological and architectural activity, greater loss ofpolarity, pseudostratifica- appearances suspicious of carcinoma (Fig 3), tion, more pronounced nuclear pleomorphism, such as are sometimes seen in biopsy specimens on October 3, 2021 by guest. Protected copyright. and more prominent nucleoli in high grade from the margins of a tumour, were excluded Figure 4: Campylobacter- dysplasia. from the study. Biopsy material with these associated chronic gastritis showing surface erosion, appearances should not be classified as dysplasia. originally diagnosed as mild dysplasia. In this case the DIAGNOSIS OF CARCINOMA foveolae show simple regenerative hyperplasia The cytological appearances ofneoplastic cells in Results (x640). After histological review ofthe original specimen seven patients were classified as having low grade and 13 as having high grade dysplasia. In 19 of the 40 cases reviewed the appearances were not considered to represent true dysplasia, and in one case the diagnosis was changed to intramucosal carcinoma. The revised diagnoses in these patients are presented in Table I. These 20 cases were originally classified as mild, moderate, or low grade dysplasia, usually by 'non-specialist' pathologists (Figs 4 and 5).

PATIENTS WITH DYSPLASIA Clinical details of the 20 patients with confirmed gastric epithelial dysplasia were obtained from the case notes. Their median age was 72 years (range 55-86 years). Eight were women and 12 men. Five of the 20 patients had been operated on for benign peptic ulceration, by means of High grade dysplasia ofthe gastric mucosa: a markerforgastrnc carcinoma 979

TABLE I Pathological conditions ofthe stomach that were TABLE II Nature ofprevious operationsfor benign peptic initially mistaken for dysplasia ulceration in patientsfound to have dysplasia in endoscopic biopsies ofthe stomach Revised diagnosis No Years between Gut: first published as 10.1136/gut.31.9.977 on 1 September 1990. Downloaded from Gastric carcinoma I Degree of operation and Chronic atrophic gastritis (intestinal metaplasia with No Operative procedure dysplasia diagnosis ofdysplasia hyperproliferative 'crypts') 7 Chronic gastritis with erosions (regenerative atypia) 6 1 Polya partial gastrectomy High grade 15 Chronic superficial gastritis (foveolar hyperplasia) I 2 Polya partial gastrectomy High grade 31 Reflux gastritis (foveolar hyperplasia) 4 3 Gastroenterostomy High grade 50 Lymphocytic gastritis (erosion with regenerative atypia) 1 4 Polya partial gastrectomy Low grade 30 Total 20 5 Truncal vagotomy and gastroenterostomy Low grade 21

Polya gastrectomy or gastroenterostomy, 15-50 years previously (Table II). unfortunately this patient did not return for follow up. The other patient had an ulcer which appeared to be benign and which healed in CLINICAL PRESENTATION response to treatment with H2 receptor antago- There is a policy of open access to endoscopy at nists. Follow up biopsies six months later the General Infirmary, but there is no screening showed continuing low grade dysplasia in the programme for asymptomatic patients. Thus all ulcer scar. the patients in this study had undergone endoscopy because they had symptoms suggestive of upper gastrointestinal disease such as pain, loss Tumour ofweight, or anaemia. The indications for endo- Two patients with high grade dysplasia had scopy are given in Table III. raised tumours. In one patient a tumour on the lesser curve was associated with liver metastases on the ultrasound scan. In the other patient the ABNORMALITIES OF THE GASTRIC MUCOSA SEEN AT tumour was seen to occupy most of the antrum: ENDOSCOPY (TABLE IV) no further specimens were taken and the patient Gastric ulcers and tumours or polyps were by far underwent a total gastrectomy with radical the commonest abnormalities seen at endoscopy. lymphadenectomy. The resected specimen contained a tumour 8 cm in diameter invading the muscularis propria, but the lymph nodes were Gastric ulceration not involved. Seven of the 13 patients with high grade dysplasia were found to have gastric ulceration at the

first endoscopy: one patient had two superficial Polyp http://gut.bmj.com/ ulcers on the lesser curve and six patients each Three patients had high grade dysplasia in had a single ulcer situated on the lesser curve. solitary polyps. One was an adenoma with a The largest ulcer was 2 cm in diameter. Only one focus of high grade dysplasia, situated adjacent Figure 5: Quiescent chronic of these ulcers was described as malignant by the to a gastroenterostomy: carcinoma was found in atrophic gastritis with endoscopist, though in another case the presence a second set of biopsy specimens taken a week intestinal metaplasia of irregular mucosa near the ulcer raised suspic- later. The resected specimen contained an initially diagnosed as mild dysplasia. Here the ion ofearly gastric cancer. Healing ofthe ulcer in adenoma 3 cm in diameter in which intramucosal on October 3, 2021 by guest. Protected copyright. hyperplastic crypts show response to treatment with H2 receptor antago- carcinoma was found. One patient with a small mild atypia and increased nists was documented in four of these patients at sessile polyp high on the posterior wall of the mitotic activity, but there is maturation at the surface of the time of the second endoscopy. stomach was found to have carcinoma at the the crypts. The appearances Two of the seven patients with low grade same site in a second set of specimens taken five are attributed to the high cell dysplasia had an ulcer on the lesser curvature. In months later. The third patient had high grade turnover that occurs in one was as an metaplastic mucosa and not patient the ulcer described erosion: dysplasia in a sessile antral polyp which at to a premalignant change endoscopy was thought to measure 1 cm in (x64). diameter: further sets of biopsy specimens taken two and three months later showed continuing high grade dysplasia. We @_ One patient had multiple tiny regenerative polyps situated on the posterior wall ofthe upper B .A* i part of the stomach. Endoscopic biopsy specimens from this region showed changes of intestinal metaplasia and regeneration and also low grade dysplasia. Further specimens taken 14 months later also showed low grade dysplasia,

TABLE III Indicationsfor endoscopy in 20 patients with gastric epithelial dysplasia Indication for endoscopy No ofpatients Upper abdominal pain 15* Anaemia 4 Melaena 1

*Four patients who presented with upper abdominal pain also complained of loss of weight. 980 Lansdown, Quirke, Dixon, Axon,Johnston

TABLE IV Findings at endoscopy in 20 patients with gastric high grade dysplasia without evidence of cancer. epithelial dysplasia Six of these seven patients underwent radical Abnormality seen Low grade High grade gastrectomy, and five ofthem were found to have at endoscopy dysplasia dysplasia cancer: four had early gastric cancer (TI NO MO) Gut: first published as 10.1136/gut.31.9.977 on 1 September 1990. Downloaded from and one had a T2 NO MO tumour. In the sixth Ulcer 2 7 Raised tumour or polyp 1 5 patient, examination of the gastrectomy speci- Plaque 0 1 men showed high grade dysplasia but there was Other (see text) 4 0 no evidence of carcinoma (the entire dysplastic area was processed for microscopic examination). The remaining patient with continuing but a third set taken after a further seven months high grade dysplasia was referred for laser showed only chronic atrophic gastritis and photocoagulation of a small antral polyp. After intestinal metaplasia. destruction of the polyp there was regression to low grade dysplasia and then no dysplasia was found in three subsequent sets of specimens Plaque taken over a period ofone year. Eighteen months One patient with high grade dysplasia had a after photocoagulation, however, a raised lesion plaque on the lesser curve. An ulcerated early was again seen at endoscopy: biopsy specimens gastric cancer was suspected by the endoscopist. showed no evidence of dysplasia, but because of Two further sets of biopsy specimens confirmed the high degree of suspicion a further set was the presence of high grade dysplasia before the taken and the presence of high grade dysplasia patient underwent gastrectomy. The entire area was confirmed. This patient is a poor operative of dysplasia was examined histologically; the risk and is returning for further biopsies before a 'ulcer' was found to be a depressed area of decision about further treatment is made. epithelium. Since no evidence of invasion of the The TNM staging of the 13 patients with high lamina propria was found the lesion was not grade dysplasia in the original set of biopsy classified as a carcinoma. specimens is summarised in Table V.

Other abnormalities Patients with low grade dysplasia The endoscopic abnormality seen in the Of the seven patients with low grade dysplasia, remaining fourpatients, all of whom had low four underwent further grade dysplasia, was mucosa' in two, endoscopy and biopsy 'atrophic and three did irregular mucosa at a gastroenterostomy in one, not. In the four patients further biopsies after 2-24 months showed no and deformity of the pyloric canal in one. evidence of dysplasia in three, while there was continuing

low grade dysplasia in one patient after six http://gut.bmj.com/ months. Of the three patients who were not CLINICAL COURSE followed up, one failed to attend for further Patients with high grade dysplasia investigation, one was discharged, and the third died ofan unrelated condition. One patient was thought to have gastric cancer at endoscopy and liver metastases were found by ultrasound scanning. One set of biopsy speci- on October 3, 2021 by guest. Protected copyright. mens was taken which showed only high grade Patients reclassified as not having dysplasia dysplasia. Since he was unfit for surgical treat- In 19 patients the diagnosis was revised to ment, no further biopsies were carried out. gastritis with regenerative atypia (Table I). None One patient with a large antral tumour, of these patients developed gastric carcinoma biopsy specimens of which showed high grade during a median follow up period of 18 months. dysplasia, underwent a total gastrectomy with radical lymphadenectomy, without undergoing further endoscopy. The resected stomach con- NUMBER OF BIOPSY SPECIMENS tained an adenocarcinoma 8 cm in diameter Because ofthe retrospective nature ofthis review which was invading the muscularis propria but it was not possible to determine how many not involving lymph nodes. biopsy specimens were taken at each endoscopy The remaining 11 patients with high grade in each patient, and when the number was dysplasia all had at least one further set ofbiopsy indicated it was not always clear whether they specimens taken at a repeat endoscopy. Four had all been taken from the abnormal area of patients were found to have cancer in follow up gastric mucosa. The number ofsets of specimens biopsies within eight months; three underwent taken in each patient was recorded. In patients radical gastrectomy, of whom two had early with high grade dysplasia in the first set of gastric cancer (Ti NO MO) and one had cancer specimens, in whom gastric cancer was diag- invading the muscularis propria (T2 NO MO). nosed on the basis of follow up biopsies, the The fourth patient, who had cancer in a polyp, cancer was always found in either the second or was unfit for surgical treatment and was referred the third set of specimens. Only one patient for laser photocoagulation: repeat biopsies at underwent gastrectomy on the basis of a single first showed no further evidence of cancer, but set of specimens that showed high grade dys- one year after laser treatment biopsy specimens plasia: this was because he had a large, malignant showed the presence of recurrent carcinoma. looking antral tumour at endoscopy. The other In seven patients follow up endoscopy and six patients who underwent gastrectomy on biopsy over 1-15 months showed continuing accountofhighgradedysplasiahad two tofive sets Highgrade dysplasia ofthegastric mucosa: a markerforgastric carcinoma 981

TABLE V Endoscopic appearance, treatment, TNM stage, and interval between diagnosis of resected for advanced cancer, areas of epithelial high grade dysplasia and gastric cancer in 13 patients who were initiallyfound to have high dysplasia can also be found.2" This strong grade dysplasia present in endoscopic biopsies association between gastric cancer and synchro-

Interval betwzeen diagnosis nous dysplasia should thus arouse strong Gut: first published as 10.1136/gut.31.9.977 on 1 September 1990. Downloaded from ofhigh grade dysplasia and has been missed Patient Endoscopic diagnosis ofgastric cancer suspicion that gastric cancer appearance Treatment TNM stage (months) when biopsy specimens show the presence only 1 Plaque Subtotal gastrectomy TO NO MO of high grade dysplasia. 2 Polyp Total gastrectomy TI NO MO <1 A problem of management arises when 3 Ulcer Subtotal gastrectomy TI NO MO 2 the 4 Ulcer Subtotal gastrectomy TI NO MO 3 repeated endoscopies and biopsies confirm 5 Ulcer Total gastrectomy TI NO MO 5 presence ofhigh grade dysplasia, but not offrank 6 Ulcer Total gastrectomy TI NO MO 6 7 Ulcer Total gastrectomy Ti NO MO 13 carcinoma. Some of the patients in the present 8 Raised tumour Subtotal gastrectomy T2 NO MO 2 study may have had a small area of invasive 9 Ulcer Subtotal gastrectomy T2 NO MO 8 10 Ulcer Subtotal gastrectomy T2 NO MO 15 cancer, within an area of high grade dysplasia, 11 Polyp Laser photocoagulation Tx Nx Mx which was missed by repeated biopsies, as borne 12 Polyp Laser photocoagulation Tx Nx Mx 5 out by the finding of invasive cancer in six of the 13 Raised tumour No treatment Tx Nx MI - seven patients who underwent gastrectomy but Note: Cancer was only diagnosed in patients 3, 4, 5, 7, 8, and 9 after examination of the resected who did not have cancer in the preoperative stomach. Patient 13 was terminally ill with disseminated cancer at the time ofdiagnosis. Patients 11 and 12 did not undergo gastrectomy: patient II had at least high grade dysplasia, and patient 12 had at endoscopic biopsy specimens. Admittedly, two least a TI cancer. of these patients had high grade dysplasia diagnosed more than a year before they underwent ofspecimens taken up to 15 months after the first gastrectomy, and they may have developed set, before surgical treatment was advised. invasive cancer during that period of follow up. Cancer detected soon after dysplasia is diagnosed probably represents synchronous disease, but TYPE OF GASTRECTOMY cancer found after many months of follow up, In each of the 10 patients who underwent with repeated biopsies, may represent true pro- gastrectomy, radical (R2) lymphadenectomy was gression. Gastric cancer has been found in performed. The gastrectomy was total in four patients with dysplasia who were followed up for patients and subtotal in six (Table V). periods of up to six years." The finding of occult cancer in patients with SURVIVAL gastric epithelial dysplasia who underwent There were no postoperative deaths. Three of gastrectomy has been described previously. the original 40 patients are known to have died. Thus, Aste et al'6 found gastric cancer in eight of The patient who had a large gastric neoplasm 13 patients with moderate or severe dysplasia with evidence of liver metastases on the ultra- who underwent gastrectomy. As in the present study, dysplasia was usually associated with a sound scan, but whose gastric biopsy specimens http://gut.bmj.com/ showed only high grade dysplasia, died in visible lesion, most commonly an ulcer, and hospital ofhis tumour. None of the patients who many of the cancers detected were 'early' and underwent gastrectomy for high grade dysplasia potentially curable. Similarly, Saraga et al6 found or gastric carcinoma has died of gastric car- cancer in 17 of 21 patients with severe dysplasia cinoma: one has died of breast carcinoma, how- within three years of the original diagnosis of ever, and another ofheart disease. severe dysplasia. In 12 of these 17 patients the carcinoma was only found on examination of the gastrectomy specimen, and in eight patients it on October 3, 2021 by guest. Protected copyright. Discussion was early gastric cancer. Previous reports of We found that most patients who were diagnosed gastric cancer developing in areas of dysplasia as having high grade dysplasia of the gastric may have underestimated the true incidence of mucosa on examination of gastric biopsy speci- gastric cancer if the whole area of dysplasia was mens either had gastric cancer or soon developed not subjected to histological examination. gastric cancer. Furthermore, the cancers that Attempts to try to distinguish three different were present in association with high grade grades ofseverity ofdysplasia may encourage the dysplasia were nearly all 'early' and thus potenti- use of the term 'mild dysplasia' when the patho- ally curable. Ten ofthe 13 patients who had high logist is presented with any biopsy specimen that grade dysplasia in their original endoscopic shows minor cytological atypia. Many of these biopsy specimens subsequently underwent cases will represent regenerative atypia rather radical gastrectomy: nine were found to have than a premalignant change and the patient will gastric carcinoma and one had high grade dys- be subjected to unnecessary surveillance and plasia. Six of the cancers resected were early anxiety. Of those patients with 'true' dysplasia gastric cancer and the other three cancers were the pathologist needs to identify a low risk group invading only into the muscularis propria but of patients who can safely be followed up by had not invaded the serosa. None of the 10 cases means of endoscopy and biopsy, and a high risk had lymph node metastases. Thus the prognosis group of patients in whom the risk of cancer is ofthese patients is much better than that of most sufficiently high to justify the use of gastrec- patients with gastric cancer. tomy. A group of patients with dysplasia of It seems likely that in some patients the 'intermediate' severity is likely to contain presence of carcinoma may have been missed in patients of low and high risk, and inclusion of the original set of biopsy specimens, which such a category therefore impedes clinical showed only an adjacent area of high grade decision making. A further difficulty is that dysplasia. In 20-40% of stomachs resected for biopsy specimens taken at endoscopy are small early gastric cancer, and up to 80% of stomachs and their interpretation may be difficult.'7 In the 982 Lansdown, Quirke, Dixon, Axon,Johnston

present study, when biopsy material was Probably only 5% of all gastric cancers arise in reviewed it was found that 16 of 23 cases polyps.22 In the present series cancer was found originally reported as mild or low grade dysplasia in follow up biopsy material from an adeno-

had been misdiagnosed. Fifteen were reclassified matous polyp in one patient. Gut: first published as 10.1136/gut.31.9.977 on 1 September 1990. Downloaded from as showing gastritis, with regenerative atypia, We have performed radical gastrectomies in but one was reclassified as frank carcinoma. In patients who had high grade dysplasia but no contrast, no set of specimens originally thought histological evidence of invasive carcinoma in to show severe or high grade dysplasia was preoperative biopsy material. It is likely that our reclassified as gastritis alone. Of the eight sets of cases of high grade dysplasia correspond to specimens originally classified as moderate dys- Nagayo's category of 'probable carcinoma,' for plasia, five were reclassified as gastritis, two as which he advises that gastrectomy be performed. high grade dysplasia, and one as low grade In Nagayo's vast experience, 'probable dysplasia. No patient whose biopsy specimens carcinomas' and 'borderline lesions' comprise were reclassified as gastritis developed gastric less than 3% of all biopsy specimens examined. carcinoma during a median period offollow up of The use of gastrectomy in these patients, how- 18 months. ever, presumably increased the number of early In this study dysplasia was most commonly gastric cancers resected and may have contrib- associated with an ulcer. This is in keeping with uted to the impressive statistics for survival after the findings ofprevious studies.66 The finding of surgery for gastric cancer in Japan. dysplasia in the margin of a peptic ulcer would In high risk patients less radical measures than suggest that the ulcer is either malignant or about aggressive surgical treatment may have to be to become malignant, but the risk of cancer considered. Thus two patients in our series, who developing in a benign ulcer is small' and many were considered unfit for gastrectomy, under- of the putative dysplastic changes at the edges of went laser photocoagulation, one for carcinoma ulcers are probably regenerative and not neo- and the other for high grade dysplasia. Although plastic. Problems in the diagnosis of dysplasia the lesions recurred in both patients after 12 to 18 and cancer in biopsy material taken from the months' follow up, we believe that the use oflaser edge of an ulcer have been highlighted by photocoagulation is a better option than simple Williams.'7 The problem is particularly acute in observation and repeated biopsy. Laser photo- the presence of erosions where regenerative coagulation has also been used by Takemoto23 to activity in residual glands ofthe deep mucosa can treat patients with early gastric cancer who were be readily mistaken for dysplasia or even considered unfit for gastrectomy. Endoluminal carcinoma.'7 'Overdiagnosis' of dysplasia in ultrasound was used to determine the depth of some series may explain why so few patients with invasion of the cancer before treatment, and to dysplasia were subsequently found to have assess the depth of destruction caused by the cancer. laser. Among 14 patients in whom the laser was It may also be the case that neoplastic and thought to have coagulated the carcinoma com- http://gut.bmj.com/ dysplastic epithelium is prone to ulceration, and pletely, no evidence of recurrent cancer was some ulcers initially thought to be benign are in found in 12 patients six to 22 months later, and fact neoplastic. Farinati et al7 discovered 10 two patients had recurrent cancer. Thus the use cancers among 144 patients with apparently of endoluminal ultrasound for the assessment benign ulcers who were followed up for 41 and follow up of early gastric cancer treated by months with repeated endoscopy and biopsy. laser photocoagulation merits further study in Most cancers were found within six months, so it selected elderly or unfit patients. on October 3, 2021 by guest. Protected copyright. is likely that the original biopsy specimens were Our series shows that a high proportion (85%) either not representative of the ulcer or were of patients with high grade dysplasia in endo- misinterpreted. scopic biopsy specimens have (or soon will have) The difficulty in distinguishing benign from gastric cancer. A high proportion (90%) of these malignant ulceration in the stomach by endo- cancers are at an early stage of invasion (TI or scopic examination is well known. This is par- T2). Our finding that cancer was invading the ticularly so for early gastric cancer, as shown by muscularis propria in three of 10 patients who Ballantyne et all9 in whose study of 14 ulcerating underwent gastric resection highlights the risk of early gastric cancers, seven were thought to be inaction - that is, ofmerely following up patients benign by naked eye inspection at gastroscopy. with high grade dysplasia by repeated endoscopy In Japan, however, where gastric cancer is found and biopsy. in its early stages more often than in the West, The present study has highlighted the diffi- malignant ulcers can be distinguished from culty that can be experienced in demonstrating benign ulcers with an accuracy rate of80% by the gastric cancer in patients presenting with high endoscopist.' It was interesting that in the grade dysplasia even when carcinoma has present study four ulcers associated with high invaded the muscularis propria. Although none grade dysplasia healed in response to H2 receptor of the patients in this series had metastases to antagonists. In each case the patient was subse- lymph nodes, the risk of lymph node metastases quently found to have gastric cancer. is directly related to the depth of invasion of Polyps of the adenomatous type are rare in the gastric cancer, being 2-4% for cancer invading stomach, comprising 10% or less of all gastric the mucosa, 20% for cancer invading the sub- polyps. Estimates ofthe proportion that progress mucosa, and over 40% for cancer invading the to gastric cancer vary from 6% to 75%.2° muscularis propria.' 24 In our own series of 27 Nakamura and Nakano2' found foci ofcarcinoma patients who had early gastric cancers resected in in 33% ofthose gastric polyps that were similar in the last 15 years, three (11%) had metastases to histological appearance to colonic adenomas. lymph nodes. We therefore consider radical High grade dysplasia ofthegastric mucosa: a markerforgastric carcinoma 983

gastrectomy with removal of the first and second 2 Meister H, Holubarsch Ch, Haferkamp 0, Schlag P, Herfarth Ch. Significance and location of atrophic gastritis and of tiers of lymph nodes to be the appropriate glandular dysplasia in benign and malignant gastric disease. operative procedure for most patients with high In: Herfarth Ch, Schlag P, eds. Gastric cancer. Berlin:

Springer-Verlag, 1979: 105-7. Gut: first published as 10.1136/gut.31.9.977 on 1 September 1990. Downloaded from grade dysplasia, but the age and general condi- 3 Grundmann E, Schlake W. Histology ofpossible precancerous tion of the patient must obviously be taken into stages in the stomach. In: Herfarth Ch, Schlag P, eds. Gastric cancer. Berlin: Springer-Verlag, 1979: 72-82. account and in some instances a less radical 4 Enterline HT, Evans GW, Mercado-Lugo R, Miller L, Fitts operation will be appropriate. Whether the WT. Malignant potential of adenomas of colon and rectum. JAMA 1962; 179: 322-30. gastrectomy should be total or subtotal depends 5 Offerhaus GJA, Stadt J, Huibregtse K, Tytgat GNJ. Endo- on the site and extent ofdysplasia in the stomach. scopic screening for malignancy in the gastric remnant: the clinical significance of dysplasia in gastric mucosa. J Clin In our series two patients had more than one Pathol 1984; 37: 748-54. focus of cancer, but in both cases the cancerous 6 Saraga E, Gardiol D, Costa J. Gastric dysplasia. A histological follow-up study. Am3rSurgPathol 1987; 11: 788-96. foci were adjacent to each other. Thus we found 7 Farinati F, Cardin F, Di Mario F, et al. Early and advanced no evidence of widespread high grade dysplasia gastric cancer during follow-up of apparently benign gastric ulcer: significance of the presence of epithelial dysplasia. in the stomach, and our data lend no support to J Surg Oncol 1987; 36: 263-7. the idea that all patients with high grade dys- 8 Cai-pu X, Wei-wen L. Follow-up observations on chronic gastritis, intestinal metaplasia and dysplasia. Chin Med J plasia should undergo total gastrectomy. If sub- (Engl) 1985; 98: 347-8. total gastrectomy is to be carried out, however, it 9 Morson BC, Sobin LH, Grundmann E, Johansen A, Nagayo T, Serck-Hanssen A. Precancerous conditions and epithelial is important that biopsy specimens be taken dysplasia in the stomach. J Clin Pathol 1980; 33: 711-21. from all regions of the stomach before operation 10 Riddell RH, Goldman H, Ransohoff DF, et al. Dysplasia in inflammatory bowel disease: standardized classification with so that areas ofabnormal mucosa may be mapped provisional clinical applications. Hum Pathol 1983; 14: 931- out accurately. 68. 11 Hermanek P, Sobin LH. TNM classification of malignant The conclusion from this study, that a histo- tumours. Berlin: Springer-Verlag, 1987. logical diagnosis of high grade is 12 Cuello C, Correa P, Zamara G, Lopez J, Murray J, Gordillo G. dysplasia Histopathology of gastric dysplasias: correlations with usually an indication for gastrectomy because of gastric juice chemistry. AmJ Surg Pathol 1979; 3: 491-500. the almost invariable association of high grade 13 Jass JR. A classification of gastric dysplasia. Histopathology 1983; 7:181-93. dysplasia with invasive carcinoma, places a 14 Ming S, Baitai A, Correa P, et al. Gastric dysplasia. Signific- heavy responsibility on the reporting patho- ance and pathologic criteria. Cancer 1984; 54: 1794-801. 15 Nagayo T. Gastric cancer preceded by severe dysplasia. Histol logist. The recognition of dysplasia, and its Histopathol 1986; 1: 171-80. distinction from atypical hyperplasia consequent 16 Aste H, Sciallero S, Pugliese V, Gennaro M. The clinical significance of gastric epithelial dysplasia. Endoscopy 1986; upon regeneration or inflammatory activity, is 18: 174-6. far from straightforward. Indeed, such changes 17 Williams GT. Early gastric cancer. In: Filipe MI, Jass JR, eds. Gastric carcinoma. Edinburgh: Churchill Livingstone, 1986: may be so bizarre as to give rise to appearances 173-96. indistinguishable from malignancy. It is obliga- 18 Isaacson P. Biopsy appearances easily mistaken for malignancy in gastrointestinal endoscopy. Histopathology 1982; 6: 377- tory therefore for the diagnosis of high grade 89. dysplasia to be confirmed by a second set of 19 Ballantyne KC, Morris DL, Jones JA, Gregson RH, Hardcastle JD. Accuracy of identification of early gastric biopsy specimens, while the biopsy specimens cancer. BrJ3Surg 1987; 74: 618-9. http://gut.bmj.com/ themselves must be reported on by very experi- 20 Sipponen P, Kekki M, Siurala M. In: Filipe MI, Jass JR, eds. Gastric carcinoma. Edinburgh: Churchill Livingstone, 1986: enced pathologists. Where expert advice is 153-71. not available locally, submission of biopsy 21 Nakamura T, Nakano G. Histopathological classification and malignant change in gastric polyps. J Clin Pathol 1985; 38: specimens showing dysplasia to a specialised 754-64. pathological panel for review should improve 22 Hermanek P. Gastric polyps and gastric cancer. In: Herfarth Ch, Schlag P, eds. Gastric cancer. Berlin: Springer-Verlag, diagnostic accuracy and reduce the risk of 1979: 147-8.

unnecessary gastrectomy. 23 Takemoto T. Laser therapy of early gastric carcinoma. on October 3, 2021 by guest. Protected copyright. Endoscopy 1986; 18 (suppl 1): 32-6. 24 Okamura T, Tsujitani S, Korenaga D, et al. Lymphadenec- 1 Nagayo T. Histogenesis and precursors of human gastric cancer. tomy for cure in patients with early gastric cancer and lymph Research and practice. Berlin: Springer-Verlag, 1988. node metastasis. AmJ Surg 1988; 155: 476-80.