Association of Prediagnostic Serum Levels of Vitamin D with Risk of Ductal Carcinoma In Situ of the Breast in the UK Biobank Cohort Study

Ductal carcinoma in situ (DCIS) of the breast, characterized by epithelial proliferation contained within the basement membrane of the ducts, is considered a non-obligate precursor of invasive breast cancer and accounts for approximately 20% of diagnosed breast cancer (1). Epidemiologic studies have shown that DCIS and invasive breast cancer share several risk factors (2). In this regard, in clinical trials, vitamin D and calcium supplementation have been associated with a sustained reduction in the risk of both invasive breast cancer and DCIS (3). In contrast, while a meta-analysis of prospective studies provided evidence for an inverse association between circulating vitamin D [25(OH)D] levels and risk of breast cancer among postmenopausal women (4), only a few studies with small sample sizes have evaluated the relationship between circulating vitamin D levels and risk of in situ breast cancer (5–7), and none has examined the association separately in pre- and postmenopausal women. A role for vitamin D in breast carcinogenesis is plausible given that it influences apoptosis, and has anti-inflammatory, antiproliferative and anti-angiogenetic effects within the tumor microenvironment (8). In the current study we examined the association between serum vitamin D and risk of DCIS using data from the UK Biobank.

The UK Biobank is a population-based prospective study, which includes 502,419 participants (54.5% women), aged 40 to 69 years and registered with the United Kingdom's National Health Service at the time of recruitment (2006–2010; ref. 9). The study received ethical approval from the North West Multi-Centre Research Ethics Committee. Details regarding both the baseline questionnaire that assessed sociodemographic, lifestyle and health related characteristics, and of blood collection, have been published (http://biobank.ctsu.ox.ac.uk/crystal/; ref. 9, 10). Circulating serum levels of 25(OH)D were measured in the baseline samples using a direct competitive chemiluminescent immunoassay (DiaSorin-Liaison XLAnalyzer - Diasorin S.p.A; https://www.ukbiobank.ac.uk/). Undetectable 25(OH)D levels were imputed as 10 nmol/L (4 ng/mL: minimum test-detectable level) if low, and as 375 nmol/L (118.8 ng/mL: maximum) if too high. Incident cancers were ascertained through the national cancer registries and classified using the International Classification of Disease, Tenth Revision (ICD-10); code D051 (DCIS), and C50 (invasive breast cancer). Deaths were ascertained through death registries. The analytical cohort included 231,203 women with no history of breast cancer (invasive or DCIS) at baseline and with 25(OH)D data.

Baseline cohort characteristics were summarized by quartiles of vitamin D. Cox proportional hazards models were used to estimate HRs and 95% confidence intervals (CI) for the association between quartiles of serum 25(OH)D and DCIS risk. In addition, the association of 25(OH)D levels (1 ng/mL increments) with DCIS risk was evaluated through a linear model across the whole spectrum of values (4–118.8 ng/mL), and via a linear spline model. The latter model allows the association of the exposure with the outcome to vary between knot-defined intervals. Various knot locations were tested to obtain a model that minimized the Akaike Information Criterion; ultimately, the model with two knots at the 50th and 75th percentiles of the 25(OH)D distribution provided similar or better results (in the premenopausal group) than the linear model with no knots. Time at risk was calculated from the baseline visit to the date of occurrence of DCIS, invasive breast cancer, death, study withdrawal, or last update (March 2021), whichever came first. Analyses were conducted in the whole cohort and by baseline menopausal status. On the basis of a previous study conducted in this cohort (2), the analyses were adjusted for age (linear and squared terms), race and ethnicity, body mass index (BMI; <18.5, 18.5–25.0, >25.0–30.0, and ≥30 kg/m²), serum calcium level (quartiles), number of live births (0, 1, 2–4, and >4), quartiles of total physical activity (metabolic equivalent/week), current use of menopausal hormone therapy (HT), current use of oral contraceptives, family history of breast cancer, history of mammogram screening, and menopausal status (2). A sensitivity analysis was conducted excluding women with less than 2 years of follow-up. Analyses stratified by baseline BMI, family history of breast cancer, and HT use were also conducted.

A total of 231,203 women with an average age of 56.2 years (SD = 8.0) were included in the study, of whom 94.0% were White, and 70.7% were postmenopausal. Over a median follow-up period of 11.3 years, a total of 1,340 DCIS cases were ascertained. Characteristics of the study subjects are summarized in Supplementary Table S1. The higher the quartile of serum 25(OH)D, the more likely were women to be older, to be White, to regularly drink alcohol, to have a lower BMI, to be postmenopausal, to currently use menopausal hormone therapy, and to have had a least one mammogram screening before enrollment in the study, and the less likely they were to be smokers, and to engage in regular physical activity.

There was no association between quartiles of serum 25(OH)D levels and risk of DCIS, overall or by menopausal status (Table 1). No linear association was observed between increasing 25(OH)D levels and DCIS risk (HR_{1ng/mL25(OH)D} = –1.00; 95% CI, 0.99–1.00). Similarly, the linear spline model did not provide a better fit to the data except among premenopausal women, in whom a model with two knots [at 18.8 and 25.0 ng/mL, corresponding to the 50th and the 75th percentiles of the 25(OH)D distribution] provided a better fit, showing an inverse association between levels of 25(OH)D within these two values and the risk of DCIS (HR_{1ng/mL25(OH)D}= 0.94; 95% CI, 0.89–0.99; P value = 0.031). Analyses conducted after excluding women with <2 years of follow-up showed similar results (Table 1). No interaction was observed between quartiles of serum vitamin D and BMI categories, breast cancer family history, and HT status with respect to DCIS risk (Table 2).

In this large prospective cohort study we found no association between circulating levels of serum 25(OH)D and risk of DCIS, overall; or by menopausal status. Although there was little variation in risk of DCIS across quartiles of 25(OH)D, among premenopausal women with 25(OH)D in the third quartile, we observed a significant 6% reduction in DCIS risk per 1 ng/mL increase in 25(OH)D. However, these results were based on a limited number of cases (n = 98) and may represent a chance finding due to multiple testing. Analyses within strata of BMI, breast cancer family history, and HT showed results similar to those observed overall.

Our results are based on a large number of DCIS cases ascertained over more than a decade of follow-up, providing substantial statistical power for the analysis. Baseline data on mammogram screening indicated that most women had had at least one mammogram screening before study enrollment (95% of women aged 50 years or older), but the study had no information on screening during follow-up. Vitamin D insufficiency is more prevalent among black women; however, since their number in the UK Biobank study is low, we could not separately test the association in this ethnic group. Several nested case–control studies with shorter follow-up than that here have reported null associations of 25(OH)D with in situ breast cancer. In the Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial, serum levels of 25(OH)D were not significantly different between 190 postmenopausal cases and an equal number of controls (5). Similar results were found in the Cancer Prevention Study-II with 103 postmenopausal case–control pairs (7), and in the Nurses’ Health Study with 109 cases and 109 controls with unspecified menopausal status (6).

Approximately 55% of the UK Biobank cohort had serum levels of 25(OH)D lower than the recommended values for overall health (20–≤50 ng/mL; ref. 11), particularly among premenopausal women. It is possible that the distribution of 25(OH)D values might have reduced the possibility of observing an association. Nevertheless, the results of the current study do not provide evidence of an association of 25(OH)D with DCIS risk.

No disclosures were reported.

Dr. T.E. Rohan is supported in part by the Breast Cancer Research Foundation (BCRF-21-140).

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.