Reproductive factors, hormone use and the risk of lung cancer among middle-aged never-smoking Japanese women: A large-scale population-based cohort study

Authors

  • Ying Liu,

    1. Epidemiology and Prevention Division, Research Center for Cancer Prevention and Screening, National Cancer Center, Tokyo, Japan
    Search for more papers by this author
  • Manami Inoue,

    Corresponding author
    1. Epidemiology and Prevention Division, Research Center for Cancer Prevention and Screening, National Cancer Center, Tokyo, Japan
    • Epidemiology and Prevention Division, Research Center for Cancer Prevention and Screening, National Cancer Center, 5-1-1 Tsukiji, Chuo-ku, Tokyo 104-0045, Japan
    Search for more papers by this author
    • Fax: +81-3-3547-8578

  • Tomotaka Sobue,

    1. Statistics and Cancer Control Division, Research Center for Cancer Prevention and Screening, National Cancer Center, Tokyo, Japan
    Search for more papers by this author
  • Shoichiro Tsugane

    1. Epidemiology and Prevention Division, Research Center for Cancer Prevention and Screening, National Cancer Center, Tokyo, Japan
    Search for more papers by this author

Abstract

Although a link between female hormonal factors and the risk of lung cancer has been suggested, few studies have examined this association in detail. We investigated the associations between reproductive factors, hormone use and the risk of lung cancer in a population-based prospective study. Self-administered questionnaires were distributed to 44,677 lifelong never-smoking women in 1990–1994 to assess menstrual and reproductive factors and hormone use. After 8–12 years of follow-up, 153 lung cancer cases were diagnosed. Relative risk (RR) and 95% confidence intervals (CI) were calculated using the Cox proportional hazards model. Age at menopause, age at menarche, number of children, age at first live birth, breast feeding and use of hormones were not associated with a risk of lung cancer, either overall or among postmenopausal women or women with natural menopause. Compared to women with both late age at menarche (≥16) and early age at menopause (≤50), those with either early age at menarche or late age at menopause had a >2-fold, significant increase in the risk of lung cancer. Induced menopausal women with experience of hormone replacement therapy had a significantly elevated risk compared to naturally menopausal women without female hormone use, with an RR of 2.40 (95% CI 1.07–5.40). These findings suggest that both endogenous and extraneous estrogen may be involved in the etiology of lung cancer. © 2005 Wiley-Liss, Inc.

A large body of research suggests an association between female hormonal factors and the risk of lung cancer in women. However, the results are quite inconsistent. With respect to early age at menopause, studies have variously shown no effect,1, 2, 3 an increase4 or a decrease5, 6 in risk; for late age at menarche, no significant effect1, 2, 3, 5, 6 or a decrease in risk;4 and for the use of hormones other than oral contraceptives, an increase,3, 6, 7 a decrease2, 8, 9 or no appreciable influence on risk. Laboratory data have also suggested that hormonal factors may be involved in the etiology of lung cancer. Estrogen (ER) and progesterone receptors (PR) were reported to be present in human lung cancers, and adenocarcinomas exhibited significantly higher expression than other lung cancer cell types.10, 11, 12 The potential role of steroids in lung carcinogenesis by steroid receptor mediation13 has led to the search for a hormonal role. To date, however, no study on the association between hormone factors and the risk of lung cancer has been reported from Japan, notwithstanding a low rate of smoking vs. a high and increasing incidence of lung cancer in Japanese women.14 Moreover, data from prospective studies, with their inherently lower susceptibility to recall and selection bias, are scarce. To our knowledge, only one prospective study has been reported, which showed a moderately increased risk among women receiving hormone replacement therapy.7

Here, we investigated the association between hormonal factors (menstrual status, reproductive factors and hormone use) and the risk of lung cancer among never-smoking women in a population-based prospective study in Japan.

Material and methods

Study cohort

The Japan Public Health Center-based Prospective Study (JPHC Study) was launched with a population-based cohort in 1990 (Cohort I), then expanded with a second cohort in 1993–1994 (Cohort II). Subjects of Cohort I were recruited from among residents of 4 Public Health Center (PHC) areas and for Cohort II from 5 PHC areas. These 9 PHC areas were located in 9 prefectures distributed in Honshu, Kyushu and Shikoku and included 27 cities, towns and villages. Study subjects were all inhabitants with Japanese nationality who lived in the study areas at the start of the follow-up and were aged 40–59 in Cohort I and 40–69 in Cohort II. This population-based cohort of 116,694 subjects, among them 59,103 women (27,397 in Cohort I and 31,706 in Cohort II) were identified using population registries maintained by local governments. Details of the cohorts are described elsewhere.15 Ethical approval was provided by the institutional review board of the National Cancer Center.

Baseline survey

A self-administered questionnaire, which included menstrual and reproductive history, hormone use, previous disease history and other lifestyle factors, was distributed to all eligible registered residents in 1990 for Cohort I and in 1993–1994 for Cohort II. Completed questionnaires were collected from 49,924 women, giving a response rate for women of 84%. We further excluded 7.3% of women with a past or current smoking habit and 1,302 women with a history of cancer at any site, leaving a total of 44,677 for analysis.

Menstrual and reproductive factors, hormone use

For both cohorts, the questions on reproductive history consisted of menstrual status, age at menarche, history of pregnancy and delivery and history of gender-specific disease (ovaritis, for example). Further data was collected from menopausal women on age and type of menopause (natural or induced). Information on female hormone use was also collected. The form of this question differed slightly between the 2 cohorts; Cohort I subjects were asked whether they had experience of using female hormone drugs, whereas Cohort II subjects were asked whether they had used female hormone drugs for contraception, the treatment of menstrual disorders or during menopause.

Follow-up

Subjects were followed until December 31, 2002. A total of 2,182 newly diagnosed cases of cancer were identified among the 44,677 never-smoking women, including 153 of lung cancer (118 adenocarcinomas, 20 others, 17 unknown). Cases of lung cancer occurring in the 2 cohorts were identified through continuous surveillance of hospital records, population-based cancer registries and death certificates. Site of origin and histologic type were coded using the International Classification of Diseases for Oncology, 3rd Edition (ICD-O-3). Data on changes in residency were obtained from residential registries. Among noncase study subjects, 1,900 women (4.3%) moved out of the study area and only 0.04% were lost to follow-up within the study period. Death certificates collected through local public health centers revealed 811 deaths from causes other than lung cancer (1.8%) within the study period.

Statistical analysis

Person-years of follow-up were calculated for each subject from the start of the study until the date of diagnosis of lung cancer, date of migration out of the study area, date of death or the end of follow-up, whichever occurred first. The Cox proportional hazards model was used to estimate age-, PHC area- and passive smoking-adjusted and multivariate-adjusted relative risks (RR) for lung cancer. The assumption for the Cox proportional hazards model has been checked by graphical assessment and found to be valid. The effect of interaction was assessed by the likelihood ratio test. Records with missing information in the corresponding categories were deleted. Passive smoking in the workplace was defined as occurring when a woman inhaled other people's smoke for more than 1 hr per day on at least 1 day per week. Women who had family members with a smoking habit when the woman was in her childhood or adolescence were classified as having experienced passive smoking during childhood. Multivariate-adjusted relative risk included further adjustment for the confounding factors of sports in leisure time (<1 time/week, 1+ time/week), alcohol consumption (nondrinker, ≤2 days/week, 3+ days/week), body mass index (<20.0, 20.0–24.9, 25.0+), green and yellow vegetable consumption (<1 day/week, 1–4 days/week, almost daily) and family history of lung cancer (no, yes). Covariates were treated as categorical variables with indicator variables representing the categories. All computations were performed using the SAS software package version 8 (SAS Institute, Cary, NC)

Results

Table I shows the baseline information and characteristics of the 2 cohorts. Because Cohort I recruited women aged 40–59 and Cohort II recruited those aged 40–69, the percentage of postmenopausal women was higher in Cohort II, whereas the percentages of induced menopausal women were similar. No difference was seen between the cohorts in the proportion of adenocarcinoma, body mass index or alcohol drinking habit. The only difference between them except age was the percentage of hormone use, with fewer women receiving hormone therapy in Cohort II. This difference was consistent across all age groups (5-year strata).

Table I. Distribution of Study and Sociodemographic Variables for Cohorts I and II
 Cohort ICohort IITotal
  • 1

    1,643 women (387 in Cohort I, 1,256 in Cohort II) without data for menopausal status were excluded.

  • 2

    3,324 women (1,222 in Cohort I, 2,002 in Cohort II) without data for hormone use were excluded.

  • 3

    2,070 women (619 in Cohort I, 1,451 in Cohort II) without data for hormone use were excluded.

No. of subjects20,11524,56244,677
Person-years211,028184,420395,448
No. of cases7776153
 Adenocarcinoma5860118
 Others12820
 Unknown71017
Family history of lung cancer (no. of subjects)425380805
Mean BMI (kg/m2)24.124.124.1
Age at baseline (%)
 40–49 years47.633.340.1
 50–59 years52.432.041.0
 60–69 years34.718.9
Alcohol drinking (%)
 < Once per week76.479.678.2
 Daily2.43.53.0
Menopausal status (%)1
 Premenopausal45.633.639.1
 Natural menopause45.957.252.0
 Induced menopause8.59.28.9
Hormone use (%)220.66.212.8
Postmenopausal women
 No. of subjects10,73215,46526,197
 Person-years112,912116,379229,291
 No. of cases5754111
 Adenocarcinoma454085
 Family history of lung cancer (no. of subjects)227230457
 Mean BMI (kg/m2)24.224.224.2
 Mean age at baseline   
 Alcohol drinking (%)   
  <Once per week81.687.685.1
  Daily4.94.24.5
Hormone use (%)321.25.712.2

Lung cancer risk according to reproduction-related factors and hormone use among never-smoking women is shown in Table II. None of the menstrual or reproductive variables analyzed showed a statistically significant association with lung cancer. A moderately increased risk was observed among induced menopausal women compared to premenopausal or naturally menopausal women. Breast feeding also seemed related to increased risk, whereas early age at menopause (≤50) showed a slightly decreased risk among naturally menopausal women. Hormone use and age at first birth of ≥23 years were associated with an elevated risk among postmenopausal women, although risk decreased toward null when analysis was restricted to naturally menopausal women. No association between risk of lung cancer and variables was seen with regard to age at menarche or parity. Duration of menstruation among naturally menopausal women was moderately related with risk, but the results and p-value for the trend were not statistically significant.

Table II. Relative Risk (RR) and 95% Confidence Intervals (CI) of Lung Cancer According to Reproduction-Related Factors and Hormone use among Never-smoking Women1
 No. of subjectsNo. of casesPerson-yearsRR (95% CI)Postmenopausal
AllNatural
No. of casesRR (95% CI)No. of casesRR (95% CI)
  • 1

    Adjusted for age, public health center (PHC) area, and passive smoking during childhood or in the workplace.

Menopausal status
 Premenopausal16,83730152,7351.00    
 Natural menopausal22,38192195,7721.11 (0.61–2.02)921.00
 Induced menopausal3,8161933,5191.62 (0.83–3.17)191.45 (0.86–2.44)
Hormone use
 No36,077113315,6281.00831.00721.00
 Yes5,2762450,9381.46 (0.92–2.32)181.45 (0.84–2.49)111.13 (0.58–2.23)
Breast feeding
 No4,9641044,5611.0061.0041.00
 Yes34,612122306,9491.64 (0.83–3.25)891.60 (0.70–3.67)741.86 (0.68–5.11)
Age at menarche (years)
 ≤1310,4232392,5991.00131.0091.00
 14–1519,80572176,0381.13 (0.70–1.83)531.17 (0.63–2.16)461.29 (0.63–2.65)
 ≥1613,08751114,0750.84 (0.48–1.45)410.79 (0.41–1.53)340.80 (0.37–1.72)
Age at menopause (years)
 ≥518,1824571,038431.00411.00
 46–5012,75347111,760400.61 (0.34–0.95)380.64 (0.41–1.01)
 ≤455,4862647,765261.02 (0.62–1.69)110.79 (0.41–1.55)
Years of menstruation
 ≤306,1953153,498311.00171.00
 31–3511,46735100,551350.89 (0.53–1.48)321.38 (0.68–2.79)
 ≥368,3373972,223391.08 (0.64–1.82)381.71 (0.83–3.50)
Parity
 0–218,80456166,2011.00401.00311.00
 3–417,84561160,4880.94 (0.64–1.37)460.96 (0.62–1.50)380.92 (0.56–1.51)
 ≥55,2172044,5300.95 (0.53–1.71)140.86 (0.43–1.71)130.82 (0.39–1.70)
Age at first live birth
 ≤229,3272283,2351.00161.00151.00
 23–2515,96356141,6471.43 (0.87–2.38)381.49 (0.82–2.70)321.30 (0.69–2.44)
 ≥2613,65854122,0621.50 (0.90–2.51)411.65 (0.91–3.01)311.23 (0.65–2.34)

Table III shows lung cancer risk according to age at menarche and menopause among postmenopausal women. Compared to women with both late age at menarche (≥16) and early age at menopause (≤50), those with early age at menarche or late age at menopause had a ≥ 2-fold, significantly increased risk of lung cancer. The p-value for interaction was 0.08, suggesting the existence of a moderate level of negative interaction. Risk did not change materially but slightly increased when analysis was restricted to naturally menopausal women or when further adjustment for hormone use was made.

Table III. Relative Risks (RR) and 95% Confidence Intervals (CI) of Lung Cancer According to Age of Menarche and Menopause among Postmenopausal Never-smoking Women1
Age at menopauseAge at menarcheP-value for interaction
≥16 years (case/person-years)≤15 years (case/person-years)
  • 1

    Adjusted for age, public health center (PHC) areas, passive smoking at the workplace and during childhood.

≤50 years18/60,34344/90,896 
1.002.15 (1.18–3.91) 
>51 years22/30,41121/36,886 
2.49 (1.30–4.79)2.20 (1.13–4.29)0.08
Natural menopausal women
 ≤50 years13/52,47333/69,265 
1.002.38 (1.19–4.75) 
 ≥51 years20/29,85721/35,722 
2.62 (1.27–5.42)2.76 (1.32–5.77)0.03

We further evaluated the interaction between hormone use and induced menopause relative to the risk of total lung cancer and adenocarcinoma of the lung (Table IV). In comparison to naturally menopausal women who had no experience of hormone use, hormone use alone and induced menopause alone were not related to the risk of total lung cancer or adenocarcinoma tumor. Induced menopausal women who had experience of hormone use, however, had a significantly elevated risk of total lung cancer and adenocarcinoma, with RR values of 2.40 (95% CI 1.07–5.40) and 2.71 (95% CI 1.12–6.58), respectively. Tests for interaction were not significant, with p-values of 0.34 and 0.41, respectively. Results for hormone use and induced menopause relative to the risk of total lung cancer and adenocarcinoma did not change substantially when the 2 cohorts were analyzed separately.

Table IV. Relative Risks (RR) and 95% Confidence Intervals (CI) of Lung Cancer Relative to Menopausal Method According to Status of Hormone use among Menopausal Never-smoking Women1
Hormone useNatural menopauseRR (95% CI)Induced menopauseRR (95% CI)
  • 1

    Adjusted for age, (public health center) (PHC) areas, passive smoking at the workplace and during childhood.

Hormone use −
 Total cases72 11 
 Person-years158,9551.0024,2231.19 (0.61–2.30)
Hormone use +
 Total cases11 7 
 Person-years21,4421.19 (0.60–2.33)6,9882.40 (1.07–5.40)
P-value for interaction   0.34
Hormone use −
 Adenocarcinoma55 9 
 Person-years158,8471.0024,2041.28 (0.60–2.73)
Hormone use +
 Adenocarcinoma8 6 
 Person-years21,4311.23 (0.59–2.58)6,9782.71 (1.12–6.58)
P-value for interaction   0.41

Further adjustment for lung cancer family history, alcohol drinking habit, green and yellow vegetable consumption and physical activity during leisure time did not alter the results appreciably.

Discussion

In our study, no statistically significant association was observed between age at menopause, age at menarche, number of children, age at first live-birth, breast feeding or use of hormones and the risk of lung cancer, either overall or among postmenopausal women or women with natural menopause. Although previous studies have examined variables in terms of age at menarche or menopause, as well as hormone use, the combined effects of these variables on the risk of lung cancer have not been evaluated. When we calculated and divided total years of menstruation of naturally menopausal women into 3 categories (<30, 31–35, ≥36 years) and examined associations with lung cancer risk, a positive but not statistically significant association with length of menstruation was suggested. Furthermore, although late age at menarche and early age at menopause were not individually associated with a significantly reduced risk of lung cancer, their combination was associated with a remarkably lowered risk. It is therefore possible that the length of exposure to estrogen at least partly accounts for the risk of lung cancer.

Early and long-term exposure to endogenous estrogen may contribute to an elevated risk of lung cancer, as may drastic changes in estrogen concentration due to intervention in menstruation and exogenous estrogen use in later life. Several previous studies have reported an increase in lung cancer risk after hysterectomy3, 6, 16 and that this was highest among women whose ovaries remained intact, suggesting a role for estrogen concentration in the etiology of lung cancer.3 Although we did not collect data on whether the ovaries, uterus or both were removed in the present study, given that 75% of women with induced menopause had a history of endometritis or myoma of the uterus, whereas 9% had ovaritis, it is reasonable to think that most women with induced menopause had undergone hysterectomy, at which time the ovaries are also usually removed. Although estrogen concentration is generally considered an etiologic factor for lung cancer among surgically menopausal women, one US study that noted an increased risk for hysterectomy suggested that hysterectomy may be associated, either directly or indirectly, with venous thrombi in the pelvic veins, which may produce multiple showers of small emboli in the lungs, resulting in localized proliferative changes in the bronchial epithelium.3 In our study, the proportion of hormone replacement therapy among surgically menopausal women was twice that in naturally menopausal women. The finding of only a slightly increased risk for lung cancer among induced menopausal women without hormone replacement therapy vs. a markedly elevated risk among induced women with hormone replacement therapy strongly suggests that risk is elevated by hormone use itself rather than any surgery-related effect. In contrast, however, hormone replacement therapy was not associated with an increased risk of lung cancer among naturally menopausal women. This apparent discrepancy might have resulted from differences in total dosages of hormones used between surgically and naturally menopausal women, particularly if both the daily dosage administered and length of administration was higher in the former, as seems likely. A second explanation may be that, with the loss of the major organ of sex hormone receptors, the uterus, opportunities for excess estrogens to bond with receptors in the lung might increase among surgically menopausal women when exogenous estrogens are used, which might in turn stimulate the epithelium of the lung to act as a lung tumor promoter. A previous study reported that the blood concentration of estrogen among women undergoing hysterectomy without removal of the ovaries was elevated and that their risk of lung cancer increased correspondingly.3 These data support our second hypothesis concerning the influence of uterus resection and imply that the lung might be an estrogen-responsive organ.

An interesting finding was a drastic and linear decrease in age at menarche occurring over a period of only 26 years. Women aged 65–69 at baseline (born in 1924–29) began menstruation at a mean age of 16.0 years. In those aged 40–44 years (born in 1945–50), in contrast, mean age had decreased to 13.6 years. This finding confirms a trend identified in a previous Japanese study17 and is the result of marked improvements in social and economic living conditions in the last century.

The predominant use of estrogen replacement therapy later in life and the observed increase in risk of lung cancer after induced menopause and use of estrogen suggest a role for exogenous estrogens in the promotion phase of carcinogenesis. However, the high risk for early menarche and late menopause, but not for a long period of menstruation alone, imply that female hormones might also be involved in the initiation stage of lung cancer. Estrogens may influence lung cancer development, either through the direct promotion of cell proliferation in the lung or as a result of an effect on lung-carcinogen metabolism or the development of lung diseases that predispose to lung cancer. Estrogens could act as promoters through a receptor-mediated mechanism. The presence of estrogen receptors (ERα and ERβ) has been reported in lung tumors18, 19 and to a lesser extent in normal lung tissue.20 Lung tumors from women are more likely to express receptors than those from men,20 and adenocarcinomas showed higher expression than squamous cell carcinomas.10 Endogenous and exogenous estrogens have also been implicated as a cause of lung cancer without receptor activation; in this case they may represent direct-acting carcinogens, after metabolic activation to catechol estrogens, which can form DNA adducts.21

There was a marked difference between the 2 cohorts in the proportion of hormone use. Hormone replacement therapy was not a common practice in Japan until the last few decades. In Cohort II, the proportion of women aged 40–49 years with experience of hormone use was almost 4 times that of women aged 60–69 years. Thus, fewer women in Cohort II, who were on average older than those in Cohort I, received hormone replacement therapy. In addition, the definition of “hormone use” differed between the 2 cohorts, partly resulting in different proportions of hormone users. Some misclassification may thus have been inevitable, although the results of hormone use and induced menopause relative to the risk of total lung cancer and adenocarcinoma of the lung showed no substantial change when the 2 cohorts were analyzed separately or in combination. Another limitation of our study is that no baseline information on the type, duration of use or dosage of female hormone use was collected and thus the association of hormone factors with lung cancer could not be confirmed by further analysis on dose-response relationships. We have no data on the percentage of women using hormones for the treatment of menstrual disorders, either during or outside of menopause. Since oral contraceptives were rarely used in Japan when the subjects were at reproductive age, the influence of oral contraceptives did not need to be considered in our study. Passive smoking was moderately associated with lung cancer risk in this subgroup (data not shown), but “passive smoking from current family members” was not elicited in our questionnaire, and so only passive smoking in the workplace and passive smoking from family members during childhood were considered as covariates for adjustment. We did not observe any interaction between passive smoking and reproductive variables or hormone use. The number of lung cancer cases was relatively low, especially after analysis was restricted to menopausal women, making the study of relatively low statistical power. Further studies are warranted to confirm our findings.

The observation of an increased risk of adenocarcinoma among induced menopausal women receiving estrogen replacement therapy implies that exogenous steroid hormones may play a role in the etiology of lung cancer in women. In addition, the contribution of endogenous estrogens, suggested by the decreased risk of lung cancer among women with early age at menopause and late age at menarche, provides evidence for our hypothesis that female hormones may be involved in the etiology of lung cancer in women. If these results are confirmed, specific attention should be given to women undergoing induced menopause with hormone replacement therapy, and the necessity of hormone replacement therapy should be carefully considered.

Acknowledgements

The members of the Japan Public Health Center-based Prospective Study (JPHC Study) Group are: S. Tsugane, T. Sobue, T. Hanaoka, M. Inoue, National Cancer Center; J. Ogata, S. Baba, T. Mannami, A.Okayama, National Cardiovascular Center; K. Miyakawa, F. Saito, A. Koizumi, Y. Sano, I. Hashimoto, Iwate Prefectural Ninohe Public Health Center; Y. Miyajima, N. Suzuki, S. Nagasawa, Y. Furusugi, Akita Prefectural Yokote Public Health Center; H. Sanada, Y. Hatayama, F. Kobayashi, H. Uchino, Y. Shirai, T. Kondo, R. Sasaki, Y. Watanabe, Nagano Prefectural Saku Public Health Center; Y. Kishimoto, E. Tanaka, M. Kinjo, T. Fukuyama, M. Irei, Okinawa Prefectural Chubu Public Health Center; K. Imoto, H. Yazawa, T. Seo, A. Seiko, F. Ito, Katsushika Public Health Center; A. Murata, K. Minato, K. Motegi, T. Fujieda, Ibaraki Prefectural Mito Public Health Center; K. Matsui, T. Abe, M. Kataoka, Niigata Prefectural Kashiwazaki Public Health Center; M. Doi, Y. Ishikawa, A. Terao, Kochi Prefectural Chuo-higashi Public Health Center; H. Sueta, H. Doi, M. Urata, N. Okamoto, F. Ide, Nagasaki Prefectural Kamigoto Public Health Center; H. Sakiyama, N. Onga, H. Takaesu, Okinawa Prefectural Miyako Public Health Center; F. Horii, I. Asano, H. Yamaguchi, K. Aoki, S. Maruyama, M. Ichii, Osaka Prefectural Suita Public Health Center; S. Matsushima, S. Natsukawa, Saku General Hospital; S. Watanabe, M. Akabane, Tokyo University of Agriculture; M. Konishi, K. Okada, Ehime University; H. Iso, Y. Honda, Tsukuba University; H. Sugimura, Hamamatsu University School of Medicine; Y. Tsubono, Tohoku University; M. Kabuto, National Institute for Environmental Studies; S. Tominaga, Aichi Cancer Center Research Institute; M. Iida, W. Ajiki, Osaka Medical Center for Cancer and Cardiovascular Disease; S. Sato, Osaka Medical Center for Health Science and Promotion; N. Yasuda, Kochi Medical School; S. Kono, Kyushu University; K. Suzuki, Research Institute for Brain and Blood Vessels Akita; Y. Takashima, Kyorin University; E. Maruyama, Kobe University; the late M. Yamaguchi, Y. Matsumura, S. Sasaki, National Institute of Health and Nutrition; and T. Kadowaki, Tokyo University.

Ancillary