gender- and site-specific differences of colorectal neoplasia relate to vitamin d

Gender- and site-specific differences of colorectal neoplasiarelate to vitamin DE. Aigner*,†,‡,1, A. Stadlmayr*,1, U. Huber-Sch€onauer*, J. Zwerina§, E. Husar-Memmer§, D. Niederseer*, M. Trauner¶,A. Heuberger**, F. Hohla*, G. Schett††, W. Patsch‡‡ & C. Datz*,‡

*Department of Internal Medicine,Oberndorf Hospital, Teaching Hospitalof the Paracelsus Medical UniversitySalzburg, Salzburg, Austria.†First Department of Medicine,Paracelsus Medical UniversitySalzburg, Salzburg, Austria.‡Obesity Research Unit, ParacelsusMedical University Salzburg, Salzburg,Austria.§First Medical Department, LudwigBoltzmann Institute of Osteology,AUVA Trauma Center Meidling,Hanusch Hospital, Vienna, Austria.¶Division of Gastroenterology andHepatology, Department of MedicineIII, Medical University of Vienna,Vienna, Austria.**Department of Surgery, OberndorfHospital, Teaching Hospital of theParacelsus Medical UniversitySalzburg, Salzburg, Austria.††Department of Internal Medicine III,Institute for Clinical Immunology,University of Erlangen-Nuremberg,Erlangen, Germany.‡‡Department of Pharmacology andToxicology, Paracelsus MedicalUniversity Salzburg, Salzburg, Austria.

Correspondence to:Dr C. Datz, Department of InternalMedicine, General HospitalOberndorf; Paracelsusstrasse 37,A-5110 Oberndorf, Austria.E-mail: [email protected] and AS contributed equally tothis manuscript as first authors.Publication dataSubmitted 28 July 2014First decision 14 August 2014Resubmitted 29 August 2014Resubmitted 15 September 2014Accepted 15 September 2014EV Pub Online 2 October 2014

This article was accepted for publicationafter full peer-review.

SUMMARY

BackgroundThe effect of vitamin D on colorectal adenomas may vary with regard togender, localisation and histological type of the lesion.

AimTo define the role of vitamin D and gender in a Caucasian cohort of sub-jects undergoing screening colonoscopy after consideration of establishedrisk factors.

MethodsOne thousand five hundred and thirty-two subjects (813 males,58.8 � 9.7 years; 719 females, 59.7 � 10.7 years) were allocated to tertiles of25-hydroxyvitamin D3 [25(OH)D3] serum concentrations. The number, locali-sation, size and histology of the detected colonic lesions were recorded.

ResultsAmong men, no association was found between vitamin D and the total num-ber, size and histological stage of adenomas at any site. In female subjects, lesswomen with adenomas were found in the highest vitamin D tertile (N = 42/239; 17.2%) as compared to the low vitamin D group (N = 60/240; 25.0%;P = 0.035). In particular, the number of women with adenomas in the proxi-mal colon was significantly lower in the highest tertile (N = 21/239, 8.8%)compared to the low vitamin D group (N = 41/240; 17.1%; P = 0.007). Therates at other sites were not different. The inverse association of vitamin Dserum concentrations with the presence of adenomas in the proximal colonwas maintained after adjustment for potential confounders. In 80 women onvitamin D supplementation, the rate of adenomas was lower compared tothose not on supplementation (3/80; 3.8%; vs. 90/719; 12.5%; P = 0.016).

ConclusionsA potential preventive effect of vitamin D on colorectal adenomas wasfound in the proximal colon in women. This observation is supported byfurther decrease of lesions in the proximal colon of women on vitamin Dsupplementation.

Aliment Pharmacol Ther 2014; 40: 1341–1348

ª 2014 John Wiley & Sons Ltd 1341

doi:10.1111/apt.12981

Alimentary Pharmacology and Therapeutics

INTRODUCTIONBesides its well-defined role in calcium and bone homo-eostasis, other nonskeletal effects of vitamin D (VD)have received increasing attention. Mostly epidemiologi-cal data suggest that lower serum VD concentrations arelinked to multiple adverse health-related outcomes suchas autoimmune disease, type 2 diabetes, cardiovasculardisease, asthma and colon cancer.1, 2

Although 1,25(OH)vitamin D3 is the active metabolitebinding to the VD nuclear receptor (VDR), determina-tion of the relatively stable precursor molecule 25(OH)vitamin D3 is commonly used to assess VD status.Higher serum concentrations of VD may decrease cancerrisk by facilitating apoptosis and differentiation and byinhibiting proliferation, invasion and neoangiogenesis.3, 4

Epidemiological studies already suggested a lower riskfor colorectal adenomas with higher sun exposure morethan 30 years ago.5 A recent meta-analysis suggested a7% risk reduction for colorectal adenoma per 10 ng/mLincrease in VD serum concentration.6 A 34% risk reduc-tion of the top vs. the lowest quintile of VD serum con-centrations was also reported.7

The chemopreventive effect of adequate VD serumconcentration may be different in men and women andalso be different at various anatomical sites of the colon,i.e. proximal colon, distal colon or rectum.8 In Japanesemen, a higher prevalence of distal colonic adenoma waslinked to low VD status.9 A reduced risk for distal ade-noma was found particularly in women with higher VDintake in a sigmoidoscopy study.10

Molecular interactions have been demonstratedbetween oestrogen and VD in the colonic epitheliumproviding a potential molecular explanation for the dif-ferences observed between men and women. In particu-lar, oestrogens may increase colonic 1-a-hydroxylase inthe colonic epithelium and thereby increasing intracellu-lar availability of the active VD metabolite. In addition,oestrogens appear to increase VD receptor levels in thecolonic epithelium.11

As these reports suggest a potentially profound effectof sex hormones on the effect of VD in carcinogenesis,we hypothesised that the associations between VD statusand colorectal polyps could differ between men andwomen in a screening cohort. We thus performed adetailed analysis of the data obtained from a Caucasiancohort undergoing screening colonoscopy. We aimed tostudy the rate, histology and localisation of the colorectallesions in men and women according to their VD statusand after consideration of confounding risk factors such

as gender, age, a family history of colorectal cancer(CRC), smoking status and impaired glucose metabolism.

MATERIAL AND METHODS

Study conceptFrom all study participants, a detailed drug and medicalhistory and a routine physical examination wereobtained. Following an overnight fast, venous blood wascollected and an oral glucose tolerance test (OGTT) wasperformed. On the following day subjects underwent col-onoscopy. The design and details of the study have beenreported previously.12 Screening colonoscopies were per-formed in subjects with average risk or with a family his-tory of CRC in the absence of any gastrointestinalsymptoms according to the recommendation of the Aus-trian Society of Gastroenterology and Hepatology(OEGGH) and the Austrian Cancer Aid.13

Study subjectsFrom a total of 891 initially screened male subjects, 59(6.6%) were excluded from the study because of incom-plete colonoscopies (N = 15), a history of previous colo-rectal polypectomy (N = 27), newly diagnosed and to dateasymptomatic inflammatory bowel disease (N = 3), otherextra-intestinal malignancies (N = 11) or systemic auto-immune diseases (N = 3; rheumatoid arthritis, systemiclupus erythematosus, autoimmune hepatitis). Fifty-two(8.9%) women of 851 initially screened were excludedbecause of incomplete colonoscopies (N = 20), a historyof previous colorectal polypectomy (N = 21), to dateasymptomatic inflammatory bowel disease (N = 2),extra-intestinal malignancies (N = 3), or systemicautoimmune disease (N = 6). Thus, the study cohort con-sisted of 832 consecutive caucasian males (aged between33 and 87 years) and 799 females (31–88 years), whounderwent colonoscopy for CRC screening according tonational screening recommendations for CRC at a singlecentre from October 2010 to January 2013. Any study par-ticipants regularly taking VD or VD in combination withcalcium at the time of colonoscopy were defined as VDusers. Most participants on VD substitution took 800 IU/day (range 400–1200 IU/day over 8.8 � 5.3 years). Eightywomen and 13 men were on current VD supplementationand these were analysed separately, hence, data from 819males (85.9%) and 719 females (89.4%) were included inthe final analysis. These were allocated to sex-specific ter-tiles according to serum VD concentrations obtained atthe time of colonoscopy. Family history of CRC was

1342 Aliment Pharmacol Ther 2014; 40: 1341-1348

ª 2014 John Wiley & Sons Ltd

E. Aigner et al.

obtained from all participants.14 The study was approvedby the local ethics committee and informed consent wasobtained from all participants.

Laboratory assessmentFull blood counts were obtained in all subjects by standardlaboratory methods. Erythrocyte sedimentation rate wasmeasured in citrate plasma. Vitamin D was measured bythe electrochemiluminescence-based Cobas e 411 analyser(TM) employing the respective Elecsys (TM) reagents(Roche Diagnostics GmbH, Mannheim, Germany). Thelaboratory performed regular quality control measure-ments, including a pooled serum sample analysis withbatches of study samples to monitor precision and identifypossible laboratory shifts over time, as well as testingduplicates in different batches. The coefficient of variationwas <6% for 25(OH)D3 between-batches analyses. Allanalyses were conducted in a blinded fashion. A standar-dised OGTT was performed with 75 g of glucose in300 mL of water. HbA1c was measured by HPLC usingAdamts H-8160 (Menarini, Florence, Italy). The homoeo-stasis model assessment [HOMA-IR; fasting insulin (lU/L) 9 fasting glucose (mmol/dL)/22.5] was used to assessinsulin resistance (IR). Type 2 diabetes was classified asuse of diabetes medication or Hba1c ≥6.5% or OGTT>11.1 mmol/L after 2 h or fasting glucose >7.0 mmol/L.

ColonoscopyThe laxative Klean–Prep� (containing macrogol 59.0 g,sodium sulphate 5.68 g, sodium bicarbonate 1.68 g, NaCl1.46 g and potassium chloride 0.74 g; Norgine, Marburg,Germany) was used for bowel preparation before colo-noscopy. Colonoscopic findings were classified as tubularadenoma, advanced neoplasia, including polyps with vil-lous or tubulovillous features, size ≥1 cm or high-gradedysplasia or carcinoma after a combined analysis of mac-roscopic and histological results.15 Hyperplastic polypswere not counted.16 Lesions were classified by location(i.e. proximal colon including caecum, ascending colonand transverse colon, distal colon ranging from the sple-nic flexure to the sigmoid and rectum alone).

Statistical analysisFor all analyses SigmaStat 3.1 or STATA 8.0 softwarepackages were used. Data are presented as mean � s.d.,unless otherwise indicated. ANOVA was used for compari-son of continuous variables. The Pearson v2 test was usedto compare rates and proportions. The Fisher’s exact testwas used for cell frequencies <10. Since the results of uni-variate analyses showed associations only in women, a

multivariate logistic regression analysis was performed inwomen to control for potential confounders.

In multivariate logistic regression analysis, the absenceor presence of adenomas in the proximal colon was used asdependent variable and age, body mass index (BMI), sea-sonal variation, glucose metabolism and VD as indepen-dent variables. Seasonal variation was defined by seasonalquarters: 1, January–March; 2, April–June; 3, July–Septem-ber; 4, October–December. In the regression model, sea-sonal quarters were coded 0, 1, 2 and 3 for the first, second,fourth and third seasonal quarter, respectively. Normaland impaired glucose metabolism including type 2 diabetesand/or impaired fasting glucose and/or impaired OGTTand was coded 0 and 1, respectively. Throughout, a two-tailed P < 0.05 was considered statistically significant.

RESULTS

Characteristics of the study cohortData from 819 males and 719 females were included inthe analysis after exclusion of subjects on VD supple-mentation (13 men and 80 women). The clinical andbiochemical characteristics of these subjects are summar-ised in Table 1. As diabetes and anti-diabetic medicationmay affect the incidence of colorectal adenomas, allcalculations were initially performed with and withoutthe exclusion of these subjects. None of the results weresignificantly different after the exclusion of subjects with

Table 1 | Clinical and biochemical characteristics ofmen and women

Variable Men Women P

SeasonalQuarters

224/246/166/183 179/220/172/157 N.S.

Age (years) 58.7 � 9.7 59.3 � 10.7 N.S.BMI (kg/m2) 27.6 � 4.1 26.7 � 5.6 <0.001Weight (kg) 86.3 � 13.9 71.5 � 15.2 <0.001Waist/hipratio

0.978 � 0.063 0.883 � 0.085 <0.001

Diabetes(y/n)

142/677 103/616 N.S.

HbA1c (%) 5.76 � 0.65 5.77 � 0.69 N.S.25 (OH)Vitamin D3

(ng/mL)*

24.6 � 10.9 22.8 � 11.6 <0.001

BMI, body mass index; HbA1c, glycated haemoglobin A1c.

* Seasonally adjusted serum concentrations. Seasonal quarterswere defined as January–March (0); April–June (1); July–Sep-tember (3); October–December (2). Numbers in bracketswere used for calculations according to the increasing VDserum concentrations measured.

Aliment Pharmacol Ther 2014; 40: 1341-1348 1343


Vitamin D and colorectal neoplasia in women

diabetes. Therefore, all results presented are from calcu-lations including the subjects with diabetes in men andwomen. Likewise, fourteen women were on hormonereplacement therapy (HRT) and the results of calcula-tions were not different when these were excluded. VDserum concentrations were significantly higher in menthan in women (Table 1). There was a significant sea-sonal variation of VD concentrations in men and womenand gender differences were maintained through all sea-sons. The lowest concentrations were measured in thefirst quarter, followed by the second, fourth and thirdquarter (P < 0.001 in men and women). Data on sea-sonal variation are summarised in Figure S1.

Prevalence, localisation and histology of colorectalneoplasias in men and womenMen and women were equally apportioned to tertiles ofVD serum concentrations and the rate of adenomas ateach colonic site, histology and size were recorded. Therate of adenomas in men was significantly higher com-pared to the rate in women (P < 0.001). The differencebetween men and women was significant in all sub-categories presented except for advanced adenomas andcancers (details not shown). In men, no differences werefound in the total number, histological results or numberof adenomas at any colonic site among the three VDcategories. However, in women the total number ofneoplastic lesions was inversely correlated with VD serumconcentrations. When investigating the localisation ofadenomas, we found that this reduction was mainly dueto a significant reduction in the rate of adenomas in theproximal colon whereas the rate of adenomas in the distalcolon or the rectum was not different among the VD ter-tiles in females. Additionally, both carcinomas weredetected in the proximal colon of women of the lowestVD tertile. The detailed summary of these analyses isgiven in Table 2. There was no association of adenomasize with VD concentrations in either men or women(data not shown). In total, 199 of 1631 subjects (12.2%)had a positive family history for CRC. Of note, a positivefamily history was not associated with colorectal polypsat any location or the total number of polyps in ourcohort in neither men nor women or in both combined.Hence, family history did not meet the criteria to beincluded as a parameter in the regression models.

We then analysed colonic lesions in females in furtherdetail. In the proximal colon (120 lesions), we found 31(25.8%) flat, 4 (3.3%) pedunculated, 83 (69.2%) sessilelesions and 2 (1.7%) carcinomas; in the distal colon (78lesions), there were 9 (11.5%) flat, 11 (14.1%) pedunculated,

58 (74.4%) sessile lesions; in the rectum (19 lesions), thecorresponding figures were 1 (5.3%), 4 (21.0%) and 14(73.7%). Thus, a higher number of flat adenomas was foundin the proximal colon (P = 0.015) compared to the distalcolon. Fifteen flat adenomas were found in the lowest tertilecompared to 6 in the highest (P = 0.043). Three subjectshad ≥2 flat lesions in the proximal colon in the low VDgroup and one in the highest tertile; only one subject had 2flat lesions in the distal colon and none in the rectum. Thesedifferences were not significant.

Prevalence of colorectal adenomas in subjects oncurrent VD supplementationNinety-three subjects (13 males and 80 females) with cur-rent VD supplementation were identified. These subjectshad higher VD serum concentrations compared to non-u-sers of VD supplementation (Table 3). Due to the smallnumber of men on supplementation all analyses wereunderpowered and the results are not reported. Amongwomen, the difference between users and non-users of VDsupplementation was not statistically significant withregard to the number of subjects with adenomas, the totalnumber of lesions, size or morphological subtypes. How-ever, in the proximal colon 3 (3.75%) of women on sup-plementation had adenomas compared to 90 (12.5%) noton supplementation (P = 0.016; Fisher’s test). Likewise,the numbers of subjects with 1 (N = 2/80 vs. 70/719) and2 or more adenomas (N = 1/80 vs. 20/719) also werelower in users (P = 047; Fisher’s test). No significant dif-ferences were detected in the distal colon or rectum.

Multivariate logistic regression analysisUnivariate correlation analysis was performed to identifypotential confounders for the association of VD with prox-imal colon adenomas in women. All variables with a sig-nificance level P < 0.05 were considered and afterchecking for plausibility the following variables wereselected for a multivariate logistic regression analysis: age,seasonal variation, BMI, glucose metabolism. VD serumconcentrations remained independently associated withthe absence/presence of proximal colon adenomas in themultivariate regression model (Table 4). Smoking status(current, past, or combined) was not related to the rate ofadenomas in men or women (data not shown) and wastherefore not included in the multivariate model.

DISCUSSIONIn this cross-sectional study in subjects undergoing theirfirst screening colonoscopy, we found a lower rate ofcolorectal adenomas in women than in men. Analysis of



E. Aigner et al.

number, histology, size and localisation revealed no asso-ciation of VD status with colorectal lesions in men, how-ever, the number of adenomas detected in the proximalcolon of women with VD concentrations in the highesttertile was significantly decreased compared to the lowestVD tertile. In addition, the number of women withadenomas of the proximal colon was lower in thosecurrently taking VD supplementation compared to VDsupplementation non-users.

Available data from epidemiological studies are incon-sistent with regard to the preventive effect of VD on theincidence of colorectal lesions. Meta-analyses providestrong evidence that on the population level higher VD

concentrations are linked to fewer adenomas17, 18 andalso CRC.19 However, the individual study results varyconsiderably with regard to the potential effect of VD atspecific colonic sites and also the number and size of ade-nomas or study endpoints (i.e. incidence adenomas vs.recurrence).20–22 In the Nurses0 Health Study, estimatedVD and calcium intake was linked to a lower incidencerate of distal adenomas.10 In our study, we observed asignificantly lower rate of adenomas with higher VDserum concentrations only in the proximal colon inwomen. The strengths of this investigation are its detailedrecording of the number, size, localisation and histology,suggesting that the supposed chemopreventive effect of

Table 2 | Summary of colorectal lesions found in men and women analysed with regard to number, histology andlocalisation. The figures in the left column refer to the respective number of adenomas and carcinomas detected.Overall, the rate of colorectal adenomas and cancers in women was lower compared to men. However, only in theproximal colon a significantly lower rate of adenomas was found in subjects considered to have higher VD serumconcentrations. The size of adenomas in the various groups is not presented as there were no differences in eithermen or women. Advanced neoplasia include adenomas with villous features (>25%), size of 1.0 cm or more, high-grade dysplasia, or early invasive cancer. Both carcinomas in women were found in the lowest VD group in theproximal colon but no difference was found in men

Men Women

VD tertile (ng/mL) Low (<19.3)Interm.

(19.3–28.6) High (>28.7) P-value Low (<16.5)Interm.

(16.5–27.1) High (>27.2) P-value

N 273 273 273 240 240 239N of lesionsAny type of lesion 92 (33.7%) 93 (34.1%) 106 (38.8%) 0.377 60 (25%) 43 (17.9%) 42 (17.2%) 0.0351 50 (18.32%) 55 (20.2%) 68 (24.9%) 39 (16.3%) 31 (12.9%) 31 (13.0%)>1 42 (15.4%) 38 (13.9%) 38 (13.9%) 0.418* 21 (8.8%) 12(5.0%) 10 (4.2%) 0.056*N, histologyTubular adenoma 81 (29.7%) 82 (30.0%) 92 (33.7%) 0.531 52 (21.7%) 39 (16.3%) 35 (14.6%) 0.0461, tubular 47 (17.2%) 49 (18.0%) 57 (20.9%) 36 (15.0%) 29 (12.1%) 27 (11.3%)>1, tubular 34 (12.5%) 33 (12.1%) 35 (12.8%) 0.820* 16 (6.7%) 10 (4.2%) 8 (3.4%) 0.100*Tubulovillous 7 (2.6%) 8 (2.9%) 8 (2.9%) 0.956 6 (2.5%) 5 (2.1%) 7 (2.9%) 0.8381, advanced 8 (2.9%) 11 (4.0%) 11 (4.0%) 9 (3.8%) 6 (2.5%) 9 (3.8)>1, advanced 4 (1.5%) 2 (0.7%) 1 (0.4%) 0.361* 1 (0.4%) 0 0 1.00*Cancer 1 (0.4%) 4 (1.5%) 1 (0.4%) 0.380 2 (0.8%) 0 0 0.332Advanced neoplasia 13 (4.8%) 15 (5.5%) 13 (4.8%) 0.902 12 (5%) 6 (2.5%) 9 (3.8%) 0.354LocalisationProximal colon 50 (18.3%) 61 (22.3%) 62 (22.7%) 0.377 41 (17.1%) 28 (11.7%) 21 (8.8%) 0.0071, proximal colon 36 (13.2%) 46 (16.9%) 50 (18.3%) 29 (12.1%) 23 (9.5%) 18 (7.5%)>1, proximal colon 14 (5.1%) 15 (5.5%) 12 (4.4%) 0.528* 12 (5.0%) 5 (2.1%) 3 (1.2%) 0.013*Distal colon 43 (15.8%) 43 (15.8%) 42 (15.4%) 0.989 27 (11.3%) 17 (7.1%) 18 (7.5%) 0.2101, distal colon 26 (9.5%) 30 (11.0%) 34 (12.5%) 22 (9.2%) 13 (5.4%) 13 (5.5%)>1, distal colon 17 (6.2%) 13 (4.8%) 8 (3.0%) 0.117* 5 (2.1%) 4 (1.7%) 5 (2.1%) 0.279*Rectum 16 (5.9%) 17 (6.2%) 24 (8.8%) 0.341 9 (3.8%) 2 (0.8%) 6 (2.5%) 0.1021, rectum 14 (5.1%) 17 (6.2%) 21 (7.7%) 9 (3.8%) 2 (0.8%) 6 (2.5%)>1, rectum 2 (0.7%) 0 (0.0%) 3 (1.1%) 0.424* 0 (0.0%) 0 (0.0%) 0 (0.0%) 0.354*

VD, vitamin D serum concentrations.

* P-value referring to the comparison for the number of adenomas in the respective regions between the first and third tertile;other P-values referring to the comparison of presence or absence of adenomas in the respective region between the first andthe third tertiles.




VD on colorectal adenoma is most pronounced in theproximal colon. The absence of associations in the malesubgroup argues that oestrogen activity may be importantfor the inverse association between VD and adenomas.Although an effect in the same direction was observedboth in the distal colon and also the rectum, these resultsmay not have reached significance due to the limitednumber of subjects in our female study population.

The higher total rate of colorectal adenomas in men inour study is in line with the recognised male predomi-nance of CRC prevalence23, 24 In addition to epidemiolog-ical findings, the use of HRT in the Women0s HealthInitiative trial was linked to a 56% risk reduction inCRC.25 In our study, HRT did not impact the results sincethe number of women using HRT was very low. This gen-der disparity has also been confirmed for advanced adeno-mas and early colorectal lesions.26, 27 Furthermore, areduction of colon cancer was found in women withhigher soy consumption which contains high amounts ofphytoestrogens.28 In an elegant intervention oestrogensubstitution for 1 month in post-menopausal womenshowed a significant increase on VDR and the VD meta-bolising enzymes CYP24A1 and CYP27B1 in the rectalmucosa.29 These findings suggest, that oestrogens can

increase both the cellular 1,25(OH)D3 concentrationtogether with its nuclear receptor in epithelial cells of thecolon, thereby augmenting the biological effectiveness ofVD. The same group of authors also provided evidencethat modulation of 1,25(OH)D3 concentrations in colonepithelium by 17b-estradiol is specific for the proximalcolon.30 Importantly, we observed generally lower VDconcentrations in women compared to men, however, arelationship of VD with colorectal adenomas was onlyobserved in women. This finding additionally supports theconcept of a permissive effect of oestrogen for VD on thedevelopment of colorectal adenomas, since no such associ-ation was observed in men despite higher VD concentra-tions. Additionally, well-recognised molecular interactionsbetween the potentially carcinogenic bile acid lithocholicacid, VD, VDR or other intestinal nuclear receptors propa-gating bile acid detoxification may underlie our observa-tions of lower rates of colon cancer precursor lesions inwomen with higher VD serum concentrations.31, 32

Notably, flat adenomas were significantly higher in theproximal colon in women, particularly in the lowest VDgroup, and both cancers in women were located in theproximal colon in the lowest VD tertile. These findings,although low in number, lend support to the speculationthat VD may play a particular role in the pathogenesis offlat lesion and cancers arising from these lesions as opposedto pedunculated lesions. Flat lesions of the proximal colonpresent with distinct biological and clinical features with

Table 3 | (a) Comparison of serum vitamin Dconcentrations in subjects on supplementation to thosenot using supplementation. In men and women, usersof supplementation had significantly higher VDconcentrations. (b) Prevalence of colorectal lesions inusers and non-users of vitamin D supplementation. Adifference in the presence of colorectal adenomas wasonly found in the proximal colon in female subjects.Any study participants regularly taking VD or VD incombination with calcium at the time of colonoscopywere defined as VD users

Non-users Users P-value

(a) VD serum concentration (ng/mL)All 23.5 � 11.3 30.1 � 11.3 <0.001Men 24.5 � 11.0 31.2 � 14.6 0.021Women 22.8 � 11.7 29.6 � 10.7 <0.001(b) WomenN 719 80Any type of lesion 144 (20.3%) 11 (13.8%) 0.177Tubular adenoma 126 (17.5%) 10 (12.5%) 0.256Tubulovillousadenoma

18 (2.5%) 1 (1.3%) 0.417

Advanced adenoma 25 (3.5%) 1 (1.3%) 0.551Cancer 2 (0.28%) 0 1.00Proximal colon 90 (12.5%) 3 (3.8%) 0.016Distal colon 62 (8.6%) 7 (8.8%) 1.00Rectum 17 (2.4%) 3 (3.8%) 0.441

Table 4 | Results of the multivariate logistic regressionanalysis adjusting for potential confounders of theassociation of VD serum concentrations with thepresence/absence of adenomas in the proximal colonin women. Seasonal quarters were defined as January–March (1); April–June (2); July–September (4); October–December (3). Numbers in brackets were used forcalculations according to the increasing VD serumconcentrations measured; glucose metabolism wascoded 0 and 1 for normal and impaired (including type2 diabetes and/or impaired fasting glucose and/orimpaired glucose tolerance), respectively

Parameter Odds ratio 95% CI P-value

Seasonal variation(quarters)

0.809 0.652–1.005 0.055

Age (years) 1.044 1.020–1.068 <0.001BMI (kg/m²) 0.959 0.917–1.002 0.060Glucose metabolism(impaired/normal)

1.666 1.017–2.730 0.043

Vitamin D serum conc.(ng/mL)

0.976 0.954–0.999 0.040



E. Aigner et al.

regard to gender differences and their malignantpotential.33, 34 The similarities between the results reportedin that study and ours additionally suggest that VD mayplay a role in the development of flat lesions. Recent studiessuggested that gender not only has an impact on the overallrisk of developing CRC but also on its localisation with apreponderance of large polyps in women in the proximalcolon.13, 35, 36 Our results suggest that VD may underliethese observations on the population level.

Inherent to the design of a cross-sectional study, ourresults are limited by single determination of VD,whereas the development of adenomas and particularlyCRC is a process of several years. However, a recentinvestigation demonstrated that VD concentrations wererelatively stable over 14 years, suggesting that subjectsdeficient in VD remain so for long periods of time.37 Anadditional limitation of our study is the fact that adetailed analysis of menopausal state of women hasnot been performed. Participants fulfilled criteria forscreening colonoscopy and were thus generally older than50 years with a mean age of 59.7 � 10.7 years, suggest-ing that the large majority of study participants werepost-menopausal. Since the assumed interaction betweenVD and oestrogen could potentially depend on the tem-poral relationship to menopause further studies on thetime course of our observations appear indicated.

Although our investigation was not designed as anintervention study, we separately analysed the data fromwomen who were on current VD supplementation. SerumVD concentrations in these subjects were elevated com-pared to those not on VD supplementation and similar tothe analysis of female VD tertiles, we observed a selectivedecrease in adenomas of the proximal colon in female VDusers compared to the non-user cohort. This finding addsfurther, although only circumstantial evidence to thepotential chemopreventive effect of VD on colorectaladenomas which appears most pronounced in the proxi-mal colon. Providing additional clinical-genetic evidencefor a selective role of VD in the pathogenesis of proximalcolon adenoma, Egan et al. demonstrated that an allelicvariation in the RXR alpha, which forms the heterodimerwith VDR, was associated with decreased risk of colorectaladenoma recurrence, particularly in proximal colon.38

In summary, we provide clinical evidence that achemopreventive effect of VD may be only found infemales. Furthermore, this effect appears to be most pro-nounced in the proximal colon and only of minor rele-vance at more distal sites. These findings suggest thatfurther research on VD and colorectal neoplasia shouldinclude analysis of gender- and site-specific mechanismsin cancer development and prevention.

AUTHORSHIPGuarantor of the article: Christian Datz.Author contributions: EA: analysis of data, drafting andwriting of the manuscript; AS: acquisition of data, draft-ing of the manuscript; UHS, FH, DN, AH, EHM: acqui-sition of data, JZ, GS: acquisition of data, critical revisionof the manuscript for important intellectual content,MT: interpretation of data and critical revision of themanuscript for important intellectual content, WP:analysis and interpretation of data, statistical analysis,outlining and revising the manuscript, CD: study conceptand design, analysis and interpretation of data, outliningand revising the manuscript. All authors approved thefinal version of the manuscript.

ACKNOWLEDGEMENTSWe gratefully acknowledge laboratory technical supportby Elke Albrecht, Monika Ratkowitsch, Carmen Winkler,Oberndorf Hospital and Angela Eich, First Departmentof Medicine, Paracelsus Medical University Salzburg.Declaration of personal interests: None.Declaration of funding interests: Support from SPAR Aus-tria to Christian Datz is gratefully acknowledged. ElmarAigner is supported by PMU-Forschungsf€orderungsfonds(E-13/17/086-AIG).

SUPPORTING INFORMATIONAdditional Supporting Information may be found in theonline version of this article:Figure S1. Seasonal variation of vitamin D levels in

men and women. Seasonal quarters are expressed as col-umns (1: January–March; 2: April–June; 3: July–Septem-ber; 4: October–December). Vitamin D levels are shownas the means in ng/mL.

REFERENCES1. Christakos S, Hewison M, Gardner

DG, et al. Vitamin D: beyond bone.Ann N Y Acad Sci 2013; 1287: 45–58.

2. Zhang X, Giovannucci E. Calcium,vitamin D and colorectal cancerchemoprevention. Best Pract Res ClinGastroenterol 2011; 25: 485–94.

3. Frampton RJ, Omond SA, Eisman JA.Inhibition of human cancer cell growthby 1,25-dihydroxyvitamin D3




metabolites. Cancer Res 1983; 43:4443–7.

4. Palmer HG, Gonzalez-Sancho JM,Espada J, et al. Vitamin D(3) promotesthe differentiation of colon carcinomacells by the induction of E-cadherinand the inhibition of beta-cateninsignaling. J Cell Biol 2001; 154:369–87.

5. Garland CF, Garland FC. Do sunlightand vitamin D reduce the likelihood ofcolon cancer? Int J Epidemiol 1980; 9:227–31.

6. Lee JE. Circulating levels of vitamin D,vitamin D receptor polymorphisms, andcolorectal adenoma: a meta-analysis.Nutr Res Pract 2011; 5: 464–70.

7. Lee JE, Li H, Chan AT, et al.Circulating levels of vitamin D andcolon and rectal cancer: the Physicians’Health Study and a meta-analysis ofprospective studies. Can Prev Res 2011;4: 735–43.

8. Jacobs ET, Hibler EA, Lance P, SardoCL, Jurutka PW. Association betweencirculating concentrations of 25(OH)Dand colorectal adenoma: a pooledanalysis. Int J Cancer 2013; 133: 2980–8.

9. Takahashi R, Mizoue T, Otake T, et al.Circulating vitamin D and colorectaladenomas in Japanese men. Cancer Sci2010; 101: 1695–700.

10. Oh K, Willett WC, Wu K, Fuchs CS,Giovannucci EL. Calcium and vitaminD intakes in relation to risk of distalcolorectal adenoma in women. Am JEpidemiol 2007; 165: 1178–86.

11. Cross HS, Nittke T, Peterlik M.Modulation of vitamin D synthesis andcatabolism in colorectal mucosa: a newtarget for cancer prevention. AnticancerRes 2009; 29: 3705–12.

12. Stadlmayr A, Aigner E, Steger B, et al.Nonalcoholic fatty liver disease: anindependent risk factor for colorectalneoplasia. J Intern Med 2011; 270:41–9.

13. Ferlitsch M, Reinhart K, Pramhas S,et al. Sex-specific prevalence ofadenomas, advanced adenomas, andcolorectal cancer in individualsundergoing screening colonoscopy.JAMA 2011; 306: 1352–8.

14. Lieberman DA, Rex DK, Winawer SJ,Giardiello FM, Johnson DA, Levin TR.Guidelines for colonoscopy surveillanceafter screening and polypectomy: aconsensus update by the US Multi-Society Task Force on ColorectalCancer. Gastroenterology 2012; 143:844–57.

15. Hassan C, Quintero E, DumonceauJM, et al. Post-polypectomycolonoscopy surveillance: European

Society of Gastrointestinal Endoscopy(ESGE) Guideline. Endoscopy 2013; 45:842–51.

16. Adams SV, Newcomb PA, Burnett-Hartman AN, White E, Mandelson MT,Potter JD. Circulating25-hydroxyvitamin-D and risk ofcolorectal adenomas and hyperplasticpolyps. Nutr Cancer 2011; 63: 319–26.

17. Yin L, Grandi N, Raum E, Haug U,Arndt V, Brenner H. Meta-analysis:serum vitamin D and colorectaladenoma risk. Prev Med 2011; 53: 10–6.

18. Gandini S, Boniol M, Haukka J, et al.Meta-analysis of observational studiesof serum 25-hydroxyvitamin D levelsand colorectal, breast and prostatecancer and colorectal adenoma. Int JCancer 2011; 128: 1414–24.

19. Yin L, Grandi N, Raum E, Haug U,Arndt V, Brenner H. Meta-analysis:longitudinal studies of serum vitamin Dand colorectal cancer risk. AlimentPharmacol Ther 2009; 30: 113–25.

20. Platz EA, Hankinson SE, Hollis BW,et al. Plasma 1,25-dihydroxy- and 25-hydroxyvitamin D and adenomatouspolyps of the distal colorectum. CancerEpidemiol Biomarkers Prev 2000; 9:1059–65.

21. Peters U, Hayes RB, Chatterjee N, et al.Circulating vitamin D metabolites,polymorphism in vitamin D receptor,and colorectal adenoma risk. CancerEpidemiol Biomarkers Prev 2004; 13:546–52.

22. Jacobs ET, Alberts DS, Benuzillo J,Hollis BW, Thompson PA, MartinezME. Serum 25(OH)D levels, dietaryintake of vitamin D, and colorectaladenoma recurrence. J Steroid BiochemMol Biol 2007; 103: 752–6.

23. Brenner H, Hoffmeister M, Arndt V,Haug U. Gender differences incolorectal cancer: implications for ageat initiation of screening. Br J Cancer2007; 96: 828–31.

24. DeCosse JJ, Ngoi SS, Jacobson JS,Cennerazzo WJ. Gender and colorectalcancer. Eur J Cancer Prev 1993; 2:105–15.

25. Hartz A, He T, Ross JJ. Risk factors forcolon cancer in 150,912postmenopausal women. Cancer CausesControl 2012; 23: 1599–605.

26. Lieberman DA, Holub J, Eisen G,Kraemer D, Morris CD. Prevalence ofpolyps greater than 9 mm in aconsortium of diverse clinical practicesettings in the United States. ClinGastroenterol Hepatol 2005; 3:798–805.

27. Parente F, Bargiggia S, Boemo C, et al.Anatomic distribution of cancers and

colorectal adenomas according to ageand sex and relationship betweenproximal and distal neoplasms in ani-FOBT-positive average-risk Italianscreening cohort. Int J Colorectal Dis2014; 29: 57–64.

28. Yan L, Spitznagel EL, Bosland MC. Soyconsumption and colorectal cancer riskin humans: a meta-analysis. CancerEpidemiol Biomarkers Prev 2010; 19:148–58.

29. Protiva P, Cross HS, Hopkins ME,et al. Chemoprevention of colorectalneoplasia by estrogen: potential role ofvitamin D activity. Cancer Prev Res2009; 2: 43–51.

30. Nittke T, Kallay E, Manhardt T, CrossHS. Parallel elevation of colonic 1,25-dihydroxyvitamin D3 levels andapoptosis in female mice on a calcium-deficient diet. Anticancer Res 2009; 29:3727–32.

31. Makishima M, Lu TT, Xie W, et al.Vitamin D receptor as an intestinalbile acid sensor. Science 2002; 296:1313–6.

32. Evans RM, Mangelsdorf DJ. NuclearReceptors, RXR, and the Big Bang. Cell2014; 157: 255–66.

33. Voorham QJ, Rondagh EJ, Knol DL,et al. Tracking the molecular featuresof nonpolypoid colorectal neoplasms: asystematic review and meta-analysis.Am J Gastroenterol 2013; 108:1042–56.

34. Rondagh EJ, Masclee AA, van derValk ME, et al. Nonpolypoidcolorectal neoplasms: genderdifferences in prevalence and malignantpotential. Scand J Gastroenterol 2012;47: 80–8.

35. Lieberman DA, Williams JL, Holub JL,et al. Race, ethnicity, and sex affect riskfor polyps >9 mm in average-riskindividuals. Gastroenterology 2014; 147:351–8.

36. Regula J, Rupinski M, Kraszewska E,et al. Colonoscopy in colorectal-cancerscreening for detection of advancedneoplasia. N Engl J Med 2006; 355:1863–72.

37. Jorde R, Sneve M, Hutchinson M, EmausN, Figenschau Y, Grimnes G. Trackingof serum 25-hydroxyvitamin D levelsduring 14 years in a population-basedstudy and during 12 months in anintervention study. Am J Epidemiol 2010;171: 903–8.

38. Egan JB, Thompson PA, Ashbeck EL,et al. Genetic polymorphisms in vitaminD receptor VDR/RXRA influence thelikelihood of colon adenoma recurrence.Cancer Res 2010; 70: 1496–504.



E. Aigner et al.

gender- and site-specific differences of colorectal neoplasia relate to vitamin d

Documents