Should Age, Sex, Subjective Symptoms in the Stomach and Past History of Peptic Ulcer be Regarded in the Evaluation of Efficacy of Serum Pepsinogen Values for Use in Gastric Cancer Mass-screening ?

The subjects were 13, 500 employees. Sera and information on age, sex, subjective symptoms in the stomach and past histories of peptic ulcer were collected. Serum pepsinogen I and II values were measured and analyzed (LPI and LPII : natural logarithm of serum pepsinogen I and II (ng/ml) respectively). Among those with stomach ache, heavy feeling of stomach or nausea, LPI and LPII levels were higher than among those without the symptoms, and it was suggested that those with subjective stomach symptoms should be dealt with separately in the evaluation of efficacy of gastric cancer mass-screening by serum pepsinogen values. Among those with a past history of duodenal ulcer, LPI and LPII levels were higher, and among those with a past history of gastric ulcer, LPI level was lower but LPII level was higher, and it was suggested that those with a past history of peptic ulcer should also be dealt with separately in the evaluation. Because serum pepsinogen values changed depending on age and sex among those without any subjective stomach symptoms or past history of peptic ulcer, it was suggested that the evaluation of efficacy of these values for mass-screening should be done separately with respect to age and sex. J Epidemiol, 1993; 3 : 71-76

The subjects were 13,500 employees. Sera and information on age, sex, subjective symptoms in the stomach and past histories of peptic ulcer were collected.
Serum pepsinogen I and II values were measured and analyzed (LPI and LPII : natural logarithm of serum pepsinogen I and II (ng/ml) respectively).
Among those with stomach ache, heavy feeling of stomach or nausea, LPI and LPII levels were higher than among those without the symptoms, and it was suggested that those with subjective stomach symptoms should be dealt with separately in the evaluation of efficacy of gastric cancer mass-screening by serum pepsinogen values. Among those with a past history of duodenal ulcer, LPI and LPII levels were higher, and among those with a past history of gastric ulcer, LPI level was lower but LPII level was higher, and it was suggested that those with a past history of peptic ulcer should also be dealt with separately in the evaluation.
Because serum pepsinogen values changed depending on age and sex among those without any subjective stomach symptoms or past history of peptic ulcer, it was suggested that the evaluation of efficacy of these values for mass-screening should be done separately with respect to age and sex. J Epidemiol, 1993; 3 : 71-76 serum pepsinogen I and II, age and sex distributions, subjective symptoms in the stomach, past history of peptic ulcer, Mass-screening for gastric cancer, false-positive ratio Pepsinogen is an inactive precursor of pepsin, a well-known gastric acid protease. Pepsinogen is classified into two immunologically different subtypes ; pepsinogen I (PGI) and pepsinogen II (PGII). The source of serum pepsinogens is gastric mucosa. PGI is mainly produced in the fundic area and PGII in the whole area of the stomach1.2).
Recently it has become easy to measure serum PGI and PGII, and the relationships between these values and the risk of gastric cancer has been shown3.4). These values have already been used in several mass-screening programs.
The efficacy of these values for use in gastric cancer mass-screening should be evaluated by the analysis using receiver operator characteristics (ROC) curve before being put into the practical use5). But such evaluation has not yet been done precisely and there seem to be some problems in the evaluation.
It is generally taken as a principle that those with some symptoms should not attend mass-screening but should take medical services clinically. However, in fact it is reported that in Japan a part of the screenees participating in mass-screening for gastric cancer have some symptoms in the stomach6.7.). So the first prob- were also observed with respect to each group. As to PGI and PI/II, the values that were less than some value (cut off value) were defined to be positive. As to PGII, the values that were more than some value(cut off value) were defined to be positive.
The calculations were done using the package program for personal computers "HALBAU"10)

RESULTS
The response rate to the questionnaire was 95.4%, and the subjects who did not respond to the questionnaire were excluded from the analysis. Age and sex distributions of total subjects which were used in the analysis is shown in Table 1. Positive rates of subjective symptoms and past history of peptic ulcer are shown in Table 2. The rates changed depending on sex and age. The relationships between serum pepsinogen values and the subjective symptoms or past history were analyzed by multiple regression analysis, and the results are shown in Table 3. LPI showed negative correlations with past history of gastric ulcer and positive ones with the other four factors, and they were statistically significant, though multiple correlation coefficients regarding subjective symptoms in the stomach were considerably small. LPII showed positive correlations with the five factors, and they were statistically significant except for that of heavy feeling of stomach.
LI/II showed negative correlations with past history of gastric ulcer and positive ones with the other four factors, and ones with past history of gastric  Table 2. Positive rates (%) of subjective symptoms in the stomach and past history of peptic ulcer.% for numbers of subjects in Table I   Table 3. Fge and sex adjusted relationships between serum pepsinogens and subjective symptoms or past history of peptic ulcer.
In addition to "factor", age (years) and sex (men-1, women=2) were explanatory variables in each multiple regression analysis. C.V.: criterion variable (LPI, LPII and L Over 70 years of age, both LPI level and LPII level among men were lower than those among women, but the differences were not statistically significant. Regarding both men and women, LI/1I decreased while age increased and no statistically significant differences of LI/II level were observed between men and women for any age groups. Positive rates when criteria were defined by PGI, PGII or PI/II are shown in Tables 6-1, 6-2, 6-3. As to both men and women, the cut off values of PGI and PI/II that showed positive rates of the same value decreased and those of PGII increased while age increased.
Among those with stomach ache, heavy feeling of stomach or nausea, LPI and LPII levels were higher than among those without the symptoms, but regarding LI/II no statistically significant differences were observed. Since the serum levels of pep-sinogens have been shown to be in proportion to excretion of pepsino-gen11,1Z', the symptoms are thought to be associated with the high pepsinogen excretion level. A part of the symptoms may be related to peptic ulcer but anoth-   duodenal ulcer tends to occur under excess of peptic factors and gastric ulcer tends to occur under weakness of defending factor15). It is widely believed that past history of peptic ulcer has no relationship with risk of gastric cancer16). But it is unknown if the relationships between serum pepsinogen levels and risk of gastric cancer among those with past history of peptic ulcer are similar to the relationships among those without it. As to this problem, a well-designed cohort study is thought to be necessary. At least it may be concluded that those with past history of peptic ulcer should be dealt with separately in the evaluation of efficacy of gastric cancer mass-screening by receiver operator characteristics (ROC) curve, because the influence of past history of peptic ulcer on serum pepsinogen values is considerably large.
Among those without subjective symptoms or past history of peptic ulcer, LPI increased slightly in proportion to age between 20 and 50, and decreased slightly over 50 of age, while level of LPII increased gradually in proportion to age. This is thought to be related to the oral-ward migration of the border between the fundic gland area and the pyloric gland area12). In the 20-49 age group, both LPI and LPII levels among men were higher than those among PI/II : serum pepsinogen I/serum pepsinogen 11 women, and the differences were statistically significant. But over 50 years of age no sexual differences were observed regarding either LPI or LPII. The reason why the sexual differences exist in younger age is unknown, but it may have something to do with the serum total protein level, which is higher among men. These results were almost similar with other studies17,18), though the subjects of the studies contain those with subjective stomach symptoms or past history of peptic ulcer.
Since the probability that healthy employees have gastric cancer is very small, positive rates when criteria are defined by PGI, PGII or PI/II correspond to falsepositive rates when screening for gastric cancer is done by the above criteria. The false-positive rates changed depending on age and sex. In another study, when criterion is that PI/II is less than 3.0 and PGI is less than 70.0, positive rates also change depending on age and sex191. Though several analyses by receiver operator characteristics (ROC) curve that have been done till now to evaluate the efficacy of serum pepsinogen values in gastric cancer mass-screening9, 17) have not