Vol. 1 , I 19- 124. /anuarv/1ehruarv 1992 Cancer Epidemiology, Biomarkers & Prevention 1 19

‘ The abbreviations used are: CIN, cervical intraepithelial neoplasia; OR,.( rude odds ratio; OR,,, adjuste(l odds ratios; CI, ci)ntidenc(’ interval.

Folate, Vitamin C, and Cervical lntraepithelial Neoplasia1

Juliet VanEenwyk,2 Faith G. Davis, and Neville Colman

Department of Epidemiology and Biostatistics, School of Public Health,

University of Illinois at Chicago, Chicago, Illinois 6061 2 [1. V., F. G. D.];

and VA Medical Center, Bronx, New York 10468, and Center for Clini-cal Laboratories, Mount Sinai Medical Center, New York, New York10029 [N. C.]

Abstract

A case-control study was designed to assess therelationship between cervical intraepithelial neoplasia(CIN) and folate in serum, red blood cells, and diet.The association between CIN and dietary vitamin Cwas also investigated. Cases were selected from womenwith biopsy-confirmed CIN. Controls were age-,race-, and clinic-matched women with normal cervical(Pap) smears. Study participants completed self-administered food frequency (n = 100 matched pairs)and health (n = 102 matched pairs) questionnaires.Fasting venous blood samples were collected for serum(n = 98 matched pairs) and red cell (n 68 matchedpairs) folate assays. Conditional logistic regressionmodels were used to estimate crude odds ratios andodds ratios adjusted for smoking, income, number ofsexual partners, frequency of cervical smear, use ofspermicidal contraceptive agents, history of genitalwarts, and Quetelet index. Dietary intake variableswere adjusted for total energy intake prior to logisticregression. A protective effect of red cell folate wasevident with adjusted odds ratios (95% confidenceintervals) of 0.1 (0.0-0.4), 0.6 (0.2-2.0), and 0.5 (0.2-1.9) for those in quartiles 4 (highest), 3, and 2compared to quartile 1 (lowest). Supporting evidencefor the protective effect of folate was provided byinverse associations between CIN and folate in bothserum and diet. An inverse association was also foundbetween CIN and dietary vitamin C with adjusted oddsratios (95% confidence intervals) of 0.2 (0.0-0.7), 0.6(0.2-1.6), and 0.6 (0.2-1.8) for those in quartiles 4, 3,and 2, respectively, compared to quartile 1. Thesefindings support dietary recommendations, such asthose of the American Cancer Society, the NationalCancer Institute, and the U.S. Dietary Guidelines,which allow for adequate intake of folate and vitaminC, both of which are found in good quantity in fruitsand vegetables. Increased consumption of legumes andwhole grains is also in accord with current dietary

Re eived 4/t 6/91.1 Funded in part by a grant froni the American Canc er Society, Illinois

Division, and I)y the State Cancer Plan of the Illinois Cancer Council.

2 To whom requests for reprints should be addressed, at Illinois Depart-ment of Public Health, Division of Health Statistics and Policy Develop-

ment, 1 01) West Rai�loIph, Suite 6-600, Chicago, IL 60601.

recommendations, and both of these types of foods aregood sources of folates.

Introduction

Folic acid, a B-group vitamin, is necessary for normal cellreplication, and cells grown in folate-deficient mediamanifest chromosomal abnormalities which correspondto those found in many types oftumor cells (1). In relationto cervical cancer, it has been noted that folate deficiencycan lead to cervical cellular changes which resembleneoplastic change (2) and that preneoplastic cervicalcellular changes among users of oral contraceptives me-gness with folate supplementation (3). Three case-controlstudies of folate consumption and cervical cancer (4-6)have found little evidence of an association betweenfolate intake and disease.

This investigation assesses the relationship betweenfolic acid and CIN.3 The premalignant condition waschosen to attenuate physiological changes which mayensue from rather than foreshadow disease. It was hy-pothesized that higher levels of serum and red cell folateand higher dietary intake of folate would be associatedwith a reduction in disease. Because folates and vitaminC are found in many of the same foods and becausevitamin C protects folates from oxidative cleavage, die-tary intake of vitamin C was also assessed.

Materials and Methods

Details of case and control selection, serum and dietarymeasurement procedures, measurement of confoundingvariables, and statistical analysis have been previouslydocumented (7, 8). These topics are reviewed below.

Case and Control Selection. Participants were recruitedfrom clinics at Cook County Hospital and University ofIllinois Hospital between April 1987 and June 1989. Cases(n = 102) were selected from women aged 18 to 49 years

with biopsy-confirmed CIN I, II, or Ill. Age- and race-matched controls were selected from women who at-tended the same clinics as the cases and whose Papsmears showed no abnormality of a severity greaten thanor equal to benign atypia. Women who had been preg-nant or lactating within the past year were excluded fromthe study because of the potential for folate depletionunder these circumstances. In this population, pregnantand postpartum women may also be at increased risk fora diagnosis of CIN, since pregnancy brings women intothe clinic where cervical smears are obtained as part ofprenatal care. Women with epilepsy on sickle cell anemiawere also excluded, since these conditions are associatedwith low blood folate and with bringing women into the

on July 25, 2020. © 1992 American Association for Cancer Research. cebp.aacrjournals.org Downloaded from

120 Folate, Vitamin C, and Cervical lnlraepithelial Neoplasia

medical system where they are likely to have cervicalsmears. Women with diabetes were exduded due to therequirenient for ci 1 0-h f,ist.

All eligible wonien were requested to participate in

the study. of 166 eligible cases, 102 were enrolled,yielding a panticipation rate of approximately 61%. Toenroll an equal number of controls, 195 eligible womenwere approached, giving a participation rate of 52%.

Measurement of Exposure. The food frequency portionof the Health Habits and History Questionnaire of theNational Cancer Institute, Division of Cancer Preventionand Control, version 2. 1 , was used to assess dietary intake

of folate and vitamin C (9). Participants were asked tocomplete this questionnaire prior to the clinic visit. The

conversion of foods on the food frequency questionnaireto nutrients was accomplished via the microcomputersoftware version 2.2, August 1989, provided by the Na-tional Cancer Institute, Division of Cancer Preventionand Control (9). The measure of dietary vitamin C ob-tamed from this procedure includes vitamin C from bothfood and vitamin supplements. There is no provision forincluding supplemental folates in the dietary folate mea-sure. Adjustment for total energy intake using the regres-sion procedure of Willett and Stampfer (10) was used tocontrol for over- and underreporting of dietary intake.Two participants failed to adequately complete the foodfrequency questionnaire, resulting in 100 matched pairsavailable for analysis.

Fasting venous blood samples were collected forradioassay of serum and red cell folate. Red cell hemo-lysates were prepared on site by the method of Gutcho(11). Serum was also aliquoted on site, and all bloodsamples were stored at -70#{176}C until shipment to thelaboratory on dry ice every 6 to 10 weeks. Assays forserum and red cell folate were conducted by modiuica-tion of the methods of Waxman et a!. (12) and Longoand Herbert (13), respectively, using Becton DickinsonSimultrac kits as reagents (14). For the folate assays, theintraassay coefficient of variation was 1 .4-4.6%, and theintenassay coefficient of variation was 3.8-8.2% for con-trol samples at the limits of sensitivity of the assay. Lowlevels were associated with the highest coefficients ofvariation. All laboratory personnel were unaware of caseor control status of the blood samples.

Failure to withdraw blood from four women resultedin serum measures for 98 matched pains. Inadequate on-site preparation of the red cell hemolysates resulted inunreliable data from the first 28 cases and 14 controls,leading to the exclusion of 28 matched pairs from thefinal red cell folate analyses. Failure to collect a lavender-top tube from one woman and missing red cell data fromone woman resulted in the exclusion of an additionaltwo pairs, leaving 68 matched observations for the redcell folate analyses.

Because more cases were enrolled at the beginningof the study and more controls enrolled toward the endof the study period, there was a disparity in the allocationof case and control blood samples to the laboratory assaygroups. To enable control for confounding due to be-tween-nun variability, two quality control samples ofpooled blood were included in 1 1 of the 16 shipments.These 1 1 batches accounted for 88% of the serum and100% of red cell samples included in the logistic models.Laboratory personnel were unaware of the inclusion ofthese samples.

Measurement of Confounders. RIrti( pants (ompleted aself-administered questionnaire whi h asked aI)out back-ground, health and pregn�in y history, smoking, and sex-ual behavior. Inforniation mroni this questionnaire was

used to assess independent contributors to risk of CIN inthis sample and to control for confounders of the disease-exposure relationship. Confounders were defined as van-ables which have been reported as risk factors in previousstudies and variables whose inclusion led to a change ofmore than 20% in the adjusted odds ratio for the nu-trients of interest.

Statistical Analysis. OR, and OR,, and 95% CIs were

estimated using the MCSTRAT program (15), which per-forms an iterative conditional maximum-likelihood fit of

a logistic regression model. Quartiles for the hematolog-cal measures and calorie-adjusted nutrient intake were

defined from the distribution of the controls. Those in

the lowest quartile (quartile 1) served as the comparisongroup. Adjusted models included independent contrib-utors to risk in this sample, as well as potential confoun-dens of the disease-exposure relationship. Tests of trendwere achieved by entering quartiles of a given nutrientinto the logistic model as different values of a singleordinal variable.

Pearson product-moment correlation coefficients for

the correlation between the natural log of the hemato-logical measures and calorie-adjusted nutrient and foodintake measures were generated using SAS procedures.

Confounding due to intenassay variability was as-sessed by including a dichotomous variable in the logisticmodels. This variable was (Treated by calculating the

mean value for the quality control samples and charact-enizing assay groups according to whethen their quality

control samples were above on below the mean.

Results

The distribution of cases and controls on demographicand nondietary risk factors associated with CIN havebeen presented (7, 8). Table 1 shows the OR,,s and 95%

CIs for the nonnutnient variables included in the multi-vanial)le conditional logistic regression models. IncreasedOR,,s were associated with current smoking status, morethan �t’te year between cervical smears, any use of con-traceptive spermicidal foams or gels, and a self-reportedhistory of genital warts. An inverse association was ob-served between OddS of disease and monthly incomebracket in $400 increments to $2000. Quartile of Que-telet index (kg/m2) and number of sexual partners werenot independent contributors to risk after adjustment forthe other variables. However, these variables were me-tamed in the final model, because they were consideredto be potential confoundens of the disease-exposurerelationship. Use of oral contraceptives and parity havebeen reported to relate to both folate status and risk forCIN. However, since these factors were not independentcontributors to risk in this sample and their inclusion inthe logistic models did not alter the adjusted estimates,they were not included in the final models. Excessivealcohol consumption is associated with lowered bloodfolates. Controlling for this variable (lid not alter thefindings and it is not included in the adjusted models.

Table 2 shows the quartiles of serum and red (Tellfolate. The number of cases and controls in each quartileand the percentage with deficiencies are also presented.

on July 25, 2020. © 1992 American Association for Cancer Research. cebp.aacrjournals.org Downloaded from

Cancer Epidemiology, Biomarkers & Prevention 121

Table 1 Adjusted odds ratios� and 95% CIs for nonnutrient variablesincluded in all adjusted logistic regression models

. . . 95% confidenceNonnutrient variable Odds ratio

interva

Current smoker 2.6 1.2-5.5(yes versus no)

Income bracket 0.5 0.3-0.8(ordinal, monthly income in

$400 increments to $2000)

Frequency of cervical smear 3.3 1 .3-8.2

(less often than annual versusannual)

Use of contraceptive spermici- 2.8 1.3-6.3dal agents

(ever used versus never used)

Self-reported history of genital 3.4 1.1 -10.8

warts(yes versus no)

Quetelet index (kg/m2) 0.7 0.5-1.0(ordinal by quartile)

Numberofsexual partners 1.0 0.6-1.7

(log,-transformed)

Odds ratio for each variable is adjusted for all of the other variables

shown.

For comparison, the percentage of women from theSecond National Health and Nutrition Evaluation Surveyin the deficiency ranges for folate are also given (16). Thewomen in this sample appear to manifest poorer folatestatus than anticipated from the findings of the SecondNational Health and Nutrition Evaluation Survey.

The OR,s, ORa5, and 95% CIs by quartile of serumand red cell folate are presented in Table 3. A decreasein odds of CIN is observed for those in the highestquartile of serum folate relative to those in the lowestquartile. However, while the ORa is smaller than the ORE,the statistical significance of the decreased OR is atten-uated in the adjusted model. The pattern of decreasingORs with increasing quartile of serum folate is also atten-uated after adjustment for nonnutrient factors. The de-creased odds of disease for those in the highest quartileof red cell folate relative to those in the lowest quartile

are apparent in both the crude and adjusted models. Thepattern of decreasing ORs with increasing quartile of redcell folate is also evident in both models.

Table 4 presents the OR,s, ORaS, and 95% CIs fordietary intake of folate and vitamin C. Decreasing ORswith increasing quartile of dietary folate and vitamin Care evident in both the crude and adjusted analyses. Fordietary vitamin C, a statistically significant decrease inthe odds ofClN is evident in both the crude and adjustedmodels for those in the highest quartile of intake relativeto those in the lowest quartile.

Associations among the vitamin assays and dietarymeasures are shown in Table 5. Among the vitaminassays, the strongest correlation is between serum andred cell folate. Intakes of dietary vitamin C and dietaryfolate are highly correlated with each other, and both ofthese measures seem to be derived primarily from theintake of total fruit and citrus fruit. In contrast, the he-matological folate measures are modestly correlated withvegetable intake.

Discussion

Issues of Bias. To assess possible bias due to differentialparticipation rates of cases and controls, information onage, ethnic origin, zip code, and type of payment formedical services was collected for all women asked toparticipate in the study. Response rates were lower (P <0.05) for controls than for cases among women age 18to 24 years and among women who paid their ownmedical expenses. Since cases and controls werematched on age, the effect of this discordance cannotbe estimated. It is not known in what direction the deficitof control women who paid for their own medical ex-penses may have influenced the ORs, since an under-representation in that group could result in an overrep-resentation in both higher (privately insured) and lower(public aid) socioeconomic groups. Adjustment for thenonnutrient factors in the logistic models should mitigatebias resulting from nonrandom distribution of cases andcontrols on socioeconomic factors resulting from differ-ential response rates.

Cases and controls were comparable on the match-ing variables, as well as on educational level, employmentstatus, marital status, history of hospitalizations, oral con-traceptive use, and age at first intercourse. After control-

Table 2 Qua rtiles of serum and red cell folate

NutrientQuartile Deficienc y levela

1 2 3 4 Frank Borderline

Serum folate level )ng/ml)Controls’

1.3-3.426

3.5-4.422

4.5-6.3

25

6.4-21.2

25

<3.0 (15%i�o)

14.3%<5.0(41%”)

61.2%

Cases#{176} 36 23 25 14 23.5% 67.4%

Red cell folate level )ng/ml)

Controls’

57-126

17

127-149

17

150-190

17

191-325

17<140 (13%’)

41.2%<160)20%’)

60.3%

Cases’ 26 18 18 6 51.5% 73.5%

a Deficiency levels are those defined by the laboratory.

b The percentages of women from the Second National Health and Nutrition Evaluation Survey who were below deficiency levels (16) are in parentheses.

C Women aged 20-44 years.

d Estimated for women aged 20-64 years (Ref. 16, p. 37).

e The number of cases and controls in each quartile and the percentage below the specified deficiency level are presented.

I Estimated for women aged 20-64 years (Ref. 16, p. 38).

on July 25, 2020. © 1992 American Association for Cancer Research. cebp.aacrjournals.org Downloaded from

122 Folate, Vitamin C, and Cervical lntraepithelial Neoplasia

Table I Odds ratios, 95% CIs, and tests ot trend for serum and red cellfolate

NutrientQuartile Test i)f trend

P1 (low) 2 3 4 (high)

Serum folateOR,95%CI

1.0 0.9 0.80.4-2.0 0.3-1.7

0.40.2-0.9

0.04

()R,�

95%CI1.0) 0.9 1.1

0.3-2.6 0.4-3.20.3

0.1-1.10.15

Red cell folate

OR,95%CI

1.0 0.7 0.8

0)3-1.7 0.3-1.9

0.2

0.1-0.7

0.03

OR,�95% CI

1.0 0.5 0.60.2- 1 .9 0.2-2.0

0.1

00b_04

0.01

., Adtusted for current smoking status, monthly personal income. Ire-qu(’ncy of cervical sniear (annual versus less often), any use ot sperniicidalcontraceptive agents, sell-reported history of genital warts, quartile of

Quetelet index, and natural log of number ot sexual partners.

b <0.05.

ling for smoking and income, cases and controls werealso comparable on number of sexual partners and parity.The similarity of the cases and controls on these factorsand the selection of all participants from clinics servingprimarily low-income individuals suggests that the sam-pIe was selected from a high-risk population. The ability

to generalize the study findings to other populations may

be limited.With control for assay group, the statistically signifi-

cant decrease in the OR, for those in the highest quartileof serum folates was attenuated. This adjustment did notsubstantively modify the dose gradient results or thefindings for ned cell folates. Because of the unequaldistribution of cases and controls in the assay groups, adichotomous variable was used to control for interassayvariability. While this procedure crudely adjusts for assaygroup, residual confounding may be present and thepossibility of bias cannot be entirely excluded.

Bias in the findings for the red cell folate may haveresulted from the exclusion of the 34 pairs from the

analyses. On the dietary and serum folate measures, thecontrols who were excluded were similar to those in-cluded. However, the distribution of excluded cases byquartile of both dietary and serum folate differed fromthat of cases included in the red cell folate analyses, withmore than the expected number of cases in the highest

quartiles and fewer than expected in the lowest quartiles.To the extent that the serum and dietary folate measuresreflect red cell folate, the findings for red cell folates maybe overstated.

Folate deficiency has been reported in patients witha diversity of malignancies (1 7). The question of whetherthis is a cause or a consequence of cancer remainsunanswered. However, it has been suggested that be-

cause tumors exhibit rapid cell multiplication, those withcancer may be at increased risk for folate deficiency (18).

Although CIN may represent an early stage of the neo-plastic process, the length of time required for the pro-gression from CIN I and II to cancer (19) argues againstrapid cell multiplication at this stage of the disease. Since79% of the cases were diagnosed with CIN I or II, itseems unlikely that the elevated ORs for those in the

T,il)le 4 Odds ratios, 95% CIs, ,uicl lists of trend for dietary intake of

Icilate .ui�l vit.iniin C

N utrientQu artile Test of trend

1 low) 2 3 4 high) P

Dietary folateOR.95% CI

1 .0 1 . 10.5-2.3

1)6

0)3 � 1 .5

0.5

01.2- 1.1

0.03

OR,�

95%CI

1.0 0.8

0.3-1.9

0.7

01.2-2.0)

0.40.1-1.1

0.07

Dietary vitamin C

OR,95% CI

1 .0) 0.7

0.3 1 .60.6

0.3 1 .30.2

O).1 0.50.00”

OR;’95% Cl

1 .0 (1.6

0.2 -1 .801.6

0.2 � 1 .60.2

0.0)’ -0.70.03

., Adjusted br calories prior to Iogistb regression using Iine,ir regression

procedures ) 1 0). The logistic moh’I inc luck’s acljustnient for currentsmoking status, monthly personal incoilic’, frequency of cervical smearannual versus less (ifteii), any use ut sI)ernliciclal cc)ntraceptive agents.

self-reported history of genital s�’arts, qu,irtile of Quetelet index, andnatural log of number of sexual p.irti�c’rs.,‘ <0.005.( <0.05.

lower quartiles of serum and red cell folate are a resultof folate sequestration by the dysplastic cells.

Similarly, while anorexia is often a symptom of those

with cancer, it has not been documented as a feature of

CIN. It is unlikely that cases exhibited either poorer folatestatus or lower intake of vitamin C as the result of dietaryintake depression in sick patients with poor appetites.

Additionally, because participants were asked to recordtheir average frequency of consumption over the last 5years, the potential for reported dietary intake reflectingchanges subsequent to the onset of disease seems un-likely. Nonetheless, given the case-control design of thisstudy, the possibility that the lower levels of serum folate,red cell folate, and dietary vitamin C for the cases resultedfrom the disease process cannot be ruled out.

Levels of serum retinyl palmitate indicated high lev-

els of compliance with the fasting requirement amongboth cases and controls. For the 98 pairs included in theserum folate assays, one control and two cases had levelsof retinyl palmitate indicative of nonfasting. Therefore,

bias due to differential compliance among cases andcontrols regarding fasting is unlikely.

Hematological and Dietary Factors. Red cell folate showsa strong, statistically significant inverse association with

CIN. Additionally, the estimates of effect for both serumand dietary folate offer supporting evidence for the roleof inadequate folate nutritional status in the developmentof CIN.

Some of the difference in the findings for the folatevariables may be attributable to characteristics of themeasurements. Greater variability is expected in theserum compared to the red cell folate measure, because

serum folate indicates short-term changes in folate bal-ance, while red cell folate reflects changes over severalmonths (16). For folates, there is a greater potential formisclassification from dietary intake measures compared

to laboratory data because of inaccuracies in reportingand converting food to nutrients, the destruction of fo-

on July 25, 2020. © 1992 American Association for Cancer Research. cebp.aacrjournals.org Downloaded from

Cancer Epidemiology, Biomarkers & Prevention 123

Table 5 Correlationa among vitamin assays” and dietary intakemeasures�

Folate

Red cell

assays

Serum

Dietar

Folate

y intake

Vitamin C

Folate assaysRed cell folate 0.61 d

Serum folate

Nutrient intake

Folate 027d 0.22” 0.55”

Vitamin C 0.14 0.22”

Food intakeTotal fruit 0.00 0.02 0.60” 0.66”

Citrus fruit -0.01 0.03 062d 065d

Total vegetables 0.21’ 0.14’ 0.14’ 0.12

Vegetables excluding 025d 0.19’ 0.19’ 0.12

rice and potatoes

a Pearson product-moment correlations.b Hematological measures have been Iog,-transformed.C Dietary intake measures were adjusted for total energy intake using

linear regression methodology (10) prior to correlation procedure.d p < 0.005.

‘ P � 0.05.

lates during food preparation, and differential absorptionof folates from diverse food sources.

The findings of this study are consistent with those

of Butterworth et a!. (3), who noted higher levels of bothserum and red cell folate in women with cervical dys-plasia versus hospital employee controls. Due to the lackof statistically significant results possibly related to therelatively small sample size of 34 cases and 40 controls,Butterworth’s findings remain suggestive. Brock et a!. (4)noted an inverse association between folate intake andcervical carcinoma in situ on crude analysis. This protec-tive effect disappeared in the adjusted model, whichincluded nonnutnient risk factors, as well as dietary cano-tene, retinol, vitamin C, and energy. However, the valid-ity of simultaneous addition of dietary folate and vitamin

C into a logistic regression model must be questioned,given the generally high correlation between the twomeasures (Table 5 and Ref. 20). Of two studies on in-vasive cervical cancer and folate intake, one (5) showed

an inverse association between dietary folate and diseasein an analysis which did not control for other risk factors,and the second (6) revealed a protective effect for folateamong heavy smokers. These two studies focused onwhite women whose socioeconomic status and risk ofdisease are not described. Therefore, these samples mayhave differed in a significant fashion from that of thepresent study, which included primarily low-income,nonwhite women from a population at high risk of

disease.Two of the many mechanisms suggested for folate

deficiency causing altered DNA and subsequent carci-nogenesis are based on the observed misincorporationof uracil into DNA in place of thymine, the de novosynthesis of which is folate dependent. One theory sug-gests that this leads to methyl-poor regions in the DNAstrand, which preclude the coiling necessary for theproper configuration of the DNA molecule (21). Thesecond hypothesis proposes that efficient DNA repairmechanisms cause chromosome breakage through me-peated excision-repair cycles, which aim at removing the

misincorporated uracil, but which are futile in the low-thymine environment of folate deficiency (22).

That folate may play an antitumonigenic role bypreventing preneoplastic epithelial cellular changes issuggested in both the Butterworth et a!. study (3) and thepreliminary results of a chemopreventive trial with menat high risk of lung cancer (23). The findings from thelatter study indicate regression of bronchial squamousmetaplasia with folic acid and vitamin B12 supplementa-tion. Folate supplementation has also been reported tobe protective against the development of colon cancerand dysplasia in patients with chronic ulcerative colitis(24).

Evidence for an inverse association between dietaryvitamin C and risk of CIN is consistent with the findingsfrom a similar investigation (25). Of an additional threestudies, one found a statistically significant inverse asso-ciation between dietary vitamin C and invasive cervicalcancer (5); one found no association between vitamin Cintake and invasive cervical cancer (6); and one wassuggestive of an increase in risk of cervical carcinoma insitu with decreased consumption of vitamin C (4). Theevidence of an inverse association between CIN andvitamin C is also consistent with the finding of lowerlevels of serum vitamin C in women with CIN than inage- and clinic-matched controls (26). This latter study,however, failed to control for several possible confoun-dens, including smoking. The suggestions of an associa-tion, combined with the inconsistencies of previous in-vestigations, indicate the need for further research onthe relationship ofvitamin C to cervical dysplasia. VitaminC functions as an antioxidant and enhances cellular im-munity, both of which may play a role in cancer preven-tion (27).

Due t� the relatively high correlations between thefolate and vitamin C measures, it is not possible todelineate the relative importance of these nutrients aspossible preventive agents in the etiology of CIN. Whenquartiles of dietary vitamin C and red cell folate areentered simultaneously into a logistic model with thenonnutnient variables, there continues to be a statisticallysignificant protective effect evident for those in the high-est quartiles of both measures, and a dose gradient isevident. The decreases in odds of CIN for those in thehighest quartiles of red cell folate and dietary vitamin Care also evident controlling for quartile of serum a-carotene, /3-carotene, lycopene, and lutein.

That dietary intake of folate is highly correlated withfruits, while the serum and red cell folate measures aremore strongly correlated with vegetables, may indicatean inadequacy in the dietary folate measure.

Although it was not part of the study design, serumvitamin B2 assays were conducted simultaneously withthe folate assays. A preliminary review of these analysessuggests that vitamin B12 may play a role in a subset ofwomen with cervical dysplasia. However, controlling forquartile ofvitamin B12 did not substantively alter the oddsratios associated with serum and red cell folate.

The findings from this study support dietary recom-mendations, such as those of the American Cancer So-ciety, the National Cancer Institute, and the U.S. DietaryGuidelines, which allow for adequate intake of folate andvitamin C, both of which are found in good quantity infruit and vegetables. Increased consumption of legumesand whole grains is also in accord with current dietary

on July 25, 2020. © 1992 American Association for Cancer Research. cebp.aacrjournals.org Downloaded from

124 Folate, Vitamin C, and Cervical lntraepithelial Neoplasia

recommendations, and both of these types of foods are

good sources of folates.

AcknowledgmentsThe authors wish Ic) acknowledge the support c)f Stanley Gall at theUniversity of Illinois Hospital, Michael Makii at Cook County Hospital.Eileen McAleer at the Bronx V.A. Hospital, and Ray Murphy at the Illinois

Department of Public Health. They have appreciated the advice and( oiiiments of Phyllis Bowen, lack GOIdE)erg, William Haenszel, Frederick

Kviz, and Victoria Persky at the University of Illinois at Chicago.

References1 . Yunis, I. I.. ,incl Soreng, A. L. Constitutive fragile sites and cancer.

Siience (Washington DC), 226: 1 199-1204, 1984.

2. Koss, [. C. Diagnostic Cytology and Its Histopathologic Bases, Ed. 3.Philadelphia: I. B. Lippincott Co., 1979.

.1. Butterworth, c. E., Hatch, K. D., Gore, H., Mueller, H., and Krumdieck,

(T. Improvement in cervical dysplasia associated with folic acid therapyin users of oral ontraceptives. Am. I. Clin. Nutr., 35: 73-82, 1982.

4. Brock. K., Berry, C., Mock, P. A., Truswell, L. A., and Brinton, L. A.Nutrients in the diet and plasma and risk of in situ cervical caner. I. NaIl.

Cancer Inst., 80: 580-585, 1988.

5. Verreault, R., (Thu. I.. Mandelson, M., and Shy, K. A case-control study

of diet and invasive cervical cancer. tnt. I. Cancer, 43: 1050- 1054, 1989.

6. Ziegler, R. C., Brinton, L. A., Flaiiiman, R. F., Lehman, H. F., Levine,

R. S., Mallin, K., Norman, S. A., Rosenthal, I. F., Trumble, A. C., and

Hoover, R. N. Diet and the risk ot invasive cervical cancer among whitewomen in the United States. Am. ). Epideniiol., 112: 432-444, 1990.

7. VanEenwyk, I. Nutrition and (Tervical lntr,sepithelial Neoplasia: A Case-

Control Study (Ph.D. dissertation). Chicago, IL: Graduate College of the

University of Illinois, 1 990.

8. VanEenwyk. I.. Davis, F. C., and Bowen, P. E. Dietary and serumcarotenoids and cervical intraepithelial neoplasia. Int. I. Cancer, 48: 34-38, 1991.

9. Block, G., Coyle, L., Smucker, R., Harlan, L., Hartman, A., and Kessler,

L. Health Habits and History Questionnaire: Diet History and Other RiskFactors. Personal Computer System Packet. Bethesda, MD: NationalCancer Institute, Division of Cancer Prevention and Control, 1 987-- 1 989.

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

1 1 . Gutcho, S. Source of error in determination of erythrocyte Dilate by

competitive binding radioassay. Clin. Chem., 24: 388-389, 1978.

12. Waxman, S., Schreiber, C., and Herbert, V. Radioisotopic assay formeasurement ofserum folate levels. Blood, 38: 219-228, 1971.

1 1. Longo. D. L., and Herbert, V. Raclioassav for s(’rum and red ( elI

folate. I. Lab. Clin. Med., 87: 138-151, 1976.

14. laiob, F., CoIii�an, N., and HerI�’rt, V. Evaluation of siniult,ineousradioassav for two vit,imins: folate and vitaii�in B- 1 2. Am. I. Clin. Nutr.,

30:616, 1977.

1 5. Naessans, I. M., Offord, K. P., S ott, W. F., md l)aoocl, S. I . TheMCSTRAT procedure. In: R. P. Hastings ed.), SUGI Supplemental Library

User’s Guide, Version 5 Ed., ip. 307- 328. Can,’, NC: SAS Institute, In.,1986.

16. Sc’nti, F. R., .10(1 t’ilch, S. lvi. Assessnient (ii tl�’ Folate Nutritional

Status of the U.S. Population Based on Data Collected in the Sec ondNational Health and Nutrition Examination Survey 1976- 1 98L). Bethesda,

MD: Life Sciences Research Offices, FASEB, 1984.

1 7. Clarnon, C. H., Feld, R., Evans, W. K., Weiner, R. S., Kramer, B. S.,Lininger, L. L.. Gardner, F. B., Wolfe, E. C., DeWs’s, 55’. D., and Hotfnian,F. A. Serum folate and vitamin B 1 2 levels in P.iti#{128}’ntswith small elI lung

ancer. Cancer (Phila.), SI: 306- 310, 1984.

1 8. F-liii, Y. H. Hunian Nutrition and [)iet Therapy. Monterey, CA: Wads-

worth, Inc., 1983.

19. Cavanagh, 0., Praphat, H. and Ruffolo, E. H. Car inoma of the uterine

cervix: some current views. Obstet. and Gynecology Annual, 10: 1 93

2 36, 1981.

20. Butterworth, C. E., and Norris, D. Folic acid and vitamin C in cervical

dysplasia. Am. I. Cliii. Nutr., :17: 332 -333, 1983.

2 1 . Krunidieck, (T. L., ,in(J Howard-Peebles, P. N. On the nature of Iiili

a (1 sensitive fragile sites in human c hromosonies: an hypothesis. Aiii. I.

Med. Genet., 16: 23 28, 1983.

22. Goulian, M., Bleile, B., and Tseng, B. Y. Methotrexate-induecl urns-inucirporation of uracil mId) DNA. Proc. NatI. Acad. Sci. USA, 77: 1 956-

1960, 1980.

23. Heimburger, 0). C., Alexander, B., Birch, R., Buttervvorth, C. E., Bailey,

W. C., and Krunidieck, C. L. Improvement in bronchial squamous meta-

plasia in smokers treated with tolate and vitamin Bt2. JAMA, 259: 1525-

1530, 1988.

24. Lashner, B. A., 1-leidenreich, P. A., Sn, C. I. ., Kane, S. V., and Hanauer,

S. B. Effect of folale supplementation on the inc idence of clyspl,isis mdcanc er in chronic ulc erative colitis. Gastroenterology, 97: 255-259, 1989

25. Wassertheil-Smoller, S., Romne�, S. L., \Vylie-Rosett, I., Slagle, S.,Miller, C., Lucido, D., Duttagupta, C., and Palan, P. R. I)ietarv vitamin C

and uterine cervical dysplasia. Am. I. Epidemiol.. I 14: 7 14 724, 1981.

26. Riininey, S. F., Duttagupta. C., Basu, I., Palan, P. R., Karp, S.. SI.mgle,S., Dwyer, A., Wassertheil-Snioller, S., And Wylie-Rosett, I. Plasma vit,i-nun C and uterine cervid al clvsplasia. Am. I. Obstet. Gv’nrnoI., I � 1 S 976980, 1985.

27. Glatthaar, B. F., Hornig, D. H., ,mncl Moser, Li. 1 he role if ,is� i)rl)i(

acid in carcinogenesis. Adv. Exp. Med. Biol.. 206: 357 377. 1986.

on July 25, 2020. © 1992 American Association for Cancer Research. cebp.aacrjournals.org Downloaded from

1992;1:119-124. Cancer Epidemiol Biomarkers Prev   J VanEenwyk, F G Davis and N Colman  Folate, vitamin C, and cervical intraepithelial neoplasia.

  Updated version

  http://cebp.aacrjournals.org/content/1/2/119

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

  .pubs@aacr.orgDepartment 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/1/2/119To request permission to re-use all or part of this article, use this link

on July 25, 2020. © 1992 American Association for Cancer Research. cebp.aacrjournals.org Downloaded from