Vitamin D levels in Swiss breast cancer survivors

Baumann Michaela, Dani Sergio U.b, Dietrich Danielc, Hochstrasser Andreasd, Klingbiel Dirkc, Mark Michael T.e, Riesen Walter F.a, Ruhstaller Thomasf, Templeton Arnoud J.g, Thürlimann Beatf a Institute for Clinical Chemistry and Haematology, St Gallen Cantonal Hospital, Switzerland b Department of Medical Oncology and Haematology, St Gallen Cantonal Hospital, Switzerland c Statistical Office, Swiss Group for Clinical Cancer Research Coordinating Centre – SAKK, Bern, Switzerland d Oncology Centre, Männedorf Hospital, Zurich, Switzerland e Department of Oncology and Haematology, Graubuenden Cantonal Hospital, Chur, Switzerland f Breast Centre St Gallen, St Gallen Cantonal Hospital, Switzerland g Department of Oncology, St Claraspital Basel and Faculty of Medicine, University of Basel, Switzerland


Introduction
Cholecalciferol (vitamin D3) is widely supplemented in breast cancer survivors because of the favourable effect of vitamin D on multiple health outcomes, including: maintained bone integrity, muscle strength and immunity; reduced inflammation, cardiovascular disease mortality and cancer risk; and improved mental function, quality of life and survival [1][2][3][4][5][6][7].Low-to moderate-quality evidence supports an antineoplastic effect of vitamin D with a more favourable prognosis in various cancers, including breast cancer [8][9][10][11][12][13][14][15][16][17].Inflammation stress can decrease the levels of 25(OH)D by increasing its conversion to the vitamin D active metabolite 1,25(OH)2D [18].Hypovitaminosis D decreases the conversion of 25(OH)D to 1,25(OH)2D in postmenopausal women [19].Long-term therapy with certain drugs, especially cytochrome P450-inducing agents commonly used in patients with cancer can decrease the levels of calcitriol through accelerated cytochrome P450-mediated degradation [20,21].Some recreational drugs, including tobacco [22,23] and alcohol [24] can modify the association between protective vitamin D levels and cancer risk.In addition, significant associations with vitamin D receptor gene (VDR) polymorphisms have been reported for prostate, breast, colorectal and skin cancer [25,26], suggesting that vitamin D and/or VDR may modulate the risk of cancer.Therefore patients with breast cancer receiving hormonal therapy, chemotherapy, radiotherapy, immunotherapy or combinations thereof, as well as breast cancer survivors and postmenopausal women in general may benefit from adequate levels of vitamin D. The present observational study aimed to investigate 25(OH)D levels and their determinants in a cohort of breast cancer survivors in eastern Switzerland.

Study population
The study population consisted of all patients taking part in a cross-sectional study on sociodemographic, lifestyle and health characteristics of Swiss patients with early breast cancer, as reported elsewhere [27].Consecutive patients from the St Gallen Breast Centre (a tertiary referral centre in eastern Switzerland) were offered study participation.Inclusion criteria were a diagnosis of early (nonmetastatic) breast cancer with preoperative, current or completed postoperative treatment and no evidence of recurrence.The study was reported according to the STROBE guidelines for observational studies and approved by the local Ethics Committee (EKSG-Ethikkommission des Kantons St. Gallen, 08/082) and all participants provided informed consent for use and analyses of baseline characteristics, sociodemographic data, questionnaires and biochemical data in this study.

Data collection
All consenting patients were asked to complete a questionnaire about patient-related and sociodemographic variables including age, education, physical activity, smoking habit, place of residence (rural area vs urban area), comorbidity, use of complementary and alternative medicine and calcium and cholecalciferol intake.Vital parameters including weight, height, body mass index (BMI), bone mineral density (BMD) and body composition were recorded for all patients on the basis of routine procedures.Tumour biological characteristics including age at diagnosis, tumour stage, histology and grade of differentiation, hormone-receptor and HER2 status, antitumor treatment (therapy modalities), menopausal status, and concurrent disease were extracted manually from the records.Cholecalciferol intake and serum concentration of the storage metabolite 25(OH)D and the active metabolite 1,25-dihydroxyvitamin D (1,25(OH) 2 D) were recorded at baseline and during a second (follow-up) visit.High-performance liquid chromatography was used to analyse 25(OH)D, (kit by Chromsystems, Gräflingen, Germany) and 1,25(OH)2D was analysed with a radioimmunassay (IDS, Tyne and Wear, UK).Since our assays did not separate the major circulating vitamin D2 and vitamin D3 metabolites and their respective epimers and isobars [28], we collectively termed them 25(OH)D and 1,25(OH)2D.Vitamin D deficiency was defined as 25(OH)D ≤50 nmol/l (≤20 ng/l); vitamin D insufficiency was defined as 25(OH)D between 50 and 75 nmol/l (20-30 ng/l), and optimal 25(OH)D levels were defined as between 75 and 100 nmol/l (30-40 ng/l), in accordance with the reference values of our assays and internationally accepted concentration intervals [29][30][31].
At the first visit, there were 133 patients already receiving daily calcium-vitamin D3 supplementation as a chewing tablet, and 3 patients already receiving vitamin D3 (cholecalciferol) supplementation alone.After 25(OH)D analysis, 12 patients showing 25(OH)D deficiency were advised to have a loading intramuscular dose of 300 000 IU cholecalciferol, and all patients with 25(OH)D insufficiency or deficiency were advised to take daily calcium-D3 supplementation (500 mg and 800 IU, respectively) as a chewing tablet.

Statistical analyses
We performed exploratory statistical analyses on the interrelatedness of various laboratory, clinical and lifestyle variables including 25(OH)D, 1,25(OH)2D, age, BMI, BMD, physical activity (including frequency of sporting activity) and place of residence (rural area vs urban area).Vitamin D was the primary outcome, and we based our statistical analyses on the storage form (25(OH)D) because this is the more abundant and stable form, and also because this is the commonly reported vitamin D assay.To check for the physiological status of vitamin D in our study population, we also considered 1,25(OH)2D in some analyses.Continuous variables were summarised as median and range, categorical as frequency and percent.We performed univariate analysis using Spearman's rho to test for correlation between 25(OH)D and the continuous variables: age, 1,25(OH)2D level, BMD and BMI.To compare subgroups (place of residence, sport activity, season of the year, calcium and/or cholecalciferol supplementation) as regards to their respective median levels of vitamin D metabolites and the correlation to a number of variables, we used the Kruskal-Wallis test (age, season of the year, calcium and cholecalciferol supplementation) and the Wilcoxon rank-sum test (place of residence).We also performed a multiple regression model with 25(OH)D level as the outcome, and age, BMI, BMD, seasons of the year, place of residence and sport activity as covariates.Trends in 25(OH)D levels as related to tumour characteristics and therapy modalities were derived from a descriptive statistical analysis (median, standard deviation [SD], correlation).Because of the exploratory nature of the study, p-values were not corrected for multiple testing.The significance level used for all analyses was p <0.05.Data were visualised as scatter plots with scatter plot smoothers or as box plots.Analyses were performed with S-PLUS 8.1 (TIBCO Software, Palo Alto, CA, USA).

Patients
Between December 2008 and September 2010, 375 patients were offered participation in the study.Twenty-eight (8%) patients declined participation (unwillingness to participate, linguistic problems) and 5 (1%) patients were excluded from the analysis because they did not fulfil the selection criteria (e.g., no history of breast cancer, metastatic disease) or had withdrawn their consent.The remaining 342 patients completed the enrolment questionnaire.Fig

Factors associated with vitamin D levels
We focused our descriptive analyses on the baseline vitamin D levels for which we had more complete laboratory data, as above stated.Figure 3 and tables 3 and 4 show the baseline levels of 25(OH)D and 1,25(OH)2D during according to the season of the year for all patients.Median 25(OH)D and 1,25(OH)2D levels showed a significant seasonality, with increased levels during the seasons of higher sun exposure as compared with those of lower sun exposure: 25(OH)D at baseline, n = 332, Kruskal-Wallis p = 0.042; 1,25(OH)2D at baseline, n = 311, Kruskal-Wallis p = 0.001.Living in a rural area was associated with a higher median 25(OH)D concentration as compared with living in an urban area: 87 nmol/l (range 16-216), n = 240, median BMI 24.6 kg/m 2 versus 46 nmol/l (range 17-162), n = 92, median BMI 24.1 kg/m 2 ; Wilcoxon p = 0.001 (fig.4).A significant correlation was observed between 25(OH)D and1,25(OH)2D levels (n = 310, Spearman's rho = 0.21, p <0.001) but the plot was very scattered with a flattening of the trend line for 25(OH)D levels higher than 80 nmol/l (fig.5).BMI was inversely related to both 25(OH)D-and 1,25(OH)2D levels (n = 311, Spearman's rho = −0.24,p <0.001 and n = 310, Spearman's rho = −0.23,p <0.001, respectively; fig.6), and regular sporting activity was associated with a higher median 25(OH)D level even when of moderate intensity (n = 328, Kruskal-Wallis p = 0.045; medians 65 nmol/l for no activity, 84 nmol/l for one and 88 nmol/l for two activities per week; table 5).
The 25(OH)D level at baseline was higher in patients already receiving oral supplementation with calcium and cholecalciferol (usually one or two chewing tablets containing 500 mg calcium plus 400 IU cholecalciferol) as compared with those patients who were not receiving supplementation.Some patients diagnosed with low 25(OH)D at the first visit took cholecalciferol supplementation there-after, but the dose and form of substitution varied among patients.Sixty-two (27%) patients, including 15 with 25(OH)D deficiency, did not take any cholecalciferol supplementation between the first and second visits.A few

Trends in 25(OH)D levels related to tumour characteristics and therapy modalities
A nonsignificant trend towards lower median 25(OH)D with increasing breast tumour grade was observed: 96   6. Patients who had undergone chemotherapy showed a nonsignificant lower median 25(OH)D level as compared with patients who had not had any chemotherapy, but the former were younger than the latter (p ˂0.001).Also, the patients undergoing  chemotherapy at the time of the first visit were on average younger than the patients who had not had chemotherapy at this time point (p ˂0.001).The patients who were under endocrine therapy at the time of the first visit tended to have higher 25(OH)D levels than the smaller number of patients who were not under endocrine therapy at this time.
The few patients who were receiving HER2-directed therapy at the time of the first visit tended to have lower 25(OH)D levels than patients who were not under HER2-directed therapy at this time.However, possibly owing to the low sample sizes and therefore the low statistical power, none of the differences in 25(OH)D and 1,25(OH)2D as related to tumour biology or therapy modality reached statistical significance.

Multiple regression analyses
The multiple regression analyses yielded results similar to the univariate models.

Discussion
In this observational study we described the vitamin D status of 332 breast cancer survivors at baseline and in 230 of them at a follow-up visit a median of 210 days thereafter.
The effects of factors known to increase the 25(OH)D serum levels, such as sunlight exposure, place of residence, physical activity and supplementation with cholecalciferol, as well as factors known to reduce the 25(OH)D serum levels such as obesity and chemotherapy [36, 37], were recorded for our cohort.
The main conclusion that cholecalciferol supplementation (or lack thereof) was the single most important factor affecting vitamin D levels in our cohort seems to be quite obvious and could have been reached without this study.However, the study showed that a substantial number of patients have suboptimal 25(OH)D levels in spite of ongoing cholecalciferol supplementation.This suggests that adequate vitamin D supplementation regimens must be tailored to individual characteristics, needs and preferences.
Cholecalciferol is generated in the skin from 7-dehydrocholesterol or obtained from the diet.Cholecalciferol undergoes a first hydroxylation step in the liver to yield 25(OH)D, which is a widely measured precursor of the hormonally active metabolite, 1,25(OH) 2D.The activation of 25(OH)D to 1,25(OH)2D occurs through a second hydroxylation step occurring in great part in the kidneys, but also in other tissues including tumours [38][39][40].The half-life of 25(OH)D varies between 2 and 5 weeks, depending on multiple factors such as vitamin D compound, ethnic background, vitamin D binding protein (DBP) genotype, kidney and liver function, disease states, pharmaceutical drugs, and the method used to measure the vitamin D metabolite [41][42][43][44][45][46][47].The terminal half-life of 1,25(OH)2D is much shorter, ranging from only 5 to 10 hours in healthy subjects [48].Despite cholecalciferol supplementation, we observed that the proportion of patients with 25(OH)D deficiency or insufficiency remained practically unchanged at about 40%, and the proportion of the patients with sufficient 25(OH)D levels remained at about 60% over time (see fig. 2).Factors such as place of residence, physical activity, BMI, supplementation compliance and cholecalciferol dosage may help explain these findings.We have seen that people living in a rural area had higher 25(OH)D levels than people living in an urban area.Also, there were many more rural dwellers (72%) than urban dwellers (28%) in our cohort.Most -if not all -patients changed neither their place of residence, their BMI, nor their physical activity during the observation period.Thus, the constancy of these variables between the baseline and the follow-up visit was associated with stable vitamin D proportions during this period.
In our series of patients, both 25(OH)D and 1,25(OH)2D showed seasonal variation, but this seasonal variation was more pronounced with 1,25(OH)2D.This finding indicates that 25(OH)D levels were not sufficient in a large number of patients in our cohort.The active metabolite 1,25(OH)2D does not generally show any seasonal variation provided that 25(OH)D concentrations are sufficient [49].
The effect of cholecalciferol supplementation was not evenly distributed.Whereas most patients experienced a rise in 25(OH)D following oral supplementation, a few patients receiving even a large cholecalciferol dose of 300 000 IU maintained low levels or did not show an adequate rise of 25(OH)D at the second visit.Of these patients, two even had a decrease of their 25(OH)D levels a couple of months after the high-dose supplementation.Also, a large proportion of patients (round 35 and 30% at the first visit and at follow-up, respectively) showed 25(OH)D levels above optimum.We thereby conclude that adequate cholecalciferol supplementation is important to maintain optimal 25(OH)D levels, but most of our patients did not have adequate cholecalciferol supplementation.Limitations to this conclusion include a lack of adjustments for a number of confounding factors, but use of a multiple regression model means that important confounding factors have been included in the analysis.
The cholecalciferol dosing regimen is a very important factor in the success of treatment in clinical trials, a point that is only just becoming apparent.The topic was first raised by B.W Hollis [50] and further explained by Hollis and Wagner (2013) [51].These concepts have been shown to be accurate in a recent meta-analysis of randomised controlled trials involving vitamin D and infection, where bolus dosing of vitamin D failed and daily dosing was a success [52].
The Australasian expert panel of 2016 [53] concluded that (i) the target serum 25(OH)D concentration should be 50 to 60 nmol/l all the year around, with a conservative upper limit <100 nmol/l; (ii) the dosing interval may need to be <2 months to have a continuous benefit; and (iii) a maintenance dose of 1000 IU/day, or an equivalent dose weekly or monthly, is sufficient for most individuals in Australasia.
A similar maintenance dose has been recommended to people living in Europe and North America.The dose effects of cholecalciferol on 25(OH)D levels are depicted in figure 8, which shows that the usual daily maintenance supplementation with 1000 IU cholecalciferol is not enough to bring any appreciable change in the 25(OH)D levels over time in 25(OH)D-deficient patients.This usual dose regimen is well below the 2000-3000 IU/day required to safely raise serum 25(OH)D to optimal levels and nor-malise parathyroid hormone (PTH) during the winter season in the northern hemisphere [57].There is currently no good reason to believe that these recommendations should not apply to Swiss breast cancer survivors.

Strengths and limitations
The observational study was originally planned as a crosssectional study without an interventional-therapeutic part.
The original study had some predefined hypotheses, e.g., higher socioeconomic status would be associated with higher vitamin D levels and a higher proportion of our  [58].The percentage of participants in each of the four 25(OH)D deficiency groups -severely deficient (<10 ng/ml), deficient (10-20 ng/ml), insufficient (21-29 ng/ml) and normal (≥30 ng/ml) -were statistically compared.About 66% of the subjects had insufficient lev- els of 25(OH)D.Normal levels of 25(OH)D were found in 26.1% of the subjects of whom 21% were males and 30.5% were females.Severely deficient levels of 25(OH)D were found in 7.98% of the total study population.In a comparison with our results, vitamin D deficient and severely deficient healthy women were roughly twice as much as in our breast cancer survivor series (about two thirds in the normal population versus one third in our breast cancer survival series).The explanation for this difference is surely the higher prevalence of preventative vitamin D supplementation in our breast cancer survivors, reflecting this general conduct of Swiss medical doctors and patients alike.It remains to be proven in future studies whether a higher prevalence of vitamin D supplementation among breast cancer survivors influences disease-free survival or not.Also, we acknowledge potential study limitations arising from basing the statistical analyses on 25(OH)D, as well as from the wide variability in the results as evident in the range of values to median figures, and from the fact that our assays did not separate the different vitamin D epimers and isobars, which may have resulted in inaccuracies in the measurement and interpretation of vitamin D status.Some of these limitations are illustrated in figure 5, which showed a weaker correlation between 25(OH)D and 1,25(OH)2D for 25(OH)D levels higher than 80 nmol/l, which may reflect either the large variation in 25(OH)D levels within the cohort or varying rates of 25(OH)D conversion to 1,25(OH)2D, due, for example, to different metabolic activities, drug interactions, genetic variation, and perhaps the presence of vitamin D epimers and isobars not accounted for in our assays.It is known that the proportions of circulating vitamin D epimers and isobars from exogenous and endogenous sources can vary significantly [59], and their metabolic activities can differ [60][61][62][63].Future studies shall approach the question whether these sources of variation have any clinical significance.On a final note of caution, the associations reported herein do not allow inference of causality, and there is no highquality evidence that the findings are relevant in terms of health outcomes in breast cancer survivors.These questions and other related questions are under scrutiny in a number of ongoing large-scale, randomised controlled clinical trials, for example, VITAL [64], FIND [65], ViDA [66], DOHealth [67], and VIDAL [68].

Conclusion
Although cholecalciferol is widely supplemented in breast cancer survivors in eastern Switzerland, a remarkable fraction of these patients had serum 25(OH)D below (40%) or above (30%) optimal levels, with only 30% of patients presenting optimal levels.As expected, both 25(OH)D and 1,25(OH)2D levels are higher in patients on cholecalciferol supplementation than in patients without supplementation, but usual supplementation regimens are not adequate to bring 25(OH)D levels to the internationally accepted optimal range for a large proportion of patients.

Figure 2 :
Figure 2: Distribution of the patients according to internationally accepted categories of serum 25-dihydroxyvitamin D (25(OH)D) levels: deficiency (<50 nmol/l), insufficiency (50-74 nmol/l), optimal level (75-100 nmol/l) and physiological level above optimum (>100 nmol/l).Note that the proportion of patients in each category did not change appreciably in the follow-up visit at median 210 days (range 87-857 days) after the first visit.
225 BMI = body mass index; SD = standard deviation Original article patients received 300 000 IU cholecalciferol as a single dose either orally or intramuscularly, whereas the majority of patients preferred to take calcium-vitamin D chewing tablets containing 500 mg calcium and 400 IU cholecalciferol once or twice daily.The 25(OH)D level in those patients with lower 25(OH)D increased after supplementation, as expected, and at the second visit were: median 25(OH)D levels, no supplementation, 65 nmol/l (range 4-161), n = 55; supplementation started before the first visit, 88 nmol/l (range 12-194), n = 121; supplementation started after the first visit 81.5 nmol/l (range 22-180), n = 38; Kruskal Wallis p = 0.002 (fig.7).Factors not associated with vitamin D levels Age, which has been shown to affect 25(OH)D levels in other studies in elderly people [32-35] had no significant correlation with the levels of serum 25(OH)D in our series of mostly middle-aged patients (n = 332, rho = −0.03,p = 0.629).No clear correlation was found between 25(OH)D levels and BMD as assessed with bone densitometry (n = 234, rho = −0.07;p = 0.257).

Figure 4 :
Figure 4: Living in a rural area was associated with a higher median 25(OH)D level as compared with living in an urban area; the same trend was observed for 1,25(OH)2D.

Figure 5 :
Figure 5: Correlation between 25(OH)D and 1,25(OH)2D.Note the large scattering of the data and the flattening of the trend line for 25(OH)D levels higher than 80 nmol/l.

Figure 7 :
Figure 7: Effect of cholecalciferol supplementation on median 25(OH)D levels.Levels of 25(OH)D were higher in patients under newly started cholecalciferol supplementation and also during continued supplementation, as compared with no supplementation.

Figure 8 :
Figure 8: Changes in serum 25(OH)D levels over time after different supplementation dose regimens during winter*.For all health endpoints considered by Bischoff-Ferrari et al [1], the most advantageous serum concentrations of 25(OH)D begin at 75 nmol/l (30 ng/ml), and the best are between 90 and 100 nmol/l (36-40 ng/ml).For most people living in urban areas, this optimal range can be reached with a supplementary cholecalciferol (vitamin D3) dose of 3000 IU daily (alternatively 20 000 IU weekly or 80 000 IU monthly; 1 IU = 0.025 µg vitamin D3; 1 µg vitamin D3 = 40 IU).A loading dose followed by maintenance doses at regular intervals can be considered on an individual base.Serum 25(OH)D determination at baseline and after supplementation may be required for dose adjustment; sub-sequential 25(OH)D determinations are usually not required.*Thedata used to prepare this graphic were obtained from previous reports [54-56].

Table 3 :
25(OH)D level according to season.
The interpretation of regression coefficients as depicted in table 7 is exemplified by the continuous variable age (average 25(OH)D decrease per age year is 0.26) and the categorical variable season (25(OH)D is on average 11.51 higher in spring than in winter).BMI and living in an urban area were significant predictors of 25(OH)D.Season and activity were significant and borderline significant, respectively, when nonsignificant age and BMD were eliminated from the model, so finally the same findings are obtained as in the univariate model (table7; supplementary tables S1, S2, S3 in appendix 1).