MORTALITY FROM GASTRIC CANCER FOLLOWING GASTRIC SURGERY FOR PEPTIC ULCER

Saturday 26 April

MORTALITY FROM GASTRIC CANCER FOLLOWING GASTRIC SURGERY FOR PEPTIC ULCER CHRISTINE P. J. CAYGILL JOHN S. KIRKHAM

MICHAEL J. HILL TIMOTHY C. NORTHFIELD

Norman Tanner Gastroenterology Unit, St James’ Hospital, London; PHLS Communicable Disease Surveillance Centre, London; and Bacterial Metabolism Research Laboratories, PHLS-CAMR, Porton Down, Wiltshire

When compared with a matched population group, 4466 ulcer patients who had had gastric surgery between 1940 and 1960 showed no difference in the risk of death from gastric cancer in the first 20 years of follow-up but a 4·5-fold increase thereafter. In duodenal ulcer patients there was an initial decrease in risk followed by a 3·7-fold increase after 20 or more years. Since the initial decrease was seen only in the gastrectomy patients and not in those who had truncal vagotomy and drainage, it may have been due to the reduction in mucosal surface. The increased risk 20 years after duodenal ulcer surgery was greater in vagotomy patients than in gastrectomy patients. In gastric ulcer patients a 3·0-fold increase in risk for the first 20 years rose to a 5·5-fold increase thereafter. After 20 years, patients treated with the Bilroth II operation were at higher risk than those treated with Bilroth I, consistent with a role for bile reflux in gastric carcinogenesis. The finding that the risk differs according to original pathology and type of operation may explain the discrepancies between previous studies.

Summary

Introduction THERE is considerable controversy over the risk of gastric cancer in patients surgically treated for peptic ulcer. Some workers suggest that there is none,’-3 while others report an increased risk,4 with a latency of 15 to 20 years,5 that is possibly greater in gastric than in duodenal ulcer.6 Most of the studies have been small (fewer than 800 patients); many had a follow-up of less than 20 years; many did not distinguish between gastric and duodenal ulcer patients or between different types of surgery. These factors may have contributed to the discrepancies. We have reported7 preliminary analyses of the overall risk of cancer of the stomach and other sites in patients who had surgery for peptic ulcer between 1940 and 1960. Here we report a more detailed analysis of the risk of gastric cancer according to type of operation and site of the original ulcer.

1986

Methods We examined the records of patients treated surgically for peptic ulcer by Mr Norman Tanner and his colleagues at the gastric clinic, St James Hospital, Balham, between 1940 and 1960. Operative treatment was either by Bilroth I (BI) or Bilroth II (polya) partial gastrectomy (BII) or by truncal vagotomy with drainage (VAG). The 5018 patients were traced through the Office of Populations, Censuses and Surveys (OPCS). Death certificates were obtained for 2768 patients and coded by OPCS to the 8th revision of the International Classification of Diseases (ICD) codes to coincide with the period of the majority of deaths. 1746 patients are still alive and are flagged at OPCS for subsequent notification of death and its cause; 504 could not be traced. Information about the site of ulcer and type of operation was available for 4466 patients (tables I andu). The incidence of gastric cancers for a population age-matched and sex-matched to our study population was calculated from the mortality rates (OPCS, unpublished data) for South-West Thames Region for a period of time as near as possible to that of the deaths in our study, this region being the final residence area for most of our study population. "Years at risk" in 5-year bands from the age at operation were calculated. Probability levels were estimated from either the Poisson or the normal approximation to the binomial distribution.

TABLE I-POPULATION DISTRIBUTION ACCORDING TO TYPE OF ULCER

DU = duodenal ulcer; GU = gastric ulcer; other = recurrent or stomal ulcer after previous surgery; %= percentage of total study population.

TABLE II-POPULATION DISTRIBUTION ACCORDING TO

TYPE OF OPERATION

0/0 percentage of total

study population. 8487 ©

930 TABU V-GASTRIC CANCER MORTALITY IN GASTRIC

Results

ULCER PATIENTS

TablesI and II show the distribution of patients according to type of ulcer and operation. The ratio of males to females was 3’ 5:1. This ratio was higher in duodenal ulcer (DU) than in gastric ulcer (GU) patients (4’9:and 2 - 1: 1, respectively), and higher in BII than in BI patients (5’6:1 and 1-7:1,

respectively). The total study population was at no greater risk of gastric cancer for the first 20 postoperative years than was the control population (table III), but at more than 20 years after operation patients were about four times more at risk. When the data were analysed in 5-year groups there was no increase in risk at any time during the first 20 years. The relation between age at operation and risk of death from gastric cancer was not

examined.

Duodenal Ulcer

.

*p<0-01.

patients with DU the risk of death from gastric cancer was less in each 5-year band for the first 20 postoperative years (table III). When the 20-year postoperative period was considered as a whole, there was a significant decrease in 20 years or more after operation, mortality (p<0’01). increased risk. however, there was a 3’7-fbld (p<0’001) In

Thus the decreased risk in the first 20 years is reversed. In BII patients there was a decreased risk within the first 20 postoperative years, followed by a 3-fold increase thereafter (table IV). In patients who had VAG there was no significant alteration in risk within the first 20 years but- a 7 -8-fold increase thereafter. The excess risk 20 or more years postoperation in all types of surgery was greater for women than for men. TABLE III-GASTRIC CANCER MORTALITY ACCORDING TO TYPE OF ULCER

tp<0-001. Gastric Ulcer In patients with GU there was a 3-fold increased risk of death from gastric cancer for the first 20 years after operation, and this increased to 5 -5-fold after 20 years (table III). In patients who had a BII operation there was a 3’0-fbld increased risk in the first 20 years, followed by an 8 -6-fold increase thereafter; in those who had a BI (the commonest operation in this group) there was a 2 -6-fold increased risk in the first 20 years followed by a 4 -0-fold increase thereafter (table V). Numbers of patients treated by VAG were too small for analysis. There was no difference in gastric cancer risk between males and females.

Discussion We have confirmed that there is an increased risk of gastric in patients treated surgically for peptic ulcer, with an apparent latency of 20 years. More detailed analysis, however, shows that the risk of mortality differs between ulcer types: in the first 20 postoperative years, it decreases in DU patients but increases in GU patients. When both sets of data are combined the two results cancel each other out because of the relative proportions of the two ulcer types. 20 or more years after operation, the risk of death from gastric cancer is increased whatever the site of the original ulcer. A study of GU patients alone would therefore show an excess risk of gastric cancer, irrespective of the follow-up, whereas one of DU patients alone would not indicate an excess risk (and might show a decreased risk) unless the follow-up was more than 20 years. Thus some discrepancies between studies may be explained. The lower risk in DU patients who had had a gastrectomy rather than a VAG may be due to the removal of a susceptible portion of the stomach. By contrast, in those who had had a VAG without gastric resection, the risk of death from gastric cancer was not changed in the first 20 postoperative years. In both BII and VAG patients the risk was increased after 20 years, particularly in the VAG group in whom it was 7 -8-fold. When the initial decrease in risk in the BII patients is taken into account the risk for them is increased 5’3-fold, which is not significantly different from that of the VAG patients. The increased risk could be due to carcinogens formed as a result of surgically-induced hypochlorhydria.8,9 Raised bacteria and nitrite levels in the operated stomach cancer

0=observed; E=expected. *p<0-05,

tp<0-01,

p<0-001.

TABLE IV-GASTRIC CANCER MORTALITY IN DUODENAL

ULCER PATIENTS

*p<0-01.

tp<0-001.

931

have been reported,9-13 and increased levels of N-nitroso have not.12,14 GU patients

some

researchers have found

compounds,9although others

have an excess risk of death from time after operation, and may have been any gastric at increased risk before surgery, owing to decreased acid secretion. It is possible that the immediate effect of gastrectomy is to lower this risk by reducing the amount of gastric mucosa available for carcinogenesis, but among the GU patients there were none who had had VAG, so we were unable to test this hypothesis. That GU patients were more at risk after a BII operation than after a BI operation indicates that there could be factors other than hypochlorhydria and the amount of gastric mucosa available for carcinogenesis: one could be bile reflux, which is greater in BII than in BI seem to

cancer at

S

patients.’

Assessment of the risk of gastric cancer after gastric surgery into account the site of the original ulcer and the of type operation. The type of operation chosen could affect the risk in the longterm. During the period 1940-1960 vagotomy was done by the truncal method and was must take

accompanied by

a

drainage operation, pyloroplasty

or

gastroenterostomy, which predispose to bile reflux. The type of drainage was not recorded in enough patients to enable comparison of the risks according to type. The modern operation of highly selective vagotomy without drainage does not cause

gastric

bile reflux and may therefore carry

a

lower risk of

cancer.

We thank the Cancer Research Campaign for financial support; Mrs Edna Burns and Miss Jane Golding for abstracting information and handling the data; the medical records department at St James’ Hospital for help with finding patient records; Mr Mike Longyear (Centre for Applied Microbiology and Research) and the public health laboratory service computer services for the programs; and the staff of the Office of Population Censuses and Surveys both at St Catherine’s House, London and at the Central Registry, Southport for tracing such a large proportion of the patients.

Correspondence should be addressd to C. P. J. C., PHLS Communicable Disease Surveillance Centre, 61 Colindale Avenue, London NW9 5HT, UK.

REFERENCES 1. Ross

AHM, Smith MA, Anderson JR, Small WP Late mortality after surgery for peptic ulcer. N Engl J Med 1982; 307: 519-22. 2. Kalina TV, Kivilaakso E. Is the risk of stump cancer increased after partial gastrectomy?. Scand J Gastroenterol 1983; 18 (suppl 86): 35-36. 3. Sandler RS, Johnson MD, Holland KL. Risk of stomach cancer after gastric surgery for benign conditions: a case-control study. Dig Dis Sci 1984; 29: 703-08. 4 Pickford IR, Craven JL, Hall R, Thomas G, Stone WD. Endoscopic examination of the gastric remnant 31-39 years after subtotal gastrectomy for peptic ulcer. Gut 1984; 35: 393-97. 5 6 7

Stahlsberg H, Taksdal S. Stomach cancer following gastric surgery for benign conditions. Lancet 1971; ii: 1175-77. Nicholls JC. Carcinoma of the stomach following partial gastrectomy for benign gastroduodenal lesions. Br J Surg 1974; 61: 244-49. Caygill CPJ, Hill MJ, Hall CN, Kirkham JS, Northfield TC. Gastric surgery as a risk

factor in human carcinogenesis. Gut 1985; 26: A553. 8. Ruddell WSJ, Bone ES, Hill MJ, Blendis LM, Walters CL. Gastric juice nitrite: a risk factor for cancer in the hypochlorhydric stomach?. Lancet 1976; ii: 1037-39. 9 Reed P, Haines R, Smith PL, et al. Gastric juice N-nitrosamines in health and gastroduodenal disease Lancet 1981; ii 550-52 10. Ruddell WSJ, Bone EL, Hill MJ, Walters CL. Pathogenesis of gastric cancer in pernicious anaemia. Lancet 1978; i: 521-23 11 Jones SM, Davies PW, Savage A. Gastric juice nitrite and gastric cancer. Lancet 1978; i: 1355. 12 Hall CN, Cook A, Darkin D, et al Evaluation of the nitrosamine hypothesis of gastric carcinogenesis in man. Gut (in press) 13. Watt PCH, Sloan JM, Donaldson JD, Patterson CC, Kennedy TL Relationship between histology and gastric juice pH and nitrite in the stomach after operation for duodenal ulcer. Gut 1984; 25: 246-52 14 Keighly MRB, Youngs D, Poxon V, et al. Intragastric N-nitrosation is unlikely to be

after operations for duodenal ulcer. Gut 1984; 25: 238-45 15 Dahm K. Cancer of the gastric stump. In: Decosse JJ, Sherlock P, eds. Gastrointestinal Cancer I. The Hague, Boston, London: Martinus Nijhoff, 1981: 165-86.

responsible for gastric carcinoma developing

THE FETAL PHONOGRAM: A MEASURE OF FETAL ACTIVITY D. G. TALBERT W. L. DAVIES D. P. SOUTHALL

N. COLLEY N. G. ABRAHAM P. FAYERS

Department of Obstetrics and Gynaecology, University College, Gower Street, London; Institute of Obstetrics and Gynaecology, Queen Charlotte’s Hospital, Goldhawk Road, London; Clinical Monitoring Department, Charing Cross Hospital, Fulham Palace Road, London; Oxford Medical, 1 Kimber Road, Abingdon, Oxon; and Medical Research Council Tuberculosis and Chest Diseases Unit, Brompton Hospital, Fulham Road, London In 12 pregnant mothers fetal sounds and infrasounds were recorded by means of a new compliance matched transducer and compared with a simultaneous ultrasound record of fetal activity. A defined pattern on the fetal phonograph correlated with 86% of the total fetal breathing detected with ultrasound, and a further distinctive pattern was associated with 90% of fetal movements. Examination of the fetal phonocardiogram when the fetus was breathing showed a significant increase in the short-term variability of both the systolic and diastolic times when compared with non-breathing episodes. The median amplitude variabilities for both the first and the second heart sounds were also significantly increased during fetal breathing. Measurement of fetal sounds and infrasounds with a compliance matched transducer offers a non-invasive method for assessment of fetal activity for long periods of time.

Summary

Introduction Talbert et ar described a new wide-bandwidth sensor to detect fetal heart sounds from the surface of the maternal abdomen. The transducer, called the ’Tapho’, for total acoustic phonography, has since been shown to detect fetal body movements.2 In the present investigation we show that it can also be used to detect fetal breathing. We have called the total range of fetal sounds and infrasounds the fetal phonogram. Several methods are currently used to evaluate fetal health, including visual analysis offetal-heart-rate records,3counting of fetal body movements,4 and observations of fetal breathing.5 Since fetal breathing, body movements, and closure of the heart valves all produce characteristic identifiable patterns, the fetal phonogram may offer a new, non-invasive method of monitoring fetal wellbeing. The aims of this study were to compare the performance of the new transducer with real-time ultrasound and to examine the relation between fetal breathing activity and heart-sound intervals and amplitudes. _

,

Subjects and Methods 12 pregnant women, from 28 to 41 weeks gestation (mean 35 weeks), gave informed consent to this study. Subjects were selected from the antenatal clinic on the basis that they were willing to have an ultrasound scan to help with our research. A diagnostic sector scanner (General Electric RT 3000) with a 3’ 5 MHz phased array transducer was used to record a transverse

section of the fetal abdomen at the level of the umbilical vein. The ultrasound image and a time code signal (Racal GRA Oil) were recorded on video tape. A tapho transducer was attached to the maternal abdomen, with an adhesive disc, at a site where the fetal heart sounds were heard