Jpn J Clin Oncol 2009;39(10)690– 694 doi:10.1093/jjco/hyp084
Epidemiology Note
Gender Differences in Stomach Cancer Survival in Osaka, Japan: Analyses Using Relative Survival Model Naomi Sato, Yuri Ito, Akiko Ioka, Masahiro Tanaka and Hideaki Tsukuma Department of Cancer Control and Statistics, Osaka Medical Center for Cancer and Cardiovascular Diseases, Osaka, Japan Received April 20, 2009; accepted June 23, 2009; published online August 17, 2009
Relative 5-year survival for stomach cancer has increased gradually in Osaka for more than two decades, while women show a small but consistently lower survival for it. We analyzed gender differences in stomach cancer survival, using relative survival model proposed by Dickman et al. Study subjects were reported stomach cancer cases diagnosed in 1975 – 99. We estimated the excess hazard ratios (EHRs) of death using Poisson’s regression model. The crude EHR for women was 1.12 [95% confidence interval (CI): 1.09 –1.14] in comparison with men. After adjustments for year and age at diagnosis, the EHR for women decreased to 1.07 (95% CI: 1.05 – 1.09), and furthermore, it reached to an insignificant level of 1.02 (95% CI: 0.99 – 1.04) after an additional adjustment for the extent of disease (localized, regional, distant and unknown). With further adjustments by histological type (intestinal, diffuse and others/unknown), method of detection (screening or not) and treatment (surgery or not), the EHR decreased to 0.97 (95% CI: 0.94 –0.99), significantly lower than the unity. These results indicate that the lower stomach cancer survival among women was attributable mainly to more advanced stages among women. The survival for women would have been a little better than for men if prognostic factors for stomach cancer had been comparable between the sexes. Inequality by the gender in taking screening, medical examination or treatment for stomach cancer was suggested to exist in Osaka, Japan. Key words: stomach cancer – gender difference – excess hazard ratio
INTRODUCTION
PATIENTS AND METHODS
Stomach cancer is one of the most common cancers in the world. Age-standardized mortality rate of stomach cancer has decreased since the 1960s in Japan, and now it is the second most common causes of death from malignant neoplasm in men after lung cancer, whereas it is still in the first place in women (1). Relative 5-year survival for stomach cancer has increased in Osaka, Japan, for more than two decades, while women show a little but consistently lower survival for it (2,3). In order to clarify the magnitude of the gender difference in stomach cancer survival and its relating factors, we carried out a population-based study in Osaka using relative survival model proposed by Dickman et al. (4).
DATA SOURCES
For reprints and all correspondence: Naomi Sato, 3-3 Nakamichi 1-chome, Higashinari-ku, Osaka 537-8511, Japan. E-mail:
[email protected]
The Osaka Cancer Registry has been operating since 1962, covering all of Osaka Prefecture (8.8 million in 2000 census), and it enables the observation of long-term trends of cancer incidences with reliable accuracy (5,6). The proportion of death certificate only cases was 10.5% in men and 14% in women for the period of 1975 – 99. In the case of multiple tumors, only a first primary was included in the survival analyses. Individual data on 72 789 reported stomach cancer cases [ICD 10th Revision, C16 (7)] were retrieved from the Registry’s database. These were cases residing in Osaka Prefecture except for Osaka City, who were diagnosed from 1975 through 1999; or cases living in Osaka City diagnosed
# The Author (2009). Published by Oxford University Press. All rights reserved.
Jpn J Clin Oncol 2009;39(10)
from 1993 through 1999. Active follow-up information on vital status 5 years after the diagnosis was available for these cases. Age at diagnosis was classified into five categories: ,44, 45 – 54, 55 – 64, 65 – 74 and 75 – 99. Stage at diagnosis was classified into the following three groups: (i) localized: cancer is confined to the original organ [TNM classification of UICC (8): T1 – 2N0M0], (ii) regional: cancer metastasizes to regional lymph nodes and/or spreads to immediately adjacent tissues (T1 – 2N1 – 3, or T3 – 4N0 – 3 and M0), (iii) distant: cancer metastasizes to distant organs or lymph nodes (Tany N anyM 1 ). Year at diagnosis was classified into five periods: 1975 – 79, 1980 – 84, 1985 – 89, 1990 – 94 and 1995 – 99. With reference to the Lauren’s classification (9), histological type of stomach cancer was classified as follows: intestinal [ICD-O-M (10): 814031-2, 82103, 821131-2, 821139, 826031-2, 826039, 84803, 84813 and 85103], diffuse (ICD-O-M: 80203, 80213, 80223, 80413, 814033, 81413, 81453, 821133, 82313, 826033, 84903 and 855033) and others/unknown (ICD-O-M: 80003, 80013, 80043, 80103, 80113, 80123, 80503, 80703, 80713 814039, 81903, 82003, 82303, 82403, 82463, 83103, 83233, 85603, 85703, 87203, 88003, 88303, 88903, 88913, 88963, 89003, 89363, 89803, 91503, 95403 and 95603). STATISTICAL ANALYSIS Distributions of the patients’ characteristics were assessed with x 2 test for categorical variables. The cumulative observed survival was estimated by using the Kaplan – Meier method. Survival time was computed from the date of the first diagnosis to the endpoint, defined as death from any cause. The closing date was defined as the date after 5 years from the first diagnosis. The relative survival was calculated as the ratio of the observed to the expected survival, which was calculated by the Ederer II method using general population life tables of Japan (11). Differences in 5-year relative survival by sex were modeled by using a multiple regression approach based on generalized linear models, which adopted the Poisson assumption for the observed number of deaths. The excess hazard ratios (EHRs) of death derived from these models quantified the extent to which the risk of death in a given group differed from that in the reference category after considering the background risk of death in the general population (4,12– 14). Differences were considered statistically significant if P values were ,0.05 in the two-sided. Data management and statistical analyses were conducted with STATA (15).
RESULTS A total of 72 789 newly diagnosed stomach cancer cases were included in this study (47 535 cases for men and 25 254 for women). The proportion of the lost-to-follow-up was 3.21% for men and 3.40% for women. Patient characteristics were
691
Table 1. Characteristics of the study subjects by sex Men Number of patients n ¼ 47 535
Women %
Number of patients n ¼ 25 254
P value %
Age (years) 0–44
3548
7.5
3406
13.5
45– 54
8161
17.2
4085
16.2
55– 64
13 656
28.7
5341
21.1
65– 74
13 878
29.2
6561
26.0
8292
17.4
5861
23.2
Localized
17 602
37.0
8325
33.0
Regional
75þ
,0.001
Extent of disease
16 378
34.5
9014
35.7
Distant
8418
17.7
4814
19.1
Unknown
5137
10.8
3101
12.3
,0.001
Histological type Intestinal
19 493
41.0
6864
27.2
Diffuse
14 362
30.2
10 401
41.2
Others/ unknown
13 680
28.8
7989
31.6
,0.001
Years at diagnosis 1975–79
5912
12.4
3657
14.5
1980–84
7577
15.9
4332
17.2
1985–89
8815
18.5
4848
19.2
1990–94
11 019
23.2
5535
21.9
1995–99
14 212
29.9
6882
27.3
,0.001
Method of detection Screening
3817
8.0
1614
6.4
Not screening/ unknown
43 718
92.0
23 640
93.6
,0.001
Treatment Surgery
36 889
77.6
18 631
73.8
Not surgery/ unknown
10 646
22.4
6623
26.2
,0.001
different between the sexes (Table 1). The mean age at diagnosis was 62.8 years for men and 62.5 years for women. Men were diagnosed at an earlier stage than women (percentage of the localized; 37.0% vs. 33.0%). Intestinal type of stomach cancer was most common among men, whereas diffuse type was most common among women. In men, more patients were diagnosed in recent years, detected through screening and treated by surgery than in women. Relative 5-year survival was 45.5% [95% confidence interval (CI): 45.0 – 46.0] for men and 41.9% (95% CI: 41.3 – 42.6) for women in the entire period. Figure 1 indicates the trend of relative 5-year survival from 1975 to 1999.
692
Gender differences in stomach cancer survival
Table 2. Relative 5-year survival of the study subjects and its 95% CI Men
All
Women
Relative 5-year survival
95% CI
Relative 5-year survival
95% CI
45.5
45.0–46.0
41.9
41.3–42.6
Age (years)
Figure 1. Time trends of relative 5-year survival of stomach cancer in Osaka, 1975– 99, by sex.
0 –44
56.5
54.8–58.1
46.1
44.3–47.8
45–54
55.9
54.8–57.0
50.5
48.9–52.1
55–64
49.9
49.0–50.8
48.4
47.0–49.7
65–74
42.3
41.3–43.2
42.3
41.0–43.6
75þ
26.6
25.4–27.8
26.4
25.0–27.7
Localized
85.2
84.5–85.9
84.8
83.8–85.7
Regional
29.5
28.8–30.3
29.5
28.5–30.5
Extent of disease
Distant
The survival has increased 1.5-fold (from 32.0% to 50.3%) for men and 1.7-fold (from 26.7% to 47.6%) for women. Although women showed a little but consistently lower survival than men, these differences in survival were getting smaller in recent years. Table 2 presents the relative 5-year survival and its 95% CI according to the patients characteristics. The survival decreased with age for both sexes. The survival for the localized disease was .84% for both sexes, and there were no remarkable differences in stage-specific survival between the sexes. Intestinal type of stomach cancer showed a better survival than diffuse type (61.1% vs. 42.7% for men and 60.3% vs. 44.8% for women). Screening-detected cases showed a better survival than the others. Cases treated by surgery also showed a better survival than the others. Table 3 shows the estimated EHRs of death and their 95% CI. The crude EHR of death for women was 1.12 (95% CI: 1.09 – 1.14), significantly higher than that for men (Model 0). The EHR for women adjusted by year and age at diagnosis was 1.07 (95% CI: 1.05 – 1.09), significantly higher than that for men (Model 1). The EHR, however, decreased to 1.02 (95% CI: 0.99 – 1.04) after a further adjustment for the extent of disease (localized, regional, distant and unknown) (Model 2). With further adjustments by histological type (intestinal, diffuse and others/unknown), method of detection (screening or not) and treatment (surgery or not), the EHR decreased to 0.97 (95% CI: 0.94 – 0.99), significantly lower than the unity (Model 3). In the model with all covariates (Model 3), the EHR decreased with the later years at diagnosis, whereas it increased with the increasing age at diagnosis. With respect to cancer stage, the EHR increased from 7.33 (95% CI: 7.01 – 7.66) for the regional to 17.01 (95% CI: 16.24 – 17.82) for the distant, when compared with the localized. Stage at diagnosis was strongly associated with prognosis. The EHR for diffuse type was 1.44 (95% CI: 1.40 – 1.49), significantly higher than that for intestinal type, and the EHR for others/
1.8
1.5–2.1
1.9
1.5–2.3
32.0
30.6–33.5
25.7
24.1–27.5
Intestinal
61.1
60.3–61.9
60.3
59.0–61.6
Diffuse
42.7
41.9–43.6
44.8
43.8–45.8
Others/ unknown
25.7
24.9–26.5
22.1
21.1–23.1
1975–79
32.0
30.7–33.3
26.7
25.2–28.2
1980–84
40.6
39.4–41.8
36.9
35.4–38.4
1985–89
46.5
45.3–47.6
44.1
42.6–45.6
1990–94
49.1
48.1–50.2
46.8
45.4–48.2
1995–99
50.3
49.4–51.2
47.6
46.3–48.8
Unknown Histological type
Years at diagnosis
Method of detection Screening
80.0
78.4–81.5
79.2
77.0–81.3
Not screening/ unknown
42.9
42.3–43.4
40.1
39.4–40.8
Surgery
53.2
52.6–53.8
51.6
50.8–52.4
Not surgery/ unknown
16.2
15.4–17.0
12.7
11.8–13.6
Treatment
CI, confidence interval.
unknown was also elevated. Histological type was suggested to be an independent predictor of survival. Screening-detected and surgically treated cases showed much lower EHR of death than the others.
DISCUSSION Our study results indicate that the consistently lower survival for stomach cancer among women was attributable to
Jpn J Clin Oncol 2009;39(10)
693
Table 3. Estimated excess hazard ratio (EHR) of death and its 95% CI Model 0 EHR
Model 1 95% CI
EHR
Model 2 95% CI
EHR
Model 3 95% CI
EHR
95% CI
Sex Men
1.00
Women
1.12
1.00 1.10–1.14
1.07
1.00 1.05– 1.09
1.02
1.00 0.99– 1.04
0.97
0.94–0.99
Years at diagnosis 1975– 79
1.00
1.00
1.00
1980– 84
0.75
0.73– 0.78
0.84
0.81– 0.87
0.91
0.88–0.94
1985– 89
0.61
0.59– 0.63
0.76
0.74– 0.79
0.86
0.83–0.89
1990– 94
0.56
0.54– 0.58
0.76
0.74– 0.79
0.86
0.83–0.89
1995– 99
0.53
0.51– 0.54
0.76
0.73– 0.78
0.85
0.82–0.88
Age (years) 0– 44
1.00
1.00
1.00
45–54
0.99
0.95– 1.04
1.02
0.98– 1.07
1.05
1.01–1.10
55–64
1.18
1.13– 1.23
1.15
1.1–1.19
1.19
1.14–1.24
65–74
1.42
1.36– 1.48
1.32
1.27– 1.37
1.30
1.25–1.35
75þ
2.33
2.23– 2.42
1.96
1.88– 2.04
1.61
1.55–1.69
Extent of disease Localized Regional Distant Unknown
1.00
1.00
8.19
7.82– 8.57
7.33
7.01–7.66
25.21
24.03– 26.44
17.01
16.24–17.82
8.99
8.53– 9.47
4.95
4.70–5.21
Histological type Intestinal
1.00
Diffuse
1.44
1.40–1.49
Others/unknown
1.73
1.68–1.78
Method of detection Screening
1.00
Not screening/unknown
2.30
2.16–2.46
Treatment Surgery
1.00
Not surgery/unknown
2.18
2.13–2.24
Model 0: follow-up time þ sex; Model 1: Model 0 þ years at diagnosis þ age; Model 2: Model 1 þ extent of disease; Model 3: Model 2 þ histological type þ method of detection þ treatment.
their more advanced stages and older ages at diagnosis, more cases of diffuse type and less cases that were screening-detected and/or surgically treated. After adjustment for these covariates, the EHR for women was significantly lower than that for men. These suggest that the prognosis for stomach cancer among women would have been a little better than that among men, if background factors such as extent of disease, histological type, method of detection and treatment had been comparable between the genders.
The findings above might be able to generalize to Japan as a whole. National estimate of relative 5-year survival for stomach cancer was 59.0% (95% CI: 58.4 – 59.6) for men and 57.0% (95% CI: 56.2 – 57.8) for women in Japan during the period of 1993 – 96 (16). According to the Comprehensive Survey of the Living Conditions of People on Health and Welfare conducted by the Ministry of Health, Labour and Welfare in 2007, the proportion of those who took stomach cancer screening in the past 1 year was 32.5% for men and 25.3% for women (17). Of those who had
694
Gender differences in stomach cancer survival
undertaken stomach cancer screening, 57.7% undertook it at their work place and 22.1% at their municipality among men, whereas 42.3% took it at their municipality and 37.5% at their work place among women. Gender difference in the opportunity for stomach cancer screening may exist not only in Osaka but also in Japan as a whole, and it may partly explain the gender difference in the stomach cancer survival for the nation. Stomach cancer screening has been conducted for all residents aged 40 years and over in Japan; thus, it is unlikely for the screening program to improve stomach cancer survival among women ,40 years. However, since the proportion of stomach cancer cases ,40 years is very small, we consider that it is important to improve the equal opportunity of the screening program for reducing the gender difference in stomach cancer survival as a whole. Intestinal type of stomach cancer was reported to have a better survival than diffuse type, based on the data from surgically operated cases (18). In our study, similar results were obtained, based on a population-based cancer registry data. Our study results suggest that histological classification by Lauren was an independent predictor of survival from the extent of disease at diagnosis. The biological factor is, however, uncertain why diffuse type of stomach cancer is more in female patients than intestinal type. Some limitations existed in our study. First, the proportion of screening-detected cases might be underestimated in the Osaka Cancer Registry database due to insufficient information in history of cancer screening. Potential misclassification from this underestimation is, however, considered to be non-differential for sex; thus, it should not have reversed our findings for the sex difference. Second, the proportion of histological type of others/unknown might not be negligible. In total, 15.2% and 18.2% were the cases without histological information among men and women, respectively. Among the cases of others, 93.2% and 91.8% were adenocarcinomas, of which degrees of differentiation were unknown among men and women, respectively; 2.0% and 3.8% were leiomyosarcoma among men and women, respectively. If we assume that all data for histological type of others/unknown had been intestinal type, the survival of male stomach cancer patients for intestinal type would have decreased to 46.5% and that of female would have decreased to 39.8%. On this extreme assumption, the survival of female stomach cancer patients for diffuse type would have exceeded that for intestinal type. However, if 70% of this histological category had been intestinal type, this reverse effect would not have occurred. Although we should keep in mind the possible bias introduced by misclassifications, our findings would not change seriously. In conclusion, the prognosis for stomach cancer among women in Osaka was confirmed to be worse than that among men. Inequality between the genders in taking screening, medical examination or treatment for stomach cancer was suggested to exist in Osaka or in Japan as a whole. Further studies are warranted to clarify this potential inequality and reasons for it.
Funding The study was supported by a Grant-in-Aid for Cancer Research from the Japanese Ministry of Health, Labour, and Welfare (20-2).
Conflict of interest statement None declared.
References 1. Ministry of Health, Labour and Welfare. Tokyo: Vital Statistics of Japan. http://ganjoho.ncc.go.jp/professional/statistics/odjrh3000000hwsa-att/ cancer_mortality(1958-2006.xls) (30 March 2009, date last accessed). 2. Ajiki W, Tsukuma H, Oshima A. Trends in cancer incidence and survival in Osaka. Gann Monogr Cancer Res 2004;51:137–63. 3. Osaka Prefectural Department of Public Health Welfare, Osaka Medical Association, Osaka Medical Center for Cancer and Cardiovascular Diseases. Annual Report of Osaka Cancer Registry: Cancer Incidence and Medical Care in Osaka in 2004 and the survival in 2000, Osaka; 2004 (In Japanese). 4. Dickman PW, Sloggett A, Hills M, Hakulinen T. Regression models for relative survival. Stat Med 2004;23:51– 64. 5. Parkin DM, Whelan SL, Ferlay J, Storm H. Cancer Incidence in Five Continents, Vols I to VIII. IARC CancerBase No. 7. Lyon: IARC 2005. 6. Curado MP, Edwards B, Shin HR, Storm H, Ferlay J, Heanue M, et al. Cancer Incidence in Five Continents, Vol. IX. IARC Scientific Publications No. 160. Lyon: IARC 2007. 7. International Statistical Classification of Diseases and Health Related Problems. 2nd edn. Geneva: World Health Organization 2004. 8. Sobin LH, Wittekind Ch. TNM Classification of Malignant Tumours (UICC). 6th edn. New York: Wiley-Liss 2002. 9. Lauren P. The two histological main types of gastric carcinoma: diffuse and so-called intestinal-type carcinoma. An attempt at a histo-clinical classification. Acta Pathol Microbiol Scand 1965;64:31–49. 10. Fritz A, Percy C, Jack A, Shanmugaratnam K, Sobin L, Parkin DM, et al. International Classification of Disease for Oncology. 3rd edn. Geneva: World Health Organization 2000. 11. Center for Cancer Control and Information Services, National Cancer Center. Tokyo: Abridged Life Tables in Japan for 1962 – 2006. http:// ganjoho.ncc.go.jp/professional/statistics/cohort01.html (11 March 2009, date last accessed). 12. Ito Y, Ioka A, Tsukuma H, Ajiki W, Sugimoto T, Rachet B, et al. Regional differences in population-based cancer survival between six prefectures in Japan: an application of relative survival models with funnel plots. Cancer Sci 2009;100:1306– 11. 13. Nahleh ZA, Srikantiah R, Safa M, Jazieh AR, Muhleman A, Komrokji R. Male breast cancer in the veterans affairs population: a comparative analysis. Cancer 2007;109:1471– 7. 14. Yu XQ, O’Connell DL, Gibberd RW, Coates AS, Armstrong BK. Trends in survival and excess risk of death after diagnosis of cancer in 1980 – 1996 in New South Wales, Australia. Int J Cancer 2006;119:894 –900. 15. StataCorp. STATA Statistical Software. College Station, TX: Stata Corporation, version 10, 2006. 16. Tsukuma H, Ajiki W, Ioka A, Oshima A. Research Group of Population-Based Cancer Registries of Japan. Survival of cancer patients diagnosed between 1993 and 1996: a collaborative study of population-based cancer registries in Japan. Jpn J Clin Oncol 2006;36:602– 7. 17. Ministry of Health, Labour and Welfare. Tokyo: Vital Statistics of Japan. http://www.mhlw.go.jp/toukei/saikin/hw/k-tyosa/k-tyosa07/xls/ kenkou.xls (30 March 2009, date last accessed). 18. Ribeiro MM, Sarmento JA, Sobrinho Simoes MA, Bastos J. Prognostic significance of Lauren and Ming classifications and other pathologic parameters in gastric carcinoma. Cancer 1981;47:780– 4.
