An interest in possible relationships between megaloblastic cells and cancerous ones dates back at least four decades. In 1954, Massey and Rubin described the persistence of abnormal gastric columnar cells in the stomachs of individuals with pernicious anemia, even after the anemia had been successfully treated with vitamin B12. Since the cytologic appearance of these cells exhibited some characteristics of both megaloblastic and cancer cells, and since several prior reports had observed an increased incidence of gastric cancer in individuals with pernicious anemia,[2-4] Massey and Rubin postulated that these abnormal cells might represent a transitional cell type between cells that characterize the atrophic gastric epithelium in pernicious anemia and gastric cancer cells.
The first studies to examine a relationship between folate-associated megaloblastosis and cancer were conducted in the 1960s. Van Niekerk noted several cytologic similarities between epithelial cells of the uterine cervix from folate-deficient women and dysplastic cervical cells. In both cases, the cells were larger than normal, had nuclei that were disproportionately large compared to the size of their cytoplasm, and had hyperchromatic chromatin. Nevertheless, for many years, there was little evidence to suggest that the cytologic similarity between megaloblastic and neoplastic cells was anything more than a coincidence.
It was not until the 1970s and '80s that investigators began to effectively question whether there might be a functional association between megaloblastic and dysplastic cells. In the 1980s, investigators at the University of Alabama conducted two prospective, controlled clinical intervention trials in individuals with dysplastic changes in the epithelium of the uterine cervix and metaplastic changes in the bronchi. Both studies observed a significant degree of attenuation in dysplasia or metaplasia after several months of supplementation with pharmacologic doses of folic acid(Drug information on folic acid) alone (Figure 1) or in conjunction with pharmacologic doses of vitamin B12. These findings lent further support to the concept that folate and dysplasia might be functionally related, although it remained unclear whether the alterations in cytology associated with supplementation were truly an attenuation of dysplasia or merely an attenuation of megaloblastosis masquerading as dysplasia.
Some studies of the uterine cervix highlight important features that characterize the clinical literature on folate and carcinogenesis. A large, recently published case-control study designed to examine the various risk factors for cervical dysplasia is one such example. Although this study did not find an overall association between red blood cell (RBC) folate levels and dysplasia, subjects with human papilloma virus-16 (HPV-16) infection of the cervix, an important risk factor for cervical carcinogenesis, had a fivefold greater risk of having dysplasia if they also had diminished RBC folate levels. This finding suggests that the enhancement of cervical carcinogenesis effected by diminished folate status requires cofactors that predispose to carcinogenesis. Thus, diminished folate status, by itself, is not a carcinogen; rather, it appears to enhance an underlying predisposition to cancer.
Another interesting feature of this study is the fact that diminished folate status was defined as an RBC folate level of < 660 nmol/L (291 ng/mL)--a value that is well within the range of folate levels conventionally accepted as normal and well above the threshold below which an individual runs the risk of megaloblastic anemia. This suggests that the effect of diminished folate status on carcinogenesis does not require frank folate deficiency as it is usually defined; rather, the data suggest that a modest diminution in folate levels is all that is necessary to produce an effect. Such a relationship also seems to exist in the colorectum.
As indicated above, much of the groundwork in this field was laid in studies of the uterine cervix. Several case-control clinical trials have examined the association between dietary folate intake and cervical neoplasia (Table 1).[8,10-14] Although several of these studies demonstrated that the relative risk of cervical neoplasia is somewhat diminished in individuals who consume larger quantities of folate, the effect uniformly fell short of statistical significance when adjustment was made for confounding variables.
This negative consensus among case-control studies in cervical neoplasia should not be construed as proof that no effect exists, as these studies suffer from several limitations: In most instances, folate status was assessed by estimating average dietary folate intake by means of food frequency questionnaires that have not been validated for folate intake. Also, none of the trials, except the study by Butterworth et al, stratified patients for HPV infections. These studies were therefore unable to explore possible synergism between diminished folate intake and HPV, and such stratification may be necessary to observe an effect of folate. Furthermore, all of the studies, except that by Butterworth et al, examined women with either carcinoma in situ or invasive cancer. These represent the latter stages of neoplasia, by which time the progression to cancer may be far enough advanced that there is little opportunity for modulation by such factors as folate status.
Two intervention trials (see Table 2)[15,16] have attempted to confirm the observations made in the University of Alabama study described above. Both intervention trials examined larger populations of women and followed the subjects over a longer period. In neither trial were the investigators able to reproduce the regression of dysplasia noted in the earlier study. However, on entry over two-thirds of the subjects in the more recent trials had the lowest grade of dysplasia--a disease stage known to have a rate of spontaneous reversion to normal of more than 60%. This certainly may have interfered with the ability of either study to observe a positive effect of folate supplementation.
In summary, the results of case-control and intervention trials in the cervix are not particularly compelling. However, the manner in which these trials were performed precludes one from making a definitive statement about whether alterations in folate status truly modulate the process of carcinogenesis in the cervix.
Clinical observations from several different groups point to an association between diminished folate status, assessed by either dietary intake or measurement of blood levels, and an enhanced risk of colorectal adenomas or cancer (Tables 3 and 4). Lashner and colleagues first made this observation in a case-control study of individuals afflicted with chronic ulcerative colitis--a disease that poses not only a 10- to 40-fold increased risk of colonic cancer but also a significant risk of folate deficiency.[19,20] Individuals who had been not been taking folate supplements over the long term had a rate of colonic neoplasia that was approximately 2-1/2-fold greater than the rate in those who had been taking supplements. Furthermore, chronic administration of sulfasalazine(Drug information on sulfasalazine), a drug commonly used for the treatment of ulcerative colitis that is well known to specifically inhibit both folate absorption[20,21] and folate metabolism, was associated with a 50% increase in the risk of dysplasia.
Both of these observations fell slightly short of statistical significance. Nevertheless, the study established the importance of examining this issue in the colon.
More recently, Lashner has confirmed these observations by prospectively comparing RBC folate levels in ulcerative colitis patients who harbor neoplastic changes in their colorectum with levels in age-, gender-, and disease-matched controls without neoplastic lesions. The mean RBC folate level was significantly lower in the individuals with neoplasia than in the controls when adjustment was made for confounding variables. It is also noteworthy that although the group harboring neoplasia had diminished RBC folate levels compared to the control group, the absolute folate level in the neoplasia group (454 ng/mL) was well within the range of values that is conventionally accepted as normal. This observation recapitulates the earlier finding made in the cervix.
Epidemiologic studies conducted in the general population generally support the hypothesis that diminished folate status is associated with an increased rate of colorectal neoplasia, although the nature of the relationship is different in each study. One study based in western New York observed that increased habitual intake of folate was associated with a decreased incidence of rectal, but not colonic, cancer, whereas two studies examining the population on the island of Majorca[25,26] observed reductions in relative risk for colonic and rectal neoplasms combined.
A study conducted in a population from western Washington state only assessed rates of colon cancer and did not observe a significant reduction in risk with increased dietary folate intake. However, a strong trend toward such an effect was present in women. Conversely, in a large study from southern California, decreased RBC folate levels were associated with a significantly higher risk of colorectal adenomas in men, but no significant effect was observed in women.
To date, some of the most compelling epidemiologic evidence to establish an association between folate status and colorectal neoplasia in the general population has come from studies that have carefully examined in-teractions between folate and other dietary components, particularly alcohol(Drug information on alcohol).[29-31] Giovannucci et al utilized large, prospectively analyzed populations to convincingly support the hypothesis that increased folate intake is inversely associated with the incidence of colorectal adenomatous polyps and of colorectal cancers. The prospective design of the two studies, the large size of the populations analyzed (25,474 and 47,931 subjects, respectively), and the effective control of several confounding dietary variables add considerable strength to the conclusions. In both studies, the superimposition of moderate to high alcohol intake greatly increased the neoplastic risk of a low-folate diet.
Similarly, using a nested case-control design among a prospectively examined cohort of men, Glynn et al observed that neither blood folate values nor dietary intake of folate alone was significantly associated with colorectal cancer. However, the combination of a high-alcohol, low-folate diet was associated with a highly significant, fourfold increased risk of cancer.
These studies in the general population are particularly important since they demonstrate an effect in people who clearly do not have other conditions predisposing them to colorectal cancer. Although the results from the epidemiologic studies pertaining to folate and colorectal cancer are not uniformly positive, the totality of the data provides a very persuasive argument for the existence of such an effect; certainly the effect is much stronger than that observed in the cervix.
Evidence from prospective, controlled, intervention trials are apt to provide the most compelling proof for a causal relationship between folate status and colorectal neoplasia, but results from such trials are just beginning to emerge. Results from two small intervention trials have been provocative, but nonetheless, inconclusive (Table 4).[32,33]
Paspatis et al randomized 60 individuals to receive either 1 mg of folic acid or a placebo daily following colonoscopic removal of an adenomatous polyp. Colonoscopic follow-up at 12 and 24 months revealed a rate of polyp recurrence in the folate group that was 40% to 60% of the rate observed in the placebo group. Nevertheless, these results fell short of statistical significance, perhaps because of the small size of the study population.
The second intervention trial, by Cravo et al, also utilized an intermediary biomarker of colorectal cancer as the primary end point, although a biochemical one. In this trial, 11 subjects with colonic polyps and 11 subjects with colorectal adenocarcinoma were randomized to receive either 10 mg or 0 mg of folic acid daily for 6 months following removal of their neoplasm. The major end point examined was genomic DNA methylation of normal-appearing rectal mucosa, which has been proposed as an intermediary marker of colorectal cancer.
Alterations in DNA methylation are commonly observed as an early biochemical phenomenon in colonic carcinogenesis.[34-38] Genomic hypomethylation of DNA has been noted uniformly in early, premalignant, colonic neoplasms,[34,35] as has hypomethylation of certain proto-oncogenes, such as K-ras. Genomic DNA hypomethylation has even been observed to occur as a widespread phenomenon in normal-appearing colonic mucosa of individuals harboring colorectal neoplasms. Therefore, in addition to possibly serving as a mechanism by which folate deficiency enhances carcinogenesis (a hypothesis that is discussed later in this review), genomic DNA methylation may be a satisfactory intermediary marker of colonic carcinogenesis.
Interestingly, DNA hypomethylation precedes most of the other known molecular events in colorectal tumorigenesis, suggesting that alterations in DNA methylation may play an active role in the evolution of carcinogenesis. Therefore, it is of interest to determine whether the DNA hypomethylation associated with early dysplasia can be modified by intervention. The DNA hypomethylation observed in early adenomatous polyps in humans is not likely a consequence of folate deficiency. Thus, it is of considerable interest that the trial by Cravo et al demonstrated that pharmacologic doses of folic acid reversed the genomic DNA hypomethylation present in the normal-appearing rectal epithelium of individuals harboring either colorectal adenomas or cancers. This observation is pertinent to the mechanism by which folate status modulates colorectal carcinogenesis and, in addition, has important implications regarding a potential role for folate in the prevention of colorectal cancer.
Lung, Esophagus, and Stomach
Chronic tobacco use is associated with diminished blood levels of folate, as well as diminished concentrations of the vitamin in the squamous mucosa of the oral cavity; this association is seen even when dietary folate intake of smokers is the same as that in nonsmoking controls.[40,41] Compared to nonsmokers, smokers also display a greater prevalence of chromosomal aberrations in oral mucosa cells.[41,42]
A single trial has examined the potential chemopreventive role of folate in lung cancer. Heimburger et al performed a prospective, placebo-controlled, intervention trial in 80 male smokers with squamous metaplasia. Both folic acid and hydroxycobalamin (vitamin B12) were concurrently administered to the experimental group. After 4 months, a significant improvement in the severity of metaplasia was observed in the experimental group but not in the controls.
Two studies performed in the tribal homelands of South Africa, which contains regions with particularly high rates of esophageal cancer, have begun to examine this issue in the esophagus.[43,44] In both studies, there appeared to be an association between diminished levels of serum folic acid and evidence of atypia and/or neoplasia on esophageal cytopathology; lower levels of vitamins A, E and B12 were also associated with dysplasia or cancer.
The sole intervention study in this organ to date was a randomized, placebo-controlled trial among 3,318 Chinese individuals with dysplastic esophageal cytopathology. The experimental group received a daily multivitamin and multimineral preparation that included 800 mcg of folic acid. Follow-up at 6 years indicated that supplementation caused a significant reversion of dysplastic cytology to normal; however, supplementation did not confer protection against esophageal cancer. It remains to be seen, therefore, whether the reversion of dysplastic cytology will translate into a significant drop in cancer rates with a longer follow-up.
A single case-control study in Italy has examined the relationship between dietary folate intake and stomach cancer. Although an inverse association was noted, this did not remain significant after correction for other dietary variables was made.
In summary, data are insufficient at this point to make any firm conclusions regarding folate and cancer in the lung, esophagus, or stomach.