Serum pepsinogen as a new marker for gastric carcinoma among young adults

2695

Serum Pepsinogen as a New Marker for Gastric Carcinoma among Young Adults Shogo Kikuchi, M.D.,* Osamu Wada, M.D.,t Kazumasa Miki, M.D.,$ Toshifilsa Nakajima, M.D.,§ Tsunehiro Nishi, M.D., 11 Osamu Kobayashi, M.D.,Y Yutaku lnaba, M.D.,* and the Research Group on Prevention of Gastric Carcinoma among Young Adults

Background. Gastric Carcinoma is relatively uncommon in Japan among persons younger than 40 years of age, but its prognosis is not favorable and it affects young adults in their most productive years. This study was performed to evaluate the validity of serum pepsinogen as a new marker for gastric carcinoma among Japanese younger than 40 years of age. Methods. Data and sera were collected from the patients (108patients younger than 40 years of age with gastric carcinoma from nine hospitals in the Kanto-

From the *Department of Hygiene, Juntendo University School of Medicine, Tokyo; the tDepartment of Hygiene and Preventive Medicine and the SFirst Department of Medicine, Faculty of Medicine, llniversity of Tokyo, Tokyo; the §Department of Surgery, Cancer Institute Hospital, Tokyo; the IIDepartment of Surgery, Mitsui Memorial Hospital, Tokyo; and the llDepartment of Surgery, Kanagawa Cancer Center Hospital, Kanagawa, Jap,an. Supported in part by a Grant-in-Aid for Developmental Scientific Research (No. 01870028) by the Japanese Ministry of Education, Science and Culture and by the Chiyoda Life-Insurance Health-Promoting Foundation. Additional members of the Research Group on Prevention of Gastric Carcinoma among Young Adults: Toshiro Konishi, M.D., and Toru Hirata, M.D. (Second Department of Surgery, Faculty of Medicine, University of Tokyo); Hiroshi Sato, M.D. (Department of Surgery, Kyoundo Hospital, Tokyo); Teruo Ishibashi, M.D. (Section of Health Administration, Asama General Hospital, Nagano, Japan); Shoichi Ichikawa, M.D., and Nobuhirci Sato, M.D. (Department of Gastroenterology, Juntendo University fkhool of Medicine); Naoyuki Okamoto, M.D. (Department of Epidemiology, Kanagawa Cancer Center Research Institute); Takuji Kubo, M.D. (Department of Surgery, Cancer Institute Hospital); and Akio Myoga, Ph.D. (Dainabot Co. Ltd., Tokyo). The authors thank Tokyo Seamen’s.Hospital (Minato-ku, Tokyo, Japan), Shibaura Seamen’s Clinic (Minato-ku, Tokyo, Japan), and Keiai Hospital (Toshima-ku, Tokyo, Japan) for assisting in the collection of sera and data. They also thank Dr. Roger Beech (Department of Public Health Medicine, UMDS, London, U.K.) and the reviewers for help in editing the manuscript and in giving helpful comments. Address for reprints: Shogo Kikuchi, M.D., Department of Hygiene, Juntendo University School of Medicine, 2-1-1 Hongo, Bunkyo-ku,.Tokyo 113, Japan. Accepted for publication January 14, 1994.

Shin’etsu area in Japan) and from the control subjects (108 hospital control subjects and 108 screening control subjects) whose sex and age (within 4 years) were matched. Pepsinogen I and pepsinogen I1 values were measured and compared between patients and control subjects by paired t test. Sensitivities and specificities when criteria were defined by pepsinogen I, pepsinogen 11, and pepsinogen 1/11 ratio were calculated. Results. The pepsinogen I and pepsinogen I1 levels among patients were higher and the pepsinogen 1/11 ratio among patients was lower than among control subjects. When a pepsinogen 11 level higher than 14.8 ng/ml was considered positive, the test showed high sensitivity (83.3% for total gastric carcinoma and 85.0% for early gastric carcinoma) and high specificity (76.9% for hospital control subjects and 75.0% for screening control subjects). Similar degrees of sensitivity and specificity were obtained with the pepsinogen 1/11 ratio. Conclusions. These results suggest that a high pepsinogen I1 level combined with a low pepsinogen 1/11 ratio may be a useful screening test for gastric carcinoma in a young population at high risk for gastric carcinoma. This impression should be confirmed by a more extensive field trial to determine whether performance of these assays promotes early diagnosis of gastric carcinoma. Cancer 1994;73:2695-702.

Key words: gastric carcinoma among young adults, serum pepsinogens, case-control study, sensitivity, specificity, early detection, diffuse type, intestinal type.

The incidence and mortality rates of gastric carcinoma are decreasing in Japan, mainly related to the decrease of intestinal-type gastric carcinoma.’ In Japan, massscreening programs for gastric carcinoma by indirect roentogenography are performed widely. Persons younger than 40 years of age are not included in these programs, because the adverse effects of X-rays are believed to outweigh the benefits of the screening in this age group.’

2696

CANCER June 1,1994, Volume 73,No. 11

Table 1. Incidence Rate of Gastric Carcinoma Regarding Age and Sex in Japan:1986 Age

(vr.)

Men

Women

10-14 15-19 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 65-69 70-74 75-79 80-84 85+

0.0 0.5 0.6 4.4 11.2 23.7 45.9 78.8 139.0 224.1 329.4 450.9 588.0 708.0 736.1 743.8

0.0 0.3 1.8 5.5 13.6 24.3 36.2 47.9 61.6 82.2 124.1 172.9 242.3 281.7 341.3 348.6

The Research Group for Population-based Cancer Registration in Japan. Cancer incidence and incidence rates in japan in 1986: estimates based on data from nine population-based cancer regishes. Jpn I Clin Oncol 1991; 21:318-23.

The prognosis of gastric carcinoma among young adults is unfavorable, because most of these tumors are of Borrmann type 4.3Gastric carcinoma is relatively uncommon among those younger than age 40 in Japan (see Table l),but it has a disproportional social impact because it affects young adults at a time in their lives when they are most likely to have young ~ h i l d r e nIt. ~ has also been observed that the decrease in frequency of diff use-type gastric carcinoma predominant among young adults is slower than the decrease in frequency of intestinal-type carcinoma predominant among older adults. Pepsinogen is an inactive precursor of pepsin, an aspartic protease. Pepsinogen is classified into two immunologically different subtypes: pepsinogen I and pepsinogen 11.’ Recently, it has become easy to measure serum pepsinogen I and pepsinogen 11, and the relationship between these values and the risk of gastric carcinoma has been The current study was performed to evaluate the validity of serum pepsinogens as markers for gastric carcinoma among young adults. Subjects and Methods

Data and sera were collected from 108 patients, 108 hospital control subjects, and 108 screening control subjects from August 1988 to November 1992. The patients were younger than 40 years of age with gastric carcinoma treated at nine hospitals in the Kanto-

Shin’etsu area in Japan. Any patients who had had a previous operation of the stomach or chemotherapy were omitted from the study. Between patients and hospital control subjects and between patients and screening control subjects, sex and age (within 4 years) were matched. Hospital control subjects were selected from hospital patients in the same area who did not have malignant disease, severe renal dysfunction, severe hepatic dysfunction, or peptic ulcer. The causes of their hospitalization included cholelithiasis, acute appendicitis, hemorrhoids, inguinal hernia, anal abscess, and traumatic bone fracture. Screening control subjects were selected from screenees in the same area who were without severe renal dysfunction, severe hepatic dysfunction, or peptic ulcer. They were asked to answer a questionnaire and to offer 1 ml serum. Informed consent was obtained from all subjects. The sera were gathered from the patients and the hospital control subjects just after their admission and before major therapy, such as an operation or chemotherapy, was started. For the screening control subjects, the residual sera of the screening were used. Pepsinogen I and pepsinogen I1 were measured by the BML Company, Ltd. (Tokyo, Japan), with respect to all sera from these subjects after freezing preservation at less than -2OOC for up to 24 months. The measurements were taken by immunoradiometric assay with Pepsinogen 1/11 Riabead Kits (Dainabot Co. Ltd., Tokyo, Japan). For the patients, information on the stage of progression and the histologic type of gastric carcinomas was collected. The 108 matched data sets were classified into two groups regarding sex (men and women), stage of progression (early and advanced), and histologic type (diffuse and intestinal) of gastric carcinomas of the patients. In the classification, patients with diffuse- and intestinal-type lesions were classified into the intestinal-type group. Natural logarithms of serum pepsinogen I, pepsinogen 11, and pepsinogen 1/11 (LPI, LPII, and LI/II, respectively) were calculated, and paired f tests for matched data were done between patients and hospital control subjects and between patients and screening control subjects with respect to total subjects and the divided groups above. Differences with P < 0.05 were considered statistically significant. Sensitivities when criteria were defined by pepsinogen I, pepsinogen 11, and pepsinogen 1/11 were calculated using the data from the patients regarding all patients and patients with early gastric carcinoma. Specificities were also calculated using the data from hospital control subjects and screening control subjects. As for pepsinogen I and pepsinogen 1/11, the values that were less than some value (cutoff value) were considered positive. As for pepsinogen 11, the values that were

Pepsinogen and Gastric Cancer of Youth/Kikuchi et al.

2697

Table 2. Age and Sex Distributions of Gastric Carcinoma Cases, Hospital Controls, and Screening Controls

Table 3. Age and Sex Distributions of Early and Advanced Gastric Carcinomas

Aue (vr)

Age (yr)

Cases 15--19 20--24 25529 30-34 35--39 40--44 Total k lospital controls 15--19 20--24 25--29 30-34 3539 40--44 Total Screening controls 1519 20-24 25--29 30-34 3539 40-44 Total

Men

Women

Total

0 2 4 14 29 0 49

0 3 5 18 33 0 59

0 5 9 32 62 0 108

1 2 6 14 22 4 49

1

2 3 15 28 48 12 108

0 2 4 14 26 3 49

1 9 14 26 8 59 1

2 6 15 34 1 5

1 4 10 29 60 4 108

more than some value (cutoff value) were considered positive. The calculations were done using the package program for personal computers, "HALBAU."~ Sensitivities and specificities when conditions were decided by two pepsinogen I vdues, pepsinogen I1 values, and pepsinogen 1/11 ratios were calculated by personal computer with respect to all possible conditions and compared with the results above. Results

Men 20-24 25-29 30-34 35-39 Total Women 20-24 25-29 30-34 35-39 Total Total (men and women)

Advanced (%)

Total (%)

0 (0.0) 2 (50.0) 8 (57.1) 15(51.7) 25 (51.0)

2 (100.0) 2 (50.0) 6 (42.9) 14 (48.3) 24 (49.0)

2 (100.0) 4 (100.0) 14 (100.0) 29 (100.0) 49 (100.0)

l(33.3) l(20.0) 2(11.1) 1 I (33.3) 15 (25.4)

2 (66.7) 4 (80.0) 16 (88.9) 22 (66.7) 44 (74.6)

3 (100.0) 5 (100.0) 18(100.0) 33(100.0) 59 (100.0)

40 (37.0)

68 (63.0)

108 (100.0)

ratio are shown in Figures 1-3. The patients and the control subjects had different distributions regarding LPII values and the LI/II ratio, whereas the difference was not clear regarding LPI values. Results of paired t tests between patients and hospital control subjects and between patients and screening control subjects are shown in Table 5. LPI and LPII values among patients were higher and LI/II ratios among patients were lower than those among hospital control subjects and those among screening control subjects. All these results were statistically significant, but differences of LPI values between patients and control subjects were relatively small compared with those of LPII and LI/II. The results for men and women were similar, except that the differences of LPI were not significant in men. Early and advanced gastric carcinoma

Table 4. Age and Sex Distributions of Intestinal Type and Diffuse Type Gastric Carcinoma Age (yr)

Age and sex distributions of patients with gastric carcinoma, hospital control subjects, and screening control subjects are shown in Table 2. Forty-nine patients were men and 59 patients were women. The distributions of the patients regarding stages of progression and histologic types are shown in Tables 3 and 4, respectively. Early gastric carcinoma was present in 37.0%of the patients. The proportion of early gastric carcinoma was larger among men (51.0%) than among women (25.4%), but the proportion was not affected by age. Diffuse-type gastric carcinoma was present in 86.1 YO of the patients, and the proportion was not affected by age or sex. The distributions of LPI and LPII values and LI/II

Early (%)

Men 20-24 25-29 30-34 35-39 Total Women 20-24 25-29 30-34 35-39 Total Total (men and women)

Intestinal (%)

Diffuse (YO)

Total (Yo)

0 (0.0) 2 (50.0) 4 (28.6) 2 (6.9) 6 (12.2)

2 (100.0) 2 (50.0) 10 (71.4) 27(93.1) 43 (87.8)

2 (100.0) 4 (100.0) 14 (100.0) 29(100.0) 49 (100.0)

0 (0.0) l(20.0) 3 (16.7) 5 (15.2) 9(15.2)

3 (100.0) 4 (80.0) 15 (83.3) 28 (84.8) 50 (84.8)

3 (100.0) 5 (100.0) 18 (100.0) 33 (100.0) 59 (100.0)

15 (13.9)

93 (86.1)

108 (100.0)

2698

CANCER June 2,2994, Volume 73,No. 11

2o

3m

1

15

8

.m c,

cp

a

‘“0 10

p

5

z 0

2.0

3.0

5.0

4.0

6.0

1.0

Logarithm of pepsinogen I (ng/ml) Cases

Hospital controls

0Screening controls

Figure 1. Distribution of serum pepsinogen I in 108 gastric carcinoma case subjects, 108 hospital control subjects, and 108 screening control subjects.

yielded similar results, except that the differences of LPI between patients with advanced gastric carcinoma and the hospital control subjects were not significant. No differences of the results regarding LPII and LI/II were observed between patients with early gastric carcinoma and advanced gastric carcinoma. Intestinal- and diffuse-type carcinoma gave similar results, except that the

differences of LPI were not significant for intestinaltype carcinoma. Sensitivities and specificities when cutoff values of pepsinogen I, pepsinogen 11, and pepsinogen 1/11 were changed are shown in Tables 6 , 7 , and 8. As for pepsinogen I, the cutoff value that gave the most favorable sensitivities and specificities was 49.5 ng/ml, where

L1 f

5

z 0

1.0

2.0

3.0

4.0

Logarithm of pepsinogen II (ng/ml) Cases

Hospital controls

0screening controls

Figure 2. Distribution of serum pepsinogen I1 in 108 gastric carcinoma case subjects, 108 hospital control subjects, and 108 screening control subjects.

2699

Pepsinogen and Gastric Cancer of Youth/Kikuchi et al.

1.o

0.0

-1.0

2.0

3.0

Logarithm of pepsinogen Upepsinogen I1 Cases

Hospital controls

0Screening controls

Figure 3 Distnbution of pepsinogen I-pepsinogen I1 ratio in 108 gastric carcinoma case subjects, 108 hospital control subjects, and 108 screening control subjects

sensitivities were 43.5% and 3!j.O% (regarding total patients and patients with earlir gastric carcinomas, respectively) and specificities were 35.2% and 25.0% (regarding hospital control subjects and screening control subjects, respectively). As for pepsinogen 11, the most favorable cutoff value was 14.8 ng/ml, where sensitivities were 83.3% and 85.0% (regarding total patients and patients with early gastric carcinomas, respectively) and specificities were 76.9% and 75.0% (regarding hospital control subjects and screening control subjects, respectively). As for pepsinogen 1/11, the most favorable cutoff value was 3.33, where sensitivities were 75.9% and 75.0% (regarding total patients and patients with early gastric carcinomas, respectively) and specificities were 76.9% and 75.0% (regarding hospital control subjects and screening control subjects, respectively). The most favorable sensitivity and specificity when two of the three markers were considered were 83.3% and 8O.6%, respectively. They were given when it was decided the test was positive if at least one of two conditions, was met: either the pepsinogen I1 value was more than 18.0 ng/ml and pepsinogen 1/11 ratio was less than 2.7. Discussion

Because in another study' regarding pepsinogen I, pepsinogen 11, and pepsinogen 1/11, natural logarithms had smaller skewness and kurtosis than crude values, paired t tests in the current study were done according to natural logarithms. Serum pepsinogen levels are believed to be influ-

enced by smoking.'" However, the influence of smoking on serum pepsinogens was not considered in the current study, because the influence was not so large to be considered. Among Japanese healthy screenees age 2070 years, the age- and sex-adjusted differences of LPI and LI/II between smokers and nonsmokers were 0.058 and 0.048, respectively, and the differences were thought not to influence sensitivity and specificity. The difference of LPII was not statistically significant (S. Kikuchi, M.D. et al., unpublished data). Serum pepsinogen levels are also believed to vary according to existence of chronic atrophic gastritis." In the current study, the influence of the chronic atrophic gastritis was not considered, although it may be of some interest to compare the pepsinogens among four groups: patients with chronic atrophic gastritis, patients without it, control subjects with it, and control subjects without it. This is because the pepsinogen 1/11 ratio has already been proven to have a strong correlation with chronic atrophic gastritis" and it was thought to be overadjustment to control for this factor. Besides, in a case-control study like the current study, whether the patients had chronic atrophic gastritis or not was not clear because in some patients gastric carcinoma might spread over the gastritis and in some patients there was a possibility that carcinoma might cause the gastritis. Pepsinogen I (LPI), pepsinogen I1 (LPII), and pepsinogen 1/11 (LI/II) all showed differences between patients and control subjects in paired t tests, but pepsinogen I differences were weak. In the analyses of validity, pepsinogen I1 (LPII) and pepsinogen 1/11 (LI/II) showed high sensitivities and specificities among young

2700

CANCER June 1,1994, Volume 73, No. 11

Table 5. Differences in Serum Pepsinogen Values Between Cases and Controls (paired t test) Case

Hospital

Total LP I 3.95 -t 0.54 3.76 k 0.38 LPII 3 . 0 0 k 0.47 2.06 2 0.71 LI/II 0.95 k 0.43 1.69 f 0.52 Men (M) and women (W) LPI M 4.01 -t 0.56 3.82 k 0.37 W 3.89 -t 0.52 3.71 k 0.39 LPII M 3.03 -t 0.45 2.12 k 0.65 W 2.98 ? 0.48 2.02 +_ 0.75 LI/II M 0.99 t 0.46 1.70 rt 0.52 w 0.91 t 0.40 1.69 k 0.53 Early (E) and advanced (A) gastric carcinomas LPI E 4.10 ? 0.47 3.68 ? 0.37 A 3.86 f 0.56 3.80 f 0.38 LPII E 3 . 1 0 t 0.45 1.95 t 0.66 A 2.94 k 0.47 2 . 1 3 f 0.73 LI/II 1.73 ? 0.55 E 0.99 f 0.31 A 0.92 t 0.48 1.67 f 0.51 Intestinal (I) and diffuse (D) type gastric carcinomas LPI 3.76 f 0.38 I 3.84 rt 0.73 D 3.97 t 0.50 3.76 f 0.38 LPII I 2.95 ? 0.55 2.05 f 0.55 2.07 f 0.73 D 3.01 ? 0.45 LI/II 1 0.88 k 0.64 1.71 ? 0.41 1.69 ? 0.54 D 0.96 k 0.38

Screening

n

C:H

c:s

3.72 ? 0.47 2.18 rt 0.65 1.55 If: 0.59

108 108 108

0.003 0.000 0.000

0.001 0.000 0.000

3.83 -t 0.63 3.64 +- 0.25

49 59

0.070 0.015

0.115 0.001

2.29 k 0.58 2.08 -t 0.69

49 59

0.000 0.000

0.000 0.000

1.54 t 0.66 1.56 ri: 0.53

49 59

0.000 0.000

0.000 0.000

3.80 t 0.71 3.68 t 0.24

40 68

0.000 0.472

0.029 0.014

2.24 f 0.64 2.14 f 0.65

40 68

0.000 0.000

0.000 0.000

1.56 k 0.69 1.54 k 0.52

40 68

0.000 0.000

0.000 0.000

3.74 ? 0.28 3.72 k 0.50

15 93

0.751 0.001

0.625 0.001

2.12 ? 0.79 2.19 ? 0.63

15 93

0.001 0.000

0.001 0.000

1.62 f 0.59 1.54 f 0.59

15 93

0.003 0.000

0.003 0.000

Values are mean 2 SD. LPI: natural logarithm of pepsinogen I (ng/ml); LPII: natural logarithm of pepsinogen 11 (ng/ml); LI/ll: natural logarithm of pepsinogen I/pepsinogen 11: Hospital: hospital control; Screening: screening control; n: number of pairs; C:H: P value between cases and hospital controls; C:S: P value between cases and screening controls.

adults. Pepsinogen I1 and pepsinogen 1/11 are expected to be good markers for gastric carcinoma. In the current study, the validity of pepsinogen I1 was a little more favorable than that of pepsinogen 1/11, Among adults older than 40 years of age, pepsinogen I and pepsinogen 1/11 are believed to be more favorable markers than pepsinogen II.l2,I3The results of the current study were incompatible with the results of the other studies. This is probably because of the subjects' age differences. The pepsinogen I1 level increases with age?13e14 and none of the other ~ t u d i e s ' ~considered *'~ age. Serum pepsinogens, especially pepsinogen I and pepsinogen 1/11, have been proven to be markers for atrophic gastritis and have been believed to be markers

only for intestinal-type gastric car~inoma,'~ which occurs as a result of intestinal metaplasia or atrophic gastriti~.'~,"In the current study, no remarkable differences of pepsinogen I, pepsinogen 11, or pepsinogen I/ I1 were observed between intestinal- and diffuse-type gastric carcinoma. The differences of P values were thought to be due to the number of pairs analyzed, and mean values were not different between the two types of gastric carcinoma. We conclude that among young adults, pepsinogen I1 and pepsinogen 1/11 can be markers for not only intestinal- type gastric carcinoma but also diffuse-type gastric carcinoma. Most tumor markers produced by a tumor itself, such as alpha-fetoprotein and carcinoembryonic anti-

Pepsinogen and Gastric Cancer of Youth/Kikuchi et al.

2701

Table 6. Sensitivities (YO)and Specificities (%) When Cutoff Values (pepsinogen I) Are Changed: Pepsinogen I (natural logarithm)

and Specificities (YO) When Table 8. Sensitivities (YO) Cutoff Values (pepsinogen 1/11) Are Changed: Pepsinogen I/Pepsinogen 11(natural logarithm)

cutoff value

Se (T)

Se (E)

SD (H)

sv (S)

cutoff value

Sc (T)

Sc (El

22.2 (3.1) 24.6 (3.2) 27.2 (3.3) 30.0 (3.4) 33.2 (3.5) 36.6 (3.6) 40.5 (3.7) 44.8 (3.8) 49.5 (3.9) 54.6 (4.0) 60.4 (4.1) 66.7 (4.2) 73.7 (4.3) 81.5 (4.4) 90.1 (4.5) 99.5 (4 6)

5.6 6.5 7.4 10.2 16.7 20.4 26.9 36.1 43.5 54.6 64.8 71.3 81.5 83.3 88.0 92.6

0.0 0.0 2.5 5.0 7.5 10.0 17.5 25.0 35.0 45.0 60.0 65.0 72.5 77.5 82.5 82.5

97.2 92.6 87.7 82.4 75.9 68.5 56.5 46.3 35.2 21.3 16.7 13.9 8.3 6.5 1.9 0.9

98.1 97.2 93.5 85.2 79.6 65.7 50.0 36.1 25.0 14.8 12.0 3.7 1.9 0.9 0.9 0.9

1.35 (0.3) 1.50 (0.4) 1.65 (0.5) 1.83 (0.6) 2.02 (0.7) 2.23 (0.8) 2.46 (0.9) 2.72 (1.0) 3.01 (1.1) 3.33 (1.2) 3.67 (1.3) 4.06 (1.4) 4.49 (1.5) 4.96 (1.6) 5.48 (1.7) 6.05 (1.8)

6.5 7.4 8.3 13.9 21.3 34.3 48.1 60.2 69.4 75.9 82.4 87.0 92.6 94.4 96.3 97.2

0.0 2.5 2.5 12.5 12.5 25.0 45.0 57.5 67.5 75.0 82.5 90.0 92.5 95.0 97.5 97.5

Values less than cutoff values are defined to be positive. Pepsinogen I (ng/ml). Se (T): sensitivities calculated using data from all gastric carcinoma cases; Se (E): sensitivities calculated using data from early gastric carcinoma cases; Sp (H): specificities calculated using data from hospital controls; Sp (S):specificities calculated using data from screening controls.

gen, are diagnostic only in the presence of large volumes of tumor tissue. These markers do not detect or predict carcinoma at an early stage. In the current study,

Table 7. Sensitivities (YO)and Specificities (%) When Cutoff Values (pepsinogen 11) Are Changed:Pepsinogen I1 (natural logarithm) cutoff value

Sc (T)

Sc (E)

33.1 (3.5) 29.9 (3.4) 27.1 (3.3) 24.5 (3.2) 22.1 (3.1) 20.0 (3.0) 18.1 (2.9) 16.4 (2.8) 14.8 (2.7) 13.4 (2.6) 12.1 (2.5) 11.O (2.4) 9.9 (2.3) 9.0 (2.2) 8.1 (2.1) 7.3 (2.0)

9.3 15.7 24.1 32.4 40.3 55.6 65.7 76.9 83.3 85.2 88.9 91.7 92.6 93.5 94.4 94.4

17.5 25.0 32.5 40.0 50.0 60.0 70.0 80.0 85.0 90.0 95.0 95.0 95.0 95.0 97.5 97.5

Sv (H) 97.2 95.4 93.5 93.5 90.7 88.0 85.2 80.6 76.9 73.1 71.3 69.4 64.8 61.1 55.6 50.9

SP (S) 97.2 95.4 95.4 93.5 90.7 88.0 82.4 79.6 75.0 72.2 70.4 65.7 61.1 57.4 51.9 45.4

Values more than cutoff values are defined tt> be positive. Pepsinogen I1 (ng/ ml). See footnotes in Table 6.

Sv (H) 99.1 99.1 99.1 99.1 96.3 95.4 89.8 88.9 84.3 76.9 72.2 69.4 68.5 64.8 56.5 50.0

SP (S) 98.1 95.4 94.4 94.4 93.5 91.7 88.9 84.3 78.7 75.0 70.4 63.9 60.2 53.7 46.3 38.9

Values less than cutoff values are defined to be positive. See footnotes in Table 6.

pepsinogen I1 and pepsinogen 1/11 showed good validity regarding early gastric carcinoma, and the validity was not inferior to that of advanced gastric carcinoma. These findings suggest that, among young adults, pepsinogen I1 and pepsinogen 1/11 are useful in the diagnosis of gastric carcinoma, including intestinal and diffuse types of this tumor. However, there remain two problems to be solved before pepsinogen I1 and pepsinogen 1/11 can be used practically. The first problem is that patients with gastric carcinoma who are screened as positive in the early stage of progression by pepsinogen I1 and pepsinogen 1/11 may be limited to those with tumors with a good prognosis, whereas tumors with a poor prognosis may be missed. If so, detection of abnormal pepsinogen I1 and pepsinogen 1/11 levels are unlikely to prevent deaths from gastric carcinoma among young adults. The second problem relates to the low positive predictive value. As mentioned in above, incidence of gastric carcinoma among young adults is low compared with older adults (see Table 1).Approximately 20-25% of the subjects will get positive results in the primary examination by pepsinogen I1 or I'I/II, and this must be confirmed by gastroendoscopy or roentogenography. Screening for gastric carcinoma must be repeated, though the appropriate time interval between pepsinogen measurements is unclear at this stage. The problem of low positive predictive value could be compensated for if the interval is prolonged, because a prolonged interval can reduce the cost of the screening system. The

2702

CANCER June 2,2994, Volume 73,No. 11

prolongation is possible if pepsinogen values are not transient. The problem of low positive predictive value could also be compensated for if it becomes clear that those with negative results in the confirmation need not receive another confirmation for a considerably long period, even though they have positive results in the examination by pepsinogens. A well-designed field trial is needed to answer these questions. This trial should determine how often one should receive pepsinogen measurements and how often those with positive results by the examination of pepsinogens have to receive confirmations by gastroendoscopy or roentgenography. Because we expected screening validity to be improved by using two markers together, sensitivities and specificities were calculated regarding all possible conditions given with two of pepsinogen I value, pepsinogen I1 value, and ratio of pepsinogens 1/11, The specificity increased from 76.9% to 80.6% without decrease of sensitivity when the pepsinogen 1/11 ratio was added to pepsinogen 11. We also expect the predictive value of the test to increase by combining it with another serum marker, such as anti-Helicobacter antibody.”

8. 9.

10.

11.

12.

13.

References 14. 1. Kato Y, Kitagawa T, Nakamura K, Sugano H. Changes in histrogic types of gastric carcinoma in Japan. Cancer 1980; 48:20847. 2. Kasugai T. Standardized method of gastric mass-survey [in Japanese]. Gastric Mass Survey 1984; 62111-3. 3. Tsunoda H,Chiba K, Nishikawa K, lnaba K, lshikawa S. Clinicopathological study of age differences for gastric carcinoma [in Japanese]. Surg Therapy 1974; 30(4):364-9. 4. Kikuchi S, Inaba Y, Wada 0, Miki K, Nakajima T, Nishi T, et al. How do we prevent the death of stomach cancer in young men [in Japanese]?]pn ] Cancer C h i 1991; 37(3):235-8.

15.

16. 17.

18.

Samloff IM. Electrophoretic demonstration of the existence of eight proteolytic fractions in human gastric mucosa. Gastroenterology 1969; 57(6):659-69. Nomura AMY, Stemmermann GN, Samloff IM. Serum pepsinogen 1 as a predictor of stomach cancer. Ann Intern Med 1980; 93:537-40. Miki K, Ichinose M, Kawamura N, Matsushima M, Ahmad HB, Kimura M, et al. The significance of low serum pepsinogen levels to detect stomach cancer associated with extensive chronic gastritis in Japanese subjects. ] p n ] Cancer Res 1989; 8O:lll-4. Yanai H, Takagi H, editors. Handbook for multivariate analysis [in Japanese]. Kyoto, Japan: Gendai-Sugaku- Sha, 1986. Kikuchi S, lnaba Y, Wada 0, Miki K, Tenjin H, Kaneko E, et al. 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 massscreening?] Epidemiol 1993; 3:71-6. Malesci A, Basilico M, Bersani M, Bonato C, Ballarin E, Ronchi G. Serum pepsinogen 1 elevation in cigarette smokers. Scnnd ] Gastroenterol 1988; 23:602-6. Samloff M, Varis K, Ihamaki T, Siurala M, Rotter Jl. Relationships among serum pepsinogen I, serum pepsinogen 11, and gastric mucosal histology. Gastoenterol 1982; 83:204-9. Huang SC, Miki K, Furihata C, lchinose M, Shimizu A, Oka H. Enzyme-linked immunosorbent assays for serum pepsinogens I and 11 using monoclonal antibodies with data on peptic ulcer and gastric cancer. Cliti Chim Acta 1988; 175:37-50. Watanabe C, Sumii K, Haruma K, Morikawa A, Okamoto S, Toyoshima H, et al. An evaluation of gastric cancer mass survey using serum pepsinogen 1and I1 test in local inhabitants [in Japanese]. Gastroenterol Mass Survey 1992; 89:92-7. Aoki K, Goto T, Kudo M, Yasui T, Shimaoka A, Misumi J. A study on the distributions of serum pepsinogens in general population [in Japanese]. ]pn Med ] 1992; 3576:43-8. Fukao A, Hisamichi S, Osato N, Fujino N, Endo N, Iha M. Correlation between the prevalence of gastritis and gastric cancer in Japan. Cancer Causes Control, 1993; 4:17-20. Nagayo T. Microscopical cancer of the stomach: a study on histogenesis of gastric carcinoma. Int ] Cancer 1975; 16:52-60. You WC, Blot WJ, Li JY, Chang YS, Jin ML, Kneller R, et al. Precancerous gastric lesions in a population at high risk of stomach cancer. Cancer Res 1993; 53:1317-21. Nomura AMY, Stemmermann GN. Helicobacter pylon and gastric cancer. !Gastroenterol Hepatol 1993; 8294-303.