predictors of toenail selenium levels in men and women1 · these values were 0.548 ± 0.101 and...

Vol. 2, 107- 1 12, March/April 1993 Cancer Epidemiology, Biomarkers & Prevention 107

Predictors of Toenail Selenium Levels in Men and Women1

Piet A. van den Brandt,2 R. Alexandra Goldbohm,Pieter van’t Veer, Peter Bode, Rudolph J. J. Hermus,and Ferd Sturmans

Department of Epidemiology, University of Limburg, Maastricht[P. A. van den B., F. S.]; Department of Nutrition, TNO-Toxicology andNutrition Institute, Zeist [R. A. G., P. van ‘I V., R. J. J. H.]; and

Interfaculty Reactor Institute, Delft University of Technology, Delft[P. B.], the Netherlands

Abstract

Potential predictors of toenail selenium levels werestudied in 1211 men and 1248 women aged 55-69years. These subjects were randomly selected cohortmembers without prevalent cancer (other than skin)participating in a prospective study on diet and cancerin the Netherlands. Information on the consideredpotential predictors (gender, age, smoking, intake ofdietary selenium and alcohol, Quetelet index) wascollected together with toenail specimens in 1986. Theaverage toenail selenium concentration wassignificantly (P < 0.001) lower in men than in women:0.547 ± 0.126 �ig/g (mean ± SD) and 0.575 ± 0.109ILg/g, respectively. The gender difference remainedsignificant after adjustment for the other variables inmultiple regression analyses. Age was not associatedwith toenail selenium levels in men or women. Aninverse association was observed with current smokingbut not with past smoking. The average toenailselenium values for male current smokers were 0.513 ±0.106 ILg/g (mean ± SD) versus 0.571 ± 0.133 �g/g formale never- or ex-smokers (P < 0.001). For womenthese values were 0.548 ± 0.101 and 0.581 ± 0.109 �g/g, respectively (P < 0.001). Dietary selenium intakewas positively associated with toenail selenium levelsin multivariate analyses (P < 0.001), but the associationwas weak (partial r 0.09). Alcohol intake andQuetelet index were not significant independentpredictors of toenail selenium. The observedassociations had similar directions in men and womenbut were stronger in men.

Introduction

Several reports on the association between low tissueselenium levels and the risk of cancer and cardiovasculardisease (1-4) have increased the interest in the role ofselenium in disease etiology during the past decade.Estimation of dietary selenium intake in epidemiological

Received 6/24/92.1 Supported by the European Commission (Europe Against Cancer) andthe Dutch Cancer Society.2 To whom requests for reprints should be addressed, at Department of

Epidemiology, University of Limburg, P. 0. Box 616, NL-6200 MD Maas-

tricht, the Netherlands.

studies is considered unreliable, since the selenium con-tent of foods may vary considerably between varieties ofthe same type of food, depending on the soil where thefood was grown (5). Therefore, epidemiological studieson the relation between selenium and chronic diseasesoften rely on biological markers of selenium status. Es-pecially in relation to diseases with long latency periodssuch as cancer, these markers should preferably reflectlong-term selenium status. Available markers include se-lenium levels in whole blood, serum, erythrocytes, urine,hair, and nails and glutathione peroxidase activity (5-7).Of these, urinary and serum selenium levels reflect short-term changes in dietary selenium intake (8, 9), and it hasbeen found that serum selenium levels may be influ-enced by the presence of(preclinical) disease (10). Long-term markers of selenium status include hair, erythro-cytes, and nails (8, 1 1), while whole blood appears totake an intermediate position in this respect betweenerythrocytes and plasma (12, 13).

In large-scale epidemiological studies among thou-sands of subjects, markers such as erythrocytes requiringinvasive sampling and specific transport and storage con-ditions are less attractive options. Hair and fingernailsmay be contaminated by selenium-containing antidan-druff shampoos (14, 15) or environmental contaminationin general and may therefore be less useful, in spite ofthe observed correlation between hair and blood sele-nium levels (3). Toenails are less prone to contaminationproblems in populations wearing shoes and because theirsurface-to-volume ratio is smaller; their usefulness as abiomarker has been investigated in the last decade. Be-cause toenails have different lengths with correspondingage differences, toenails clipped from all toes at a singletime provide a time-integrated measure of selenium in-take over several months (6). Higher toenail seleniumlevels have been observed in subjects living in selenifer-ous areas as South Dakota compared to residents ofBoston or New Zealand, which have low selenium levelsin the soil (1 1). Longnecker et a!. (16) observed correla-tion coefficients of 0.91 and 0.89 for selenium levels intoenails with those in whole blood and serum, respec-tively, among residents of South Dakota and Wyoming.Also, elevated selenium levels were found in the toenailsof subjects consuming dietary selenium supplements(17). In a recent study in South Dakota in a population

with widely varying selenium intakes, strong correlationswere observed between selenium intake measurementsfrom duplicate meal portions and selenium levels in

toenails, as well as in serum and whole blood (18).The observations on the potential value and sensitiv-

ity of toenail selenium concentrations combined withfeasibility considerations have stimulated the collectionof toenail clippings in epidemiological studies (19-21).We have started a prospective cohort study on diet andcancer among men and women that includes toenail

on March 17, 2021. © 1993 American Association for Cancer Research. cebp.aacrjournals.org Downloaded from

http://cebp.aacrjournals.org/

108 Predictors of Toenail Selenium Levels in Men and Women

3 R. A. Goldbohm et al., unpublished manuscript.

clippings as a biological marker of selenium status (22).Before analyzing the relationship between selenium andthe risk of cancer it is important to identify potentialdeterminants of toenail selenium levels in men andwomen that may act as confounders in subsequent anal-yses of selenium and cancer risk. Hunter et a!. (17)recently concluded that smoking, age, and use of sele-nium supplements were predictive of toenail seleniumlevels among U.S. nurses, while alcohol and dietary se-lenium intake were not. Swanson et a!., however, did notobserve a relationship with age in men and women (18).The purpose of our study was to examine whether as-

sociations between toenail selenium and age, smoking,alcohol, Quetelet index, and selenium intake in the Neth-erlands do exist in women and to evaluate whether theserelationships would also hold for men.

Materials and Methods

Subjects. The study population is derived from an ongo-ing prospective cohort study on diet and cancer that wasstarted in September 1986. The cohort (n = 120, 852) of55-69-year-old men (48.2%) and women (51.8%) origi-nates from 204 municipal population registries. At base-line, the cohort members completed a self-administered

questionnaire on diet and potential confounding van-ables and provided toenail clippings. For efficiency rea-sons a case-cohort approach is being used for the analysisof the cohort study (22), requiring processing of ques-tionnaires and toenail clippings of a random subcohort(n = 3500) and incident cancer cases only. For thepresent study on potential predictors of toenail seleniumwe have used data of the subcohort only. Prevalentcancer cases other than skin tumors were excluded fromthis group, leaving 3346 subjects (1630 men, 1716women). Of these, toenail clippings had been providedby 1247 men (76.5%) and 1322 women (77.0%). Prob-lems with the detection of toenail selenium (interferenceby other elements such as calcium) occurred in 16 ofthese 2569 samples. An additional 94 specimens wereexcluded because the specimen weighed less than 10mg, which would yield unreliable selenium measure-ments. Thus, toenail selenium data on 2459 subjects(1211 men, 1248 women) were available for analysis.

Potential Predictors of Toenail Selenium. The consideredpotential predictors of toenail selenium were: gender,age, smoking habits (type of tobacco and amountsmoked), alcohol consumption, Quetelet index, dietaryselenium intake as calculated from food consumption,and intake of selenium supplements. Information on thepredictors other than gender and age was obtained fromthe baseline questionnaire. The food questionnaire hasrecently been validated.3 The dietary selenium intakewas estimated by multiplying the average daily intake offoods with their selenium content. We used data on theselenium content of Dutch foods which were collectedfor an earlier case-control study on diet and breast cancer(21). Of the dietary questionnaires, about 7% could notbe used for nutrient intake calculation because of missingor inconsistent data.

Determination of Toenail Selenium Levels. The toenailselenium analyses were carried out by the lnterfaculty

Reactor Institute at Delft University (Delft, the Nether-lands). Toenails were first cleared by scratching off anydebris with a quartz knife. After ultrasonic cleaning withacetone for 15 mm, distilled water for 10 mm, andacetone for 1 5 mm, the specimens were freeze-dried for15 h to eliminate any humidity variations between runs.The selenium content of the toenails was measured byinstrumental neutron activation analysis of the metastableselenium-’77 isotope. The specimens were irradiated for17 s in a thermal flux of 1.2 X 1015 neutrons.s’.cm2.After a decay time of 20 s, -y-radiation of 77�’Se wasmeasured for 60 s. The accuracy of the method waschecked by analysis of a certified bovine liver standard(Standard Reference Material 1 577a of the U.S. NationalBureau of Standards). For 26 determinations, a meanvalue (± SD) of 0.70 ± 0.04 zg/g selenium was observed

against a certified value of0.71 ± 0.04 �zg/g. The precisionof the method was- evaluated by duplicate seleniummeasurements of specimens from 27 randomly selectedsubjects; the coefficient of variation was 6.6%.

Data Analyses. Because of some skewedness to the right,data on toenail selenium concentrations and dietary se-

lenium intake were normalized with a loge transforma-

tion. Selenium intake data were adjusted for energyintake by the residual method (23). The relationshipbetween toenail selenium and its potential predictorswas tested in bivaniate and multivaniate analyses. Anal-yses were carried out for men and women separatelyand afterwards were combined, if appropriate. First,mean toenail selenium concentrations in the variousstrata of the potential predictors were compared using at test. To investigate the influence of various potentialpredictors simultaneously, multiple regression analysiswas carried out, with toenail selenium as the dependentvariable and the potential predictors as independentvariables. Two-sided P values are reported throughout.

Results

Before excluding toenail specimens with weights below1 0 mg, the specimen weights ranged from 1 .25 to 280.59

mg in men and from 1.50 to 442.31 mg in women. Withthe aforementioned exclusion, the average weight of thetoenail specimens was 91.0 ± 57.2 mg (mean ± SD) inmen and 70.2 ± 47.4 mg in women. In Table 1 the meanselenium levels in toenails of men and women are pre-sented separately. Overall, men were found to havesignificantly (P < 0.001) lower mean selenium levels intoenails than women: 0.547 ± 0.126 pg/g versus 0.575 ±

0.109 �Lg/g, respectively.Table 1 also shows the associations between toenail

selenium and the considered potential predictors of toe-nail selenium levels for each gender. In the age rangestudied (55-69 years) no association between toenailselenium level and age existed in men or women. Smok-ing, on the other hand, showed a strong relationship withtoenail selenium, especially among men. Current ciga-rette smokers have significantly (P < 0.001 ) lower toenail

selenium levels than persons who never smoked, andselenium levels in men decrease with increasing amountssmoked. This relationship does not hold for women;female subjects smoking 10-19 cigarettes/day have thelowest toenail selenium concentrations. The Spearmanrank correlation coefficient between (untransformed)

toenail selenium and number of cigarettes smoked in the



Cancer Epidemiology, Biomarkers & Prevention 109

Table I Toenail selenium levels (g g/g) according to various characteristi cs among 121 1 me n and 1248 women in the Netherla nds, 1986

CharacteristicMen Women

n� Mean ± SD P�’ n� Mean ± SD P”

All subjects 121 1 0.547 ± 0.126 1248 0.575 ± 0.109

Age (years)55-59 467 0.544±0.113 -C 474 0.575±0.109 -

60-64 426 0.548±0.148 0.791 429 0.570±0.104 0.57365-69 318 0.551 ±0.112 0.304 345 0.580±0.114 0.626

Smoking statusNever 119 0.576±0.105 - 744 0.583±0.112 -

Only cigar/pipe 46 0.535 ± 0.1 13 0.02 1

Ex-cigaretle 636 0.568 ± 0.136 0.313 250 0.577 ± 0.101 0.557

Current cigarette1 -9/day 57 0.518 ± 0.097 <0.001 77 0.566 ± 0.087 0.24710-19/day 147 0.508±0.094 <0.001 93 0.537±0.112 0.00120+/day 152 0.497 ± 0.081 <0.001 65 0.551 ± 0.096 0.030

Alcohol intake (g/day(

0 170 0.565±0.193 - 366 0.573±0.113 -

1-4 235 0.543±0.090 0.271 404 0.580±0.114 0.438

5-14 314 0.556±0.126 0.776 209 0.572±0.100 0.908

15-29 261 0.538±0.120 0.072 103 0.570±0.106 0.82130+ 159 0.538±0.096 0.134 45 0.557±0.082 0.403

Quetelet index (kg/m2)<20 35 0.516±0.106 0.150 57 0.558±0.137 0.34520-24 579 0.550±0.140 - 591 0.571 ±0.113 -

25-29 513 0.545±0.104 0.733 444 0.584±0.101 0.014

�30 47 0.551 ± 0.086 0.558 109 0.570 ± 0.093 0.75 1

Dietary selenium (pg/day), quintiles

(energy-adjusted)

1 (�47.5) 253 0.529 ± 0.095 - 213 0.550 ± 0.097 -

2 (>47.5 to �53.3) 227 0.548 ± 0.169 0.167 235 0.581 ± 0.106 0.002

3 (>53.3 to �59.1) 216 0.543 ± 0.126 0.282 243 0.585 ± 0.129 0.0024 (>59.1 to �66.7) 232 0.561 ± 0.1 19 0.001 237 0.575 ± 0.104 0.015

51>66.7) 220 0.562±0.116 0.001 250 0.577±0.096 0.006

respectively.Due to missing questionnaire data, numbers may not add up to 1211 and 1248,

b � test between strata, based on log,Aransformed toenail selenium levels.

, -- , reference category.

combined group of current smokers and never-smokers

was -0.29 for men (P < 0.001) and -0.1 3 for women (P

< 0.001 ). Men who smoke only cigars or a pipe also havesignificantly lower toenail selenium levels than never-smokers. The corresponding Spearman correlation coef-

ficient for these 155 subjects was -0.20 (P < 0.014) inthis case. Ex-smokers of cigarettes have somewhat lower

selenium levels than never-smokers, but this differenceis not significant in men or women. When current smok-ens were contrasted with never or ex-smokers, the aver-

age toenail selenium values for male current smokers

were 0.513 ± 0.106 �tg/g versus 0.571 ± 0.133 �zg/g for

male never- on ex-smokers (P < 0.001). For women these

values were 0.548 ± 0.101 versus 0.581 ± 0.109 �zg/g,respectively (P < 0.00 1).

Alcohol intake does not show a consistent relation-ship with toenail selenium concentration. Subjects drink-

ing more than 30 g alcohol/day have lower selenium

levels than nondninkers (not statistically significant in men

nor women). However, males drinking 5-14 g alcohol/day show somewhat higher toenail selenium levels than

males drinking 1 -4 g alcohol daily. Women drinking 1-4g alcohol/day even show somewhat higher selenium

levels than nondninkers, but the levels decrease with

higher alcohol consumption.

Subjects with a Quetelet index lower than 20 kg/rn2exhibit lower toenail selenium levels in men and womenthan in the referent category (20-25 kg/rn2), although thedifferences are not significant. Although women with aQuetelet index between 25 and 29 have significantlyincreased toenail selenium levels compared to the ref-erent category, the levels are decreased in obese women(Quetelet index � 30 kg/rn2). In men the toenail seleniumlevels are comparable in the upper three Quetelet indexcategories.

Selenium supplement use was not reported by anyof the subjects studied; therefore, its relationship withtoenail selenium was not evaluated. Since the estimatedselenium intake was positively correlated with total en-ergy intake (r = 0.58), the relationship between toenailselenium and dietary selenium was evaluated after ad-justing the latter for energy intake. After dividing theenergy-adjusted selenium intake into quintiles (for menand women combined) there was a positive trend withtoenail selenium levels. For men, the mean toenail sele-nium levels in the upper two quintiles of intake weresignificantly higher than in the lowest quintile. Forwomen, the mean toenail selenium levels in the secondand third quintile were also significantly higher than inthe bottom qumntile, but the levels in the upper twoqumntiles were somewhat lower than in the second and



1 10 Predictors of Toenail Selenium Levels in Men and Women

., Dummy variables (coded as 0, 1 ) were introduced for the smoking categories, with never- or ex-smokers as baseline category.S Dummy variables were introduced for the drinking categories, with nondrinkers as baseline category.

Table 2 Mean (± SD) daily energy-adjusted intake of selenium (jag) in men and women separately according to smoking status

MenSmoking status

0 Mean ± SD

Women

P� n Mean ± SD

All subjects 1 148 57.0 ± 12.7 1 178 58.7 ± 14.0

Never smoked 1 13 58.1 ± 12.9 701 58.4 ± 14.0

Only cigar/pipe 43 58.6 ± 15.0 0.874

Ex-cigarette 613 58.0 ± 12.6 0.972 240 60.0 ± 13.2 0.096

Current cigarette1 -9/day 54 56.8 ± 1 1 .8 0.586 73 60.4 ± 1 7. 1 0.458

10-19/day 138 55.9± 13.2 0.150 90 58.1 ± 13.8 0.788

20+/day 142 54.3 ± 12.2 0.014 62 55.3 ± 14.3 0.065

I test based on lo&-transformed toenail selenium values.b Reference category.

third quintile. The Spearman correlation coefficient be- is significantly lower (P = 0.003) than in women: 57.0 ±

tween toenail selenium and energy-adjusted dietary se- 12.7 j.�g/day (men) versus 58.7 ± 14.0 �tg/day (women).

lenium intake was 0.11 (P < 0.001). The results of both bivariate regression and multiple

To investigate whether the lower toenail selenium regression analyses, where the effects of several predic-

levels in smokers might possibly be due to a decreased tors are controlled simultaneously, are shown in Table 3.

mntakeofseleniumwealsocalculatedtheaverageenergy- Because the previous analyses in both genders had not

adjusted selenium intake according to smoking habits revealed any large differences between men and women

and tested the differences in both men and women in the associations with potential predictors, the regres-

(Table 2). The results in Table 2 indicate that male current sion analyses were carried out for the combined group.

smokers indeed consume less selenium per day (after The multivariate analysis yielded a significant effect of

adjustment for energy intake) than those who never gender, while no effect of age was observed. The nega-

smoked; the difference increases with increasing amount tive association between smoking on toenail seleniumsmoked and is significant for those smoking 20 cigarettes/ levels remained significant in the regression model, al-day or more. In women, no significant differences were though not for subjects who smoke only cigars or a pipe.

observed, although heavy smokers consume less sele- Also, the energy-adjusted selenium intake showed an

nium. Table 2 also shows the mean energy-adjusted independent positive association with toenail seleniumselenium intake values in men and women without re- levels in this population. The partial correlation coeffi-gard to smoking status. The mean selenium intake in men cient between the two variables was 0.09, indicating a

Table 3 Predictors of toenail selenium levels (�zg/g, log.,-transformed) in bivariate and multiple regression analysis in 2459 men and women, theNetherlands, 1986

Bivariate regression Multiple regression

13 se(f3) PI� se(�3) P

Intercept -0.858

Gender (M = 0, F = 1) 0.051 0.008 <0.001 0.034 0.009 <0.001

Age (10 years) 0.005 0.009 0.571 -0.010 0.010 0.308

Smoking habits�Only cigar/pipe1-9 cigts/day10-19 cigts/day20+ cigts/day

-0.074

-0.053-0.109-0.1 16

0.029

0.0170.0130.014

0.01 1

0.002<0.001<0.001

-0.043-0.043-0.106-0.093

0.0300.0170.0140.014

0.151

0.013<0.001<0.001

Alcohol intake (g/day)”

<15�15

-0.007-0.041

0.0100.012

0.493<0.001

-0.006-0.011

0.010

0.012

0.5320.343

Quetelet index (5 kg/m2) 0.015 0.006 0.020 0.001 0.007 0.870

Selenium intake (log. pig/day) (energy-adjusted) 0.101 0.018 <0.001 0.082 0.018 <0.001

R2 0.07; adtusled R2 = 0.07



Cancer Epidemiology, Biomarkers & Prevention 1 1 1

weak relationship. Although the effect is in the antici-pated direction, its significance (P < 0.001 ) is merely theresult of the large number of individuals studied. Theobserved regression coefficient of 0.082 for the Iog,�-transformed values in the multivariate model implies, forexample, a 4% increase in toenail selenium level whenthe median of the top qumntile of selenium intake (73.8jzg/day) is contrasted with the median of the bottomqumntile of intake (43.3 j.tg/day). Alcohol intake and Que-telet index were not significant predictors of toenailselenium in this model. Together, the independent pre-dictors gender, smoking habits, and selenium intake ex-plained 7% of the variance in toenail selenium levels inthis population.

Regression analyses conducted separately for menand women (results not shown here) revealed similardirections of the associations with smoking and seleniumintake, although the associations were somewhatstronger in men.

DiscussionThe average toenail selenium values that we observedamong women (0.58 �g/g) are somewhat lower thanother recent estimates (0.65 �g/g) from the Netherlands(21), butthe estimates ofselenium intake in both studieswere virtually similar, using the same food tables. Thedifference in toenail levels might be due to differencesin analytic conditions, used in different nuclear researchreactors. Among U.S. nurses toenail selenium levelsaround 0.80 �zg/g were observed (1 7), whereas substan-tially higher values (averaging 1.17 j�g/g) were foundamong residents of a seleniferous area in South Dakota(18). Similar values were found by Morris et a!. (11), whoalso observed an average value of 0.26 .tg/g for residentsof New Zealand. In Greece values around 0.54 jzg/g havebeen observed in fingernails (24). The fact that men havelower toenail selenium levels than women was also ob-served by Swanson et a!. (18). Other studies did not showa clear gender difference regarding fingernail (24), wholeblood, or plasma selenium levels (25). However, Lloyd eta!. (25) did find significantly lower selenium values inerythrocytes among men.

Age was not a predictor of toenail selenium in ourstudy. Various investigators have found an inverse asso-ciation between selenium status parameters and age (17,25, 26), but this observation is not consistent (18, 24, 27).In the study of Dickson and Tomlinson (26), the inverseassociation between serum selenium and age was re-stricted to subjects under 55 years of age, while Lloyd eta!. (25) only observed lower levels in persons 56 yearsold or older, and Bratakos et a!. (24) noted an inversetrend only in subjects 40 years old or older. The absenceof an age effect in our study may also be due to thelimited age range in our cohort population (55-69 years).

Smoking was an independent predictor of loweredtoenail selenium levels in both men and women. Ingeneral, a negative dose-response relationship existedwith the amount smoked, as was observed earlier inwomen (17). Swanson et a!. (18) also observed a strongnegative association of smoking with toenail seleniumlevels but not with serum or whole blood selenium.Negative relationships with smoking have been observedfor whole blood and serum selenium (28). Lloyd et a!.(25) found significantly decreased levels of selenium in

whole blood, plasma, and erythrocytes among smokingmen, but for smoking women only in plasma for thosewho drank alcohol daily. This is in accordance with ourobservation that the smoking effect is more pronouncedin men than in women. It is unclear whether smokinginduces increased selenium utilization (e.g., as an activesite of glutathione peroxidase) or influences seleniumincorporation in nail keratin. Our data indicate that cur-rent smokers have lower energy-adjusted selenium in-takes compared to never-smokers, but in the multipleregression analysis smoking remained an independentpredictor. Swanson et a!. (18) also observed significantlylower selenium intakes among smokers, although noadjustment was made for energy intake. We also corn-pared selenium intakes of smokers and nonsmokers with-out adjustment for energy intake; the unadjusted intakesin ex- or in current smokers were not significantly lowerthan in those who never smoked.

Our estimates of selenium intake are substantiallylower than the intakes measured by Swanson et a!. (18)among South Dakota residents (174 pg/day) with theduplicate portion technique. These investigators ob-served a very strong independent effect of dietary sele-nium on toenail selenium levels. Our intake estimatesare also lower than the intakes estimated by Hunter eta!. (17) with a questionnaire among U.S. nurses. Theestimation of dietary selenium intake based on question-naires or interviews is difficult because of large variationsin selenium contents within varieties of the same food,depending on soil conditions. When a single seleniumfood table is used, the estimation problem is likely toaggravate if a study is conducted in a country like theUnited States as opposed to the Netherlands (17). Al-though we observed a positive association between se-lenium intake and toenail levels, in contrast to Hunter eta!. (1 7), the partial correlation coefficient between dietaryand toenail selenium was still only 0.09 in our study,indicating a weak positive relationship which reachedsignificance only because of the large numbers of sub-jects studied. We could not evaluate the influence ofdietary methionine on toenail selenium levels (which wasrecently found to be a mediating factor in the depositionof selenium in nails of rats; Ref. 29) because informationon methionine intake was not available.

Although in bivariate regression analysis of men andwomen combined, subjects drinking 15 g of alcohol ormore per day showed significantly lower toenail seleniumlevels, alcohol intake was not an independent predictorof toenail selenium in multivariate analyses with smokingin the model. Earlier studies have revealed decreasedserum selenium levels among alcoholics (30, 31). Theabsence of an alcohol effect in the dose range we havestudied is in agreement with other studies using toenailselenium (1 7) or whole blood or erythrocyte selenium(25). With regard to plasma selenium in men (but not inwomen), Lloyd et a!. (25) observed decreased levels innonsmokers who were daily drinkers compared to non-drinkers and weekend drinkers.

In bivariate analyses there was an indication that theQuetelet index is positively associated with toenail sele-nium levels, but it was not an independent predictor inmultiple regression analyses. Swanson et a!. (18) sug-gested that controlling for differences in lean body massinstead of weight (or relative weight) would diminish theeffect of gender on toenail selenium levels because of



1 12 Predictors of Toenail Selenium Levels in Men and Women

the storage of selenium in muscle tissue (32, 33). Thisadjustment is difficult in practice, however, because dataon lean body mass are typically unavailable. The presentresult does not support a need to control for Queteletindex as a confounder in studies on selenium and obe-sity-associated diseases.

In conclusion, smoking was the most important pre-dictor of toenail selenium levels in both men and women.The inverse relationship with smoking appeared to bestronger in men than in women. Gender itself was alsoan independent predictor of toenail selenium, with menshowing lower values than women. Dietary seleniumintake (excluding supplements) showed a significantlyweak positive association with toenail selenium levels,despite the well-known lack of reliability of question-naire-based assessment of selenium intake (5). Age, al-cohol intake, and Quetelet index were not independentpredictors of toenail selenium levels. In any epidemio-logical analysis relating toenail selenium to risk of smok-ing associated diseases, adjustment for smoking habits isindicated.

Acknowledgments

The authors would like to thank Sacha van de Crommert, Patricia Florax,Mari#{235}lle Bethlehem, Jolanda Nelissen, and Judith van den Brink for their

assistance in the conduct of this study and Frank van Paassen and Anneke

Ammerlaan for the selenium measurements.

References1. Willett, W. C., Morris, I. S., Pressel, S., et al. Prediagnostic serum

selenium and risk of cancer. Lancet, 2: 130-134, 1983.

2. Keshan Disease Research Group. Observations on effect of sodiumselenite in prevention of Keshan disease. Chin. Med. J. (EngI. Ed.), 92:471-476, 1979.

3. Chen, X., Yang, G., Chen, J., et al. Studies on the relations of seleniumand Keshan disease. Biol. Trace Elem. Res., 2: 91-107, 1980.

4. Schrauzer, G. N., White, D. A., and Schneider, C. J. Cancer mortality

correlation studies. Ill: Statistical associations with dietary selenium in-

takes. Bioinorg. Chem., 7: 23-34, 1977.

5. Levander, 0. A. The need for measures of selenium status. I. Am.

CoIl. Toxicol., 5: 37-44, 1986.

6. Willett, W. Nutritional Epidemiology. New York: Oxford UniversityPress, 1990.

7. Hulka, B. S., Wilcosky, T. C., and Griffith, I. D. Biological Markers inEpidemiology. New York: Oxford University Press, 1990.

8. Levander, 0. A. Considerations on the assessment of selenium status.Fed. Proc., 44: 2579-2583, 1985.

9. Robinson, I. R., Robinson, M. F., Levander, 0. A., et al. Urinary

excretion of selenium by New Zealand and North American humansubjects on differing intakes. Am. J. Clin. Nutr., 41: 1023-1031, 1985.

10. Broghamer, W. L., McConnell, K. P., and Blotcky, A. L. Relationship

between serum selenium levels and patients with carcinoma. Cancer(Phila.),37: 1384-1388, 1976.

11. Morris, J. S., Stampfer, M. J., and Willett, W. Dietary selenium in

humans; toenails as an indicator. Biol. Trace Elem. Res., 5: 529-537,

1983.

12. Thompson, C. D., Robinson, M. F., Campbell, D. R., et al. Effect of

prolonged supplementation with daily supplements of selenomethionineand sodium selenite on glutathione peroxidase activity in blood of NewZealand residents. Am. J. Clin. Nutr., 36: 24-31, 1982.

13. Taylor, P. R., Longnecker, M. P., Levander, 0. A., et al. Seasonalvariation in selenium (Se) status among free-living persons in South

Dakota. Fed. Proc., 46: 882, 1987.

14. Gordus, A. Factors affecting the trace-metal content of human hair.I. Radioanal. Chem., 15: 229-243, 1973.

15. Davies, T. S. Hair analysis and selenium shampoos. Lancet, 2: 935,

1982.

16. Longnecker, M. P., Taylor, P. R., Levander, 0. A., et al. Selenium in

diet, blood, and toenails in relation to human health in a seleniferousarea. Am. J. Clin. NuIr., 53: 1288-1294, 1991.

1 7. Hunter, D. J., Morris, I. S., Chute, C. G., et al. Predictors of selenium

concentration in human toenails. Am. I. Epidemiol., 132: 114-122, 1990.

1 8. Swanson, C. A., Longnecker, M. P., Veillon, C., et al. Selenium intake,age, gender, and smoking in relation to indices of selenium status of

adults residing in a seleniferous area. Am. I. Clin. NuIr., 52: 858-862,1990.

19. Hunter, D. I., Morris, I. S., Stampfer, M. I., et al. A prospective studyof selenium status and breast cancer risk. JAMA, 264: 1 1 28- 1 1 3 1 , 1990.

20. Van Noord, P. A. H., Colette, H. I. A., Maas, M. I.. e1 a!. Seleniumlevels in nails of premenopausal breast cancer patients assessed predi-agnostically in a cohort-nested case-referent study among women

screened in the DOM project. InI. I. Epidemiol., /6: 318-322, 1987.

21. van ‘I Veer, P., van der Wielen, P. 1.. Kok, F. I., et al. Selenium indiet, blood, and toenails in relation to breast cancer: a case-control study.Am. J. Epidemiol., 131: 987-994, 1990.

22. van den Brandt, P. A., Goldbohm, R. A., van ‘I Veer, P., et al. A large-scale prospective cohort study on diet and cancer in the Netherlands. I.

Clin. Epidemiol., 43: 285-295, 1990.

23. Willett, W., and Stampfer, M. I. Total energy intake: implications forepidemiologic analyses. Am. I. Epidemiol., 124: 17-27, 1986.

24. Bratakos, M. S., Kanaki, H. C., Vasiliou Waite, A., et al. The nutritionalselenium status of healthy Greeks. Sci. Total Environ., 91: 161 -1 76, 1990.

25. Lloyd, B., Lloyd, R. S., and Clayton, B. F. Effect of smoking, alcohol,and other factors on the selenium status of a healthy population. I.

Epidemiol. Community Health, .37: 213-217, 1983.

26. Dickson, R. C., and Tomlinson, R. H. Selenium in blood and human

tissues. Clin. Chim. Ada, 16: 311-321, 1967.

27. McAdam, P. A., Smith, D. K., Feldman, E. B., et a!. Effect of age, sex,

and race on selenium status of healthy residents of Augusta, Georgia.Biol. Trace Elem. Res., 6: 3-9, 1984.

28. Ellis, N., Lloyd, B., Lloyd, R. S., et al. Selenium and vitamin E in

relation to risk factors for coronary heart disease. I. Clin. Pathol., 37: 200-

206, 1984.

29. Salbe, A. D., and Levander, 0. A. Effect of various dietary factors on

the deposition of selenium in the hair and nails of rats. I. NuIr., 120: 200-

206, 1990.

30. Korpela, H., Kumpulainen, I., Luoma, P. V., et a!. Decreased serum

selenium in alcoholics as related to liver structure and function. Ani. I.

Clin. Nutr., 42: 147-151, 1985.

31. Aaseth, I., Thomassen, Y., Alexander, I.. et a!. Decreased serum

selenium in alcoholic cirrhosis. N. EngI. I. Med., 303: 944-945, 1980.

32. Behne, D., and Wolters, W. Distribution of selenium and glulalhione

peroxidase in the rat. I. Nutr., 113: 456-461, 1983.

33. Schroeder, H. A., Frost, D. V., and Balassa, J. I. Essential trace metalsin man: selenium. I. Chronic Dis., 23: 227-243, 1970.



1993;2:107-112. Cancer Epidemiol Biomarkers Prev P A van den Brandt, R A Goldbohm, P van't Veer, et al. Predictors of toenail selenium levels in men and women.

Updated version

http://cebp.aacrjournals.org/content/2/2/107

Access the most recent version of this article at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cebp.aacrjournals.org/content/2/2/107To request permission to re-use all or part of this article, use this link



http://cebp.aacrjournals.org/cgi/alerts

mailto:[email protected]



predictors of toenail selenium levels in men and women1 · these values were 0.548 ± 0.101 and...

Documents