Abstract

Various studies suggest that vitamin D may reduce breast cancer risk. Most studies assessed the effects of dietary intake only, although endogenous production is an important source of vitamin D. Therefore, the measurement of serum 25-hydroxyvitamin D [25(OH)D] better indicates overall vitamin D status. To assess the association of 25(OH)D serum concentrations with post-menopausal breast cancer risk, we used a population-based case–control study in Germany, which recruited incident breast cancer patients aged 50–74 between 2002 and 2005. Information on sociodemographic and breast cancer risk factors was collected by personal interview. For this analysis, we included 1394 cases and 1365 controls, matched on year of birth and time of blood collection. Conditional logistic regression was used to calculate odds ratios (ORs) for breast cancer adjusted for potential confounders. Serum 25(OH)D concentration was significantly inversely associated with post-menopausal breast cancer risk. Compared with the lowest category (<30 nM), OR [95% confidence intervals (CI)] for the higher categories of 25(OH)D (30–45, 45–60, 60–75 and ≥75 nM) were 0.57 (0.45–0.73), 0.49 (0.38–0.64), 0.43 (0.32–0.57) and 0.31 (0.24–0.42), respectively ( Ptrend < 0.0001). Analysis using fractional polynomials indicated a non-linear association. The association was stronger in women never using menopausal hormone therapy (HT) compared with past and current users ( Pinteraction < 0.0001). Our findings strongly suggest a protective effect for post-menopausal breast cancer through a better vitamin D supply as characterized by serum 25(OH)D measurement, with a stronger inverse association in women with low serum 25(OH)D concentrations (<50 nM).

Introduction

Epidemiological and experimental studies suggest an inverse association between vitamin D and cancer of different sites, including breast cancer ( 1–6 ). Most epidemiological studies have assessed the effects of dietary intake only, although endogenous production after sun exposure is the main source of vitamin D ( 6 ). Both vitamin D from diet and endogenous production is converted via two consecutive hydroxylation steps to 25-hydroxyvitamin D [25(OH)D] in the liver and to the biologically active form of vitamin D 1,25-dihydroxyvitamin D [1,25(OH) 2 D] in the kidneys. In addition to its essential role in calcium homeostasis and bone metabolism, 1,25(OH) 2 D has a wide range of non-classical actions that include induction of cell differentiation, inhibition of cell growth and regulation of apoptosis in normal and malignant cells, including human breast cancer cells ( 7–10 ). 1,25(OH) 2 D exerts its anticarcinogenic actions via the vitamin D receptor (VDR) modulating the transcription of target genes such as p21, p27, c-fos and c-myc ( 1 ).

Although not biologically active, there is overall agreement that 25(OH)D is the appropriate biomarker with which to measure vitamin D status in humans. In contrast to 1,25(OH) 2 D, 25(OH)D is not tightly regulated and better reflects vitamin D stores obtained from both sunlight exposure and ingested vitamin D over longer periods ( 11–14 ). Recent studies have demonstrated that conversion to the biologically active form of vitamin D 1,25(OH) 2 D also occurs in extrarenal tissues such as colon, prostate and breast ( 15–17 ). In addition, a recent cell study reported expression of CYP27B1, the enzyme that converts 25(OH)D to active 1,25(OH) 2 D, in mammary cells as well as growth inhibition of mammary cells by 25(OH)D, thus linking vitamin D status to breast cancer risk ( 18 ).

Data from epidemiologic studies assessing the association between dietary intake of vitamin D and breast cancer risk have been inconclusive ( 5 , 19–25 ). Thus far, only four studies have assessed the relationship between serum vitamin D metabolites [25(OH)D or 1,25(OH) 2 D)] and breast cancer risk ( 2 , 4 , 26 , 27 ). Of two small hospital-based case–control studies, one found a significantly inverse association for 25(OH)D and breast cancer risk ( 4 ), whereas the other study revealed no association for 25(OH)D but an inverse association for 1,25(OH) 2 D ( 27 ). A case–control study nested in a voluntary health check up showed no association between 1,25(OH) 2 D and breast cancer risk ( 26 ). To date, the largest case–control study including 701 cases nested in the Nurses’ Health Study (NHS) ( 2 ) reported a non-significant decrease in breast cancer risk both with higher serum 25(OH)D and with 1,25(OH) 2 D concentrations.

To our knowledge, the present study including 1394 cases is by far the largest population-based case–control study assessing the relationship of serum 25(OH)D concentrations and breast cancer risk in post-menopausal women. Because estrogens are known to increase concentrations of vitamin D-binding protein, VDR and serum 25(OH)D ( 28–34 ), we additionally examined effect modification by risk factors including hormone-related variables such as menopausal hormone therapy (HT) and number of pregnancies.

Materials and methods

Study population and data collection

We used data from a large population-based case–control study (MARIE study, Mamma Carcinoma Risk factor Investigation) carried out in two regions in Germany, the city of Hamburg and the Rhein–Neckar–Karlsruhe (R-N-K) region. The study was approved by the ethics committees of both the University of Heidelberg and the University of Hamburg and conducted in accordance with the Declaration of Helsinki. All study participants gave informed consent. Cases were eligible if they had a histologically confirmed primary invasive or in situ breast cancer diagnosed between 01 January 2001 and 30 September 2005 in Hamburg and between 01 August 2002 and 31 July 2005 in the R-N-K region. Further inclusion criteria were age between 50 and 74 and being a resident of one of the study regions. Cases were identified through frequent monitoring of hospital admissions, surgery schedules and pathology records. Clinical and pathological characteristics of the patients were abstracted from hospital and pathology records. Of the 5970 eligible patients who could be contacted, 3919 (65.6%) participated, whereas 2051 (34.4%) declined participation or did not respond to invitation letters.

Two controls per case were randomly selected from lists of residents provided by population registries and frequency matched by year of birth and study region to the cases. Of the 17 093 controls who met the inclusion criteria, 7421 (43.4%) participated, 7521 (44.0%) refused to participate and 2151 (12.6%) did not respond.

All participants were interviewed by trained personnel to obtain information on sociodemographic factors, anthropometric measures, data on lifetime HT exposure, including information on timing and duration, type, dose and brand name of HT, and other potential breast cancer risk factors. An alphabetical list with photographic depictions of >300 HT products prescribed in Germany over the past 35 years was presented to women as a memory aid recalling their HT history. In addition, participants reported their dietary habits during the past 12 months with a self-administered validated 176 items’ food frequency questionnaire ( 35 ). Nutrient intake was calculated with the German food composition table BLS II.3 (Bundesinstitut für Gesundheitlichen Verbraucherschutz und Veterinärmedizin, Berlin, Germany).

Measurement of 25(OH)D

For quantification of 25(OH)D in serum, we used the OCTEIA 25(OH)D enzyme immunoassay (Immunodiagnostic Systems Limited, Boldon, UK). Serum samples were stored in aliquots at −80°C until measurement. Samples were analyzed in a single batch between November 2006 and January 2007. The coefficient of variation was 3.4% for intraassay determination and 7.6% for interassay determination. Recovery was analyzed by spiking a high with a low serum concentration sample (98.7 and 26.3 nM, respectively) and comparing the observed with the expected values. When spiking the high- and low-concentration samples in a 1:1, 3:1 and 1:3 ratio, recovery was 105, 99 and 96%, respectively. We measured 235 random samples (8.5%) in duplicate. The average absolute deviation from the mean between two duplicates was 2.2%.

Data analysis

Women were defined as post-menopausal if they reported their last natural menstrual bleeding at least 12 months before the reference date, a bilateral oophorectomy or cessation of menses due to radiation or chemotherapy. In addition, women >55 years with unclear menopausal status due to hysterectomy or hormone use were also considered post-menopausal (since the 90th percentile for age at menopause for women with natural menopause was 55 years). Women under age 55 with unclear menopausal status and women who were clearly pre-menopausal were excluded from the analysis. A total of 3464 invasive or in situ breast cancer cases and 6657 controls, of which 1559 cases and 3008 controls came from the R-N-K region, were classified as post-menopausal. For this analysis, we identified 1400 post-menopausal cases from the R-N-K region for whom we had blood samples (90% of the cases from the study region) and randomly selected 1400 post-menopausal controls from the same study region with matching on time of blood collection in four categories (January–March, April–June, July–September and October–December) and year of birth (continuous). After exclusion of hemolytic serum samples, our final study population consisted of 1394 post-menopausal cases and 1365 post-menopausal controls.

We assessed the association of serum 25(OH)D and post-menopausal breast cancer risk by means of logistic regression with stratification by year of birth (continuous) and time of blood collection (four categories). Odds ratio (OR) estimates and their 95% confidence intervals (CIs) were calculated assessing 25(OH)D concentration both as continuous (per 10 nM increment) and as categorical variable divided into five categories (<30, 30–45, 45–60, 60–75, ≥75 nM). This categorization closely corresponds to the classification into quintiles according to the distribution in the control group. We present OR estimates adjusted for age at menopause (<47, 47–51, 52–55, ≥56, unknown), body mass index (BMI) (<22.5, 22.5–25, 25–30, ≥30 kg/m2), education level (low, middle, high), first-degree family history of breast cancer (yes, no, unknown), history of benign breast disease (yes, no), number of pregnancies (≥28th week) (0, 1, 2, ≥3), age at menarche (<12, 12–14, ≥15), breastfeeding history (ever, never), total number of mammograms (0, 1–4, 5–9, ≥10, unknown), smoking status (never, past, current) and use of menopausal HT (never, past, current). Women reporting HT use for ≤3 months before the reference date were considered ‘never’ users and women with >3 months were considered ‘ever’ users. Among the ever users, women were defined as ‘current’ users of HT if they reported use within 6 months before the reference date; otherwise they were defined as ‘past’ users. Adjustment on type of menopause, type of HT, age at first birth, alcohol consumption, contraceptive use, physical activity, dietary intake of calcium and dietary intake of vitamin D did not affect the OR estimates substantially and were therefore not included in the fully adjusted models. Test for linear trend was performed using the median values in each category as an ordinal variable. To further examine dose–response relation and non-linearity of the log OR function for predictive 25(OH)D concentrations, we used the method of fractional polynomials ( 36 ). The continuous 25(OH)D variable was entered into the multivariate logistic regression model via a set of defined transformations [ x−2 , x−1 , x−0.5 , x0.5 , x2 , x3 and log( x )], allowing a maximum of two terms (including the untransformed variable) in the model. The function that best fitted the data was selected on the basis of the −2 log likelihood of the respective model. We additionally fitted the multivariate logistic regression models with restricted cubic splines for 25(OH)D concentration treated as a continuous variable ( 37 ). We specified four knot points at 25(OH)D concentrations of 9.7, 36, 51.5 and 70.7 representing the minimum, the 25th, 50th and 75th percentiles in the control group, respectively. Other knot points were also specified but did not change the shape of the OR function.

We assessed effect modification by testing for multiplicative interaction. We included an interaction term of the continuous variable of interest [25(OH)D] and potential interaction variables in the fully adjusted model and evaluated statistical significance with the likelihood ratio test. Differences in cases and controls were evaluated with the χ 2 test. All tests were two sided and considered to be statistically significant if P ≤ 0.05. All statistical analyses were carried out using the statistical software SAS 9.1 (SAS Institute, Cary, NC).

Results

The mean age of cases and controls was 63.6 and 63.5 years, respectively. Characteristics, including sociodemographic variables and main risk factors for breast cancer, of the 1394 post-menopausal cases and 1365 post-menopausal controls are shown in Table I . Cases and controls differed significantly with respect to age at menopause, family history of breast cancer in at least one first-degree relative, benign breast disease history, number of pregnancies, breastfeeding history, age at menarche, number of mammograms and use of menopausal HT ( Table I ).

Table I.

Characteristics and risk factors for post-menopausal breast cancer in cases and matched controls in the study population

Characteristics  Cases ( N  = 1394)   Controls ( N  = 1365)  P
 n n  
Age at diagnosis/recruitment (years) 
    50–54 91 6.5 86 6.3 0.84 
    55–59 300 21.5 272 19.9  
    60–64 447 32.1 457 33.5  
    65–69 380 27.3 372 27.3  
    ≥70 176 12.6 178 13.0  
BMI (kg/m 2 )  
    <22.5 544 39.0 536 39.3 0.75 
    22.5–25 470 33.7 436 32.0  
    25–30 322 23.1 333 24.4  
    ≥30 58 4.2 659 4.3  
    Missing values     
Educational level 
    Low 905 64.9 923 67.6 0.12 
    Middle 299 21.5 290 21.3  
    High 190 13.6 152 11.1  
Age at menopause (years) 
    <47 146 10.5 210 15.4 <0.01 
    47–51 399 28.6 384 28.1  
    52–55 243 17.4 196 14.4  
    ≥56 63 4.5 65 4.8  
    Unknown 543 39.0 510 37.3  
First-degree family history of breast cancer 
    No 1102 79.1 1118 81.9 0.01 
    Yes 232 16.6 181 13.3  
    Unknown 60 4.3 66 4.8  
Benign breast disease 
    No 848 61.0 976 72.8 <0.01 
    Yes 542 39.0 383 27.2  
    Missing values    
Number of pregnancies (≥28th week) 
    0 199 14.3 163 11.9 <0.01 
    1 377 27.0 311 22.8  
    2 514 36.9 555 40.7  
    ≥3 304 21.8 336 24.6  
Age at menarche (years) 
    <12 126 9.0 120 8.8 0.04 
    12–14 938 67.3 862 63.3  
    ≥15 330 23.7 380 27.9  
    Missing values     
Ever breastfeeding 
    No 550 39.5 457 33.5 <0.01 
    Yes 844 60.5 908 66.5  
Number of mammograms in total 
    0 192 13.8 175 12.8 <0.01 
    1–4 590 42.3 720 52.8  
    5–9 344 24.7 317 23.2  
    ≥10 250 17.9 145 10.6  
    Unknown number 18 1.3 0.6  
Use of HT 
    Never 524 38.0 595 44.0 <0.01 
    Past 288 20.9 346 25.6  
    Current (≤6 months) 567 41.1 411 30.4  
    Missing values 15  13   
Smoking 
    Never 876 62.8 806 59.1 0.05 
    Past 317 22.8 365 26.7  
    Current 201 14.4 194 14.2  
Alcohol consumption (g/day) 
    0 247 17.7 222 16.3 0.51 
    >0–18 977 70.1 983 72.0  
    ≥19 170 12.2 160 11.7  
Physical activity in quintiles (MET; h/week) b 
    <145.3 281 20.3 273 20 0.41 
    145.3–179.3 293 21.1 272 20  
    179.4–213.1 303 21.8 273 20  
    213.2–257.0 243 17.5 273 20  
    ≥257.1 268 19.3 272 20  
    Missing values    
Hormonal receptor status of the tumor c 
    ER 
    Positive 984 76.4    
    Negative 304 23.6    
    PR 
    Positive 842 65.4    
    Negative 445 34.6    
Time of blood collection 
    January–March 319 22.9 311 22.8 0.99 
    April–June 292 21.0 286 21.0  
    July–September 402 28.8 392 28.7  
    October–December 381 27.3 376 27.5  
Characteristics  Cases ( N  = 1394)   Controls ( N  = 1365)  P
 n n  
Age at diagnosis/recruitment (years) 
    50–54 91 6.5 86 6.3 0.84 
    55–59 300 21.5 272 19.9  
    60–64 447 32.1 457 33.5  
    65–69 380 27.3 372 27.3  
    ≥70 176 12.6 178 13.0  
BMI (kg/m 2 )  
    <22.5 544 39.0 536 39.3 0.75 
    22.5–25 470 33.7 436 32.0  
    25–30 322 23.1 333 24.4  
    ≥30 58 4.2 659 4.3  
    Missing values     
Educational level 
    Low 905 64.9 923 67.6 0.12 
    Middle 299 21.5 290 21.3  
    High 190 13.6 152 11.1  
Age at menopause (years) 
    <47 146 10.5 210 15.4 <0.01 
    47–51 399 28.6 384 28.1  
    52–55 243 17.4 196 14.4  
    ≥56 63 4.5 65 4.8  
    Unknown 543 39.0 510 37.3  
First-degree family history of breast cancer 
    No 1102 79.1 1118 81.9 0.01 
    Yes 232 16.6 181 13.3  
    Unknown 60 4.3 66 4.8  
Benign breast disease 
    No 848 61.0 976 72.8 <0.01 
    Yes 542 39.0 383 27.2  
    Missing values    
Number of pregnancies (≥28th week) 
    0 199 14.3 163 11.9 <0.01 
    1 377 27.0 311 22.8  
    2 514 36.9 555 40.7  
    ≥3 304 21.8 336 24.6  
Age at menarche (years) 
    <12 126 9.0 120 8.8 0.04 
    12–14 938 67.3 862 63.3  
    ≥15 330 23.7 380 27.9  
    Missing values     
Ever breastfeeding 
    No 550 39.5 457 33.5 <0.01 
    Yes 844 60.5 908 66.5  
Number of mammograms in total 
    0 192 13.8 175 12.8 <0.01 
    1–4 590 42.3 720 52.8  
    5–9 344 24.7 317 23.2  
    ≥10 250 17.9 145 10.6  
    Unknown number 18 1.3 0.6  
Use of HT 
    Never 524 38.0 595 44.0 <0.01 
    Past 288 20.9 346 25.6  
    Current (≤6 months) 567 41.1 411 30.4  
    Missing values 15  13   
Smoking 
    Never 876 62.8 806 59.1 0.05 
    Past 317 22.8 365 26.7  
    Current 201 14.4 194 14.2  
Alcohol consumption (g/day) 
    0 247 17.7 222 16.3 0.51 
    >0–18 977 70.1 983 72.0  
    ≥19 170 12.2 160 11.7  
Physical activity in quintiles (MET; h/week) b 
    <145.3 281 20.3 273 20 0.41 
    145.3–179.3 293 21.1 272 20  
    179.4–213.1 303 21.8 273 20  
    213.2–257.0 243 17.5 273 20  
    ≥257.1 268 19.3 272 20  
    Missing values    
Hormonal receptor status of the tumor c 
    ER 
    Positive 984 76.4    
    Negative 304 23.6    
    PR 
    Positive 842 65.4    
    Negative 445 34.6    
Time of blood collection 
    January–March 319 22.9 311 22.8 0.99 
    April–June 292 21.0 286 21.0  
    July–September 402 28.8 392 28.7  
    October–December 381 27.3 376 27.5  
a

χ 2 test.

b

MET, metabolic equivalents.

c

Data on ER and PR status were available for 1288 and 1287 invasive tumor cases ( in situ tumors excluded), respectively.

Median serum concentration of 25(OH)D was 44.9 and 51.5 nM in cases and controls, respectively ( P < 0.0001). We found a significant inverse association between serum concentration of 25(OH)D and risk of post-menopausal breast cancer. Compared with the lowest category (<30 nM), the ORs (95% CI) for higher serum concentrations (30–45, 45–60, 60–75 and ≥75 nM) were 0.57 (0.45–0.73), 0.49 (0.38–0.64), 0.43 (0.32–0.57) and 0.31 (0.24–0.42), respectively ( Ptrend < 0.0001) ( Table II ). We additionally analyzed 25(OH)D as a continuous variable in the multivariate model and found a significantly reduced risk for post-menopausal breast cancer of 0.88 (0.85–0.91) per 10 nM increment in serum 25(OH)D ( Table II ). We further examined the shape of the risk function using fractional polynomials and found a non-linear association between the log OR and the 25(OH)D concentration ( Pnon-linearity = 0.0002). The model with the polynomial ( x + 1) −0.5 best fitted the data. The resulting function, OR( x ) = exp[10.4 × ( x + 1) −0.5 ], taking the median value of the reference category from the categorical analysis [24.3 nM 25(OH)D in the controls] as the reference, is displayed in Figure 1 . Our data suggest a more pronounced inverse association between 25(OH)D and breast cancer risk in the lower concentration range [<50 nM 25(OH)D] and a flattening of the risk function with increasing 25(OH)D concentrations. There was a good fit between the curve and the OR estimates of the categorical analysis. Using restricted cubic spline regression, the shape of the risk function was similar, but model fit was inferior compared with the model using fractional polynomials.

Table II.

ORs for post-menopausal breast cancer by serum 25(OH)D concentration

Serum 25(OH)D (nM) Cases Controls  Crude model a  Adjusted model b 
n n OR (95% CI) OR (95% CI) 
Categorized 
    <30 345 24.8 218 16.0 
    30–45 354 25.4 327 23.9 0.65 (0.52–0.82) 0.57 (0.45–0.73) 
    45–60 300 21.5 308 22.6 0.56 (0.44–0.71) 0.49 (0.38–0.64) 
    60–75 186 13.3 218 16.0 0.49 (0.37–0.64) 0.43 (0.32–0.57) 
    ≥75 209 15.0 294 21.5 0.39 (0.30–0.50) 0.31 (0.24–0.42) 
Ptrend     <0.0001 <0.0001 
As continuous variable 
    Per 10 nM increment 1394  1365  0.89 (0.87–0.92) 0.88 (0.85–0.91) 
Serum 25(OH)D (nM) Cases Controls  Crude model a  Adjusted model b 
n n OR (95% CI) OR (95% CI) 
Categorized 
    <30 345 24.8 218 16.0 
    30–45 354 25.4 327 23.9 0.65 (0.52–0.82) 0.57 (0.45–0.73) 
    45–60 300 21.5 308 22.6 0.56 (0.44–0.71) 0.49 (0.38–0.64) 
    60–75 186 13.3 218 16.0 0.49 (0.37–0.64) 0.43 (0.32–0.57) 
    ≥75 209 15.0 294 21.5 0.39 (0.30–0.50) 0.31 (0.24–0.42) 
Ptrend     <0.0001 <0.0001 
As continuous variable 
    Per 10 nM increment 1394  1365  0.89 (0.87–0.92) 0.88 (0.85–0.91) 
a

Conditional logistic regression stratified by time of blood collection and year of birth.

b

Conditional logistic regression stratified by time of blood collection and year of birth adjusted for age at menopause, first-degree family history of breast cancer, history of benign breast disease, number of pregnancies (≥28th week), age at menarche, breastfeeding history, total number of mammograms, use of HT, BMI, education level, smoking status; due to missing values, 40 observations were not included in the adjusted model.

Fig. 1.

ORs for post-menopausal breast cancer by 25(OH)D concentrations using fractional polynomial as dose–response analysis. The resulting function OR( x ) = exp[10.4 × ( x + 1) −0.5 ] is displayed, setting the median of the lowest category from the categorical analysis as reference [24.3 nM 25(OH)D]. The solid squares and respective bars represent the ORs and 95% CIs of the categorical analysis (30–45, 45–60, 60–75 and ≥75 nM). For graphical illustration categorical OR estimates were displayed at the median value in the controls of each category, at 24.3 (reference), 38.0, 52.2, 67.1 and 88.7 nM, respectively.

Fig. 1.

ORs for post-menopausal breast cancer by 25(OH)D concentrations using fractional polynomial as dose–response analysis. The resulting function OR( x ) = exp[10.4 × ( x + 1) −0.5 ] is displayed, setting the median of the lowest category from the categorical analysis as reference [24.3 nM 25(OH)D]. The solid squares and respective bars represent the ORs and 95% CIs of the categorical analysis (30–45, 45–60, 60–75 and ≥75 nM). For graphical illustration categorical OR estimates were displayed at the median value in the controls of each category, at 24.3 (reference), 38.0, 52.2, 67.1 and 88.7 nM, respectively.

Following previous reports on the interaction between the estrogen and the vitamin D endocrine system, we evaluated effect modification by use of HT. Median serum 25(OH)D concentrations were higher in women currently using HT compared with past or never users. However, these differences were more prominent in cases than in controls ( Table III ). In women who reported never having used HT, breast cancer risk reduction per 10 nM increment in 25(OH)D was greater [OR (95% CI) = 0.78 (0.73–0.83)] than in past and current hormone users [OR (95% CI) = 0.86 (0.79–0.94) and 0.96 (0.91–1.02), respectively] ( Table III ). The interaction between 25(OH)D concentration and use of HT was statistically significant ( Pinteraction < 0.0001) and independent of the type of HT used (estrogen-only or combined estrogen–progestin therapy). The effect modification by HT use also did not differ by estrogen receptor (ER) or progesterone receptor (PR) status of the tumor.

Table III.

ORs a for post-menopausal breast cancer by serum 25(OH)D concentration by use of HT

Serum 25(OH)D (nM) Never HT use Past HT use Current HT use 
 Ca/Co OR (95% CI) Ca/Co OR (95% CI) Ca/Co OR (95% CI) 
Median 37.3/47.6  45.8/53.4  51.5/54.0  
Categorized n  n  n  
    <30 193/125 58/46 89/44 
    30–45 130/149 0.51 (0.36–0.74) 82/71 0.62 (0.33–1.14) 140/105 0.57 (0.35–0.92) 
    45–60 101/123 0.45 (0.30–0.68) 59/83 0.36 (0.19–0.68) 136/98 0.61 (0.37–1.01) 
    60–75 56/81 0.30 (0.19–0.49) 48/72 0.35 (0.18–0.69) 80/63 0.60 (0.34–1.05) 
    ≥75 44/117 0.18 (0.11–0.30) 41/74 0.25 (0.12–0.51) 122/101 0.49 (0.29–0.83) 
     Ptrend  <0.0001  0.0001  0.05 
As continuous variable b 
    Per 10 nM increment 524/595 0.78 (0.73–0.83) 288/346 0.86 (0.79–0.94) 567/411 0.96 (0.91–1.02) 
Serum 25(OH)D (nM) Never HT use Past HT use Current HT use 
 Ca/Co OR (95% CI) Ca/Co OR (95% CI) Ca/Co OR (95% CI) 
Median 37.3/47.6  45.8/53.4  51.5/54.0  
Categorized n  n  n  
    <30 193/125 58/46 89/44 
    30–45 130/149 0.51 (0.36–0.74) 82/71 0.62 (0.33–1.14) 140/105 0.57 (0.35–0.92) 
    45–60 101/123 0.45 (0.30–0.68) 59/83 0.36 (0.19–0.68) 136/98 0.61 (0.37–1.01) 
    60–75 56/81 0.30 (0.19–0.49) 48/72 0.35 (0.18–0.69) 80/63 0.60 (0.34–1.05) 
    ≥75 44/117 0.18 (0.11–0.30) 41/74 0.25 (0.12–0.51) 122/101 0.49 (0.29–0.83) 
     Ptrend  <0.0001  0.0001  0.05 
As continuous variable b 
    Per 10 nM increment 524/595 0.78 (0.73–0.83) 288/346 0.86 (0.79–0.94) 567/411 0.96 (0.91–1.02) 

Ca, cases; Co, controls.

a

Conditional logistic regression stratified by time of blood collection and year of birth adjusted for age at menopause, first-degree family history of breast cancer, history of benign breast disease, number of pregnancies (≥28th week), age at menarche, breastfeeding history, total number of mammograms, BMI, education level, smoking status; due to missing values, 40 observations were not included in the model.

b

Pinteraction < 0.0001.

The effect of 25(OH)D on breast cancer risk was also modified by number of pregnancies ( Pinteraction = 0.004). The OR (95% CI) for breast cancer per 10 nM increment in 25(OH)D in women with less than two pregnancies was 0.91 (0.87–0.96), while the association was stronger in women with two [0.86 (0.82–0.92)] and three or more pregnancies [0.81 (0.73–0.89)] ( Table IV ).

Table IV.

ORs a for post-menopausal breast cancer by serum 25(OH)D by number of pregnancies

Serum 25(OH)D (nM) 0–1 pregnancies 2 pregnancies ≥3 pregnancies 
 Ca/Co OR (95% CI) Ca/Co OR (95% CI) Ca/Co OR (95% CI) 
Median 45.9/50.1  45.2/53.6  42.3/49.7  
Categorized n  n  n  
    <30 139/76 119/82 87/60 
    30–45 136/119 0.42 (0.27–0.65) 134/123 0.72 (0.47–1.08) 84/85 0.51 (0.29–0.88) 
    45–60 122/107 0.40 (0.26–0.63) 104/127 0.50 (0.33–0.78) 74/74 0.46 (0.25–0.82) 
    60–75 75/70 0.39 (0.24–0.65) 76/88 0.51 (0.31–0.82) 35/60 0.28 (0.14–0.55) 
    ≥75 104/102 0.36 (0.22–0.58) 81/135 0.31 (0.19–0.49) 24/57 0.23 (0.11–0.47) 
     Ptrend  0.0006  <0.0001  <0.0001 
As continuous variable b 
    Per 10 nM increment 576/474 0.91 (0.87–0.96) 514/555 0.86 (0.82–0.92) 304/336 0.81(0.73–0.89) 
Serum 25(OH)D (nM) 0–1 pregnancies 2 pregnancies ≥3 pregnancies 
 Ca/Co OR (95% CI) Ca/Co OR (95% CI) Ca/Co OR (95% CI) 
Median 45.9/50.1  45.2/53.6  42.3/49.7  
Categorized n  n  n  
    <30 139/76 119/82 87/60 
    30–45 136/119 0.42 (0.27–0.65) 134/123 0.72 (0.47–1.08) 84/85 0.51 (0.29–0.88) 
    45–60 122/107 0.40 (0.26–0.63) 104/127 0.50 (0.33–0.78) 74/74 0.46 (0.25–0.82) 
    60–75 75/70 0.39 (0.24–0.65) 76/88 0.51 (0.31–0.82) 35/60 0.28 (0.14–0.55) 
    ≥75 104/102 0.36 (0.22–0.58) 81/135 0.31 (0.19–0.49) 24/57 0.23 (0.11–0.47) 
     Ptrend  0.0006  <0.0001  <0.0001 
As continuous variable b 
    Per 10 nM increment 576/474 0.91 (0.87–0.96) 514/555 0.86 (0.82–0.92) 304/336 0.81(0.73–0.89) 

Ca, cases; Co, controls.

a

Conditional logistic regression stratified by time of blood collection and year of birth adjusted for age at menopause, first-degree family history of breast cancer, history of benign breast disease, age at menarche, breastfeeding history, total number of mammograms, use of HT, BMI, education level, smoking status; due to missing values, 40 observations were not included in the model.

b

Pinteraction = 0.004.

The association between 25(OH)D concentration and breast cancer risk was not significantly modified by age, smoking status, age at menarche, age at menopause, BMI, alcohol, family history of breast cancer, breastfeeding history, history of benign breast disease, physical activity or dietary intake of calcium. In addition, no modification by ER or PR status of the tumor or combinations of both was observed. We further examined the association of serum 25(OH)D and breast cancer risk within strata of dietary vitamin D intake. The association remained inverse and no statistical interaction was observed. In addition, dietary intake was not correlated with serum 25(OH)D ( r = 0.03, P = 0.14).

Median difference between time of diagnosis and time of blood collection in cases was 66 days. In a sensitivity analysis, we evaluated potential effects of diagnosis or cancer therapy on circulating 25(OH)D concentration by excluding patients close to diagnosis. Compared with the lowest category (<30 nM), the ORs (95% CI) for the highest category (≥75 nM) were 0.31 (0.24–0.42) for all cases and 0.32 (0.23–0.43) and 0.39 (0.26–0.58), respectively, when excluding cases with <15 days (26% of all cases) and cases with <6 months (63% of all cases) between time of diagnosis and time of blood collection.

Discussion

In this population-based case–control study, the concentration of serum 25(OH)D was significantly inversely associated with the risk of post-menopausal breast cancer. To date, only three studies have assessed the association of breast cancer risk and 25(OH)D ( 2 , 4 , 27 ). Our findings support the recently reported results from a case–control study nested in the NHS ( 2 ). The authors found an inverse, although not significant, association between breast cancer risk and both 25(OH)D and 1,25(OH) 2 D metabolites restricted to post-menopausal women only. Lowe et al. ( 4 ) also observed a significant risk reduction of breast cancer in the highest category of 25(OH)D compared with the lowest in predominantly post-menopausal women. In contrast, Janowsky et al. ( 27 ) reported no association of breast cancer with serum 25(OH)D but with 1,25(OH) 2 D levels. However, the authors included both pre- and post-menopausal women and used hospital-based controls, who may have had less sun exposure than population-based controls. This might explain the conflicting results for 25(OH)D and 1,25(OH) 2 D as less sun exposure has a greater impact on 25(OH)D status than on the more tightly regulated 1,25(OH) 2 D concentration ( 14 ). Recently, a case–control study found lower levels of 1,25(OH) 2 D but not 25(OH)D in breast cancer patients ( 38 ). However, this study was very small and no risk estimates were reported.

Results regarding the association between vitamin D metabolites and breast cancer risk are not easy to compare because of different metabolite levels in the study populations and reference groups used for calculating relative risks. In our study, median serum 25(OH)D concentrations of 44.9 and 51.5 nM for cases and controls, respectively, were comparable with recently published data from the German National Health Survey for women in this age range ( 39 ). However, 25(OH)D concentrations were relatively low compared with other studies in the USA and the UK reporting mean concentrations of >80 nM ( 2 , 4 ). We therefore repeated our analysis calculating OR estimates for breast cancer risk comparing women with concentrations >100 nM 25(OH)D to women with concentrations <50 nM 25(OH)D, which is comparable with the categorization in the NHS ( 2 ). The observed inverse association was considerably stronger in our study population compared with that found in the NHS [OR (95% CI) = 0.44 (0.30–0.65) and 0.73 (0.49–1.07), respectively].

There is ample evidence from cellular and animal studies linking 25(OH)D to breast cancer. This includes the known anticarcinogenic effects of vitamin D with regard to apoptosis, cell differentiation and proliferation, growth inhibition of human mammary epithelial cells by both 1,25(OH) 2 D and 25(OH)D, the presence and expression of the VDR and CYP27B1, the enzyme that converts 25(OH)D to active 1,25(OH) 2 D, in mammary cells, and the uptake of the vitamin D-binding protein–25(OH)D complex in mammary cells in vitro ( 40 ). However, little is known about the delivery of 25(OH)D to mammary cells and therefore further investigations into the mechanisms of delivery, uptake and action of 25(OH)D in mammary cells are needed.

Our findings of a non-linear relationship between 25(OH)D concentrations and the log OR for breast cancer using fractional polynomials are of special interest. In our study population, the protective effect was strongest in women with low 25(OH)D concentrations (<50 nM) and was reduced substantially with higher 25(OH)D concentration. These are important public health findings for women in low 25(OH)D concentration ranges and indicate subgroups where vitamin D supplementation may be most relevant for breast cancer prevention. Due to the low 25(OH)D concentrations in our study population, we were able to provide important additional information to further understand the dose–response relationship of vitamin D and breast cancer risk.

We found a statistically significant modification of the association of breast cancer risk and serum 25(OH)D concentrations with a stronger effect in women never having used HT as compared with HT users. Accordingly, results from the NHS showed some evidence for a stronger effect in non-HT users, although power was low for their comparisons ( 2 ). In women using HT, taking oral contraceptives or with increased endogenous estrogen levels, higher concentrations of vitamin D metabolites have been reported ( 32 , 33 , 41 , 42 ). The underlying mechanisms include the estrogen-induced activation of renal 1α-hydroxylase, inhibition of 24-vitamin D hydroxylase and the up-regulation of the VDR resulting in increased serum concentrations and activity of 25(OH)D and 1,25(OH) 2 D ( 33 , 41 , 43 , 44 ). We also found higher 25(OH)D serum concentrations in women currently using HT compared with past or never users, and this difference was not attributable to differences in age, BMI, physical activity or other known breast cancer risk factors. Serum 25(OH)D concentrations in women never using HT were significantly lower in cases than in controls ( P < 0.0001), whereas no significant case–control differences were observed in women currently using HT. We hypothesize that HT may induce an increase in 25(OH)D only below a certain threshold level of 25(OH)D (median: 37.3 and 47.6 nM for cases and controls never using HT, respectively), resulting in differential effects in cases and controls in our study population and in effect modification by HT on the association of 25(OH)D with breast cancer risk. In line with this hypothesis, use of HT was significantly associated with 25(OH)D using a linear regression model in women with low serum 25(OH)D concentrations (<50 nM, β = 2.2, P = 0.004) but not in women with high serum concentrations (>50 nM, β = −2.2, P  = 0.21). To our knowledge, there is, however, no data biologically supporting this threshold hypothesis.

We also found a stronger effect of 25(OH)D on breast cancer risk in women with increasing number of pregnancies. Parity is known to be associated with changes in sex hormone level resulting in a potential lower lifetime estrogen exposure ( 45 , 46 ). Cell studies on the interaction of estrogens, anti-estrogens and 1,25(OH) 2 D in tumor cell lines have been conducted. In cell studies, 1,25(OH) 2 D has been shown to be considerably more anticarcinogenic in combination with anti-estrogens ( 47 , 48 ). In addition, in the presence of estradiol, 1,25(OH) 2 D has been shown to stimulate proliferation at low doses and inhibit proliferation at higher doses, whereas in a non-estrogen environment, only anti-proliferative effects have been observed ( 48 ). Therefore, both, the interaction of 25(OH)D with use of HT and with the number of pregnancies, could be explained by interactions with the estrogen metabolism, i.e. a lower lifetime exposure to estrogens resulting in a potential stronger anticarcinogenic effect of vitamin D. However, the complex regulation of vitamin D in the human body and its interactions with the estrogen metabolism in breast cancer cells deserve further investigation.

We further analyzed whether the effect of vitamin D on breast cancer differed by receptor status of the tumor and did not find differential effects. Recent data from the NHS suggest an inverse association for ER−/PR−, but not for ER+/PR+ and ER+/PR− tumors, although power was low for their analysis ( 2 ). Because the effect of HT on breast cancer is known to be stronger in receptor-positive tumors ( 49 , 50 ), we additionally analyzed potential differences in the modification of the breast cancer–25(OH)D association by hormone use in ER+ or PR+ tumors only. No evidence was found for a stronger effect modification in ER+ or PR+ tumors.

With regard to the contribution of dietary vitamin D to serum 25(OH)D, few foods naturally contain vitamin D (i.e. fish and eggs) and fortification of food with vitamin D as well as high fish consumption is not common in Germany. Therefore, a low contribution of dietary vitamin D to the serum 25(OH)D concentration in our study population seems plausible.

Our study may have methodological limitations due to the retrospective case–control design. We are aware that a cancer diagnosis may change dietary or behavioral habits (less dietary intake of vitamin D, less sun exposure), which may influence 25(OH)D concentrations. Nevertheless, short-term changes in diet or behavioral habits are less likely to influence serum concentration because the half-life of 25(OH)D is rather long (2–3 weeks), the correlation between dietary vitamin D and serum 25(OH)D in this study was low ( r = 0.03) and plasma levels of 25(OH)D are fairly consistent over time ( 51 ). Although a notable change in 25(OH)D concentration after chemotherapeutic treatment was not observed in two studies ( 52 , 53 ), cancer therapy following diagnosis may affect circulating 25(OH)D concentration. However, the sensitivity analysis excluding patients <6 months between diagnosis and blood collection showed no notable change in the risk estimates.

Selection bias due to the participation rate of cases and controls is also unlikely to have biased our results in terms of 25(OH)D status. Information collected via short questionnaire from subjects who refused participation indicated that the study participants were better educated and reported hormone use more frequently than non-participants. However, these differences between study participants and non-participants answering the short questionnaire only were similarly observed both for cases and controls and therefore unlikely to have biased our results. Unfortunately, no information on 25(OH)D status-, diet- or vitamin D-related variables like outdoor activity was available for the non-participants with short questionnaire information.

Finally, we address the concern of possible measurement errors with respect to 25(OH)D status. Several reports emphasize the measurement variation introduced by different assays and different laboratories ( 14 , 54 , 55 ). Nevertheless, in the international Vitamin D Quality Assessment Scheme, which monitors the performance of vitamin D assays, the IDS enzyme immunoassay used in our study gave comparable results to the gold standard high-performance liquid chromatography method ( 56 ). In addition, intra- and interassay variation in 25(OH)D measurement as well as the average absolute deviation from the mean between two duplicate samples were very low in our analysis, reassuring the validity of the biomarker measurement.

In summary, we found a significant inverse association between serum 25(OH)D concentration and post-menopausal breast cancer risk. In addition, the relationship was non-linear, suggesting a stronger effect in women with low 25(OH)D concentrations as compared with women with higher concentrations. Use of HT and number of pregnancies were identified as modifiers of the association between serum 25(OH)D and breast cancer risk. The association was stronger in women who had never used HT compared with women with past or current use and in women with increasing number of pregnancies; however, these findings need confirmation in further studies. Our findings strongly suggest a protective effect for post-menopausal breast cancer risk through a better vitamin D supply as characterized by measurement of serum 25(OH)D.

Funding

Deutsche Krebshilfe e. V. (70-2892-BR I); Deutsche Forschungsgemeinschaft, Graduiertenkolleg 793 to S.A.

Abbreviations

    Abbreviations
  • BMI

    body mass index

  • CI

    confidence interval

  • ER

    estrogen receptor

  • HT

    hormone therapy

  • NHS

    Nurses' Health Study

  • OR

    odds ratio

  • 25(OH)D

    25-hydroxyvitamin D

  • 1,25(OH) 2 D

    1,25-dihydroxyvitamin D

  • PR

    progesterone receptor

  • R-N-K

    Rhein–Neckar–Karlsruhe

  • VDR

    vitamin D receptor

We are grateful to all participants for their help and patience, to all hospitals that recruited patients for this study and to the collaborating pathologists who provided detailed histologic diagnoses and immunohistochemical examination results on tumor samples. We also thank S.Behrens, R.Birr, W.Busch, U.Eilber, B.Kaspereit, N.Knese and K.Smit for valuable technical assistance.

Conflict of Interest Statement : None declared.

References

1.
Colston
KW
, et al.  . 
Mechanisms implicated in the growth regulatory effects of vitamin D in breast cancer
Endocr. Relat. Cancer
 , 
2002
, vol. 
9
 (pg. 
45
-
59
)
2.
Bertone-Johnson
ER
, et al.  . 
Plasma 25-hydroxyvitamin D and 1,25-dihydroxyvitamin D and risk of breast cancer
Cancer Epidemiol. Biomarkers Prev.
 , 
2005
, vol. 
14
 (pg. 
1991
-
1997
)
3.
Welsh
J
Vitamin D and breast cancer: insights from animal models
Am. J. Clin. Nutr.
 , 
2004
, vol. 
80
 (pg. 
1721S
-
1724S
)
4.
Lowe
LC
, et al.  . 
Plasma 25-hydroxy vitamin D concentrations, vitamin D receptor genotype and breast cancer risk in a UK Caucasian population
Eur. J. Cancer
 , 
2005
, vol. 
41
 (pg. 
1164
-
1169
)
5.
Shin
MH
, et al.  . 
Intake of dairy products, calcium, and vitamin D and risk of breast cancer
J. Natl Cancer Inst.
 , 
2002
, vol. 
94
 (pg. 
1301
-
1311
)
6.
Hansen
CM
, et al.  . 
Vitamin D and cancer: effects of 1,25(OH)2D3 and its analogs on growth control and tumorigenesis
Front. Biosci.
 , 
2001
, vol. 
6
 (pg. 
D820
-
D848
)
7.
Eisman
JA
, et al.  . 
Normal and malignant breast tissue is a target organ for 1,25-(0H)2 vitamin D3
Clin. Endocrinol. (Oxf)
 , 
1980
, vol. 
13
 (pg. 
267
-
272
)
8.
Saez
S
, et al.  . 
William L. McGuire Memorial Symposium. 1,25(OH)2D3 modulation of mammary tumor cell growth in vitro and in vivo
Breast Cancer Res. Treat.
 , 
1993
, vol. 
27
 (pg. 
69
-
81
)
9.
Jensen
SS
, et al.  . 
Inhibitory effects of 1alpha,25-dihydroxyvitamin D(3) on the G(1)-S phase-controlling machinery
Mol. Endocrinol.
 , 
2001
, vol. 
15
 (pg. 
1370
-
1380
)
10.
Trump
DL
, et al.  . 
Anti-tumor activity of calcitriol: pre-clinical and clinical studies
J. Steroid Biochem. Mol. Biol.
 , 
2004
, vol. 
89–90
 (pg. 
519
-
526
)
11.
Clemens
TL
, et al.  . 
Serum vitamin D2 and vitamin D3 metabolite concentrations and absorption of vitamin D2 in elderly subjects
J. Clin. Endocrinol. Metab.
 , 
1986
, vol. 
63
 (pg. 
656
-
660
)
12.
Reichel
H
, et al.  . 
The role of the vitamin D endocrine system in health and disease
N. Engl. J. Med.
 , 
1989
, vol. 
320
 (pg. 
980
-
991
)
13.
Haddad
JG
Jr
, et al.  . 
Acute administration of 25-hydroxycholecalciferol in man
J. Clin. Endocrinol. Metab.
 , 
1976
, vol. 
42
 (pg. 
284
-
290
)
14.
Zerwekh
JE
The measurement of vitamin D: analytical aspects
Ann. Clin. Biochem.
 , 
2004
, vol. 
41
 (pg. 
272
-
281
)
15.
Friedrich
M
, et al.  . 
Analysis of the vitamin D system in cervical carcinomas, breast cancer and ovarian cancer
Recent Results Cancer Res.
 , 
2003
, vol. 
164
 (pg. 
239
-
246
)
16.
Welsh
JE
, et al.  . 
Vitamin D-3 receptor as a target for breast cancer prevention
J. Nutr.
 , 
2003
, vol. 
133
 (pg. 
2425S
-
2433S
)
17.
Zehnder
D
, et al.  . 
Extrarenal expression of 25-hydroxyvitamin D(3)-1 alpha-hydroxylase
J. Clin. Endocrinol. Metab.
 , 
2001
, vol. 
86
 (pg. 
888
-
894
)
18.
Kemmis
CM
, et al.  . 
Human mammary epithelial cells express CYP27B1 and are growth inhibited by 25-hydroxyvitamin D-3, the major circulating form of vitamin D-3
J. Nutr.
 , 
2006
, vol. 
136
 (pg. 
887
-
892
)
19.
McCullough
ML
, et al.  . 
Dairy, calcium, and vitamin D intake and postmenopausal breast cancer risk in the Cancer Prevention Study II Nutrition Cohort
Cancer Epidemiol. Biomarkers Prev.
 , 
2005
, vol. 
14
 (pg. 
2898
-
2904
)
20.
John
EM
, et al.  . 
Vitamin D and breast cancer risk: the NHANES I Epidemiologic follow-up study, 1971–1975 to 1992. National Health and Nutrition Examination Survey
Cancer Epidemiol. Biomarkers Prev.
 , 
1999
, vol. 
8
 (pg. 
399
-
406
)
21.
Witte
JS
, et al.  . 
Diet and premenopausal bilateral breast cancer: a case-control study
Breast Cancer Res. Treat.
 , 
1997
, vol. 
42
 (pg. 
243
-
251
)
22.
Potischman
N
, et al.  . 
Intake of food groups and associated micronutrients in relation to risk of early-stage breast cancer
Int. J. Cancer
 , 
1999
, vol. 
82
 (pg. 
315
-
321
)
23.
Levi
F
, et al.  . 
Dietary intake of selected micronutrients and breast-cancer risk
Int. J. Cancer
 , 
2001
, vol. 
91
 (pg. 
260
-
263
)
24.
Simard
A
, et al.  . 
Vitamin D deficiency and cancer of the breast: an unprovocative ecological hypothesis
Can. J. Public Health
 , 
1991
, vol. 
82
 (pg. 
300
-
303
)
25.
Abbas
S
, et al.  . 
Dietary vitamin d and calcium intake and premenopausal breast cancer risk in a German case-control study
Nutr. Cancer
 , 
2007
, vol. 
59
 (pg. 
54
-
61
)
26.
Hiatt
RA
, et al.  . 
Prediagnostic serum vitamin D and breast cancer
J. Natl Cancer Inst.
 , 
1998
, vol. 
90
 (pg. 
461
-
463
)
27.
Janowsky
EC
, et al.  . 
Association between low levels of 1,25-dihydroxyvitamin D and breast cancer risk
Public Health Nutr.
 , 
1999
, vol. 
2
 (pg. 
283
-
291
)
28.
van Hoof
HJ
, et al.  . 
Relationship between free and total 1,25-dihydroxyvitamin D in conditions of modified binding
Eur. J. Endocrinol.
 , 
2001
, vol. 
144
 (pg. 
391
-
396
)
29.
Lips
P
Vitamin D deficiency and secondary hyperparathyroidism in the elderly: consequences for bone loss and fractures and therapeutic implications
Endocr. Rev.
 , 
2001
, vol. 
22
 (pg. 
477
-
501
)
30.
Sowers
MR
, et al.  . 
Parameters related to 25-OH-D levels in a population-based study of women
Am. J. Clin. Nutr.
 , 
1986
, vol. 
43
 (pg. 
621
-
628
)
31.
Nesby-O'Dell
S
, et al.  . 
Hypovitaminosis D prevalence and determinants among African American and white women of reproductive age: Third National Health and Nutrition Examination Survey, 1988–1994
Am. J. Clin. Nutr.
 , 
2002
, vol. 
76
 (pg. 
187
-
192
)
32.
Harris
SS
, et al.  . 
The association of oral contraceptive use with plasma 25-hydroxyvitamin D levels
J. Am. Coll. Nutr.
 , 
1998
, vol. 
17
 (pg. 
282
-
284
)
33.
Aarskog
D
, et al.  . 
Effect of estrogen on vitamin D metabolism in tall girls
J. Clin. Endocrinol. Metab.
 , 
1983
, vol. 
57
 (pg. 
1155
-
1158
)
34.
Schwartz
B
, et al.  . 
Estrogen controls expression and bioresponse of 1,25-dihydroxyvitamin D receptors in the rat colon
Mol. Cell. Biochem.
 , 
2000
, vol. 
203
 (pg. 
87
-
93
)
35.
Bohlscheid-Thomas
S
, et al.  . 
Reproducibility and relative validity of energy and macronutrient intake of a food frequency questionnaire developed for the German part of the EPIC project. European Prospective Investigation into Cancer and Nutrition
Int. J. Epidemiol.
 , 
1997
, vol. 
26
 
(Suppl. 1)
(pg. 
S71
-
S81
)
36.
Royston
P
, et al.  . 
The use of fractional polynomials to model continuous risk variables in epidemiology
Int. J. Epidemiol.
 , 
1999
, vol. 
28
 (pg. 
964
-
974
)
37.
Heinzl
H
, et al.  . 
Gaining more flexibility in Cox proportional hazards regression models with cubic spline functions
Comput. Methods Programs Biomed.
 , 
1997
, vol. 
54
 (pg. 
201
-
208
)
38.
de Lyra
EC
, et al.  . 
25(OH)D3 and 1,25(OH)2D3 serum concentration and breast tissue expression of 1alpha-hydroxylase, 24-hydroxylase and Vitamin D receptor in women with and without breast cancer
J. Steroid Biochem. Mol. Biol.
 , 
2006
, vol. 
100
 (pg. 
184
-
192
)
39.
Hintzpeter
B
, et al.  . 
Vitamin D status and health correlates among German adults
Eur. J. Clin. Nutr
 , 
2007
40.
Welsh
J
Vitamin D and prevention of breast cancer
Acta Pharmacol. Sin.
 , 
2007
, vol. 
28
 (pg. 
1373
-
1382
)
41.
Gallagher
JC
, et al.  . 
Effect of estrogen on calcium absorption and serum vitamin D metabolites in postmenopausal osteoporosis
J. Clin. Endocrinol. Metab.
 , 
1980
, vol. 
51
 (pg. 
1359
-
1364
)
42.
Peters
U
, et al.  . 
Circulating vitamin D metabolites, polymorphism in vitamin D receptor, and colorectal adenoma risk
Cancer Epidemiol. Biomarkers Prev.
 , 
2004
, vol. 
13
 (pg. 
546
-
552
)
43.
Buchanan
JR
, et al.  . 
The effect of endogenous estrogen fluctuation on metabolism of 25-hydroxyvitamin D
Calcif. Tissue Int.
 , 
1986
, vol. 
39
 (pg. 
139
-
144
)
44.
Liel
Y
, et al.  . 
Evidence that estrogens modulate activity and increase the number of 1,25-dihydroxyvitamin D receptors in osteoblast-like cells (ROS 17/2.8)
Endocrinology
 , 
1992
, vol. 
130
 (pg. 
2597
-
2601
)
45.
Bernstein
L
Epidemiology of endocrine-related risk factors for breast cancer
J. Mammary Gland Biol. Neoplasia
 , 
2002
, vol. 
7
 (pg. 
3
-
15
)
46.
Bernstein
L
, et al.  . 
Estrogen and sex hormone-binding globulin levels in nulliparous and parous women
J. Natl Cancer Inst.
 , 
1985
, vol. 
74
 (pg. 
741
-
745
)
47.
Vink-van Wijngaarden
T
, et al.  . 
Inhibition of breast cancer cell growth by combined treatment with vitamin D3 analogues and tamoxifen
Cancer Res.
 , 
1994
, vol. 
54
 (pg. 
5711
-
5717
)
48.
Love-Schimenti
CD
, et al.  . 
Antiestrogen potentiation of antiproliferative effects of vitamin D3 analogues in breast cancer cells
Cancer Res.
 , 
1996
, vol. 
56
 (pg. 
2789
-
2794
)
49.
Chen
WY
, et al.  . 
Association of hormone replacement therapy to estrogen and progesterone receptor status in invasive breast carcinoma
Cancer
 , 
2004
, vol. 
101
 (pg. 
1490
-
1500
)
50.
Ursin
G
, et al.  . 
Does menopausal hormone replacement therapy interact with known factors to increase risk of breast cancer?
J. Clin. Oncol.
 , 
2002
, vol. 
20
 (pg. 
699
-
706
)
51.
Feskanich
D
, et al.  . 
Plasma vitamin D metabolites and risk of colorectal cancer in women
Cancer Epidemiol. Biomarkers Prev.
 , 
2004
, vol. 
13
 (pg. 
1502
-
1508
)
52.
Kailajarvi
ME
, et al.  . 
Serum bone markers in breast cancer patients during 5-fluorouracil, epirubicin and cyclophosphamide (FEC) therapy
Anticancer Res.
 , 
2004
, vol. 
24
 (pg. 
1271
-
1274
)
53.
Gao
Y
, et al.  . 
The effects of chemotherapy including cisplatin on vitamin D metabolism
Endocr. J.
 , 
1993
, vol. 
40
 (pg. 
737
-
742
)
54.
Binkley
N
, et al.  . 
Assay variation confounds the diagnosis of hypovitaminosis D: a call for standardization
J. Clin. Endocrinol. Metab.
 , 
2004
, vol. 
89
 (pg. 
3152
-
3157
)
55.
Hollis
BW
Editorial: the determination of circulating 25-hydroxyvitamin D: no easy task
J. Clin. Endocrinol. Metab.
 , 
2004
, vol. 
89
 (pg. 
3149
-
3151
)
56.
Carter
GD
, et al.  . 
How accurate are assays for 25-hydroxyvitamin D? Data from the international vitamin D external quality assessment scheme
Clin. Chem.
 , 
2004
, vol. 
50
 (pg. 
2195
-
2197
)