Search PubMed⌕ Search

Biomedical subjects

R K Ross

Publications and source records attributed to R K Ross.

At least 91 records · Page 5Linked to original sources
PubMed1996

Morbidity and mortality in relation to cigarette smoking in Shanghai, China. A prospective male cohort study.

OBJECTIVE: To evaluate prospectively the health risk of cigarette smoking in middle-aged men in Shanghai, China. DESIGN: Prospective cohort study with annual follow-up. PARTICIPANTS: A total of 18 244 male residents of Shanghai, China, enrolled in the study during January 1, 1986, through September 30, 1989, and actively followed via annual visits. RESULTS: By September 30, 1993, 852 deaths and 554 incident cancer cases were identified during the follow-up period, which averaged 5.4 years per subject. The overall incidence rate for cancer was 568 per 100 000 man-years, with the 3 leading sites being lung (146/100 000), stomach (116/100 000), and liver (81/100 000). Forty-one percent of all deaths were from cancer. Stroke was the most frequent cause of death unrelated to cancer, with an age-adjusted rate 4.2 times higher than that of US white men (201/100 000 vs 48/100 000), followed by ischemic heart disease, with an age-adjusted rate one-fifth that of US white men (69/100 000 vs 366/100 000). Compared with lifelong nonsmokers, the relative risks in heavy smokers (20 or more cigarettes per day) after adjustment for alcohol consumption were 2.2 for any incident cancer, 9.4 for incident lung cancer, 6.7 for head and neck cancer, and 1.8 for liver cancer. In terms of mortality, heavy smokers were at a 60% greater risk of death relative to lifelong nonsmokers; there was a 2.3-fold excess risk of death from cancer and 2-fold to 3-fold excess risk of death from heart disease. CONCLUSIONS: Cigarette smoking is an important predictor of risk of cancer and mortality in men in Shanghai. Among the study subjects, 36% of all cases of cancer and 21% of all deaths could be attributed to cigarette smoking.

Cause of Death↗
PubMed1996

Genetic susceptibility to cancer from exogenous and endogenous exposures.

The past four decades of epidemiological research have yielded valuable information on the risks of populations to environmental exposures such as tobacco, asbestos, and dietary components. Prevention efforts have been focused on large-scale population-based interventions to minimize exposure to such external carcinogens. While some cancers are beginning to show a decline from changing environmental exposures, hormone-related cancers, such as breast and prostate, are becoming more prevalent. The development of these cancers appears to be closely related to endogenous exposures to circulating steroid hormones. Although prevention trials using antihormone agents are proving successful in some instances, the long-term control of these cancers necessitates a clearer understanding of the metabolism and transport of the relevant hormone in vivo. The revolution in molecular biology has provided powerful genetic tools for evaluating mechanisms of cancer causation as well as the potential to better define individual susceptibility. Using tobacco exposure as an example, we and others have demonstrated that polymorphisms in genes controlling aromatic amine metabolism provide at least a partial explanation for ethnic and individual susceptibility to bladder cancer. Similar studies have examined genetic polymorphisms in the metabolism of tobacco smoke and lung cancer risk, red meat and colorectal cancer, and aflatoxin and liver cancer. Our current studies have pursued a similar paradigm of genetic polymorphism and individual cancer susceptibility in prostate and breast carcinogenesis. We are evaluating polymorphisms in the steroid 5 alpha-reductase type II and androgen receptor genes in relation to prostate cancer based on the evidence that intracellular dihydrotestosterone is the critical "carcinogen." We are pursuing genetic polymorphisms affecting estradiol metabolism, including those in the 17 beta-hydroxysteroid dehydrogenase 2 and estrogen receptor genes as they relate to susceptibility to breast cancer. The potential role of a polymorphism in the cytochrome P450c 17 alpha gene in both breast and prostate cancers is also being examined.

Adult↗
PubMed1996

Bladder cancer epidemiology and pathogenesis.

The two major causes of bladder cancer, cigarette smoking and occupational exposure to arylamines, have been recognized for 4 decades. Recently, attention has turned increasingly toward molecular approaches that can help predict which individuals exposed to these factors are at greatest risk of developing bladder cancer and can help explain the wide variation in bladder cancer risk among populations with similar prevalences of exposure to these two known causes. The mechanism for the relationship between smoking and bladder cancer risk has not yet been fully elucidated, but most likely is due to the presence in cigarette smoke of low levels of the same carcinogenic arylamines known to induce bladder cancer in occupational settings. These arylamines undergo metabolic transformation and the activated forms can form adducts with DNA, a probable essential step in the carcinogenic process. Alternatively, these arylamines can be detoxified. Two of the enzymes involved in this detoxification, glutathione S transferase mu and N-acetyltransferase, are genetically controlled. There is now ample epidemiological evidence that variations in the genes that code for these enzymes can affect bladder cancer risk among individuals exposed to arylamines. Levels of arylamine adducts are strongly correlated with cigarette "dose" and can be substantially modified by these detoxifying enzymes.

Carcinogens↗
PubMed1996

Risk factors for in situ breast cancer.

Few data exist on risk factors for in situ breast carcinoma. We examined risk factors for in situ breast carcinoma in data from two population-based case-control studies of breast cancer conducted among female residents of Los Angeles County. Cases with in situ or invasive disease were identified through the cancer registry for Los Angeles Country in the 1980s. We included all cases ages 40 years or younger diagnosed over a 5.5-year period and all cases ages 55-64 years diagnosed over a 3-year period. Control subjects were individually matched to cases by age (+/-3 years), neighborhood of residence, and, for younger controls, parity (nulliparous versus parous). The analysis included 233 cases with in situ cancer, 2057 cases with invasive cancer, and 2203 controls. Odds ratios (ORs) and 95% confidence intervals (CIs) were adjusted for screening mammography and established risk factors. In general, risk factors for in situ breast cancer were similar to those for invasive disease in this population. In premenopausal women, however, the risk of in situ breast cancer decreased with increasing body mass index, whereas for invasive disease, body mass index was unrelated to risk. In addition, in postmenopausal women with known age at menopause, use of unopposed estrogen replacement therapy was associated with increased risk of in situ disease [OR for ever use of estrogen alone was 1.60 (95% CI, 1.00-2.58)], whereas for invasive disease the OR was 1.23 (95% CI, 1.00-1.50). A similar difference was seen for combined hormone replacement therapy. Unmeasured increased screening among estrogen or combined replacement hormone users compared with nonusers could account for some of the association of in situ breast cancer risk with hormone replacement use.

Adult↗
PubMed1996

Clonal analysis of bilateral breast cancer.

Forty-nine pairs of bilateral breast tumors (41 synchronous and 8 asynchronous cases) were examined for X-chromosome inactivation status and p53 mutations to address the issue of their clonality. Among 12 cases that were informative for the trinucleotide repeat polymorphism in exon 1 of the androgen receptor gene on the X chromosome, 3 cases were found to have different alleles of the locus inactivated in the right and left breast tumors, indicating that the two tumors arose from distinct transformed cells. Thirteen tumors (13%) from 11 women (22%) contained somatic mutations in exons 5-8 of the p53 gene. In two cases, both breast tumors harbored p53 mutations, but the specific mutations were not identical. Seven synchronous and two asynchronous cases had p53 mutations in one tumor only. A germ line p53 mutation at codon 248, one of the most common p53 mutations in Li-Fraumeni syndrome, was observed in one case. Immunohistochemical analysis of p53 protein with a monoclonal antihuman p53 antibody showed concordant positivity between the right and left tumors in three bilateral breast cancer cases. Our results suggest that at least some bilateral breast tumors originate from distinct cells, but that some bilateral breast tumors may be related through a common p53 abnormality.

Adult↗
PubMed1995

A cohort study of serum testosterone and hepatocellular carcinoma in Shanghai, China.

A nested case-control study of HCC based on a cohort of 18,244 middle-aged men in Shanghai, China, who had been followed for an average of 5.3 years, was conducted. Our hypothesis dealt with the possible role of testosterone in the etiology of HCC, which shows a minimum of a 2- to 3-fold male excess in all populations world-wide. Seventy-six incident cases of HCC and 410 control subjects drawn from the cohort and individually matched to the cases by age (within 1 year), time of blood sample collection (within 1 month) and neighborhood of residence were assessed for serum HBsAg, anti-HBc, anti-HBs, anti-HCV and testosterone. Among controls, serum testosterone levels were similar between those who had no markers of HBV infection, those who were positive for anti-HBs only and those who were positive for anti-HBc but negative for HBsAg. However, the geometric mean level of testosterone in HBsAg-positive controls was 21% higher relative to HBsAg-negative controls and the difference was statistically significant (2-sided p = 0.0006). Relative to controls, HCC cases had a significantly higher mean level of testosterone at the time of recruitment (570 vs. 485 ng/dl, 2-sided p = 0.0005), but the difference was explicable on the basis of a higher proportion of HBsAg-positive individuals among cases than controls (p = 0.42 after adjustment for HBsAg status).

Age Factors↗
PubMed1995

Cancer incidence among Filipinos in Los Angeles County, 1972-1991.

Although Filipinos are the second largest Asian subgroup in the United States, little is known about their patterns of cancer incidence. We have examined cancer incidence rates among Filipinos living in Los Angeles County from 1972 through 1991 and evaluated their risk of non-localized cancer relative to non-Hispanic whites. Although Filipinos have substantially lower cancer rates than non-Hispanic whites in Los Angeles County, their incidence rates of liver cancer, cancer of the nasopharynx and thyroid cancer are higher. Filipino men and women had somewhat more advanced cancer at diagnosis, on average, than non-Hispanic whites, though no statistically significant differences were observed for sites where early detection methods exist. Our data suggest that Filipino-Americans would benefit from increased screening for cervical and breast cancer as well as efforts to interrupt transmission of hepatitis B virus to reduce liver cancer incidence.

Aged↗
PubMed1995

Correlation of estrogen levels between successive pregnancies.

To date, no study has attempted to correlate hormone levels between successive pregnancies in the same woman. Serum levels of total estradiol and total free estradiol have been measured between the eighth and the 17th weeks in the first and second full-term pregnancies of 34 white women participating in the Collaborative Perinatal Study, 1959-1965. Intraindividual interpregnancy Pearson's product moment correlations were calculated. Partial correlation coefficients (adjusted for gestational age) for log pregnancy total estradiol and total free estradiol were 0.78 and 0.73, respectively (p < 0.001). These findings provide evidence that levels of pregnancy estradiol are significantly and strongly correlated in successive pregnancies of the same woman. This phenomenon can provide an explanation for the higher concordance of breast cancer incidence between two sisters than between a mother and daughter in the familial pattern of breast cancer.

Breast Neoplasms↗
PubMed1995

Genetic variability of the human SRD5A2 gene: implications for prostate cancer risk.

Elevated dihydrotestosterone levels have been suggested to increase the risk of prostate cancer. The human SRD5A2 gene encodes the type II steroid 5 alpha-reductase, which converts testosterone to the more bioactive compound dihydrotestosterone. We have determined the distribution of a dinucleotide repeat in low-risk Asian-Americans, high-risk African-Americans, and intermediate-risk non-Hispanic Whites. We found this marker to be more polymorphic than previously reported, with some alleles being specific to African-Americans. Genetic variants of the SRD5A2 gene may play a role in predisposition to prostate cancer and in explaining the substantial racial/ethnic variability in risk.

3-Oxo-5-alpha-Steroid 4-Dehydrogenase↗
PubMed1995

The CAG and GGC microsatellites of the androgen receptor gene are in linkage disequilibrium in men with prostate cancer.

The androgen receptor genotype was determined in the white blood cell DNA of 45 African-American, 39 non-Hispanic white, and 39 Asian (Chinese, Japanese) normal subjects and 68 patients with prostate cancer (57 whites), all of whom were residents of Los Angeles County. For each subject, we measured the number of repeats in the polymorphic CAG and GGC microsatellites of exon 1 of the androgen receptor gene. In normal subjects, the distributions of CAG and GGC microsatellites differed significantly among the races (two-sided P = 0.046 and < 0.0005, respectively). The prevalence of short CAG alleles (< 22 repeats) was highest (75%) in African-American males with the highest risk for prostate cancer, intermediate (62%) in intermediate-risk non-Hispanic whites, and lowest (49%) in Asians at very low risk for prostate cancer. High-risk African-Americans also had the lowest frequency (20%) of the GGC allele with 16 repeats; the comparable values for intermediate-risk whites and low-risk Asians were 57% and 70%, respectively. Consistent with the interracial variation in CAG and GGC distributions, there was an excess of white patients with < 22 CAG and not-16 GGC repeats relative to white controls (relative risk, 2.1; one-sided P = 0.08). We observed no association (linkage) between the two microsatellites among normal subjects. On the other hand, there was a statistically significant negative association between the numbers of CAG and GGC repeats among the prostate cancer patients studied (two-sided P = 0.008). Among the 47 subjects with short CAG alleles (< 22 repeats), 43% had long GGC alleles (> 16 repeats) whereas only 15% of the 20 subjects with long CAG alleles (> or = 22 repeats) had long GGC alleles. Overall, our data suggest a possible association between CAG and GGC microsatellites of the androgen receptor gene and prostate cancer development.

Adult↗
PubMed1995

Prostate cancer: trends in mortality and stage-specific incidence rates by racial/ethnic group in Los Angeles County, California (United States).

Between 1976 and 1988 in the United States, the secular trends to age-adjusted incidence rates of prostate cancer were significantly different by racial/ethnic group (P < 0.001), and increased significantly only among non-Hispanic Whites at a rate of 2.7 percent (95 percent confidence interval [CI] = 2.3-3.1%) annually. While incidence rates of regional disease increased significantly (7.7 percent to 11.3 percent annually) among all racial/ethnic groups during this period, localized disease increased significantly only among non-Hispanic Whites, by 1.8 percent (CI = 1.4-2.3%) annually. Prostate cancer mortality in Los Angeles County (California) remained constant among Hispanics, non-Hispanic Whites, and Asians, but increases 1.6 percent (CI = 0-3.2%) annually among Blacks. While the increase in localized disease rates of non-Hispanic Whites may be due to increased detection of asymptomatic disease, this apparently has not occurred among other racial/ethnic groups in Los Angeles County. The secular increase in regional disease rates among all racial/ethnic groups without a concurrent increase in mortality (except Blacks), suggests increased accuracy of staging rather than a true increase in incidence may account for these trends. Adjusted for socioeconomic status, year and age at diagnosis, Black and Hispanic men were at significantly higher risk of being diagnosed with non-localized disease (odds ratio = 1.39 and 1.24, respectively) than were non-Hispanic Whites.

Black or African American↗
PubMed1995

Diet and breast cancer in Shanghai and Tianjin, China.

Various aspects of adult diet have been linked to breast cancer development. These include intake of fat (risk factor), and intake of fibre, soy protein and vitamins A, C and E (protective factors). Results of previous studies have been inconsistent. We examined the possible associations between breast cancer and various indices of nutrient and food intake in two Chinese populations who are at relatively low risk for breast cancer (one-fifth the rate in US white women). Two case-control studies of breast cancer were conducted in the cities of Shanghai and Tianjin, China. In Shanghai, 534 women aged 20-69 years with histologically confirmed breast cancer were recruited, whereas in Tianjin 300 women aged 20-55 years with histologically confirmed breast cancer were interviewed. All controls were community controls who were individually matched to the cases by sex and age (case-control ratio = 1:1). All interviews were conducted in person. Findings from the two studies were similar, although the diets in Shanghai and Tianjin were different in many respects. Cases and controls were similar in their consumption of soy protein, measured either in absolute levels or as percentages of total protein. Overall, all components of dietary fat (saturated fat, monounsaturated fat, polyunsaturated fat) showed a modest, non-significant association with breast cancer after adjustment for energy intake and other non-dietary risk factors for breast cancer. Intake of crude fibre, carotene and vitamin C, on the other hand, exhibited strong, statistically significant inverse associations with breast cancer risk. The last three indices were highly correlated, rendering it impossible to disentangle their individual effects; they were closely associated with intake of green vegetables in the two study populations. Vitamin E intake was unrelated to breast cancer risk in Shanghai and Tianjin. In the multivariate logistic regression model which included all non-dietary risk factors for breast cancer and energy intake, Shanghai women in the lowest tertile of crude fibre intake and highest tertile of fat intake had a 2.9-fold increased risk for breast cancer relative to those in the highest tertile of crude fibre intake and lowest tertile of fat intake. The comparable relative risk in Tianjin women was 2.4. Our data indicate a strong protective effect against breast cancer development with intake of foods rich in fibre, vitamin C and carotene. Our results are also compatible with dietary fat having a modest, positive effect on breast cancer risk within the range of exposure experienced by women in China.(ABSTRACT TRUNCATED AT 250 WORDS)

Adult↗
PubMed1995

Evidence of an X-linked or recessive genetic component to prostate cancer risk.

We used data from a population-based cohort study of blacks, Hispanics, Japanese and whites to examine the frequency of prevalent prostate and breast cancer by family history status of first-degree relatives (parents and siblings). Independent of race, the age-adjusted relative risk for prevalent prostate cancer in subjects with affected brothers was approximately two times that in subjects with affected fathers (P < 0.00005). No such excess risk for breast cancer was observed among subjects with affected sisters compared to those with affected mothers (age- and race-adjusted relative risk = 1.10, P = 0.34). The magnitude of the relative risk for prostate cancer in sibling- versus parent-affected groups was significantly different from that of the comparable relative risk for breast cancer (P < 0.00005). An excess risk of prostate cancer in men with affected brothers compared to those with affected fathers is consistent with the hypothesis of an X-linked, or recessive, model of inheritance.

Breast Neoplasms↗
PubMed1995

A urinary marker of alcohol intake.

Previously, one of us (B. K. T.) developed an assay that measures levels of free ethanol and ethanol conjugates in urine and showed that the mean levels of these ethanol markers in confirmed alcoholics were at least 20-fold higher than those levels in control subjects. In this study, we assessed the relationship of these biomarkers with self-reported levels of alcohol intake in a multiethnic sample of Los Angeles County residents who were male and over the age of 35 years (n = 128; 40 non-Hispanic whites, 46 blacks, 17 Chinese, and 25 Japanese). Regardless of race, the mean levels of free, bound, and total (free plus bound) ethanol were lowest in nondrinkers, intermediate in weekly drinkers, and highest in daily drinkers (P = 0.0001 in all three statistical tests of differences in the three biomarkers). Stepwise discriminant analysis showed that of the three potential biomarkers, total ethanol best discriminates between the three classes of drinkers (non, weekly, and daily), and that additional inclusion of either free or bound ethanol in the discriminant function had negligible effect. Overall, mean level of total ethanol was 2.2 times higher in weekly than in nondrinkers; daily drinkers, in turn, showed a 4.2-fold increase in mean total ethanol relative to weekly drinkers. However, there was no correlation between any of the three biomarkers and self-reported level (in grams of ethanol) of average consumption in either weekly or daily drinkers whose mean intake was about 13 and 42 g of ethanol/day, respectively. As the level of urinary free ethanol and ethanol conjugates showed extraordinary differences among racial groups for a given level of self-reported ethanol intake, the data suggest possible interracial differences in the in vivo elimination rate of ethanol; this latter finding needs to be confirmed in larger studies.

Adult↗
PubMed1995

Glutathione S-transferase M1 genotype affects aminobiphenyl-hemoglobin adduct levels in white, black and Asian smokers and nonsmokers.

Cigarette smoking is the major cause of bladder cancer in men in the United States, and the arylamines contained in cigarettes smoke, including 4-amino-biphenyl (4-ABP), are believed to play an important role in the induction of bladder cancer among smokers. N-acetylation, which is catalyzed by the genetically controlled hepatic N-acetyltransferase enzyme displaying two phenotypes (slow versus rapid), is a detoxification pathway for arylamines with regard to bladder carcinogenesis. In Los Angeles, CA, non-Hispanic white (white), black, and Asian males have comparable smoking habits and yet dramatically different risks of bladder cancer (31 of 100,000 in whites, 16 of 100,000 in blacks, and 13 of 100,000 in Chinese and Japanese). Previously, we have demonstrated that the prevalence of slow acetylators (the high-risk phenotype) was highest in whites (54%), intermediate in blacks (34%), and lowest in Asians (14%). We also showed that mean 3- and 4-ABP hemoglobin adduct levels were significantly higher in cigarette smokers relative to nonsmokers, and that the level increased with increasing number of cigarettes smoke/day. Most importantly, slow acetylators consistently exhibited higher mean levels of ABP hemoglobin adducts relative to rapid acetylators, regardless of race and level of cigarette smoking. We assessed 151 residents of Los Angeles County (CA) who were either white, black, or Asian (Chinese or Japanese) and over the age of 30 years for their glutathione S-transferase M1 (GSTM1) genotype (null versus non-null), acetylator phenotype (slow versus rapid), levels of 3- and 4-ABP hemoglobin adducts, and current use of tobacco products. Whites (27%) had the highest prevalence of the highest risk profile (slow acetylator, GSTM1 null), followed by blacks (15%) and Asians (2.7%), and the difference was statistically significant (P = 0.006). Whites also had less than one-half the prevalence of the "protective" profile (rapid acetylator, GSTM1 non-null) relative to blacks and Asians (23 versus 57%; P = 0.0001). Regardless of race and level of cigarette smoking, mean levels of 3- and 4-ABP hemoglobin adducts were higher in subjects possessing the higher risk (GSTM1/acetylator profile. Mean level of 4-ABP hemoglobin adduct (adjusting for race, cigarette smoking, and acetylator phenotype) was significantly higher in subjects possessing the GSTM1-null versus GSTM1-non-null genotype (46.5 versus 36.0 pg/g Hb; P = 0.037). The comparable difference in mean levels of 3-ABP hemoglobin adduct was borderline significant (1.6 versus 1.1 pg/g Hb; P = 0.07). Thus, our results suggest that GSTM1 is involved in the detoxification of 3- and 4-ABP and may contribute to the racial variation in bladder cancer incidence among white, black, and Asian males in Los Angeles, CA.

Adult↗
PubMed1995

MeIQx (2-amino-3,8-dimethylimidazo [4,5-f]quinoxaline): metabolism in humans and urinary metabolites in human populations.

The metabolism of the heterocyclic amines has been extensively studied in rodents and limited studies have been conducted in nonhuman primates. A study has been undertaken on the metabolism of 2-amino-3,8-dimethylimadazo[4,5-f]quinoxaline (MeIQx) in human subjects consuming normal cooked foods. A variety of fish and meat products has been cooked in several ways and fed to human volunteers. Urine was collected and analyzed according to methods developed for this purpose. Earlier rodent studies using radioactive MeIQx had suggested that the principal urinary excretion products included unmetabolized MeIQx, MeIQx sulfamate and MeIQx N-glucuronide. Conditions were developed for HPLC analysis of the free amine and its conjugates in human urine and for total hydrolysis of the conjugates to the free amine. Extraction and cleanup from urine were made possible by the availability of suitable monoclonal antibodies to MeIQx which could be used for immunoaffinity chromatography. For sensitive detection of the amounts present in human urine, gas chromatographymass spectrometry (GC-MS) was used in the negative chemical ionization mode. These studies have demonstrated that the distribution of metabolites in humans strongly resembles that in the rat. The sulfamate and N-glucuronide appear to be predominant human metabolites, while no evidence could be found of N-acetyl MeIQx. Subsequent to the studies just described, the methods developed were applied to urines collected in an epidemiological study on aromatic amine metabolism in Los Angeles. The study includes African American, White, and Asian people who have been phenotyped for caffeine acetylator status.

Adult↗
PubMed1994

Physical exercise and reduced risk of breast cancer in young women.

BACKGROUND: Epidemiologic evidence strongly suggests that cumulative exposure to ovarian hormones is a determinant of breast cancer risk. Because physical activity can modify menstrual cycle patterns and alter the production of ovarian hormones, it may reduce breast cancer risk; yet few epidemiologic studies have assessed this relationship. PURPOSE: The major objective of this study was to determine whether young women (aged 40 and younger) who regularly participated in physical exercise activities during their reproductive years had a reduced risk of breast cancer. METHODS: Using a case-control design, we conducted personal interviews of a total of 545 women (aged 40 and younger at diagnosis) who had been newly diagnosed with in situ or invasive breast cancer between July 1, 1983, and January 1, 1989, and a total of 545 control subjects. Case patients and control subjects were individually matched on date of birth (within 36 months), race (white), parity (nulliparous versus parous), and neighborhood of residence. Lifetime histories of participation in physical exercise activities on a regular basis were obtained during the personal interview. RESULTS: After adjustment for potential confounding factors, we found that the average number of hours spent in physical exercise activities per week from menarche to 1 year prior to the case patient's diagnosis was a significant predictor of reduced breast cancer risk (two-sided P for trend < .0001). The odds ratio (OR) of breast cancer among women who, on average, spent 3.8 or more hours per week participating in physical exercise activities was 0.42 (95% confidence limits [CLs] = 0.27, 0.64) relative to inactive women. The effect was stronger among women who had had a full-term pregnancy. Comparing most active (> or = 3.8 hours/wk of exercise) women to inactive women, the ORs were 0.28 (95% CL = 0.16, 0.50) for parous and 0.73 (95% CL = 0.38, 1.41) for nulliparous women. CONCLUSIONS: Most previously identified risk factors for breast cancer are reproductive and menstrual events that cannot be readily altered. The protective effect of exercise on breast cancer risk in the women whom we studied suggests that physical activity offers one modifiable lifestyle characteristic that may substantially reduce a woman's lifetime risk of breast cancer. IMPLICATIONS: Whether the protective effects of exercise on breast cancer risk are due to alterations in ovarian function and whether they extend into women's menopausal years need to be established. Our results suggest that implementation of regular physical exercise programs as a critical component of a healthy lifestyle should be a high priority for adolescent and adult women.

Adolescent↗