Folate Intake and the Risk of Upper Gastrointestinal Cancers

A Systematic Review and Meta-analysis

Martin Tio; Juliana Andrici; Michael R Cox; Guy D Eslick

Disclosures

J Gastroenterol Hepatol. 2014;29(2):250-258. 

In This Article

Results

From 3477 citations screened by our search, we identified a total of 36 observational studies that met our inclusion criteria (Fig. 1). Table 1 shows selected characteristics of the identified studies.[18–53]Table 2 shows selected subgroup analyses. Two studies[29,46] were non-English papers, and were translated by an independent translator. Sixteen studies[19,21–23,29,30,32,35,38,39,41,42,45–47,52] did not report absolute levels of folate intake, and so meta-regression was not undertaken because of the low proportion of studies available to analyze.

Figure 1.

Study selection flowchart.

Esophageal Cancer

Ten case–control studies[18,22,23,26,31,32,41,46,51,53] were identified for the folate and esophageal cancer search. Nine studies[18,22,23,26,31,32,41,46,53] reported on dietary folate, while one study[51] reported on serum folate. Prospective studies, and studies that analyzed total folate intake, were not identified in the search. Four studies[18,22,46,53] reported results on any histological form of esophageal cancer, two studies[31,41] reported separate results for both esophageal squamous cell carcinoma and esophageal adenocarcinoma, three studies[26,32,51] reported esophageal squamous cell carcinoma results, and one study[23] reported esophageal adenocarcinoma results.

Figure 2 shows the meta-analysis of nine studies comprising of 2574 cases with a total of 9254 individuals on dietary folate and the risk of any histological type of esophageal cancer. High dietary folate intake was associated with a statistically significant decreased risk of any histological type of esophageal cancer, with an OR of 0.59 (95% CI 0.51–0.69). There was no significant heterogeneity (I2 = 21.1%, P = 0.24) or publication bias (Eggers = 0.09). Although there was no significant heterogeneity, one study by Jessri et al.[32] had a substantially different OR from the remainder of the studies. The meta-analysis with Jessri et al. removed was not significantly changed, with an OR of 0.60 (95% CI 0.52–0.69).

Figure 2.

Dietary folate intake and esophageal cancer.

When stratified by esophageal squamous cell carcinoma, the meta-analysis of four case–control studies[26,31,32,41] comprising of 829 cases with a total of 3977 individuals showed a statistically significant decreased risk with an OR of 0.63 (95% CI 0.44–0.89). There was no significant heterogeneity (I2 = 47.7%, P = 0.13). There was no significant publication bias (Eggers = 0.12).

When stratified by esophageal adenocarcinoma, three case–control studies[23,31,32,41] comprising of cases 1019 with a total of 3546 individuals showed a statistically decreased risk with an OR of 0.57 (95% CI 0.43–0.76). There was no significant heterogeneity (I2 = 44.9%, P = 0.16). There was no significant publication bias (Eggers = 0.85).

Gastric Cancer

Seventeen studies[18,20,23,25,30,34,35,37–43,45,52,53] (4 prospective studies and 13 retrospective studies) were identified in the folate and gastric cancer risk search. Sixteen studies reported dietary folate intake.[18,20,23,25,30,34,35,37–43,52,54] One study[35] reported total folate intake, and one study[52] reported plasma folate.

Figure 3 shows the meta-analysis of 16 studies[18,20,23,25,30,34,35,37–43,45,53] (3 prospective studies and 13 retrospective studies) comprising of 4414 cases with a total of 209 689 individuals on dietary folate and the risk of gastric cancer. Dietary folate intake had no statistically significant association with the risk of gastric cancer, with a pooled OR of 0.94 (95% CI 0.78–1.14). There was significant heterogeneity (I2 = 55.1%, P = 0.003). A sensitivity analysis did not find any single study that was significantly contributing to the heterogeneity. There was no significant publication bias (Eggers = 0.31).

Figure 3.

Dietary folate and the risk of gastric cancer.

When stratified by prospective studies, the meta-analysis of three studies[20,25,35] comprising of 776 cases with a total of 197 159 individuals showed no significant association, with an OR of 1.19 (95% CI 0.92–1.54). There was no significant heterogeneity (I2 = 0.0, P = 0.70). Although publication bias was present (Eggers = 0.01), the trim-and-fill method found no difference between actual and adjusted results, indicating that the publication bias had little effect on the actual results.

When stratified by retrospective studies, the meta-analysis of 13 studies[18,23,30,34,37–43,45,53] comprising of 3638 cases with a total of 12 530 individuals showed no significant association, with an OR of 0.87 (95% CI 0.70–1.09). There was significant heterogeneity (I2 = 56.2%, P = 0.01). A sensitivity analysis did not find any single study that was significantly contributing to the heterogeneity. There was no significant publication bias (Eggers = 0.43).

Pancreatic Cancer

Twelve studies[19,21,24,28,29,33,36,44,47–50] were identified in the folate and pancreatic cancer search. Eight studies[19,21,28,33,36,44,48,50] (five prospective studies and three retrospective studies) were found that reported on dietary folate. Four studies[28,36,44,48] (three prospective studies and one retrospective study) were found that reported on total folate. Four studies[24,29,47,49] (three prospective studies and one retrospective study) were found that reported on plasma folate.

Figure 4 shows the meta-analysis of eight studies comprising of 2209 cases with a total of 295 526 individuals on dietary folate and the risk of pancreatic cancer. High dietary folate intake was associated with a statistically significant decreased risk of pancreatic cancer, with an OR of 0.66 (95% CI 0.49–0.89). There was significant heterogeneity (I2 = 70.6%, P = 0.001). There was no significant publication bias (Eggers = 0.19). A sensitivity analysis found that Keszei et al.[33] was significantly contributing to the heterogeneity. The meta-analysis result with Keszei et al. removed was not substantially changed, with an OR of 0.60 (95% 0.47–0.77), and with no significant heterogeneity (I2 = 46.9%, P = 0.08).

Figure 4.

Dietary folate intake and the risk of pancreatic cancer.

When stratified by prospective studies, the meta-analysis of five studies[33,36,44,48,50] comprising of 1247 cases with a total of 291 958 individuals showed no significant association, with an OR of 0.65 (95% CI 0.38–1.11). There was significant heterogeneity (I2 = 80.1%, P < 0.0001). There was no significant publication bias (Eggers = 0.08). A sensitivity analysis showed that Keszei et al. was significantly contributing to the heterogeneity. The meta-analysis with Keszei et al. removed showed a statistically significant decreased risk of pancreatic cancer, with an OR of 0.54 (95% CI 0.27–0.78), with no significant heterogeneity (I2 = 36.1%, P = 0.20).

When stratified by retrospective studies, the meta-analysis of three studies[19,21,28] comprising of 962 cases with a total of 3568 individuals showed a statistically significant decreased risk, with an OR of 0.67 (95% CI 0.49–0.91). There was no significant heterogeneity (I2 = 42.0%, P = 0.18). There was significant publication bias (Eggers = 0.01). The trim-and-fill method found that the adjusted results became non-significant, indicating that publication bias had a substantial effect on the actual results.

The meta-analysis of four studies[28,36,44,48] comprising of 1259 cases with a total of 261 727 individuals showed no association between high total folate intake and pancreatic cancer, with an OR of 0.69 (95% CI 0.47–1.03). There was significant heterogeneity (I2 = 62.7%, P = 0.04). There was no publication bias (Eggers = 0.32). The removal of the single retrospective study[28] to stratify by prospective studies only did not substantially change the result, with an OR of 0.67 (95% CI 0.35–1.27).

The meta-analysis of four studies[24,29,47,49] comprising of 839 cases with a total of 2215 individuals showed no association between plasma folate levels and pancreatic cancer, with an OR of 0.73 (95% CI 0.47–1.13). There was significant heterogeneity (I2 = 67.1%, P = 0.03). There was no publication bias (Eggers = 0.87). The removal of the single retrospective study to stratify by prospective studies did not substantially change the result, with an OR of 0.80 (95% CI 0.44–1.45).

processing....