Introduction

Despite a substantial decrease in the incidence and mortality of carcinoma of the cervix over the past 40 years, this disease remains a significant health problem in Canada. In 1995, there will be approximately 1300 newly diagnosed cases of cervical cancer and 370 deaths.(1) Introduction of the Pap smear in the early 1950s had a major impact on reducing the incidence of invasive lesions and deaths; however, recent evidence suggests that incidence and mortality rates are no longer declining.(1)

As a screening tool the Pap smear has poor sensitivity,(2) and mass screening should only be accompanied by the development of information systems and strict quality control measures.(3) Further research into the etiology and prevention of cervical cancer is needed, and primary prevention may become a desirable alternative to early detection and treatment.

A number of risk factors associated with cervical cancer have been identified, including cigarette smoking, parity, oral contraceptive use, multiple sexual partners and infection with human papilloma virus (HPV).(4) The influence of various nutritional factors, such as folate, vitamin C, vitamin E and carotenoids, on the cervical epithelium has also been explored.

This review provides a critical appraisal and synthesis of the literature relating intake of folate to the risk of developing cervical neoplasia. It is recognized that the neoplastic process is complex and is influenced by a variety of factors, which are commonly genetic and/or environmental in origin. When studying nutrition and cervical cancer, similarly, it should be noted that dietary micronutrients likely exert their effects in a concerted fashion rather than independently. Due to difficulties in measuring interaction between micronutrients, epidemiologic studies have largely focused on individual, rather than interdependent, effects. For this reason, the current review is limited to the separate role of folate (although whenever possible the independent effects of other micronutrients that have been studied are presented).

Folate, a member of the B group of vitamins, is the dietary form of this vitamin. When it exists in supplemental form (as in multivitamin preparations), it is known as folic acid. The role of folate in human carcinogenesis and in the treatment of cancer has been extensively studied over the past 40 years. A growing body of epidemiologic evidence suggests that folate deficiency may play a role in the occurrence of cervical cancer, possibly through an interaction with other risk factors.

Since folate deficiency may have different effects depending on the stage of dysplasia, it is useful to understand the natural progression of cervical lesions to invasive disease. Cervical cytology is conventionally described in terms of grades of dysplasia or cervical intraepithelial neoplasia (CIN). CIN I represents very mild to mild dysplasia; CIN II, moderate; and CIN III, severe dysplasia. Dysplastic epithelium, especially CIN I, can revert to normal: this process is known as regression. Although the literature reports conflicting rates of regression, up to 60% of mildly dysplastic lesions can spontaneously revert to normal within a relatively short period of time (6-12 months).(5)

There is actually little distinction between CIN III and carcinoma in situ. Low-grade lesions (atypia and CIN I) have minimal involvement of abnormal cells in the epithelial wall, whereas high-grade lesions show increasing involvement of cells exhibiting nuclear atypia and loss of cellular differentiation. Once the dysplastic cells invade the basal epithelium, invasive carcinoma is present.(6) It is not known at what stage nutritional factors may have an impact. Therefore, in this report, the effects of folate on cervical dysplasia and invasive disease are considered separately.

Methods

Studies were identified using a MEDLINE search of the period January 1984 to May 1995 (inclusive), searching the reference lists of retrieved articles and through direct communication with experts in the field. Unpublished studies were not considered.

The database was searched using the National Library of Medicine's Medical Subject Heading (MeSH) "cervical neoplasia" (which included these more specific terms: "cervical intraepithelial neoplasia," "dysplasia" and "invasive cancer" or "carcinoma") and the subject term "folate" (including folic acid and all forms of this vitamin).

The list of citations was refined by excluding animal studies, studies that were not published in English and studies that did not address folate or folic acid as a nutritional factor in the development of cervical neoplasia. Twenty pertinent articles were retrieved, of which eleven involved original research. A critical appraisal of these 11 studies is summarized in tables 1 and 2.

Folate and Cervical Dysplasia

The first evidence of the folate-cervical cancer link was brought into focus in 1973 by Whitehead et al., who observed megaloblastic features in cervical epithelial cells in women taking oral contraceptive steroid hormones.(7) Even though systemic signs of deficiency were not apparent, the cytologic changes disappeared within three weeks of oral folate supplementation. It was postulated that folate deficiency occurred at the end-organ level due to hormonal stimulation, and thus the concept of localized vitamin deficiency was introduced.

The effect of oral folic acid supplementation on the course of cervical dysplasia was evaluated in two clinical intervention trials.(8,9) In 1982, Butterworth et al. reported that daily supplementation with folic acid for three months in 22 subjects was associated with an improvement in dysplasia score compared to the 25 placebo-treated control subjects.(8) Although the study design was theoretically strong, the following methodologic problems were identified.

• The sample size was small (47 subjects).

• The results were not adjusted for the effect of known risk factors or other confounding variables.

• The participation rate was low, with only 47 out of the 78 enrolled subjects completing the protocol.

• Information on the compliance and drop-out rates was not provided.

A serum vitamin assay revealed that red blood cell (RBC)-folate levels were lower in oral contraceptive pill users as compared to non-users. However, these results may be unreliable since baseline levels of serum and RBC-folate levels were not obtained, blood samples were drawn at variable times during the study and the control group (hospital workers) was not comparable to the cases.

Subjects received three Pap smears over the course of the study that were categorized from 1 to 5 according to the severity of the cytology (1: less severe; 5: carcinoma in situ). An average dysplasia score was then computed for each group and the scores were compared both within and between groups using a t-test. The validity and reliability of this scoring system were not discussed.

A second double-blind randomized controlled trial was conducted by Butterworth et al. in 1992.(9) A total of 235 subjects with CIN I or II were randomized to receive either 10 mg of folic acid or a placebo daily for six months. After adjusting for potential confounders, there were no significant differences between supplemented and unsupplemented subjects regarding dysplasia status, biopsy results or prevalence of HPV infection. The findings were felt to be reliable since the sample size and selection of study subjects were reliable. High rates of regression of cervical cytology in both the placebo and folic acid-supplemented groups (at 66.3% and 63.7% respectively) may have contributed to the lack of significant differences.

A number of case-control studies have examined the relationship between folate deficiency and the development of cervical dysplasia.(10-12) Butterworth et al. conducted such a study on women of childbearing age attending a county health department clinic for routine health examination or family planning advice.(13) All women with abnormal Pap smears were referred to a colposcopy clinic where a second Pap was performed along with a gynecological exam, a smear for pathogens, scrapings for HPV and blood sampling for 12 nutritional indices including RBC-folate.

This study was the first of its kind to show an interaction between level of folate and risk factors for cervical dysplasia. Although the risk of developing cervical dysplasia was not associated with plasma nutrient levels, a low level of RBC-folate enhanced the effect of HPV infection. The adjusted odds ratio for HPV-16 was 1.1 among women with folate levels above 660 nmol/L, but it was 5.1 (95% confidence interval [CI] = 2.3-11) among women with lower levels.

Of all the articles reviewed, this was the only study that attempted to verify the classification of cases and controls. Smears referred from the family planning clinic that were negative at colposcopy were reassessed and false positives were excluded. To further reduce the likelihood of false positives, cases who did not have a colposcopically visible lesion were excluded. Of the 237 potential controls, 1 was reclassified as a case and 66 were excluded due to inconclusive cytologic or colposcopic evidence. Unfortunately, since histology specimens were not obtained, estimates of Pap smear sensitivity and specificity could not be given.

VanEenwyk et al. conducted another case-control study assessing the possible link between CIN and folate in serum, red blood cells and diet.(10) The association between vitamin C and CIN was also investigated. After adjusting for potential confounding factors, RBC-folate and vitamin C were found to be protective. Micronutrient levels were divided into quartiles with 4 being the highest and 1 the lowest. For RBC-folate, the odds ratios (with 95% CIs) comparing quartiles 4, 3 and 2 to quartile 1 were 0.1 (0-0.4), 0.6 (0.2-2.0) and 0.5 (0.2-1.9), respectively.

For vitamin C, the odds ratios were 0.2 (0-0.7), 0.6 (0.2-1.6) and 0.6 (0.2-1.8) for quartiles 4, 3 and 2 compared to quartile 1. These findings should be interpreted with caution since the possibility of referral and non-response bias were not excluded. Cases and controls were recruited from two hospital clinics, and it is unknown whether these patients were representative of the community. If study subjects differed from the population from which they were drawn, then the true effects of folate may have been either underestimated or overestimated.

Only 102 of the 166 potential cases participated in the study (61%), and the response rate for the controls was even lower at 52%. The authors did address the possibility of non-response bias by collecting information on the age, ethnic status, zip code and mode of payment for medical services of all participants. Lower response rates were found among women aged 18-24 years and among those who paid for their own medical expenses. Since cases and controls were matched according to age, non-response in the younger group should not have affected the results. The lower response in women paying for their own medical expenses could have led to a selection bias though, with overrepresentation of higher (privately insured) and lower (public aid) socio-economic groups in the sample.

Another retrospective study by McPherson revealed an association between folate intake and the risk of cervical dysplasia.(11) Information on diet, demographics and known risk factors for cervical cancer was collected by questionnaire on 75 cases and 84 controls. Women consuming less than 397 mcg per day of folate were 2.66 times more likely to develop dysplasia than women who consumed more than 397 mcg per day. Since the investigation had only been reported in abstract form at the time of this review, a critical appraisal of the study methods was not performed.

Buckley et al. conducted a less rigorous case-control study of the relationship between dietary micronutrients (folate, vitamin C, vitamin E, retinol and carotenoids) and cervical dysplasia in American Indian women.(12) Low intake of vitamin C, vitamin E and folate was associated with an increased risk of CIN. Comparing the lowest level of intake to the highest, the odds ratios (95% CIs) for folate, vitamin C and vitamin E were 3.31 (1.87-5.84), 3.03 (1.78-5.15) and 1.65 (1.07-2.53), respectively.

Despite the excellent response rate for both cases and controls ( >90%), the following methodologic problems were identified.

• Cases were selected according to a history of dysplasia in the past year. Since dysplastic lesions can regress during this period of time, misclassification bias may have been introduced.

• The instrument (questionnaire) used to measure dietary intake was not validated.

• Results were not adjusted for possible confounding factors (demographic, behavioural, known risk factors). For example, the control group was of higher socio-economic status (SES) than cases. Therefore, SES may have confounded the association between dietary micronutrients and cervical dysplasia.

A summary of the folate-dysplasia studies is presented in Table 1. The first clinical intervention trial by Butterworth et al.(8) revealed an improvement in dysplastic changes with folic acid supplementation; however, this study had significant methodologic limitations. A second and more methodologically sound randomized controlled trial by the same investigator did not reveal an association between folate and cervical cancer.(9)

The most rigorously conducted case-control study had negative findings.(13) The case control studies by VanEenwyk et al.(10) and Buckley et al.(12) were methodologically weaker, but showed that folate may protect against cervical cancer.

In view of the inconsistent findings and weak methodology in a number of these studies, a conclusion regarding the role of folate in the development of early cervical carcinogenesis cannot be made. Further well-conducted research addressing problems with misclassifaction bias and validation of study instruments must be conducted before this question can be answered.

Folate and Invasive Cervical Cancer

It is unknown whether folate status is important early on when cervical abnormalities are first being initiated or whether levels of folate affect the progression of carcinoma in situ to invasive carcinoma. Invasive cervical cancer has the potential to metastasize and/or invade other organs and is treated promptly with either surgery, radiation or both. Unlike CIN then, this disease cannot be studied using the design of a randomized clinical trial. The relationship between folate (either in the diet or in serum) and cervical carcinoma has mainly been evaluated using the case-control design. This report presents the findings of four such investigations.

Potishman et al. evaluated whether low levels of serum folate were associated with increased risk of invasive cervical cancer in a Latin American population.(14) All women with newly diagnosed cervical carcinoma from four major treatment centres (Bogota, Colombia; Mexico City, Mexico; Costa Rica; and Panama) were invited to participate. Each case from Bogota and Mexico City had two age-matched hospital controls. For Costa Rica and Panama, cases were matched with one hospital and one community control of the same age. Ninety-five percent of eligible subjects agreed to participate and were interviewed regarding their medical, reproductive, demographic and dietary status.

Mean levels of folate were similar for cases and controls. After adjustment for possible confounders, there was no association between level of serum folate and risk of cervical cancer, and an interaction with known risk factors was not observed.

The negative findings in this study may be attributed to the fact that serum folate (a short-term marker of folate status) was measured instead of RBC-folate. The latter is less affected by fluctuations in diet and is, overall, a more reliable measure of folate stores. Given that carcinoma of the cervix has a relatively long latency period, it is unlikely that recent changes in the diet would affect the development of this disease. Measurement of long-term stores of folate may have been more useful and may have produced different results.

Using the same population and methodology as described in the above study, Potishman et al. also examined the association between dietary intake of certain micronutrients (folate, vitamin C, carotenoids and retinol) and invasive cervical cancer.(15) Daily intakes of specific nutrients were derived from dietary assessments; however, multivitamin supplementation was not measured.

A slightly lower risk of cervical cancer was observed with the highest consumption of fruit juices (odds ratio = 0.9, p = 0.008), although a dose-response relationship was not seen with increasing levels of consumption. In terms of specific nutrients, folate was not associated with risk of cervical carcinoma. A trend of decreasing risk was evident for vitamin C and carotenoids.

Despite the positive aspects of this study (high compliance rate and detailed collection of data on risk factors), interpretation of the results is limited. Further analyses revealed that dietary associations were driven mainly by reductions in risk from two areas (Colombia and Mexico) and by women of higher socio-economic status, introducing the possibility of selection bias.

Dietary questionnaires require extensive validation in order to minimize or eliminate the potential for measurement error. Since the questionnaire in this study was not validated, micronutrient intakes may have been either overestimated or underestimated and the protective effect of folate also may have been underestimated.

Ziegler and colleagues studied the association between diet and incident invasive cervical cancer in white women in five areas in the United States: Birmingham, Alabama; Chicago, Illinois; Denver, Colorado; Philadelphia, Pennsylvania; and Miami, Florida.(16) A questionnaire was administered in the home to 271 cases and 502 age-matched controls to obtain data on demographic, dietary and known risk factors. Diet was assessed using the methods of the First and Second National Health and Nutrition Surveys (NHANES I and II). Multivitamin use in the preceding 20 years was also determined.

After adjustment for potential confounders, an association between cervical carcinoma and the four micronutrients studied (folate, vitamin C, vitamin A or carotenoids) was not observed. Multivitamin use for more than 15 years appeared to be protective and a high intake of vitamin C was protective for heavy smokers.

Methods used in this study were quite rigorous, participation rates were high (73% for cases and 75% for controls), power was sufficient and the results were adjusted for potential confounders. Recall bias was unlikely since cancer patients would have to report eating more than they did to obscure the findings and this would be very unusual. Dietary assessment was felt to be adequate since the study instrument (a semi-quantitative food frequency questionnaire used in NHANES I and II) was found to be valid and reproducible in other studies.

A fourth population-based case-control study by Verrault et al. examined the relation of diet, especially intake of vitamins A, C and E and folate, to the risk of invasive cervical cancer.(17) Cases were 189 women diagnosed with invasive cervical cancer between 1979 and 1983 in three counties in the Seattle area. Controls (n = 227) were selected from the community by telephone through random digit dialing. After adjusting for possible confounders, intake of folate and vitamin A was not found to be related to risk of cervical cancer. There was, however, an inverse relationship between intake of vitamins C and E and risk of cervical cancer. High consumption of fruit juices and green and yellow vegetables was also found to be protective.

One limitation of this study was the lack of validation of the food frequency questionnaire. The authors offered no information on the accuracy of this instrument even though it apparently had been used in previous similar studies. We cannot be sure then that the food group and micronutrient measurements reflected the true dietary status of the subjects. In addition, socio-economic status, a known risk factor for cervical cancer, was not assessed in this study (although education level was controlled for in the analysis).

In summary (see Table 2), one study was very well conducted,16 and there were few serious methodologic problems with the other three investigations.(14,15,17) The findings from these studies are consistent and show that folate does not appear to protect against invasive cervical cancer.

Conclusion

Researchers have had a long-standing interest in the nutritional epidemiology of cervical neoplasia. A variety of micronutrients have been explored, and, although this review focused on the role of folic acid, it is recognized that cytologic changes in the cervical epithelium are likely influenced by a combination of dietary factors. Studies evaluating the association between folic acid and cervical neoplasia have been complicated by inaccuracies in dietary measurement and incorrect classification of cases and controls using the Pap smear.

Studies examining the association between folic acid and cervical dysplasia have had inconsistent findings. Supplementation with folic acid was not found to be protective in a recently conducted randomized controlled trial.(9) A well-conducted case-control study revealed an interaction between low RBC-folate levels and HPV-16 on the risk of cervical cancer.(13) Another case-control study by VanEenwyk et al.(10) had significant methodologic problems, but showed an inverse association between folate levels and risk of cervical cancer. Although conclusive evidence of folate as a protective factor has not been demonstrated, dysplasia of the cervix may result from an interaction between specific dietary micronutrients and other risk factors such as HPV. Studies with more rigorous methodology need to be conducted to clarify the folate-dysplasia relationship.

Intake of folate does not appear to protect against invasive carcinoma of the cervix. All four reasonably well-conducted case-control studies had negative findings. Although this relationship may still exist, it seems more plausible that dietary micronutrients exert their effects during the early stages of carcinogenesis rather than later, when invasive aggressive tumours have developed.

A comprehensive assimilation of the research examining the role of other dietary micronutrients was beyond the scope of this review, but in many of the above studies, intake of other nutrients, especially vitamin C, appeared to protect against cervical cancer. Therefore, the individual and conjoint effects of a variety of dietary micronutrients on the cervical epithelium also deserves further evaluation.

References

1. National Cancer Institute of Canada. Canadian Cancer Statistics 1995. Toronto: NCIC, 1995.

2. Fahey MT, Irwig L, Macaskill P. Meta-analysis of Pap test accuracy. Am J Epidemiol 1995;141(7):680-9.

3. Mills CJ. Toward the implementation of organized screening [workshop report]. Chronic Dis Can 1993;14(3):110-2.

4. Miller AB, Anderson G, Brisson J, Laidlaw J, Le Pitre N, Malcolmson P, et al. Report of a national workshop on screening for cancer of the cervix. Can Med Assoc J 1991;145(10):1301-25.

5. Nasial K, et al. Behaviour of mild cervical dysplasia during long-term follow-up. Obstet Gynecol 1986;67:665-9.

6. Potishman N. Nutritional epidemiology of cervical neoplasia. American Institute of Nutrition 1993:424-9.

7. Whitehead N, Reyner F, Lindenbaum J. Megaloblastic changes in the epithelium. JAMA 1973;226(12):1421-4.

8. Butterworth CE Jr, Hatch KD, Gore H, Mueller H, Krumdieck CL. Improvement in cervical dysplasia associated with folic acid therapy in users of oral contraceptives. Am J Clin Nutr 1982;35:73-82.

9. Butterworth CE, Hatch KD, Soong SJ, Cole P, Tamura T, Sauberlich H, et al. Oral folic acid supplementation for cervical dysplasia: a clinical intervention trial. Am J Obstet Gynecol 1992;166(3):803-9.

10. VanEenwyk J, Davis FG, Colman N. Folate, vitamin C and cervical intra-epithelial neoplasia. Cancer Epi Bio Prev 1992;1:119-24.

11. McPherson RS. Nutritional factors and the risk of cervical dysplasia [abstract]. In: Proceedings and abstracts of papers presented at the twenty-second annual meeting of the Society for Epidemiological Research; 1989 Jun 14-16; Birmingham (Alabama). Am J Epidemiol 1989;130(4):830.

12. Buckley, DI, McPherson RS, North CQ, Becker TM. Dietary micronutrients and cervical dysplasia in southwestern American Indian women. Nutr Cancer 1992;17:179-85.

13. Butterworth CE Jr, Hatch KD, Macaluso M, Cole P, Sauberlich HE, Soong S-J, et al. Folate deficiency and cervical dysplasia. JAMA 1992;267:528-33.

14. Potishman N, Brinton LA, Laiming VA, Reeves WC, Brenes MM, Herrero R, et al. A case-control study of serum folate levels and invasive cervical cancer. Cancer Res 1991;51:4785-9.

15. Herrero R, Potishman N, Brinton LA, Reeves WC, Brenes MM, Tenorio F, et al. A case-control study of nutrient status and invasive cervical cancer. Am J Epidemiol 1991;134:1335-46.

16. Ziegler RG, Brinton LA, Hamman RF, Lehman HF, Levine RS, Mallin K, et al. Diet and the risk of invasive cervical cancer among white women in the United States. Am J Epidemiol 1990;132(3):432-45.

17. Verrault R, Chu J, Mandelson M, Shy K. A case-control study of diet and invasive cervical cancer. Int J Cancer 1989;43:1050-4.

Author Reference

Maureen Carew, Resident in Community Medicine and Epidemiology, University of Ottawa, c/o Christina J Mills, Laboratory Centre for Disease Control, Health Canada, Tunney's Pasture, Postal locator: 0601E2, Ottawa, Ontario K1A 0L2