OUP user menu

Dietary fiber and the risk of precancerous lesions and cancer of the esophagus: a systematic review and meta-analysis

Helen G Coleman, Liam J Murray, Blanaid Hicks, Shivaram K Bhat, Ai Kubo, Douglas A Corley, Chris R Cardwell, Marie M Cantwell
DOI: http://dx.doi.org/10.1111/nure.12032 474-482 First published online: 1 July 2013

Abstract

Dietary fiber has several anticarcinogenic effects and is thought to be protective against esophageal cancer. The aim of this systematic review was to quantify the association between dietary fiber and the risk of esophageal cancer by investigating histological subtypes of esophageal cancer and the stage at which fiber may influence the carcinogenic pathway. Systematic search strategies were used to identify relevant studies, and adjusted odds ratios (ORs) were combined using random-effects meta-analyses to assess the risk of cancer when comparing extreme categories of fiber intake. Ten relevant case-control studies were identified within the timeframe searched. Pooled estimates from eight studies of esophageal adenocarcinoma revealed a significant inverse association with the highest fiber intakes (OR 0.66; 95% confidence interval [CI] 0.44–0.98). Two studies also identified protective effects of dietary fiber against Barrett's esophagus. Similar, though nonsignificant, associations were observed when results from five studies of fiber intake and risk of squamous cell carcinoma were combined (OR 0.61; 95%CI 0.31–1.20). Dietary fiber is associated with protective effects against esophageal carcinogenesis, most notably esophageal adenocarcinoma. Potential methods of action include modification of gastroesophageal reflux and/or weight control.

  • Barrett's esophagus
  • dietary fiber
  • esophageal adenocarcinoma
  • esophageal squamous cell carcinoma

Introduction

Esophageal adenocarcinoma is a cancer of columnar-type glandular epithelium that occurs primarily in the lower third of the esophagus, just above the esophagogastric junction.1 It differs from the other major histological type of esophageal cancer, i.e., squamous cell carcinoma, which affects the native squamous cell epithelium in the esophagus.2 Although esophageal cancer is relatively uncommon, the incidence of esophageal adenocarcinoma has increased sharply in Europe3,4 and the United States,57 in contrast to the static or declining incidence of esophageal squamous cell cancer.

These conflicting epidemiological trends likely reflect different risk factors for the two histological subtypes of esophageal cancer and their precursor conditions. The vast majority, if not all, of esophageal adenocarcinomas arise from Barrett's esophagus; among patients with this condition, the risk of esophageal adenocarcinoma is 10- to 40-fold higher than it is among the general population.8,9 Risk factors that are well documented for esophageal adenocarcinoma include male sex, gastroesophageal reflux, tobacco smoking, increased body mass index, and abdominal obesity, but not alcohol consumption,1,1013 whereas both smoking and alcohol consumption are related to the etiology of esophageal squamous cell carcinoma.14,15 The timing of the obesity epidemic has been attributed to the rising incidence of esophageal adenocarcinoma and is synonymous with changes in dietary intake and physical activity levels.16,17

One such dietary change is the increased consumption of processed foods, which often have reduced dietary fiber content.18 Foods rich in dietary fiber are known to have several anticarcinogenic properties, such as the ability to lower levels of circulating markers of inflammation (e.g., interleukin 6),19 the ability to remove carcinogenic or damaged cells from the esophageal epithelium, and the ability to alter the glycemic response.20 As such, studies conducted to date have generally found an inverse association between fiber intake and the risk of esophageal cancer.21 However, it is unclear at which stage of the disease process fiber may act defensively against carcinogenesis or whether this protective association may differ by histological subtype, given the diverse etiologies of adenocarcinomas and squamous cell carcinomas.

The aim of this review was to quantify the relationship between dietary fiber and the two distinct histological subtypes of esophageal cancer along with a precursor lesion of esophageal adenocarcinoma, Barrett's esophagus.

Methods

Study selection

The electronic databases Ovid Medline® (US National Library of Medicine, Bethesda, Md., USA), Embase® (Reed Elsevier PLC, Amsterdam, Netherlands), and Web of Science® (Thomson Reuters, New York, N.Y., USA) were searched for relevant studies that included at least one keyword or Medical Subject Heading from each of the following: 1) dietary fibre/fiber, fibre/fiber, cellulose, cereal fibre/fiber, fruit fibre/fiber, vegetable fibre/fiber, soluble fibre/fiber, insoluble fibre/fiber; 2) (o)esophageal neoplasm(s), (o)esophageal cancer, (o)esophageal adenocarcinoma, (o)esophageal squamous cell carcinoma, Barrett('s) (o)esophagus, columnar lined epithelium, specialised intestinal metaplasia, (o)esophageal squamous dysplasia. The search strategy excluded reviews but did not impose any language restrictions. Observational studies published up to 20 January 2012 were considered for inclusion if they described a study population in which dietary fiber intake had been measured and information had been reported regarding incidence of esophageal cancer as adenocarcinoma, squamous cell carcinoma, esophageal squamous dysplasia, or Barrett's esophagus to give a statistical risk estimate and corresponding 95% confidence interval (CI) (or sufficient information to allow these to be calculated). Potentially relevant articles were screened independently by two of six reviewers, initially by abstract and then by full text where necessary to determine if selected articles met the inclusion criteria, with any discrepancies resolved by discussion. At least one reviewer in each pairing had previous experience with systematic reviews (HC, LM, MC). The reference lists of identified articles and recent e-publications on PubMed were also searched for any additional studies.

Data extraction

Two reviewers (HC, BH) extracted information on study design and location, sample size, number of cases of esophageal lesions or esophageal cancer, population demographics, exclusion criteria, method of dietary assessment, fiber intake, adjustments for confounders, and results from each publication. Where a study reported on multiple types of dietary fiber, data on intake and results were extracted separately for each type, based on a priori knowledge of the differing definitions of dietary fiber.

Statistical analysis

Meta-analyses were conducted separately for esophageal adenocarcinoma and squamous cell carcinoma by comparing cancer risk in the highest reported category of fiber intake with that in the lowest reported category. Most studies categorized intake using tertiles or quartiles. Adjusted odds ratios (ORs) and 95%CIs were combined and weighted to produce pooled ORs using a random-effects model, rather than a fixed-effects model, in order to account for anticipated heterogeneity between studies.22 The extent of heterogeneity in pooled estimates was investigated using the chi-squared test and the I2 statistic.23 Too few studies presented data in a suitable format to conduct linear dose-response meta-analyses or pooled estimates of risk of Barrett's esophagus. Relative risks were plotted against their corresponding standard errors to produce funnel plots, which were checked for asymmetry in order to investigate publication bias in each meta-analysis.24 Separate analyses were conducted for population-based and hospital-based case-control studies, and tests for interaction by study design were conducted.25 Further sensitivity analysis was conducted to evaluate excessive influence from studies by removing each individual study and assessing the resulting effect estimates and measures of heterogeneity. Stratified analysis had also been planned for fiber originating from cereal, fruit, or vegetable foods, but too few studies reported this information to allow meaningful risk estimates to be calculated. Statistical analysis was conducted using Intercooled Stata version 11.0 (StataCorp 2005, College Station, Texas, USA).

Results

Of 825 articles screened, 10 were eligible for inclusion in this review (Figure 1), some of which published results on more than one outcome.2635 Three studies were specifically excluded because they reported risk of esophageal cancer as one entity rather than for each histological subtype,3638 while a further two studies were excluded because 95%CIs could not be calculated.39,40 The characteristics of studies investigating dietary fiber and the risk of esophageal adenocarcinoma (n = 8) or its precursor condition, Barrett's esophagus (n = 2), are presented in Table 1. As shown in Table 1, fewer studies have explored dietary fiber in relation to the risk of esophageal squamous cell carcinoma (n = 5). No studies were identified that explored dietary fiber and the risk of esophageal squamous dysplasia.

Figure 1

Flowchart for selection of studies that assessed the relationship between dietary fiber and risk of esophageal lesions. Observational studies were considered for inclusion if they 1) described a study population in which dietary fiber intake had been measured, 2) reported information regarding incidence of esophageal cancer as either adenocarcinoma, squamous cell carcinoma, esophageal squamous dysplasia, or Barrett's esophagus, and 3) provided statistical risk estimates and corresponding 95%CIs (or sufficient information to allow these to be calculated).

View this table:
Table 1

Characteristics of studies assessing the relationship between dietary fiber and the risk of esophageal cancer and premalignant conditions

graphic

All studies originated from population-based or hospital-based case-control studies, and all utilized food frequency questionnaires as a dietary assessment tool (Table 1). All studies adjusted for age, sex (where applicable), and energy intake as confounders; most also accounted for smoking, alcohol intake, education, or another indicator of socioeconomic status. Adjustment for other confounders varied between studies, and only the most recent studies considered the effect of Helicobacter pylori infection, which reflects the more recent knowledge of this as a protective factor in the etiology of esophageal adenocarcinoma.

Esophageal adenocarcinoma

Figure 2 shows results for all studies comparing the risk of esophageal adenocarcinoma in subjects with the highest versus the lowest intakes of total dietary fiber. The pooled risk estimate demonstrated a significant inverse association (OR 0.66; 95%CI 0.44–0.98), but there was evidence of considerable heterogeneity between studies (I2 = 83%; P < 0.001). This heterogeneity was not explained by the contribution of any individual study. Even after removal of the study by Kabat et al.,33 which had markedly different results, the heterogeneity remained (I2 = 72%; P = 0.001), despite the strengthening of the protective association (OR 0.55; 95%CI 0.40–0.76) that was observed. Moreover, heterogeneity was not reduced in stratified analysis by study design, although the protective effect of fiber was much stronger in population-based case-control studies (OR 0.50; 95%CI 0.32–0.70; I2 = 74%; P = 0.004) than in hospital-based case-control studies only (OR 1.02; 95%CI 0.51–2.05; I2 = 86%; P = 0.001), and a formal interaction test was borderline significant (P for interaction = 0.08). There was no suggestion of publication bias in the overall analysis by either inspection of the funnel plot or by Egger's test (P = 0.51).

Figure 2

Meta-analysis of risk of esophageal cancer risk in the highest versus the lowest category of total dietary fiber intake.

*Test for heterogeneity: χ2 = 40.5, df = 7, P < 0.001; I2 = 83% (95%CI 67–91%).

**Test for heterogeneity: χ2 = 23.8, df = 3, P < 0.001; I2 = 87% (95%CI 70–95%).

Only two studies explored fiber from specific food groups and the risk of esophageal adenocarcinoma.27,30 As shown in Table 2, vegetable and cereal fiber was largely inversely associated with the risk of esophageal adenocarcinoma, although statistical significance was not always achieved. No consistent associations were observed between fruit fiber intake and the development of esophageal adenocarcinoma.

View this table:
Table 2

Summary of results of highest versus lowest category of fruit, vegetable, and cereal dietary fiber intakes and risk of esophageal cancer

ReferenceHistological subtypeRelative risk, highest vs. lowest category of fiber intake (95%CI)
Fruit fiber
Wu et al. (2007)27Adenocarcinoma0.45 (0.3–0.7)
Terry et al. (2001)30Adenocarcinoma1.70 (1.0–2.9)
Terry et al. (2001)30Squamous cell carcinoma0.80 (0.4–1.4)
Vegetable fiber
Wu et al. (2007)27Adenocarcinoma0.39 (0.2–0.6)
Terry et al. (2001)30Adenocarcinoma0.80 (0.5–1.5)
Terry et al. (2001)30Squamous cell carcinoma0.60 (0.3–1.2)
Cereal fiber
Wu et al. (2007)27Adenocarcinoma0.83 (0.5–1.3)
Terry et al. (2001)30Adenocarcinoma0.70 (0.4–1.2)
Terry et al. (2001)30Squamous cell carcinoma1.30 (0.7–2.2)

Barrett's esophagus

Two studies, one originating from a sample within the United States34 and another conducted in Ireland,26 investigated the role of dietary fiber in the development of Barrett's esophagus. Both studies identified a significant protective effect against Barrett's esophagus in the region of a 60–66% reduced risk for those with the highest versus those with the lowest total fiber intake. The US study specifically reported significant inverse associations for fruit and vegetable fiber, but not for fiber originating from grain or beans.34

Esophageal squamous cell carcinoma

As shown in Figure 2, the pooled estimate comparing the risk of squamous cell carcinoma of the esophagus in individuals with the highest versus the lowest intakes of fiber revealed a nonsignificant inverse association (OR 0.61; 95%CI 0.31–1.20). Again, significant heterogeneity was present (I2 = 85%; P < 0.001), which could not be attributed to individual studies and did not become attenuated in stratified analysis by study design. Results similar to those of the main analysis were observed in restricted analysis of the population-based study (OR 0.72; 95%CI 0.35–1.45; I2 = 69%; P = 0.04) and the hospital-based study (OR 0.51; 95%CI 0.11–2.29; I2 = 94%; P < 0.001). There was no indication of publication bias in the funnel plots of studies of total fiber intake and squamous cell carcinoma of the esophagus (Egger's test, P = 0.70).

Only one study investigated fiber sources by food groups30; intake of fruit and vegetable fiber was inversely associated with the risk of esophageal squamous cell carcinoma, though no associations were evident for intake of cereal fiber (Table 2).

Discussion

This systematic review suggests that dietary fiber is inversely associated with esophageal carcinogenesis; the evidence, however, is more consistent for adenocarcinoma than for squamous cell carcinoma of the esophagus. There was considerable heterogeneity between the studies in the meta-analyses. This was partially (though not fully) attributable to a stronger protective association in population-based studies than in hospital-based studies. The heterogeneity, however, likely influences only the magnitude of the association, not whether a protective association exists, given that seven of the eight adenocarcinoma studies had significant values or trends for inverse associations. No specific food source of dietary fiber could be attributed to the protective effects shown.

A highly significant inverse association between esophageal adenocarcinoma and dietary fiber intake was identified, suggesting that individuals with the highest dietary fiber intakes have an approximately 30% lower risk of cancer. The 2007 World Cancer Research Fund (WCRF)/American Institute for Cancer Research (AICR) global report on nutrition and cancer risk21 determined that only limited evidence was available showing foods containing dietary fiber may protect against esophageal cancer. That review, which included studies investigating the overall risk of esophageal cancer, was based on nine case-control studies (seven of which are included in the current review), one cohort study, and two ecological studies.21 However, the 2007 WCRF/AICR report did not produce pooled risk estimates for fiber intake by separate histological subtypes of esophageal cancer. The present review additionally incorporated evidence from studies of Barrett's esophagus and dietary fiber,26,34 which suggests that this protective effect is seen even at earlier stages in the carcinogenic pathway. A similar magnitude of association was identified in pooled estimates of dietary fiber and esophageal squamous cell carcinoma, but this was not statistically significant.

An inverse association between fiber intake and esophageal cancer is biologically plausible. Phytic acid, which is present in high-fiber foods, has been demonstrated to decrease cellular proliferation and promote apoptosis, resulting in the inhibition of esophageal adenocarcinoma cell growth in vitro,41 but has yet to be explored in the etiology of squamous cell carcinoma. Reduced concentrations of interleukin 6, a marker of systemic inflammation, have also been associated with high-fiber diets and may be involved in tumor development.19 Dietary fiber may also encourage the perfunctory removal of damaged cells and/or carcinogens from the esophageal lining.40 All of these mechanisms, however, are generic in their anticarcinogenic effects and therefore do not clearly explain the more consistent protective effect against adenocarcinomas, compared with squamous cell carcinomas, of the esophagus. Other protective effects of fiber that may be of specific etiological importance for esophageal adenocarcinoma include a reduced risk of gastroesophageal reflux symptoms4244 and a mediation of the glycemic response as a result of the reduced rate of gastric emptying and, hence, glucose absorption.20 Likewise, a diet high in fiber may aid weight control by promoting satiety, delaying gastric emptying, and postponing the onset of hunger.45 Both high-glycemic-index diets26 and high body fatness46 are directly associated with esophageal adenocarcinoma to a greater extent than with squamous cell carcinoma. It would be beneficial for future studies to investigate the association between fiber intake and the risk of esophageal lesions by measures of body fatness in order to address this hypothesis.

In the present analysis, an attempt was made to isolate specific food groups contributing to fiber intake that may be associated with the risk of esophageal neoplasm, but too few studies investigated food groups to allow sufficient conclusions to be drawn. High consumption of fruits and vegetables has been repeatedly linked with a reduced risk of esophageal squamous cell carcinoma, esophageal adenocarcinoma, and Barrett's esophagus,21,4749 but the protective component(s) of this heterogeneous food group has yet to be fully established. The content of antioxidant vitamins accounts for at least some of the improved defense against esophageal carcinogenesis in individuals with high fruit and vegetable intakes,50 but the fiber content of these foods may provide additional benefits. Alternatively, it is plausible that the observed association between risk of esophageal cancer and fiber intake is confounded by these coexisting nutrients in high-fiber foods. Determination of associations by specific sources of dietary fiber in future studies would help to disentangle these associations and allow formulation of targeted dietary guidelines that may aid cancer prevention.

The primary strength of this review lies in its ability to determine the risk of different histological subtypes of esophageal cancer on the basis of dietary fiber intake. This is important given the widely disparate etiologies of these conditions. Further investigations of dietary fiber and squamous cell carcinoma and dysplasia of the esophagus are particularly warranted, given the smaller number of studies included.

The heterogeneity observed in meta-analyses could not be accounted for by study design or individual studies. Specifically, the Kabat et al.33 study differed from other studies in that it had fewer items on the food frequency questionnaire and was unable to adjust for body mass index; however, since removal of the Kabat et al.33 study did not reduce the heterogeneity estimates observed in pooled analysis, the different results cannot be attributed to these differences alone, since it is also possible that the findings were due to chance. It is possible that the definition of dietary fiber – which is the subject of much debate, has changed considerably over time, and varies internationally51– may have been responsible for the heterogeneity seen. For example, the United Kingdom primarily classifies dietary fiber using the Englyst method, which measures nonstarch polysaccharides via a chemical method.52 Conversely, most other countries use dietary fiber values obtained via the Association of Official Analytical Chemists gravimetric method,5356 which additionally incorporates lignin and resistant starch. Despite recommendations that dietary fiber components always be quantified,57 the majority of studies included in this systematic review did not do so. Coupled with the approximate 20-year time span and the variety of countries represented in the studies included, it is likely, but can only be speculated, that some of the heterogeneity observed may be attributed to the varying components of dietary fiber described. None of the dietary assessment tools used were specifically developed for dietary fiber intake, and the food groupings included in the food frequency questionnaires may have misclassified dietary fiber components. A further limitation of the present review is that all collated evidence originated from case-control studies, which are known to be inferior to cohort studies in terms of nutritional assessment due to the potential for recall and response bias.58,59

Conclusion

To conclude, this systematic review found that the protective effects of dietary fiber appear to be more consistent for esophageal adenocarcinoma than for squamous cell carcinoma. Potential biological mechanisms for this association include reduction of gastroesophageal reflux, promotion of normal body weight, and modification of the glycemic response.

Acknowledgments

The authors would like to thank Mr Darren Ward and Ms Roisin O'Neill, Queen's University Belfast, who aided study selection in the initial stages of the review.

Funding

No funding support was received for this review.

Declaration of interest

The authors have no competing interests to declare.

References

View Abstract