Abstract
Brassica vegetable consumption (e.g., Chinese cabbage) provides isothiocyanates (ITC) and other glucosinolate derivatives capable of inducing Phase II enzymes [e.g., glutathione S-transferases (GSTM1, GSTT1, and GSTP1) and NADPH quinine oxidoreductase] and apoptosis, altering steroid hormone metabolism, regulating estrogen receptor response, and stabilizing cellular proliferation. Asian populations consuming large amounts of Brassica have a lower breast cancer incidence compared with Western populations; however, the association between Brassica consumption and breast cancer risk is uncertain. It is difficult to estimate glucosinolate exposure and degradation in humans, possibly limiting epidemiological investigations of Brassica and cancer associations. We conducted a case control investigation of breast cancer in Shanghai, China, using urinary ITC levels as a biological measure of glucosinolate intake and degradation in populations with habitual Brassica intake. A representative subgroup of 337 cases providing presurgery, fasting, and first-morning urine specimens was one-to-one matched (age, menopausal status, date of urine collection, and day of laboratory assay) to population controls. Urinary ITC levels were inversely associated with breast cancer [odds ratio (OR) Quartile 1 = 1 (ref); ORQ2 = 0.9, 95% confidence interval (0.6, 1.4); ORQ3 = 0.7, (0.5, 1.1); ORQ4 = 0.5, (0.3, 0.8), adjusted for age, menopausal status, soy protein, fibroadenoma history, family breast cancer, physical activity, waist-to-hip ratio, body mass index, age at menarche, and parity in conditional logistic model]. This protective association persisted within post and premenopausal women. In contrast, total Brassica intake estimated from a food frequency questionnaire was not associated with breast cancer. Trends in the association between urinary ITC and breast cancer were more consistent with homozygous deletion of GSTM1 or GSTT1, the AAgenotype of GSTP1 (A313G), or with the C allele of NADPH quinine oxidoreductase (C609T), although interactions were not statistically significant. In conclusion, greater Brassica vegetable consumption, as measured by the urinary ITC biomarker, was associated with significantly reduced breast cancer risk among Chinese women.
INTRODUCTION
Breast cancer incidence is lower in Asian populations, suggesting that lifestyle practices and dietary habits affect breast cancer risk (1). In addition to greater soy and tea intake, and less fat intake, Asian populations habitually consume vegetables of the Brassica genus (e.g., cabbage, bok choy, cauliflower, and turnip). These vegetables are a source of glucosinolates, an N-hydroxysulfate with a sulfur-linked β-glucose and a variable side chain containing either an alkyl, alkenyl, aromatic, indolyl, or perhaps other moiety (2). With chewing or cutting, the plant cell wall ruptures, and the enzyme myrosinase is released, cleaving the glucose from the glucosinolate. The gut microflora may have a level of myrosinase-like activity and the capacity to further degrade glucosinolates (3). The remaining aglucone intermediate is unstable and further degrades to yield a number of biologically active molecules, including sulforaphane and I3C.3 These compounds are generally grouped and labeled as ITCs and indoles.
In animal models of breast cancer, tumor growth is inhibited by Brassica consumption, or ITC or I3C administration (4, 5, 6, 7, 8). The ITCs are both inducers of, and substrates for, Phase II enzymes, including GSTs and NQO1 (9, 10, 11). These detoxifying enzymes may protect cells against cancer initiation by neutralizing endogenous and exogenous electrophiles in breast tissue. Furthermore, the ITCs from Brassica have been found to induce apoptosis (10, 12, 13, 14, 15, 16). The persistence, distribution, and excretion of ITCs may depend on GST activity (17, 18, 19, 20, 21), and genetic polymorphisms in GSTM1, GSTT1, GSTP1, or perhaps other Phase II enzymes, such as NQO1, may limit the ability of ITCs to stabilize cellular proliferation and induce apoptosis in breast tissues. The indoles also induce apoptosis and regulate cellular proliferation (22, 23, 24, 25, 26). In breast tumor and other cells, indoles down-regulate expression of the estrogen-responsive pS2 and cathepsin-D genes and induce p21 expression consistent with inhibition of proliferation and G1-S stasis (27, 28, 29, 30). Recent preliminary investigations in PC-3 prostate cancer cells suggest that indoles down-regulate the epidermal growth factor receptor and reduce activity of the antiapoptotic proto-oncogene AKT (31). Furthermore, indole exposure may shift estrogen metabolism to favor the catechol estrogens with less affinity for the estrogen receptor (32) and exhibit tamoxifen-like properties (29, 33).
However, in humans, the relationship between Brassica consumption and breast cancer risk is uncertain. Investigations have found null associations (34, 35, 36, 37), protective but not statistically significant associations (38, 39, 40), and statistically significant protective associations (41). In one sense, these studies are consistent, because Brassica has not been associated with increased breast cancer risk. However, there may be several explanations for the inability to distinguish a “null” from “protective” association. Across these studies, population Brassica consumption may be below a biologically effective level, the measured dietary intake may not be during an etiologically relevant time period, or the analyses may be confounded. Furthermore, FFQs may not measure the sources of variability in ITC and indole exposure, including cultivar, the consumption of less common vegetables, vegetable preparation methods, or storage conditions (42, 43, 44, 45, 46).
To address dietary assessment limitations, a biomarker of dietary Brassica has been developed to compliment existing FFQ assessment methods common in epidemiological investigations. The ITCs from Brassica are excreted in urine as dithiocarbamates (47) and may be measured by HPLC after deconjugation and a cyclocondensation reaction with 1,2-benzenedithiol (47, 48). This urinary ITC marker provides a measure of total ITC exposure for each subject and has been positively correlated with habitual Brassica intake in an Asian study population (19). Previously, this urinary ITC marker was successful in identifying a protective interaction between Brassica, Phase II enzymes, and lung cancer in Chinese men (49). In this study, we investigated the association between breast cancer, urinary ITC levels, and the interactions with GSTM1, GSTT1, GSTP1, and NQO1 genotypes among women living in Shanghai, China.
MATERIALS AND METHODS
Participants.
Details of the SBCS, recruitment protocols, and data collection protocols have been reported previously (50, 51). Briefly, the SBCS is a population-based case control study conducted among Chinese in urban Shanghai, China. Eligible breast cancer patients were diagnosed during the period August 1996 to March 1998, between 25 and 64 years of age, and permanent residents of Shanghai, China. Cases were identified through our rapid case ascertainment system and supplemented by the population-based Shanghai Cancer Registry. There were 1601 eligible breast cancer cases identified for the study, and interviews were completed from 1459 (91.1%) of eligible cases. The major reasons for nonparticipation were refusal (n = 109, 6.8%), death before interview (n = 17, 1.1%), and inability to locate (n = 17, 1.1%). Two senior pathologists confirmed each cancer diagnosis through the review of tumor slides.
Community controls were randomly selected from the female general population using the Shanghai Resident Registry, a registry of all adult residents in urban Shanghai. Only women who lived at the listed address during the study period were eligible. Controls were frequency matched by 5-year age categories using projections of the age distribution of breast cancer patients, and in-person interviews were completed on 90.3% of the 1724 eligible controls (refusals: n = 166, 9.6%; death or previous cancer diagnosis: n = 2, 0.1%). Informed consent was obtained from all participants.
A urine sample was collected from 98.7% of cases and 99.8% of controls. After urine collection, 125 mg of ascorbic acid were added to ∼100 ml of urine to prevent oxidation of labile compounds, and samples were immediately transported on ice (0°C-4°C) to the central laboratory for processing and long-term storage at −70°C within 6 h of collection.
Data Collection.
A structured questionnaire was used to obtain information on demographics, reproductive history, hormone use, dietary habits, disease history, physical activity, tobacco and alcohol use, weight history, and family history of cancer for each participant. Weight, height, and the circumferences of the waist and hips were measured by trained staff. Habitual dietary intake regarding the past 5 years was measured by a validated FFQ specifically designed to measure intake of foods commonly consumed in Shanghai, China. Nutrient scores were computed using the Chinese Food Composition Table (52). Previous analysis of the SBCS found positive association with age, education, family history, menopausal age, age at first live birth, WHR, and BMI and inverse associations with age at menarche and exercise (50).
Study Design.
We conducted an individually matched case control study within the SBCS to increase the overall comparability of cases and controls in studying quantitative biomarkers. Preliminary sample size calculations suggested that 350 case control pairs would provide reasonable statistical power. Cases in the ITC substudy were SBCS cases providing a fasting, first-morning urine specimen collected before any cancer treatment or surgery. For each case, a control was selected from the pool of controls completing the study, individually matched to cases by age (±3 year), menopausal status, and date of sample collection and interview (±30 days). Successful matches were completed for 350 case control pairs for urinary isothiocyanate analysis. Thirteen cases were found not to meet study inclusion criteria after we conducted the urinary ITC assays; thus, our analysis included 337 case control pairs. Table 1 compares these 337 cases and 337 controls to cases and controls in the SBCS across established breast cancer risk factors. By chance, ITC substudy cases were 1 year older at menarche, and 6% more controls exercised. Overall, across almost all demographic and reproductive parameters, the ITC substudy participants were generally comparable with the SBCS population.
Analysis of Urinary Isothiocyanates.
The method for analysis of urinary total ITCs and their thiol metabolites described in detail (48) was slightly revised. Briefly, frozen urine samples were thawed and vortexed; 1-ml samples were placed in 2-ml glass vials and centrifuged (2800 rpm) for 15 min to sediment suspended matter, then placed on ice. Triplicate aliquots of 100 μl of clarified urine were carefully pipetted into 2 ml of HPLC vials (Chromacol, Inc., Trumbull, CT) containing 600 μl of a degassed 2-propanol solution of 10 mm 1,2-benzenedithiol (Lancaster Synthesis, Inc., Waldham, NM) and 500 μl of degassed 0.1 m potassium phosphate (pH 8.5). The reaction mixtures in capped vials were vortexed and incubated for 2 h at 65°C in a shaking water bath. The reaction mixtures were cooled and centrifuged (2800 rpm, 20 min) before analysis of the reaction product 1,3-benzendithiol-2-thione by HPLC.
After incubation, samples were analyzed by reverse-phase HPLC using a Waters μBondapak C18 (150 × 3.9 mm) with a Waters C18 guard column and detection wavelength of 365 nm. The mobile phase consisted of a mixture of methanol and H2O (7:3 volume for volume) with a flow rate of 1.75 ml/min. A Shimadzu model SCL-10A controller, dual LEC-10AS pumps, and SIL-10A autosampler (Shimadzu Scientific Instruments, Inc., Columbia, MD) were used; Axxiom 727 software (Axxiom Chromatography, Inc., Moorpark, CA) was used to collect and integrate HPLC data. Concurrent triplicate standards of the N-acetyl conjugate of PEITC-NAC were prepared in 20 mm phosphate buffer (pH 5.0) at concentrations of 0 (H2O), 5, 15, and 25 μm; concurrent standards were analyzed with each batch of urine samples. PEITC-NAC was prepared, and purity and structure were verified by nuclear magnetic resonance and HPLC in the Organic Synthesis Laboratory (53). A standard curve [1–100 μm PEITC-NAC in 20 mm phosphate buffer (pH 5.0), in triplicate], prepared and analyzed weekly, was used for quantification of urinary total ITC concentrations.
All laboratory analyses were blind to the case control status. Each case and matched control were handled as a pair and analyzed on the same day to reduce variability. The interbatch CV [(CV = SD/mean) × 100] was 3.4%. The intrabatch CV across five ITC standards (2, 5, 10, 15, and 100 μm) was 9.64, 6.64, 5.57, 5.11, and 3.84%, respectively.
Determination of Urinary Creatinine.
Urine samples (100 μl) were diluted to 2 ml with deionized H2O, briefly vortexed, then pipetted into a 500-μl sample cup for analysis using a Vitros 500 Clinical Chemistry Analyzer (Johnson & Johnson Clinical Diagnostics, Rochester, NY). High- and low-level human urine creatinine controls were concurrently analyzed to insure that the instrument was performing satisfactorily.
Genotyping Methods.
Genomic DNA was extracted from buffy coat fractions using the Puregene DNA isolation Kit (Gentra Systems, Minneapolis, MN) following the manufacturer’s protocol. DNA concentration was measured by PicoGreen dsDNA Quantitation Kit (Molecular Probes, Eugene, OR). Five to 10 ng of genomic DNA were used for each PCR. The laboratory staff was blind to the identity of the subject. Quality control samples (water, CEPH 1347-02 DNA, as well as blinded and unblinded DNA samples) were included in genotyping assays.
The NQO1 C609T genetic polymorphism, reported to have lower activity (54), was evaluated by the PCR-RFLP method. The primers for the PCR reaction were: F: 5′-TCC TCA GAG TGG CAT TCT GC-3′ and R: 5′-TCT CCT CAT CCT GTA CCT CT-3′. Each PCR product was subjected to HinfI digestion. The C→T substitution at nucleotide 609 creates a HinfI restriction site. The PCR product (230 bp) with C allele was digested to two fragments (195 and 35 bp), whereas the PCR product with T allele was digested to three fragments (151, 44, and 35 bp). A multiplex PCR protocol was used to analyze simultaneously for the presence or absence of GSTM1 and GSTT1 genes (55). The Albumin gene was used as an internal control. Although these assays did not distinguish between heterozygote- and homozygote-positive genotypes, they conclusively identify the null genotypes. The GSTP1 A313G polymorphism Ile105Val is located within the substrate-binding site of GSTP1, and substitution for the G allele is believed to have differential affinity for electrophilic compounds (56). The interaction between this GSTP1 polymorphism and colorectal cancer has been described recently; thus, our analysis will extend GSTP1 function across Asian study populations and cancer sites (57). This polymorphism (58) was determined by PCR-RFLP method following the method reported previously (59).
Statistical Analysis.
Urinary ITC concentrations were standardized to creatinine levels (micromoles of ITC/milliliters of urine/milligrams of creatinine) to adjust for variability in urine volume. A paired t test was used to compare cases and controls across continuously scaled measurements, and conditional logistic regression (ORs and 95% CIs) was used to compare matched cases and controls across categorically scaled breast cancer risk factors. The nonparametric Wilcoxon sign rank test compared urinary ITC levels or Brassica intake across case control pairs while avoiding evaluation of log-transformed data. The Wilcoxon rank-sum test was used to compare urinary ITC levels across Phase II enzyme genotypes, whereas the Kruskal-Wallis test was used to compare ITC levels by stage of breast cancer.
Quartile categories of urinary ITC were determined using the control series distribution. ORs and 95% CIs were calculated using multivariable conditional logistic regression (SAS, version 8.2). Results were almost identical using an unconditional logistic model, with inclusion of matching covariates, and results from the conditional modeling approach are reported. Covariates included as potential confounders include those reproductive, behavioral, genetic, and body size measures found to be associated with breast cancer risk (Table 2). Gene–ITC interactions were evaluated by inserting the corresponding cross-product terms into the model. Statistical significance of the interaction terms was evaluated using the log-likelihood test; however, the multiplicative interaction may be conservative, and thus, we also describe overall patterns of associations between ITC levels and breast cancer across genotypes. Tests for trend between increasing categories of urinary ITC levels and breast cancer were determined by the significance of a continuous variable representing each participant’s ITC category inserted into the logistic model.
RESULTS
In this individually matched case control investigation, premenopausal subjects (64%) averaged 43 years of age (range: 28–56 years), whereas postmenopausal subjects (36%) averaged 56 years of age (range: 40–64 years). Cases had a higher BMI and waist-to-hip ratio and lower activity levels (Table 2). Patterns in reproductive indexes were consistent with greater lifetime estrogen exposure among cases. Few subjects used tobacco [cases: n = 4 (1%); controls: n = 7 (2%)], and only three cases and seven controls worked in an agricultural or rubber/plastics industry. Homozygous deletion of GSTT1, the AA genotype of GSTP1, or the TT genotype of NQO1 were marginally associated with breast cancer. Cases and controls had similar energy, total fat, and soy protein intakes (data not shown).
Cases reported nonsignificantly less habitual Brassica intake (cases: median = 77.4 grams/day; controls: median = 81.5 grams/day; P = 0.16). Consistently, cases had significantly lower urinary ITC levels compared with controls (Table 3). Case control differences in urinary ITC levels were slightly larger with deletion of GSTM1 or GSTT1 or with the G or C alleles of GSTP1 and NQO1, respectively. Within either the cases or controls, urinary ITC levels were fairly stable across categories of each genotype. Urinary ITC levels differed the most with homozygous deletion of GSTT1 among controls; however, this difference was not statistically significant (P = 0.22). Urinary ITC levels did not vary with stage of breast cancer diagnosis [median ITC levels: stage 0/1: 1.7 (n = 82), stage 2A/B: 1.7 (n = 209), stage 3/4: 1.8 (n = 35), and stage unknown: 1.7 (n = 12); P = 0.62].
Habitual Brassica intake estimated by FFQ was not consistently associated with breast cancer (Table 4), although there was a marginally significant protective trend among postmenopausal women attributable to the highest level of self-reported Brassica intake and breast cancer. In contrast, participants categorized to the highest quartile of urinary ITC excretion were 50% less likely to be diagnosed with breast cancer [OR = 0.5, 95% CI (0.3, 0.8); Table 4], with a consistent trend across lower ITC categories. Removal of subjects currently using tobacco or working in agriculture, rubber, or plastics had no effect on these results. Furthermore, adjustment for GSTM1, GSTT1, GSTP1, or NQO1 genotypes did not substantively alter the results. This protective association persisted within pre and postmenopausal women. Further adjustment for Phase II enzyme genotypes did not effect the observed associations among premenopausal women, although trends among the smaller group of postmenopausal women were less stable [ORQ1 = 1 (ref), ORQ2 = 0.6 (0.2, 1.6), ORQ3 = 0.9 (0.3, 2.8), and ORQ4 = 0.5, (0.2, 1.7), adjusted for Phase II genotypes and other covariates].
To investigate the effects of Phase II enzyme activity, the association between ITC and breast cancer was determined for each genotype (Table 5). No statistically significant multiplicative interaction was observed. Trends appeared to be stronger or more consistent within the GSTM1-null, GSTT1-null, GSTP1-AA, and NQO1 C allele genotypes, and higher urinary ITC levels were necessary to observe a protective association among subjects with the G allele of GSTP1 or TT genotype of NQO1. The protective trend with the C allele of NQO was statistically significant among premenopausal women (Ptrend = 0.04).
DISCUSSION
Laboratory research suggests that Brassica consumption reduces breast cancer risk, perhaps through induction of detoxifying Phase II enzymes (60), interaction with estrogen metabolism or the estrogen signaling pathway (29, 30, 61, 62), induction of apoptosis (10, 22, 23, 24, 25, 26), and modified expression of cell cycle regulators (27, 63). However, full-scale epidemiological studies have been challenged to develop an exposure index for Brassica ITCs and indoles. We investigated the association between breast cancer and urinary ITC levels, a biomarker of overall Brassica vegetable consumption and an estimate of exposure to at least one phytochemical group of interest. In our study, greater urinary ITC excretion was associated with lower pre and postmenopausal breast cancer.
Breast carcinogenesis requires years, and a biomarker-breast cancer association requires inference from a single biomarker measurement to a habitual dietary pattern. Although ITCs are excreted within 1–3 days after a single Brassica meal (64), this biomarker may provide an index of habitual Brassica intake because Brassica consumption levels are high and consumed with great frequency in China. Groups with steady-state Brassica consumption would have a steady-state ITC excretion reflecting typical glucosinolate exposure for that group. Seow et al. (19) found habitual Brassica intake, averaging ∼40 grams/day, was favorably associated with urinary ITC levels among Chinese living in Singapore. In our investigation, participants reported consuming an average of 92 grams/day Brassica during the previous 5 years, and almost all urine specimens had detectable ITC level (3% nondetects; 12 cases and 8 controls). We also have found that habitual Brassica intake estimated from our FFQ significantly increased with urinary ITC levels (65), reflecting a traditional diet with strong links to regional agriculture. We could not rule out a possible contribution of side-stream smoke exposure to urinary ITC levels; however, few participants used tobacco, and the insensitivity of the ITC assay to tobacco smoke thiols would further reduce the effects of any tobacco smoke exposure (66). Urine collection protocols were standardized across cases and controls, with the collection of first-morning, fasting, urine specimens, and minimizing case control differences caused by recent intake.
One of the strengths of the urinary ITC dietary biomarker is that it provides an estimate of Brassica consumption independent of recall bias or other potential FFQ reporting errors (67, 68, 69). Most FFQs are not designed to measure a narrowly defined food group such as Brassica. Less frequently consumed vegetables, although potentially potent, may not be on the food list, and it is not possible to have portion-size guides for each food item. Furthermore, urinary ITC levels provide a measure of the internalized exposure to Brassica ITCs and perhaps other Brassica phytochemicals, accounting for variability in glucosinolate levels across species of Brassica, myrosinase-like activity, cooking methods, storage, plant size and age, and weather and soil conditions (42, 70). Our observation of a consistent protective association between urinary ITC levels and breast cancer was in contrast to the weaker associations observed with total Brassica intake, suggesting that the urinary ITC biomarker may indeed be a better measure of habitual Brassica intake in an Asian population.
Phase II enzyme activity may affect breast cancer risk (71, 72) or ITC excretion (19). However, we found no evidence of confounding because of variability in Phase II genotypes between cases and controls. Any associations between Phase II enzyme genotypes and breast cancer were weak, and in contrast to Seow et al. (19), urinary ITC levels were somewhat higher among GSTT1-null subjects. No statistically significant interactions were identified, but there were several patterns to consider. Protective trends were more consistent with the GSTM1 null, GSTT1 null, GSTP1-AA, and NQO1-C genotypes. Phase II enzyme function may contribute to the biological response to Brassica by affecting the transport of phytochemicals to target tissues and the neutralization of electrophilic species (21). Additionally, ITCs transported to breast tissue may induce breast NQO1 expression (73) to neutralize transitory semiquinones from 2- and 4-hydroxy catechol (74, 75, 76). Speculation aside, the nonsignificant patterns observed will require confirmation.
Study results using indole or ITC in animal models of chemoprevention have been mixed. Tumor status was dependent on the ITC or indole dose, congener profile, time of administration, animal model, and tumor-inducing agent (77, 78, 79, 80, 81). In humans, short-term I3C administration did not produce adverse effects (82, 83), although the long-term effects on health remain unknown. In contrast, Brassica consumption is nontoxic, inexpensive, and provides a complex exposure to glucosinolates and nutrients, possibly providing protection against many common cancers (84, 85, 86). Greater overall fruit and vegetable consumption may not be sufficient to reduce breast cancer risk (39). However, there is accumulating evidence that Brassica vegetables hold potential in breast cancer prevention. We found that urinary ITC levels, a glucosinolate biomarker in Asian populations, were significantly associated with reduced breast cancer risk in pre and postmenopausal women.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Supported by NIH Grants RO3CA89845, R01CA64277, and P30 CA68485.
The abbreviations used are: I3C, indole-3-carbinol; ITC, isothiocyante; GST, glutathione S-transferase; HPLC, high-performance liquid chromatography; FFQ, food frequency questionnaire; WHR, waist-to-hip ratio; SBCS, Shanghai Breast Cancer Study; NQO1, NADPH quinine oxidoreductase; BMI, body mass index; CV, coefficient of variation; PEITC, phenethyl isothiocyante; OR, odds ratio; CI, confidence interval.
A. . | Cases . | . | . | Controls . | . | . | ||||
---|---|---|---|---|---|---|---|---|---|---|
. | ITC median (IQR)a . | SBCS median (IQR) . | P . | ITC median (IQR) . | SBCS median (IQR) . | P . | ||||
BMI | 23.2 (4.0) | 23.2 (4.3) | 0.90 | 22.5 (4.3) | 22.8 (4.4) | 0.29 | ||||
WHR | 0.81 (0.07) | 0.81 (0.07) | 0.49 | 0.80 (0.07) | 0.80 (0.07) | 0.45 | ||||
Age (years) | 47 (12) | 47 (11) | 0.81 | 46 (12) | 46 (14) | 0.35 | ||||
Age at menarche (years) | 15 (3) | 14 (3) | 0.03 | 15 (3) | 15 (3) | 0.31 | ||||
Age at first live birth (years)b | 27 (5) | 27 (5) | 0.97 | 26 (4) | 26 (4) | 0.88 | ||||
Age of menopause (years)c | 49 (6) | 49 (5) | 0.72 | 48 (5) | 48 (5) | 0.79 | ||||
Number of children | 1 (1) | 1 (1) | 0.77 | 1 (1) | 1 (1) | 0.55 |
A. . | Cases . | . | . | Controls . | . | . | ||||
---|---|---|---|---|---|---|---|---|---|---|
. | ITC median (IQR)a . | SBCS median (IQR) . | P . | ITC median (IQR) . | SBCS median (IQR) . | P . | ||||
BMI | 23.2 (4.0) | 23.2 (4.3) | 0.90 | 22.5 (4.3) | 22.8 (4.4) | 0.29 | ||||
WHR | 0.81 (0.07) | 0.81 (0.07) | 0.49 | 0.80 (0.07) | 0.80 (0.07) | 0.45 | ||||
Age (years) | 47 (12) | 47 (11) | 0.81 | 46 (12) | 46 (14) | 0.35 | ||||
Age at menarche (years) | 15 (3) | 14 (3) | 0.03 | 15 (3) | 15 (3) | 0.31 | ||||
Age at first live birth (years)b | 27 (5) | 27 (5) | 0.97 | 26 (4) | 26 (4) | 0.88 | ||||
Age of menopause (years)c | 49 (6) | 49 (5) | 0.72 | 48 (5) | 48 (5) | 0.79 | ||||
Number of children | 1 (1) | 1 (1) | 0.77 | 1 (1) | 1 (1) | 0.55 |
B. . | Cases . | . | . | . | . | Controls . | . | . | . | . | ||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
. | n . | % . | n . | % . | P . | n . | % . | n . | % . | P . | ||||||||
Exercise regularly (no) | 71 | 21% | 273 | 19% | 0.32 | 104 | 31% | 392 | 25% | 0.03 | ||||||||
Premenopausal | 219 | 65% | 990 | 64% | 0.95 | 216 | 64% | 990 | 64% | 0.87 | ||||||||
Breast cancer in first-degree relative | 11 | 3% | 54 | 4% | 0.87 | 6 | 2% | 38 | 2% | 0.55 | ||||||||
Ever had breast fibroadenoma | 27 | 8% | 140 | 9% | 0.41 | 15 | 4% | 78 | 5% | 0.67 | ||||||||
Current tobacco user | 4 | 1% | 10 | 1% | 0.31 | 7 | 2% | 36 | 2% | 0.99 | ||||||||
Education | ||||||||||||||||||
No formal education | 16 | 5% | 37 | 3% | 0.67 | 15 | 5% | 67 | 5% | 0.80 | ||||||||
Elementary school | 28 | 8% | 96 | 9% | 29 | 9% | 102 | 9% | ||||||||||
Middle + high school | 248 | 74% | 836 | 75% | 259 | 77% | 915 | 75% | ||||||||||
Professional, college, and above | 45 | 13% | 153 | 13% | 31 | 9% | 138 | 11% |
B. . | Cases . | . | . | . | . | Controls . | . | . | . | . | ||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
. | n . | % . | n . | % . | P . | n . | % . | n . | % . | P . | ||||||||
Exercise regularly (no) | 71 | 21% | 273 | 19% | 0.32 | 104 | 31% | 392 | 25% | 0.03 | ||||||||
Premenopausal | 219 | 65% | 990 | 64% | 0.95 | 216 | 64% | 990 | 64% | 0.87 | ||||||||
Breast cancer in first-degree relative | 11 | 3% | 54 | 4% | 0.87 | 6 | 2% | 38 | 2% | 0.55 | ||||||||
Ever had breast fibroadenoma | 27 | 8% | 140 | 9% | 0.41 | 15 | 4% | 78 | 5% | 0.67 | ||||||||
Current tobacco user | 4 | 1% | 10 | 1% | 0.31 | 7 | 2% | 36 | 2% | 0.99 | ||||||||
Education | ||||||||||||||||||
No formal education | 16 | 5% | 37 | 3% | 0.67 | 15 | 5% | 67 | 5% | 0.80 | ||||||||
Elementary school | 28 | 8% | 96 | 9% | 29 | 9% | 102 | 9% | ||||||||||
Middle + high school | 248 | 74% | 836 | 75% | 259 | 77% | 915 | 75% | ||||||||||
Professional, college, and above | 45 | 13% | 153 | 13% | 31 | 9% | 138 | 11% |
IQR, interquartile range.
Among women with any live births.
Among postmenopausal women only.
. | . | Cases . | . | Controls . | . | OR . | 95% CI . | ||
---|---|---|---|---|---|---|---|---|---|
. | . | n . | % . | n . | % . | . | . | ||
BMI | ≥30 | 15 | 4% | 10 | 3% | 1.6 | 0.7, 3.6 | ||
WHR | ≥0.85 | 72 | 21% | 50 | 15% | 1.6 | 1.1, 2.4 | ||
Leisure activity | No activity | 71 | 21% | 104 | 31% | 1.8 | 1.2, 2.6 | ||
Age at menarche | <11 yrs | 5 | 1% | 1 | <1% | 5.0 | 0.6, 43 | ||
Age at menopausea | ≥54 yrs | 13 | 19% | 11 | 18% | 1.3 | 0.3, 5.9 | ||
Age at first live birthb | >30 yrs | 73 | 22% | 52 | 18% | 1.5 | 1.0, 2.2 | ||
Number of children | ≤1 | 231 | 69% | 219 | 65% | 1.5 | 0.9, 2.4 | ||
Family breast cancer | Yes | 11 | 3% | 6 | 2% | 1.8 | 0.7, 4.9 | ||
Fibroadenoma history | Yes | 27 | 8% | 15 | 4% | 1.9 | 1.0, 3.6 | ||
Current smoker | Yes | 4 | 1% | 7 | 2% | 0.6 | 0.2, 2.0 | ||
GSTM1c | Null (−) | 172 | 53% | 161 | 53% | 1.0 | 0.7, 1.3 | ||
GSTT1 | Null (−) | 151 | 46% | 113 | 38% | 1.4 | 1.0, 1.9 | ||
GSTP1 | GA/GG | 116 | 36% | 87 | 29% | 1.5 | 1.0, 2.0 | ||
NQO1 | TT | 65 | 21% | 38 | 13% | 1.4 | 0.9, 2.3 |
. | . | Cases . | . | Controls . | . | OR . | 95% CI . | ||
---|---|---|---|---|---|---|---|---|---|
. | . | n . | % . | n . | % . | . | . | ||
BMI | ≥30 | 15 | 4% | 10 | 3% | 1.6 | 0.7, 3.6 | ||
WHR | ≥0.85 | 72 | 21% | 50 | 15% | 1.6 | 1.1, 2.4 | ||
Leisure activity | No activity | 71 | 21% | 104 | 31% | 1.8 | 1.2, 2.6 | ||
Age at menarche | <11 yrs | 5 | 1% | 1 | <1% | 5.0 | 0.6, 43 | ||
Age at menopausea | ≥54 yrs | 13 | 19% | 11 | 18% | 1.3 | 0.3, 5.9 | ||
Age at first live birthb | >30 yrs | 73 | 22% | 52 | 18% | 1.5 | 1.0, 2.2 | ||
Number of children | ≤1 | 231 | 69% | 219 | 65% | 1.5 | 0.9, 2.4 | ||
Family breast cancer | Yes | 11 | 3% | 6 | 2% | 1.8 | 0.7, 4.9 | ||
Fibroadenoma history | Yes | 27 | 8% | 15 | 4% | 1.9 | 1.0, 3.6 | ||
Current smoker | Yes | 4 | 1% | 7 | 2% | 0.6 | 0.2, 2.0 | ||
GSTM1c | Null (−) | 172 | 53% | 161 | 53% | 1.0 | 0.7, 1.3 | ||
GSTT1 | Null (−) | 151 | 46% | 113 | 38% | 1.4 | 1.0, 1.9 | ||
GSTP1 | GA/GG | 116 | 36% | 87 | 29% | 1.5 | 1.0, 2.0 | ||
NQO1 | TT | 65 | 21% | 38 | 13% | 1.4 | 0.9, 2.3 |
Postmenopausal women only, <48 yrs of age at menopause as referent.
Age at first birth >30 years or no live births.
Frequencies may be <337 case control pairs because of missing genotype data.
. | . | Cases . | . | Controls . | . | P a . | ||
---|---|---|---|---|---|---|---|---|
. | . | n . | Median . | n . | Median . | . | ||
ITC | 337 | 1.71 | 337 | 2.31 | <0.01 | |||
Geneb | ||||||||
GSTM1 | + | 152 | 1.73 | 141 | 2.31 | 0.24 | ||
− | 172 | 1.71 | 161 | 2.50 | <0.01 | |||
GSTT1 | + | 171 | 1.77 | 187 | 2.31 | 0.15 | ||
− | 151 | 1.65 | 113 | 3.00 | 0.01 | |||
GSTP1 | AA | 208 | 1.76 | 217 | 2.31 | 0.06 | ||
AG/GG | 116 | 1.64 | 87 | 2.42 | 0.04 | |||
NQO1 | CC/CT | 251 | 1.71 | 249 | 2.31 | 0.02 | ||
TT | 65 | 1.68 | 38 | 2.07 | 0.55 |
. | . | Cases . | . | Controls . | . | P a . | ||
---|---|---|---|---|---|---|---|---|
. | . | n . | Median . | n . | Median . | . | ||
ITC | 337 | 1.71 | 337 | 2.31 | <0.01 | |||
Geneb | ||||||||
GSTM1 | + | 152 | 1.73 | 141 | 2.31 | 0.24 | ||
− | 172 | 1.71 | 161 | 2.50 | <0.01 | |||
GSTT1 | + | 171 | 1.77 | 187 | 2.31 | 0.15 | ||
− | 151 | 1.65 | 113 | 3.00 | 0.01 | |||
GSTP1 | AA | 208 | 1.76 | 217 | 2.31 | 0.06 | ||
AG/GG | 116 | 1.64 | 87 | 2.42 | 0.04 | |||
NQO1 | CC/CT | 251 | 1.71 | 249 | 2.31 | 0.02 | ||
TT | 65 | 1.68 | 38 | 2.07 | 0.55 |
P from Wilcoxon’s sign rank test for paired ITC data or rank-sum test for ITC by genotype analysis.
Genotype frequencies may be <337 case control pairs because of missing genotype data.
. | Cases . | Controls . | ORa . | 95% CI . | ORb . | 95% CI . | |
---|---|---|---|---|---|---|---|
ITC | All subjects | ||||||
Q1 | 104 | 87 | 1.0 | 1.0 | |||
Q2 | 97 | 87 | 0.9 | (0.6, 1.3) | 0.9 | (0.6, 1.4) | |
Q3 | 84 | 87 | 0.7 | (0.4, 1.1) | 0.7 | (0.5, 1.1) | |
Q4 | 56 | 87 | 0.5 | (0.3, 0.8) | 0.5 | (0.3, 0.8) | |
Ptrend < 0.01 | Ptrend < 0.01 | ||||||
Postmenopausal subjects | |||||||
Q1 | 36 | 35 | 1.0 | 1.0 | |||
Q2 | 35 | 36 | 0.7 | (0.3, 1.4) | 0.7 | (0.3, 1.7) | |
Q3 | 29 | 28 | 0.7 | (0.3, 1.7) | 0.7 | (0.3, 1.9) | |
Q4 | 22 | 29 | 0.5 | (0.2, 1.3) | 0.6 | (0.2, 1.7) | |
Ptrend = 0.20 | Ptrend = 0.38 | ||||||
Premenopausal subjects | |||||||
Q1 | 68 | 52 | 1.0 | 1.0 | |||
Q2 | 62 | 51 | 0.9 | (0.5, 1.6) | 0.9 | (0.5, 1.7) | |
Q3 | 55 | 59 | 0.7 | (0.4, 1.3) | 0.7 | (0.4, 1.2) | |
Q4 | 34 | 58 | 0.4 | (0.2, 0.8) | 0.5 | (0.2, 0.9) | |
Ptrend = 0.01 | Ptrend = 0.01 | ||||||
Brassica | All subjects | ||||||
Q1 | 90 | 88 | 1.0 | 1.0 | |||
Q2 | 85 | 87 | 0.9 | (0.6, 1.4) | 0.9 | (0.6, 1.5) | |
Q3 | 96 | 86 | 1.1 | (0.7, 1.7) | 1.2 | (0.8, 1.9) | |
Q4 | 70 | 87 | 0.7 | (0.5, 1.2) | 0.8 | (0.5, 1.3) | |
Ptrend = 0.36 | Ptrend = 0.79 | ||||||
Postmenopausal subjects | |||||||
Q1 | 31 | 26 | 1.0 | 1.0 | |||
Q2 | 34 | 30 | 1.0 | (0.4, 2.4) | 1.2 | (0.4, 3.2) | |
Q3 | 32 | 32 | 0.7 | (0.3, 1.8) | 1.2 | (0.4, 3.3) | |
Q4 | 25 | 40 | 0.4 | (0.2, 1.0) | 0.4 | (0.1, 1.1) | |
Ptrend = 0.02 | Ptrend = 0.07 | ||||||
Premenopausal subjects | |||||||
Q1 | 59 | 62 | 1.0 | 1.0 | |||
Q2 | 51 | 57 | 0.8 | (0.5, 1.4) | 0.8 | (0.4, 1.4) | |
Q3 | 64 | 54 | 1.2 | (0.7, 2.0) | 1.2 | (0.7, 2.0) | |
Q4 | 45 | 47 | 0.9 | (0.5, 1.7) | 1.0 | (0.5, 1.8) | |
Ptrend = 0.78 | Ptrend = 0.79 |
. | Cases . | Controls . | ORa . | 95% CI . | ORb . | 95% CI . | |
---|---|---|---|---|---|---|---|
ITC | All subjects | ||||||
Q1 | 104 | 87 | 1.0 | 1.0 | |||
Q2 | 97 | 87 | 0.9 | (0.6, 1.3) | 0.9 | (0.6, 1.4) | |
Q3 | 84 | 87 | 0.7 | (0.4, 1.1) | 0.7 | (0.5, 1.1) | |
Q4 | 56 | 87 | 0.5 | (0.3, 0.8) | 0.5 | (0.3, 0.8) | |
Ptrend < 0.01 | Ptrend < 0.01 | ||||||
Postmenopausal subjects | |||||||
Q1 | 36 | 35 | 1.0 | 1.0 | |||
Q2 | 35 | 36 | 0.7 | (0.3, 1.4) | 0.7 | (0.3, 1.7) | |
Q3 | 29 | 28 | 0.7 | (0.3, 1.7) | 0.7 | (0.3, 1.9) | |
Q4 | 22 | 29 | 0.5 | (0.2, 1.3) | 0.6 | (0.2, 1.7) | |
Ptrend = 0.20 | Ptrend = 0.38 | ||||||
Premenopausal subjects | |||||||
Q1 | 68 | 52 | 1.0 | 1.0 | |||
Q2 | 62 | 51 | 0.9 | (0.5, 1.6) | 0.9 | (0.5, 1.7) | |
Q3 | 55 | 59 | 0.7 | (0.4, 1.3) | 0.7 | (0.4, 1.2) | |
Q4 | 34 | 58 | 0.4 | (0.2, 0.8) | 0.5 | (0.2, 0.9) | |
Ptrend = 0.01 | Ptrend = 0.01 | ||||||
Brassica | All subjects | ||||||
Q1 | 90 | 88 | 1.0 | 1.0 | |||
Q2 | 85 | 87 | 0.9 | (0.6, 1.4) | 0.9 | (0.6, 1.5) | |
Q3 | 96 | 86 | 1.1 | (0.7, 1.7) | 1.2 | (0.8, 1.9) | |
Q4 | 70 | 87 | 0.7 | (0.5, 1.2) | 0.8 | (0.5, 1.3) | |
Ptrend = 0.36 | Ptrend = 0.79 | ||||||
Postmenopausal subjects | |||||||
Q1 | 31 | 26 | 1.0 | 1.0 | |||
Q2 | 34 | 30 | 1.0 | (0.4, 2.4) | 1.2 | (0.4, 3.2) | |
Q3 | 32 | 32 | 0.7 | (0.3, 1.8) | 1.2 | (0.4, 3.3) | |
Q4 | 25 | 40 | 0.4 | (0.2, 1.0) | 0.4 | (0.1, 1.1) | |
Ptrend = 0.02 | Ptrend = 0.07 | ||||||
Premenopausal subjects | |||||||
Q1 | 59 | 62 | 1.0 | 1.0 | |||
Q2 | 51 | 57 | 0.8 | (0.5, 1.4) | 0.8 | (0.4, 1.4) | |
Q3 | 64 | 54 | 1.2 | (0.7, 2.0) | 1.2 | (0.7, 2.0) | |
Q4 | 45 | 47 | 0.9 | (0.5, 1.7) | 1.0 | (0.5, 1.8) | |
Ptrend = 0.78 | Ptrend = 0.79 |
Crude ORs.
Adjusted ORs for soy protein, fibroadenoma, family breast cancer, leisure activity, WHR, BMI, age at menarche, and number of children.
Urinary ITC categories . | . | . | . | . | . | . | |||||
---|---|---|---|---|---|---|---|---|---|---|---|
Gene . | . | Q1 . | Q2 . | Q3 . | Q4 . | P trend . | |||||
GSTM1 | + | 1.0 | 0.9 (0.5, 1.8) | 0.7 (0.3, 1.5) | 0.6 (0.3, 1.4) | 0.20 | |||||
− | 1.0 | 1.0 (0.5, 1.9) | 0.9 (0.4, 1.7) | 0.5 (0.2, 0.9) | 0.05 | ||||||
Pint = 0.82 | |||||||||||
GSTT1 | + | 1.0 | 1.1 (0.6, 1.9) | 0.8 (0.4, 1.5) | 0.6 (0.3, 1.3) | 0.20 | |||||
− | 1.0 | 0.7 (0.3, 1.4) | 0.9 (0.4, 2.1) | 0.4 (0.2, 0.9) | 0.03 | ||||||
Pint = 0.44 | |||||||||||
GSTP1 | AA | 1.0 | 0.8 (0.5, 1.5) | 0.7 (0.4, 1.4) | 0.6 (0.4, 1.3) | 0.18 | |||||
AG/GG | 1.0 | 1.0 (0.4, 2.7) | 1.0 (0.4, 2.5) | 0.5 (0.2, 1.2) | 0.12 | ||||||
Pint = 0.98 | |||||||||||
NQO1 | CC/CT | 1.0 | 0.9 (0.5, 1.5) | 0.8 (0.5, 1.6) | 0.6 (0.3, 1.1) | 0.09 | |||||
TT | 1.0 | 1.7 (0.5, 5.9) | 1.9 (0.6, 6.7) | 0.5 (0.1, 1.8) | 0.51 | ||||||
Pint = 0.43 |
Urinary ITC categories . | . | . | . | . | . | . | |||||
---|---|---|---|---|---|---|---|---|---|---|---|
Gene . | . | Q1 . | Q2 . | Q3 . | Q4 . | P trend . | |||||
GSTM1 | + | 1.0 | 0.9 (0.5, 1.8) | 0.7 (0.3, 1.5) | 0.6 (0.3, 1.4) | 0.20 | |||||
− | 1.0 | 1.0 (0.5, 1.9) | 0.9 (0.4, 1.7) | 0.5 (0.2, 0.9) | 0.05 | ||||||
Pint = 0.82 | |||||||||||
GSTT1 | + | 1.0 | 1.1 (0.6, 1.9) | 0.8 (0.4, 1.5) | 0.6 (0.3, 1.3) | 0.20 | |||||
− | 1.0 | 0.7 (0.3, 1.4) | 0.9 (0.4, 2.1) | 0.4 (0.2, 0.9) | 0.03 | ||||||
Pint = 0.44 | |||||||||||
GSTP1 | AA | 1.0 | 0.8 (0.5, 1.5) | 0.7 (0.4, 1.4) | 0.6 (0.4, 1.3) | 0.18 | |||||
AG/GG | 1.0 | 1.0 (0.4, 2.7) | 1.0 (0.4, 2.5) | 0.5 (0.2, 1.2) | 0.12 | ||||||
Pint = 0.98 | |||||||||||
NQO1 | CC/CT | 1.0 | 0.9 (0.5, 1.5) | 0.8 (0.5, 1.6) | 0.6 (0.3, 1.1) | 0.09 | |||||
TT | 1.0 | 1.7 (0.5, 5.9) | 1.9 (0.6, 6.7) | 0.5 (0.1, 1.8) | 0.51 | ||||||
Pint = 0.43 |
Acknowledgments
We thank Dr. Wanging Wen for consultation in the statistical analysis.