Elevated Serum Pepsinogen I and II Levels Differ as Risk Factors for Duodenal Ulcer and Gastric Ulcer

Transcription

1 GASTROENTEROLOGY 1986;90:570-6 Elevated Serum Pepsinogen I and II Levels Differ as Risk Factors for Duodenal Ulcer and Gastric Ulcer I. MICHAEL SAMLOFF, GRANT N. STEMMERMANN, LANCE K. HEILBRUN, and ABRAHAM NOMURA Veterans Administration Medical Center, Sepulveda, California, and the Japan-Hawaii Cancer Study, Kuakini Medical Center. Honolulu, Hawaii We investigated the possibility that serum pepsinogen I (PC I) and pepsinogen II (PC II) levels might differ as risk factors for duodenal ulcer and gastric ulcer. From 1967 to 1970, serum was obtained from 7498 Japanese men in Hawaii, and the cohort was followed up to 1981 for the development of peptic ulcer disease. Pepsinogen I and PC II levels in stored serum were significantly higher in the subjects who developed duodenal ulcer (n = 43) or gastric ulcer (n = 115) than in 212 contral subjects. The linear trend in risk of each type of ulcer was highly significant for both PC I and PC II. An elevated serum PC I level (2:130 p.,g/l), however, was associated with about a threefold higher odds ratio for duodenal ulcer than for gastric ulcer (8.37 vs. 2.83), whereas an elevated PC II level (2:30 p.,g/l) was associated with about a threefold higher odds ratio for gastric ulcer than for duodenal ulcer (18.21 vs. 6.49). In contrast, the PC I/PC II ratio was significantly lower in the gastric ulcer than in the control and duodenal ulcer cases, and showed a significant linear trend in risk only for gastric ulcer. In addition, a PC IIPC II ratio of <4.0, which has been shown previously to be indicative of chronic gastritis, was associated with an almost -fold higher odds ratio for gastric ulcer than for duodenal ulcer (7.35 vs. 0.79). The results indicate that an elevated serum PC I level is a major risk factor for duodenal ulcer, whereas an elevated se- Received March 7, Accepted August 22, Address requests for reprints to: I. Michael Samlo f, M.p., Research Service (151), Veterans Administration Medical Center, Plummer Street, Sepulveda, California This work was supported by contract CP 660 and grant CA from the National Cancer Institute, and by grant AM from the National Institute of Arthritis, Diabetes, Digestive and Kidney Diseases. The authors thank the following institutions for their helpful cooperation: Castle Medical Center, Kaiser Medical Center, Queen's Medical Center, St. Francis Hospital, Straub Clinic and Hospital, TripIer Medical Center, and Wahiawa General Hospital. rum PC II level and a low PC I/PC II ratio are major risk factors for gastric ulcer. A number of studies during the past three decades have shown that serum pepsinogen levels are elevated in patients with duodenal ulcer (1-11), and two studies have found that an elevated level is associated with an increased risk for developing this disorder (12,13). These findings are consistent with the concept that a high serum pepsinogen level reflects an increased secretory capacity of the gastric mucosa to secrete acid and pepsin and, in turn, that gastric hypersecretion is a major factor in the pathogenesis of duodenal ulcer (14). Many of these same studies have shown that serum pepsinogen levels are also elevated in patients with gastric ulcer (3-11). However, the cause of hyperpepsinogenemia in gastric ulcer is not clear. In this disorder gastric acid output is usually within the normal range, and it is generally accepted that decreased mucosal resistance secondary to gastritis is a major factor in the pathogenesis of gastric ulcer (14,15). In all of the above studies, serum pepsinogen levels were determined by enzymatic assay. It is now known, however, that the acid proteinase activity of serum is derived from two immunologically distinguishable aspartic proteinases, pepsinogen I (PC I) and pepsinogen II (PG II), and that these differ partially in their cellular origins; PG I is synthesized principally by the chief cells in fundic gland mucosa, whereas PG II is produced by these cells and by the pyloric glands in the gastric antrum (16,17). Recent studies by Ichinose et al. (18) and by us (unpublished) using radioimmunoassays have shown that serum PG I and PG II levels are higher Abbreviations used in this paper: PG I, pepsinogen I; PG II, pepsinogen II.

2 March 1986 SERUM PEPSINOGENS AND PEPTIC ULCER 571 than normal in both duodenal ulcer and gastric ulcer patients, but that in gastric ulcer patients, the ratio of PC I to PC II is lower than in duodenal ulcer patients. The latter observation is of interest because a low serum PC IIPC II ratio has been found to be indicative of chronic gastritis (19). Because gastritis is almost always found in patients with established gastric ulcer (20-25)' we speculated that a low PC IIPC II ratio might also be a subclinical marker of increased risk for developing this disorder. In this study, we investigated the possibility that serum PC I and PC II levels and the PC IIPC II ratio might differ in their abilities to predict the development of duodenal ulcer and gastric ulcer. The population st;udied, Japanese men in Hawaii, is particularly we1i suited to the study of gastric ulcer risk because the incidence of gastric ulcer in this population is about threefold greater than that of duodenal ulcer (26,27). Methods The original cohort consisted of 11,148 Hawaiian Japanese men who had been identified by the Honolulu Heart Program through World War II Selective Service Registration files in 1965 (28). Subsequently, 8006 men volunteered to participate in a study designed to identify attributes that might predict the development of heart disease. Between 1967 and 1970 serum was obtained from 7498 members of the cohort and was stored at -20 e. Surveillance of the cohort for peptic ulcer disease was accomplished by review of discharge records of the hospitals on Oahu. Cases were accessioned as incident cases if an initial diagnosis of duodenal ulcer or gastric ulcer was established by radiologic, endoscopic, or pathologic examination after serum had been collected. During the surveillance period, which ended in August 1981, gastric ulcer was detected in 115 subjects, duodenal ulcer in 43 subjects, and combined duodenal and gastric ulcer in 17 subjects. Because we estimate that the rate of emigration of cohort men from Oahu has been only 1.5/00 per year, the identification of patients discharged from hospital with a diagnosis of ulcer should be almost complete. The control group consisted of 212 members of the cohort who had a negative history of peptic ulcer, gastric carcinoma, and gastric surgery before obtaining the blood sample, and in whom neither peptic ulcer nor gastric carcinoma developed during the surveillance period. Of these, 174 had been selected previously as age-matched controls for a companion study of gastric cancer. Because this group was, on the average, 3.6 yr older than the ulcer group as of the date of blood collection, an additional 38 subjects were randomly selected from an age-stratified younger subgroup of the cohort to obtain a closer age match between the ulcer and control groups. Serum PG I and PG II levels were determined by radioimmunoassay as described previously (29). The serum samples from the ulcer and control groups were transferred to coded tubes before assay. The normal range of serum PG I in this laboratory is from 25 to 0 /Lg/L. A recent correlative study of serum PG I and PG II levels and gastric mucosal histology has shown that a normal serum PG II level can best be defined as that value that yields a PG I1PG II ratio of >4.50 (19). Thus, although a serum PG II level of 22 /Lg/L is effectively the upper limit of normal (0/22 = 4.54), the same level would be abnormal if the serum PG I level was 50 /Lg/L (50/22 = 2.27). Approximate quartiles of the duodenal ulcer cases were used to establish ranges of each pepsinogen variable for statistical testing. Nonparametric methods of statistical analysis were used because of a positive skew in the distributions of serum PG I and PG II levels. The Kruskal Wallis test (30) was used for the simultaneous comparison of more than two pepsinogen distributions. To control the type I error rate, multiple comparisons of interest were performed by the non parametric rank sum method of Dunn (31). Odds ratios for gastric and duodenal ulcer were derived from logistic regression models (32) with adjustment for age. The odds ratio estimates the risk of disease in a particular group (e.g., a pepsinogen quartile) relative to the risk in a reference group. Tests for a linear trend in the logit of risk (33) were obtained from logistic regression models using continuous ungrouped pepsinogen levels and age. All logistic models were fitted using iterative maximum likelihood methods (34). Results Premorbid Serum Pepsinogen Levels Premorbid median and mean levels of serum PC I and PC II were significantly higher in the duodenal ulcer and gastric ulcer groups than in the control group (Table 1). Serum PC I was higher in the duodenal ulcer cases, whereas serum PC II was higher in the gastric ulcer cases. The PC IIPC II ratio was higher in the duodenal ulcer group than in the control group, but the difference was not statistically significant. In contrast, the PC IIPC II ratio was significantly lower (p < 0.001) in the gastric ulcer cases than in the duodenal ulcer and control cases. The median and mean levels of each pepsinogen variable in subjects with combined duodenal and gastric ulcer were higher than in controls, but none of the differences was statistically significant (Table 1). Because the combined ulcer group included only 17 patients, it was not analyzed further. The median time interval between blood collection and diagnosis was 5.75 yr in the duodenal ulcer cases and 5.04 yr in the gastric ulcer cases. An analysis of the effect of time between blood collection and diagnosis on pepsinogen levels in the gastric ulcer cases is shown in Table 2. Serum PC II was higher, while the PC IIPC II ratio was lower in the gastric ulcer cases than in controls at each time interval. The difference in serum PC II levels was statistically significant (p < 0.001) at each time interval.

3 572 SAMLOFF ET AL. GASTROENTEROLOGY Vol. 90, No.3 Table 1. Median and Mean Levels of Serum Pepsinogen I, Pepsinogen II, and the Pepsinogen I1Pepsinogen II Ratio in Ulcer Cases and Controls Serum PG I Serum PG II Median, Median, PG I1PG II ratio mean, SE mean, SE Median, Diagnosis No. (lig/l) pa No. (f.lg/l) p No. mean, SE p Duodenal ulcer < < > Gastric ulcer < < < Duodenal and gastric > > >0.50 ulcer Controls Four-sample test b p < P < P < PG 1, pepsinogen I; PG II, pepsinogen II. a Significance of the difference between the distribution in each ulcer type and controls (Reference 31). b Simultaneous comparison of all four distributions (Reference 30). The smaller number of duodenal ulcer patients precluded a similar statistical analysis, but the median level of each pepsinogen variable was higher in the duodenal ulcer cases than in controls at the same time intervals. Serum Pepsinogen Levels and Ulcer Risk The quartile values of each pepsinogen variable and their respective age-adjusted odds ratibs and 95% confidence limits for duodenal ulcer and gastric ulcer are shown in Tables 3-5. Serum PC I and PC II showed highly significant trends in the logit of risk for both duodenal ulcer and gastric ulcer (Tables 3 and 4). In contrast, the PC I1PC II ratio showed a significant trend only for gastric ulcer (Table 5). To facilitate presentation, the relationships between pepsinogen quartiles and odds ratios for duodenal ulcer and gastric ulcer are shown in Figure 1. Duodenal ulcer. The odds ratio for duodenal ulcer rose with increasing levels of serum PC I and PC II (Figure 1A). For PC I and PC II levels of 2:::130 p,g/l and 2:::30 p,g/l, respectively (values corresponding to their respective fourth quartiles), the odds ratios were 8.37 and In contrast, a PC I!PC II ratio of <4.0 was associated with an odds ratio of only Gastric ulcer. The risk relationships for gastric ulcer (Figure 1B) differed markedly from those found for duodenal ulcer. First, increased levels of serum PC II were associated with a steep and essentially linear increase in the odds ratio for gastric ulcer. At the highest quattile of serum PC II, the odds ratio was (vs for duodenal ulcer). Second, the dose-response curve for serum PC I reached a Table 2. Median Levels of Serum Pepsinogen I, Pepsinogen II, and the Pepsinogen I1Pepsinogen II Ratio by Time Interval From Blood Collection to Diagnosis of Gastric Ulcer Serum PG I Serum PG II Time interval Median Median PG IIPG II ratio (yr) No. (f.lg/l) po No. (f.lg/l) P No. Median p < < < > > < < < < >0. ;?: < < >0.50 Controls Five-sample test b p = P < P = PG I, pepsinogen I; PG II, pepsinogen II. 0 Significance of the difference between the distribution at each time interval and the distribution in controls (Reference 31). b Simultaneous comparison of all five distributions (Reference 30).

4 March 1986 SERUM PEPSINOGENS AND PEPTIC ULCER 573 Table 3. Age-Adjusted Odds Ratios of Duodenal Ulcer and Gastric Ulcer by Level of Serum Pepsinogen I Serum PG I (llg/l) < :130 Total No. of subjects Controls Odds ratio a % 95% confidence Odds confidence interval ratio interval P < O.OOOlb 1.00 (1.17,7.41) 1.72 (0.92, 3.21) (1.82, 12.34) 3.39 (1.78, 6.46) (3.12,22.49) 2.83 (1.30,6.15) P < O.OOlb, duodenal ulcer;, gastric ulcer. a From logistic regression model containing three indicator variables to represent the pepsinogen I quartiles and age at blood sample as a covariate. b Test for linear trend in the values of ge[risk/(1 - risk)], the logit of risk. plateau at its third quartile and had a maximal odds ratio of about 3, approximately one-third of the highest odds ratio found for duodenal ulcer. Third, a decrease in the PG IIPG II ratio was associated with a progressive increase in the odds ratio for gastric ulcer. For a PG IIPG II ratio of <4.0, the odds ratio was 7.35 (vs for duodenal ulcer). Discussion It is reasonable to assume that the morphologic and functional abnormalities of the gastric mucosa that have been described in patients with established duodenal ulcer and gastric ulcer precede the onset of these diseases. There is, however, little direct evidence for this assumption. One line of evidence is that a high serum pepsinogen level is both a feature of established duodenal ulcer (1-11) and a risk factor for developing duodenal ulcer (12,13). Another is the finding that, in some families with a prominent history of duodenal ulcer, an elevated serum PG I level is inherited as an autosomal dominant trait and is a subclinical marker of increased susceptibility for developing this disease (35). Hyperpepsinogenemia is also present in patients with established gastric ulcer (3-11), but there has been no evidence that an elevated serum pepsinogen level is a risk factor for this disorder. This study shows that in Hawaiian Japanese men, elevated serum PG I and PG II levels are risk factors for both duodenal ulcer and gastric ulcer, but that the degree of risk associated with these variables and with the ratio of PG I to PG II differs in strength or direction, or both, for each type of ulcer. The odds ratio associated with an elevated serum PG I level was about threefold greater for duodenal ulcer than for gastric ulcer (8.37 vs. 2.83), whereas the odds ratio associated with an elevated serum PG II level was about threefold greater for gastric ulcer than for duodenal ulcer (18.21 vs. 6.49). In addition, a PG I1PG II ratio of <4.0 was associated with an increased odds ratio for gastric ulcer, but with a decreased odds ratio for duodenal ulcer (7.35 vs. 0.79). Hyperpepsinogenemia in duodenal ulcer probably reflects an increased chief cell mass. The latter has not been shown directly, but can be implied from the finding that duodenal ulcer is associated with an increased parietal cell mass (36), and with increased gastric output of both acid and pepsin (37,38). In this study, we found that serum PG I and PG II levels were significantly higher in the subjects who developed duodenal ulcer than in control subjects,but Table 4. Age-Adjusted Odds Ratios of Duodenal Ulcer and Gastric Ulcer by Level of Serum Pepsinogen II Serum PG II (llg/l) < :30 Total 9 39 No. of subjects Controls Odds ratio a % 95% confidence Odds confidence interval ratio interval 1.00 (0.81, 5.40) 6.17 (2.93, 12.97) (2.16, 16.61) (4.89, 28.22) (2.21, 18.82) (7.79,42.56) P = b P < O.OOOlb, duodenal ulcer;, gastric ulcer. a From logistic regression model containing three indicator variables to represent the pepsinogen II quartiles and age at blood sample as a covariate. b Test for linear trend in the values of ge[risk/(1 - risk)], the logit of risk.

5 574 SAMLOFF ET AL. GASTROENTEROLOGY Vol. 90, No.3 Table 5. Age-Adjusted Odds Ratios of Duodenal Ulcer and Gastric Ulcer by Level of the Pepsinogen I1Pepsinogen II Ratio 95% 95% PG IIPG II No. of subjects Odds confidence Odds confidence ratio Controls ratio" interval ratio interval 2: (1.63, 12.29) 4.42 (1.42, 13.73) (0.75, 7.29) 6.99 (2.48, 19.71) < (0.32, 1.95) 7.35 (3.29, 16.44) Total P = b P < O.OOOlb, duodenal ulcer;, gastric ulcer; PG I, pepsinogen I; PG II, pepsinogen II. "From logistic regression model containing three indicator variables to represent the quartiles of the pepsinogen IIII ratio and age at blood sample as a covariate. b Test for linear trend in the values of ge[risk/(1 - risk)]. the logit of risk. that the ratio of PG I to PG II did not differ significantly from the control value. This pattern is similar to that found in patients with established duodenal ulcer (18), and is in keeping with the evidence that the chief cells are normally the major source of PG I and PG II (19). Thus, the available data suggest that comparable elevations of serum PG I and PG II are subclinical markers of increased risk for duodenal ulcer because this pattern of hyperpepsinogenemia reflects acid-pepsin hypersecretion. In contrast, the pattern of hyperpepsinogenemia in the subjects who developed gastric ulcer was characterized by a significant decrease in the PG IIPG II ratio due to a relatively greater increase in serum PG II than PG I. This serum pepsinogen pattern has also been found in patients with established gastric ulcer (18), a disorder that is almost always associated with gastritis (20-25)' and in subjects without gastric ulcer in whom gastritis was documented by endo- scopic biopsy (19). Thus, the available evidence suggests that the combination of a high serum PG II level and a low PG IIPG II ratio is a subclinical marker of gastritis. As such, the finding in this study that a high serum PG II level and a low PG IIPG II ratio were associated with a markedly increased risk for developing gastric ulcer is concordant with the concept that gastritis precedes gastric ulcer and is a major risk factor for this disease (14,15). Indeed, the lack of a temporal relationship between the premorbid pattern of serum pepsinogen levels and the diagnosis of gastric ulcer (Table 2) implies that gastritis precedes gastric ulcer by many years. It also appears from this study that gastritis may protect against the development of duodenal ulcer. Although the PG IIPG II ratio exhibited a V-shaped risk relationship to duodenal ulcer, a distinctly low ratio «4.0) was found to be associated with a slightly decreased risk for this disease :;:; o 0:: UI " o 8 4 A. Duodenal Ulcer Gastric Ulcer , , , , , First Second Third fourth First Second Third fourth Quartile 8 4 PC II Figure 1. Risk relationships between serum pepsinogen variables and odds ratios for duodenal ulcer (A) and gastric ulcer (B). The range of values for the quartiles of each variable is given in Tables 3-5. The quartiles of the PG IIPG II ratio are shown in reverse, i.e., the lowest quartile is plotted as the fourth quartile. A. The odds ratios for duodenal ulcer are similar for elevated levels of serum PG I and PG II. The odds ratio associated with a low PG IIPG II ratio is less than unity. B. The highest odds ratios for gastric ulcer are associated with an elevated serum PG II level and a low PG IIPG II ratio.

6 March 1986 SERUM PEPSINOGENS AND PEPTIC ULCER 575 The strong positive association that was found between the level of serum PG II and the odds ratio for gastric ulcer deserves further mention. Although serum PG II levels are elevated in subjects with gastritis (19), there is no evidence that the magnitude of the elevation defines an "ulcer-prone" type of gastritis. Chronic gastritis is an age-related disorder that is common in the general population (39,40), and particularly in the Japanese (41) who are at high risk for both gastric ulcer and gastric cancer (42). Yet, most persons with gastritis do not develop a gastric ulcer. Further correlative studies of gastric mucosal histology and serum PG I and PG II levels in gastric ulcer patients may elucidate the reason for the marked positive association that has been found between serum PG II and gastric ulcer risk. References 1. Mirsky la, Futterman P, Kaplan S. Blood plasma pepsinogen. II. The activity of the plasma from "normal" subjects, patients with duodenal ulcer, and patients with pernicious anemia. J Lab Clin Med 1952;40: Chinn AB. Studies on a blood serum proteolytic enzyme with particular reference to gastric secretory function. Gastroenterology 1953;25: Spiro HM, Ryan AE, Jones CM. The utility of the blood pepsin assay in clinical medicine. N Engl J Med 1955;253: Hirschowitz BI. Pepsinogen in the blood. J Lab Clin Med 1955;46: Hoar CS, Browning JR. Plasma pepsinogen in peptic-ulcer disease and other gastric disorders. A clinical and laboratory investigation. N Engl J Med 1956;255: Edwards K, Jepson RP, Wood KF. Value of plasma pepsinogen estimation. Br Med J 1960;1: Lee T. Studies of plasma pepsinogen. J Formosan Med Assoc 1960;59: Anscombe AR. Plasma pepsinogen: normal and abnormal secretion. Ann R Coll Surg Engl 1964;35: Lombarts AJPF, Peters HJ. Routine determinations of serum pepsinogens. Clin Chim Acta 1972;36: Naganna B, Balasundaram D, Venkaiah KR. Plasma pepsinogen in health and gastroduodenal disorders. J Indian Med Assoc 1972;58: Saez-Alquezar A, Marchese MA, Bezerra TVS, Andreoli JC, Parra DC, Pontes JF. Serum pepsinogen: correlation of techniques and analysis of levels found in different diseases of the stomach and duodenum. Clin Chim Acta 1978;90: Weiner H, Thaler M, Resiser MF, Mirsky IA. Etiology of duodenal ulcer. I. Relation of specific psychological characteristics to rate of gastric secretion (serum pepsinogen). Psychos om Med 1957;19: Niederman JC, Spiro HM, Sheldon WHo Blood pepsin as a marker of susceptibility to duodenal ulcer disease. Arch Environ Health 1964;8: Ippoliti A, Walsh J. Newer concepts in the pathogenesis of peptic ulcer disease. Surg Clin North Am 1976;56: Rhodes J. Etiology of gastric ulcer. Gastroenterology 1972; 63: Samloff 1M. Cellular localization of group I pepsinogens in human gastric mucosa by immunofluorescence. Gastroenterology 1971;61: Samloff 1M, Liebman WM. Cellular localization of the group II pepsinogens in human stomach and duodenum by immunofluorescence. Gastroenterology 1973;65: Ichinose M, Miki K, Furihata C, et al. Radioimmunoassay of serum group I and group II pepsinogens in normal controls and patients with various disorders. Clin Chim Acta 1982; 126: Samloff 1M, Varis K, Ihamaki T, Siurala M, Rotter JI. Relationships among serum pepsinogen I, serum pepsinogen II and gastric mucosal histology. A study in relatives of patients with pernicious anemia. Gastroenterology 1982;83: Hebbel R. Chronic gastritis: its relation to gastric and duodenal ulcer and to gastric carcinoma. Am J Pathol 1943;19: Du Plessis DJ. Pathogenesis of gastric ulceration. Lancet 1965;i: Stemmermann GN, Hayashi T. Intestinal metaplasia of the gastric mucosa: a gross and microscopic study of its distribution in various disease states. J Nat! Cancer Inst 1968;41: Gear MWL, Truelove SC, Whitehead R. Gastric ulcer and gastritis. Gut 1971;12: Thomas E, Hall P, Hislop IG. Observations on the histology of the gastric mucosa in chronic gastric ulcer. Dig Dis Sci 1972; 17: MacDonald WC. Correlation of mucosal histology and aspirin intake in chronic gastric ulcer. Gastroenterology 1973;65: May JM. Report on the geography of peptic ulcers. Schweiz Jur Alleg Path Bakteriol 1958;21: Haenszel W, Kurihara M. Studies of Japanese migrants: I. Mortality from cancer and other diseases among Japanese in the United States. J Nat! Cancer Inst 1968;40: Worth RM, Kagan A. Ascertainment of men of Japanese ancestry in Hawaii through World War II selective service registration. J Chronic Dis 1970;23: Samloff 1M. Pepsinogens I and II. Purification from gastric mucosa and radioimmunoassay in serum. Gastroenterology 1982;82: Kruskal WH, Wallis WA. Use of ranks in one-criterion analysis of variance. J Am Statis Assoc 1952;47: Dunn OJ. Multiple comparisons using rank sums. Technometrics 1964;6: Cox DR. The analysis of binary data. London: Methuen & Co. Ltd., Schlesselman JJ. Case-control studies: design, conduct, analysis. New York: Oxford University Press, Harrell F. The LOGIST procedure, SAS supplemental library user's guide. Cary, N.C.: SAS Institute Inc., 1983: Rotter JI, Sones JW, Samloff 1M, et al. Duodenal ulcer disease associated with elevated serum pepsinogen I: an inherited autosomal dominant disorder. N Engl J Med 1979;300: Cox A. Stomach size and its relation to chronic peptic ulcer. Arch Pathol 1952;54: Grossman MI, Kirsner JB, Gillespie IE. Basal and histalogstimulated gastric secretion in control subjects and in patients with peptic ulcer or gastric cancer. Gastroenterology 1963; 45: Petersen H, Myren J. Pentagastrin dose-response in peptic ulcer disease. Scand J GastroenteroI1975;: Siurala M, Iskoski M, Varis K, Kekki M. Prevalence of gastritis in a rural population. Bioptic studies of subjects studied at random. Scand J GastroenteroI1968;3:

7 576 SAMLOFF ET AL. GASTROENTEROLOGY Vol. 90, No Kreuning J, Bosman FT, Kuiper G, Van der Wal AM, Lindeman J. Gastric and duodenal mucosa in "healthy" individuals. An endoscopic and histopathological study of 50 volunteers. J Clin PathoI1978;31: Stemmermann GN, Ishidate T, Samloff 1M, et al. Intestinal metaplasia of the stomach in Hawaii and Japan: a study of its relation to serum pepsinogen I. gastrin, and parietal cell antibodies. Am J Dig Dis 1978;23: Segi M, Fujisaku S, Kurihara M. Mortality for gastric and duodenal ulcer in countries and its geographical correlation to mortality for gastric and intestinal cancer. Schweiz Z Path Bakteriol 1959;22: