Abstract
Although it seems biologically plausible for iron to be associated with gastric carcinogenesis, the evidence is rather heterogeneous and insufficient to lead to any conclusions. To further investigate the relationship between body iron status and gastric cancer risk, we conducted a nested case-control study in the multi-centric European Prospective Investigation into Cancer and Nutrition (EPIC) study. The study included 456 primary incident gastric adenocarcinoma cases and 900 matched controls that occurred during an average of 11 years of follow-up. We measured pre-diagnostic serum iron, ferritin, transferrin, and C-reactive protein, and further estimated total iron-binding capacity (TIBC) and transferrin saturation (TS). Odds ratios (OR) and 95% confidence intervals (CI) for the risk of gastric cancer by iron metrics were estimated from multivariate conditional logistic regression models. After adjusting for relevant confounders, we observed a statistically significant inverse association between gastric cancer and ferritin and TS indices (ORlog2=0.80, 95% CI=0.72-0.88; and OR10%increment=0.87, 95% CI=0.78-0.97, respectively). No statistical differences were found by gastric cancer localization (cardia and non-cardia) or histological (diffuse or intestinal) type. TIBC increased risk of overall gastric cancer (OR50µg/dl=1.13, 95% CI=1.02-1.2) and also with non-cardia gastric cancer and intestinal type. Additional analysis suggests that time since diagnosis of gastric cancer and pepsinogen levels could modify these findings. In conclusion our results showed a decreased risk of gastric cancer related to body iron status, measured by serum iron and ferritin. Further investigation is needed to clarify the role of iron in gastric carcinogenesis.
Keywords: iron homeostasis, gastric cancer, nested case-control study
Introduction
Almost one million new cases of gastric cancer were estimated to occur in 2012, making it the fifth most common malignancy in the world and the third leading cause of cancer death worldwide(1). Infection with Helicobacter pylori (Hp) is the strongest established risk factor for gastric non-cardia adenocarcinoma, but only a minority of infected individuals develop cancer of the stomach (1). Tobacco smoking (2) is another recognized risk factor for gastric cancer Although the evidence relating the risk of gastric cancer to consumption of red meat is considered to be limited by the World Cancer Research Fund & the American Institute for Cancer Research (3),newer research suggests that red meat intake could be also a risk factor for gastric cancer (4). One of the hypotheses behind meat intake and increased risk of gastric cancer relates to dietary iron, especially the heme type (5). Heme iron could increase endogenous formation of N-nitroso compounds, which are well-known carcinogens (6). Furthermore, redox cycling of iron is closely related to the production of oxygen reactive species (ROS) such as hydroxyl radicals, a highly reactive molecule able to induce lipid peroxidation and oxidative damage to DNA (7). If iron plays a role in carcinogenesis, subjects that have elevated total body iron stores or iron overload could be at a greater risk of developing cancer. Hereditary hemochromatosis is the most severe clinical expression of iron overload and hemochromatosis gene mutations has been associated with gastric cancer in the EPIC study (8).
Iron, an essential element for human life but also toxic when in excess (7,9), has a very well-regulated metabolism (10). Only a small percent of dietary iron is absorbed, and dietary iron is not necessarily related to body iron status (11). The latter is frequently assessed by measuring unbound (free) iron, ferritin, total iron binding capacity (TIBC) and transferrin saturation (TS). Iron is mainly bound to hemoglobin and about 30% of iron is stored in the form of ferritin. Most circulating iron is bound to transferrin and TIBC indirectly measures the extent to which transferrin is saturated. Transferrin saturation is a direct measure of the transferrin binding sites that are occupied (12). Ferritin, representing the levels of iron in liver, spleen and bone marrow, is decreased in iron deficiency and increased in iron overload, liver disease, infection, inflammation and other disorders (13).
A recent systematic review and meta-analysis gathered information on iron (intake and status) and different types of cancer risk (14). Although it seems biologically plausible for iron to be associated with gastric carcinogenesis, the evidence is rather heterogeneous and insufficient to lead to any conclusions. Only one of the studies included in this review studied body iron status and gastric cancer risk (15). It measured iron biomarkers in 341 male Finnish cases and 341 matched controls, and found an inverse (but not statistically significant) relationship between iron metrics (ferritin, iron, TS and TIBC) and non-cardia gastric cancer (NCGC) risk. Two cohorts studying Japanese subjects, one in Japan (16) and the other in Hawaii (17) had previously reported a significant decrease in risk of cancer of the stomach associated with lower levels of serum ferritin. Regarding transferrin, a significant decreased risk was observed in men (but not in women) in a Finnish cohort (18) and a decreased but no significant risk was found in the two Japanese cohorts (16, 17). While a significant increase in the risk was found in an American cohort only among women (12).
To further investigate the relationship between body iron status and gastric carcinogenesis, we conducted a nested case-control study in the multi-centric European Prospective Investigation into Cancer and Nutrition (EPIC) study as measured by markers of body iron status.
Material and Methods
Study subjects
The subjects in this study were participants from the European Prospective Investigation into Cancer and Nutrition (EPIC) study, and were selected according to a nested case-control design. The methodology and rationale behind the EPIC study has been described elsewhere (19). In summary, EPIC is a multicentre prospective cohort of over 500.000 participants, recruited between 1992 and 2000 in 23 centres from 10 European countries (Denmark, France, Germany, Greece, Italy, the Netherlands, Norway, Spain, Sweden and the United Kingdom). Participants provided information on diet and lifestyle factors, and anthropometric data and blood samples were collected at recruitment. This study was approved by the Ethical Committees at the International Agency for Research on Cancer (IARC) and in each of the EPIC centres.
Case identification and control selection
Case identification was based on record linkage with population cancer registries except in France, Germany, Greece and Naples, where the centres have used a combination of methods such as health insurance records, cancer and pathology hospital registries, and active follow up with study subjects. Cancer of the stomach included cancers coded as C16 according to the 10th Revision of the International Classification of Diseases (ICD-10). Cases for our study were subjects with no previous cancer, newly diagnosed with primary gastric adenocarcinoma during the follow-up, from recruitment through 2010 depending on the study centre. A total of 665 gastric adenocarcinoma cases were diagnosed during this follow-up period. We excluded 113 cases from Denmark since the biological samples were not available at the time of the study. Furthermore, we excluded 6 cases from Norway because of the given small number of cases and 75 cases from Malmö (Sweden) since the centre decided not to participate in this study. In total, 471 incident primary gastric adenocarcinoma cases were included. They were also classified according to both anatomic location (cardia and noncardia) and Lauren histological type (intestinal and diffuse).
For each case, two controls were randomly selected among cohort subjects alive and cancer free at the time of diagnosis of the case, matched by centre, sex, age at baseline (±2.5 years) and date of blood collection (±45 days). According to these criteria, 942 controls were selected.
From the 471 cases and 942 controls there were available biological samples for 460 cases and 905 controls. We then proceeded to exclude 9 subjects, including 4 cases and 5 controls, because of non-valid biomarker measurements and incomplete risk sets. We finally had 456 cases and 900 controls for the analyses.
Laboratory procedures and measurements of biomarkers of iron status
Serum ferritin levels were measured by electrochemiluminescence immunoassay (ECLIA) by an Elecsys analyzer (Roche Diagnostics, Mannheim, Germany). Serum iron was measured by immune (chemiluminescence) assay and serum transferrin by immunoturbidimetric assay using a Modular Analytics P800 chemistry analyzer (Roche Diagnostics, Mannheim, Germany). All these analyses were carried out at the “Laboratory de Referència Sud de Catalunya”(Tarragona, Spain).
Complementary biomarkers
Serum vitamin C was measured owing to its role as an enhancer of iron absorption (especially the inorganic iron fraction). Serum vitamin C was measured using liquid chromatography coupled to mass spectrometry. Vitamin C was extracted from the serum by liquid-liquid extraction using an acidified organic solvent mixture, and quantified in a 1290 UHPLC Series Liquid Chromatograph coupled to a 6490 QqQ-MS/MS (Agilent Technologies). The technique uses 50 µl of serum and was performed in the Center for Omic Sciences (COS), Universitat Rovira i Virgili (URV), Reus (Tarragona, Spain).
Since ferritin can be increased in inflammatory conditions or infections, high sensitive C-reactive protein (hsCRP) was measured in order to rule out inflammation as a cause of increased levels of ferritin by immunoturbidimetric assay on an Modular Analytics P800 chemistry analyzer (Roche Diagnostics, Mannheim, Germany). hsCRP was measured simultaneously with the biomarkers of iron status, using the same aliquot.
Other factors
Anthropometric measurements, smoking status, physical activity and education level, were collected in the cohort baseline questionnaire; measured height and weight were then used to compute the body mass index (BMI, kg/m2). Data on habitual dietary intake such as meat and vitamin C were collected from validated country-specific dietary questionnaires in the cohort. Heme iron intake was estimated multiplying the estimated heme iron content by the mean daily intake of related food sources (meat and fish) for each subject (5). Pepsinogen 1 level was determined by ELISA using the kits from Biohit (Helsinki, Finland). We considered pepsinogen 1 levels <22 μg/l as an indicator of severe chronic atrophic gastritis (SCAG). This information was available for 123 cases and 240 controls. Moreover, we assessed Hp serology by Immunoblot using the HELICOBLOT 2.1 kit (Genelab Diagnostics, Singapore). Detailed information can be found elsewhere (21).
Statistical Analysis
Indicators of body iron status tested as potentially related to gastric cancer risk were ferritin, serum iron, TIBC and TS. Since most of these variables were not normally distributed, we used the median and interquartile range to describe them. Dietary variables (daily intake of meat, total iron and heme iron) were expressed as nutrient density per 2000 kcal (e.g., [g/kcal] x 2000). The distributions of baseline characteristics in cases and controls were compared through Chi-square test for categorical variables, Student’s t-test (normally distributed) and Wilcoxon rank sum tests (non-normally distributed) for continuous variables. Correlations between variables were calculated using Pearson’s test for normally distributed variables and Spearman test if not normally distributed.
Associations between each iron biomarker and gastric cancer risk were assessed by the odds ratio (OR) and 95% confident intervals (CI) estimated by conditional logistic regression. Main exposure variables were used as both continuous and categorical in the models. Since ferritin showed a non-normal distribution, we proceeded to a log2 transformation beforehand when analysed it as continuous variable; the OR is interpreted as the increase in risk of gastric cancer related to a doubling of the ferritin concentration. Quartile cut-off points for iron biomarker variables were calculated using the distribution among controls. P for trend was calculated by treating categorical variables as continuous in regression analyses. Parsimonious models were built by testing each covariate using the likelihood ratio test after elimination of each variable. The final model included educational level (no formal education, primary school, technical or professional training, university and not specified), cigarette smoking (never smoker, former smoker, current smoker and not specified), and heme iron intake. We tested for other potential confounders including hsCRP because of the role of inflammation on iron biomarkers levels, but as they did not change the estimated effect >10%, we did not include them in the final model. In the subset of subjects with measured values of plasma pepsinogen 1, we further assessed the relationship between gastric cancer risk and body iron status biomarkers including pepsinogen 1 in the multivariate model.
We conducted subgroup analysis according to tumour localization: gastric cardia cancer (GCC) vs. NCGC and histology (intestinal vs. diffuse) using conditional logistic regression, and applying the Wald test to assess heterogeneity of the OR. Furthermore, we assessed the effect of biomarkers of iron status on gastric cancer risk according to the time elapsed from the date of blood drawing to date of incidence. We also explored the interaction or modifying effect of Hp infection, plasma levels of vitamin C, alcohol, tobacco smoking, sex, and BMI by means of stratified analysis, using unconditional regression models including the matching variables, and testing these interactions with the likelihood ratio (LR) test.
Results
Of the 456 gastric cancer cases, 116 were classified as GCC, 236 as NCGC and for 104 cases the site was unknown. Histological subtype was similar between the proportion of intestinal and diffuse tumours, with 154 classified as diffuse, 149 intestinal and 153 cases were unclassified. Table 1 describes the baseline characteristics of our study subjects. Gastric cancer overall had a significantly higher waist-to-hip ratio, proportion of current smokers, percentage of Hp positivity, but had a significantly lower proportion of more educated subjects. Cases also had a greater intake of heme iron and processed meat. Considering markers of iron body status, cases had significantly lower levels of ferritin and transferrin saturation. When looking into descriptive analysis by subgroups, all of these differences concerning iron indices among cases and controls were only found in the NCGC type, and additionally showed significantly higher levels of TIBC among cases.
Table 1. Baseline characteristics of cases and matched controls of EURGAST study.
| Characteristics | All gastric cancer | Gastric cardia | Gastric noncardia | Diffuse | Intestinal | ||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Controls (n = 900) | Cases (n = 456) | pa | Cases (n = 116) | pa | Cases (n = 236) | pa | Cases (n = 154) | pa | Cases (n = 149) | pa | |
| BMI (kg/m2), mean (SD) | 26.76 (3.89) | 27.04 (4.01) | 0.212 | 26.69 (3.71) | 0.65 | 27.01 (3.90) | 0.5351 | 27.04 (4.35) | 0.3611 | 27.63 (3.59) | 0.6462 |
| Waist-to-hip ratio, mean (SD) | 0.88 (0.09) | 0.90 (0.09) | <0.05 | 0.92 (0.09) | 0.11 | 0.89 (0.08) | 0.1237 | 0.89 (0.09) | 0.0437 | 0.91 (0.09) | 0.4789 |
| Smoking status, n (%) | |||||||||||
| Never | 418 (46.44) | 177 (38.82) | 31 (26.72) | 104 (44.07) | 68 (44.16) | 53 (35.57) | |||||
| Former | 296 (32.89) | 134 (29.39) | 41 (35.34) | 57 (24.15) | 33 (21.43) | 49 (32.89) | |||||
| Current | 177 (19.67) | 136 (29.82) | 40 (34.48) | 73 (30.93) | 50 (32.47) | 45 (30.20) | |||||
| Unknown | 9 (1.00) | 9 (1.97) | <0.001 | 4 (3.45) | <0.001 | 2 (0.85) | 0.004 | 3 (1.95) | 0.006 | 2 (1.34) | 0.08 |
| Physical activity, n (%) | |||||||||||
| Inactive | 273 (30.33) | 160 (35.09) | 37 (31.90) | 88 (37.29) | 51 (33.12) | 56 (37.58) | |||||
| Moderately inactive | 282 (31.33) | 139 (30.48) | 36 (31.03) | 72 (30.51) | 51 (33.12) | 42 (28.19) | |||||
| Moderately active | 187 (20.78) | 80 (17.54) | 21 (18.10) | 45 (19.07) | 29 (18.83) | 32 (21.48) | |||||
| Active | 149 (16.56) | 72 (15.79) | 21 (18.10) | 30 (12.71) | 21 (13.64) | 19 (12.75) | 0.5 | ||||
| Missing | 9 (1.00) | 5 (1.10) | 0.27 | 1 (0.86) | 0.7 | 1 (0.42) | 0.422 | 2 (1.30) | 0.353 | ||
| Education level, n (%) | |||||||||||
| None | 74 (8.22) | 38 (8.33) | 4 (3.45) | 25 (10.59) | 11 (7.14) | 19 (12.75) | |||||
| Primary school completed | 344 (38.22) | 198 (43.42) | 42 (36.21) | 117 (49.58) | 70 (45.45) | 70 (46.98) | |||||
| Technical/professional school | 189 (21.00) | 104 (22.81) | 34 (29.31) | 44 (18.64) | 36 (23.38) | 28 (18.79) | |||||
| Secondary school | 103 (11.44) | 43 (9.43) | 14 (12.07) | 15 (6.36) | 13 (8.44) | 14 (9.40) | |||||
| Longer education | 164 (18.22) | 54 (11.84) | 14 (12.07) | 31 (13.14) | 20 (12.99) | 14 (9.40) | |||||
| Not specified | 26 (2.89) | 19 (4.17) | <0.05 | 8 (6.90) | 0.65 | 4 (1.69) | 0.07 | 4 (2.60) | 0.102 | 4 (2.68) | 0.1 |
| Helicobacter pylori, n (%) | |||||||||||
| Positive | 366 (65.59) | 254 (86.10) | 50 (66.67) | 153 (95.03) | 100 (90.91) | 96 (85.71) | |||||
| Negative | 192 (34.41) | 41 (13.90) | <0.001 | 25 (33.33) | 0.15 | 8 (4.97) | <0.001 | 10 (9.09) | <0.001 | 16 (14.29) | 0.0 |
| Gastric atrophy, n (%) | |||||||||||
| Positive | 9 (3.75) | 10 (8.13) | 1 (4.00) | 8 (11.59) | 4 (9.09) | 4 (8.51) | |||||
| Negative | 231 (96.25) | 113 (91.87) | 0.08 | 24 (96.00) | 0.15 | 61 (88.41) | 0.056 | 40 (90.91) | 0.186 | 43 (91.49) | 0.2 |
| Dietary intake (median, IQR) | |||||||||||
| Total energy (kcal/day) | 2052.16 (1625.68–2504.70) | 2022.27 (1666.15–2563.47) | 0.8 | 2006.07 (1619.08–2508.66) | 0.92 | 2034.56 (1679.79–2586.93) | 0.71 | 2109.52 (1677.13–2619.21) | 0.516 | 1999.46 (1587.33–2552.21) | 0.63 |
| Total iron (mg/2,000 kcal) | 12.67 (11.01–14.42) | 12.62 (10.93–14.42) | 0.68 | 12.11 (10.53–14.02) | 0.7 | 12.80 (10.94–14.68) | 0.61 | 12.84 (10.95–14.52) | 0.72 | 13.43 (11.52–15.04) | 0.42 |
| Heme iron (mg/2,000 kcal) | 1.11 (0.75–1.58) | 1.20 (0.79–1.71) | <0.05 | 1.09 (0.70–1.54) | 0.18 | 1.28 (0.84–1.80) | 0.0068 | 1.23 (0.82–1.71) | 0.06 | 1.47 (0.92–180) | 0.01 |
| Red meat (g/2,000 kcal) | 33.94 (17.58–54.17) | 36.54 (18.80–56.83) | 0.13 | 38.36 (18.09–63.52) | 0.11 | 34.96 (20.50–54.38) | 0.304 | 35.76 (20.79–53.73) | 0.21 | 38.70 (23.39–60.34) | 0.17 |
| Processed meat (g/2,000 kcal) | 23.70 (10.16–42.31) | 26.79 (12.37–47.79) | <0.05 | 25.66 (12.73–46.64) | 0.28 | 27.76 (12.41–47.94) | 0.04 | 31.21 (13.79–49.31) | 0.009 | 25.75 (13.64–49.86) | 0.03 |
| Vitamin C (mg/2,000 kcal) | 113.09 (80.45–158.77) | 106.86 (75.05–157.82) | 0.4 | 100.13 (74.32–143.15) | 0.32 | 106.86 (67.99–162.07) | 0.63 | 102.97 (74.45–157.81) | 0.99 | 106.57 (72.44–163.51) | 0.83 |
| Serum markers (median, IQR) | |||||||||||
| Ferritin (ng/ml) ♂♀ | 104.45 (57.32–193.8) | 86.2 (44.82–161.5) | <0.001 | 98.92 (57.66–189.8) | 0.31 | 82.93 (35.01–140.7) | <0.001 | 83.06 (41.28–143.5) | 0.13 | 85.78 (37.69–179.1) | 0.46 |
| Ferritin (ng/ml) ♂ | 138.15 (76.71–251.5) | 110.5 (64.52)–219.2) | <0.05 | 132.8 (69.8–217.9) | 0.37 | 98.67 (48.42–192.4) | 0.003 | 109.4 (67.69–251.5) | 0.68 | 129.4 (57.66–269.5) | 0.24 |
| Ferritin (ng/ml) ♀ | 74.88 (40.75–131) | 60.69 (29.46–103.9) | <0.05 | 62.82 (33.29–122.6) | 0.44 | 61.85 (29.46–109.8) | 0.3 | 62.16 (30.34–101.7) | 0.08 | 53.72 (26.24–97.75) | 0.12 |
| Iron (µg/dl) | 111 (89–140) | 107 (84–133) | 0.13 | 114 (90–139) | 0.9 | 104 (83.5–130) | 0.01 | 105 (85–132) | 0.33 | 105.5 (81.5–136.5) | 0.70 |
| Transferrin saturation (%) | 29.68 (23.34–37.77) | 28.34 (22.20–35.73) | <0.05 | 30.92 (24.87) | 0.8 | 27.94 (21.80–33.70) | 0.02 | 28.31 (22.45–33.84) | 0.29 | 27.88 (20.90–36.6) | 0.32 |
| TIBC (µg/dl) | 371.8 (341.77–410.41) | 383.24 (343.2–423.28) | 0.09 | 376.09 (343.2–411.12) | 0.44 | 384.67 (344.63–427.57) | 0.04 | 379.66 (346.06–418.99) | 0.77 | 387.53 (338.91–430.43) | 0.06 |
| CRP (mg/dl) | 0.2 (0.2–0.3) | 0.2 (0.2–0.3) | 0.16 | 0.2 (0.2–0.4) | 0.03 | 0.2 (0.2–0.3) | 0.97 | 0.2 (0.2–0.3) | 0.75 | 0.2 (0.2–0.4) | 0.97 |
| Serum vitamin C (µg/dl) | 7.25 (4.80–9.57) | 6.80 (4.21–9.10) | 0.08 | 6.39 (3.57–8.63) | 0.07 | 7.02 (4.61–9.40) | 0.26 | 7.24 (4.73–9.41) | 0.35 | 6.59 (4.56–8.88) | 0.97 |
p values were calculated excluding subjects with missing values and by comparing cases with their own controls; data not shown for variables on which cases and controls were matched.
Table 2 shows the correlations between iron biomarkers, dietary iron, and dietary sources of iron. Ferritin was correlated with serum iron (r=0.22, p<0.001) and TS (r=0.36, p<0.001), and inversely correlated with TIBC (r= -0.42, p<0.001). As expected, TS was inversely correlated with TIBC (r= -0.43, p<0.001). Dietary variables, such as red meat, total iron, heme iron and vitamin C did not show correlations with iron indices except for processed meat which showed a positive correlation with ferritin (r=0.22, p<0.001).
Table 2. Correlation matrix for biomarkers of iron status and dietary variables among control subjects.
| Ferritin (ng/ml) | Iron (µg/dl) | Transferrin saturation (%) | TIBC (µg/dl) | CRP (mg/dl) | Serum vitamin C (µg/dl) | Total iron (mg) | Heme iron (mg) | |
|---|---|---|---|---|---|---|---|---|
| Ferritin (ng/ml) | 1 | |||||||
| Iron (µg/dl) | 0.22 (<0.001) | 1 | ||||||
| Transferrin saturation (%) | 0.36 (<0.001) | –0.05 (0.16) | 1 | |||||
| TIBC (µg/dl) | –0.42 (<0.001) | –0.05 (0.16) | –0.43 (<0.001) | 1 | ||||
| CRP (mg/dl) | 0.08 (0.01) | –0.11 (0.001) | –0.13 (<0.001) | 0.02 (0.48) | 1 | |||
| Serum vitamin C (µg/dl) | 0.0059 (0.86) | 0.05 (0.10) | 0.03 (0.38) | 0.07 (0.04) | –0.10 (0.003) | 1 | ||
| Total iron (mg/day) | 0.04 (0.16) | 0.05 (0.11) | 0.04 (0.20) | 0.04 (0.26) | –0.03 (0.34) | 0.003 (0.92) | 1 | |
| Heme iron (mg/day) | 0.1151 (<0.001) | 0.02 (0.56) | 0.04 (0.19) | –0.03 (0.38) | –0.002 (0.94) | –0.11 (0.001) | 0.62 (<0.001) | 1 |
| Red meat (mg/day) | 0.1097 (0.001) | 0.06 (0.05) | 0.05 (0.11) | 0.005 (0.87) | 0.06 (0.07) | –0.0002 (0.99) | 0.42 (<0.001) | 0.63 (<0.001) |
| Processed meat (mg/day) | 0.22 (<0.001) | –0.02 (0.60) | 0.02 (0.59) | –0.09 (0.005) | 0.02 (0.44) | –0.05 (0.16) | 0.26 (<0.001) | 0.44 (<0.001) |
| Vitamin C (mg/day) | –0.11 (0.001) | 0.004 (0.90) | –0.02 (0.63) | 0.10 (0.004) | 0.02 (0.50) | 0.21 (<0.001) | 0.44 (<0.001) | 0.19 (<0.001) |
Non-normally distributed data—Spearman correlation; normally distributed data—Pearson correlation.
Table 3 presents the results of the conditional logistic regression models for iron indices and all gastric cancer as well as gastric cancer subgroups for site and histology respectively. Comparing the highest to the lowest quartiles of iron biomarkers we found statistically inverse associations between overall risk of gastric cancer and levels of ferritin, serum iron and TS indices (OR=0.38, 95% CI=0.25-0.57; OR=0.62, 95% CI=0.44-0.88; and OR=0.60, 95% CI=0.43-0.86, respectively). Using ferritin as a continuous variable (in the log2 scale), the observed OR was 0.80 (95% CI=0.72-0.88), indicating a 20% decrease in risk of gastric cancer for each doubling of the concentration of ferritin (ng/mL). An increment of 10% of TS led to a reduction of 13% in all gastric cancer risk and 22% in NCGC. Comparing those in the highest to those in the lowest quartile, TIBC was positively associated with all gastric cancer (OR=1.65, 95% CI=1.16-2.36) and with NCGC (OR=2.01, 95% CI=1.21-3.32 Furthermore, a 50µg/dl increment of TIBC yielded an OR of 1.22 (95% CI=1.06-1.41) in NCGC and 1.29 (95% CI=1.06-1.56) for the histologic subtype of intestinal tumours. There were no associations between iron indices and GCC.
Table 3. ORs and 95% CI for the association between gastric cancer and iron metrics in the EURGAST study.
| All gastric cancer | Gastric cardiaa | Gastric noncardiaa | Diffuseb | Intestinalb | |||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Controls | Cases | ORc | 95% CI | ptrend | Cases | OR | 95% CI | ptrend | Cases | OR | 95% CI | ptrend | Cases | OR | 95% CI | ptrend | Cases | OR | 95% CI | ptrend | |
| Ferritin (ng/ml) | |||||||||||||||||||||
| ≤57.09 | 225 | 147 | Referent | 32 | Referent | 84 | Referent | 45 | Referent | 48 | Referent | ||||||||||
| 57.09–104.3 | 225 | 120 | 0.70 | 0.50–0.97 | 26 | 0.66 | 0.32–1.36 | 59 | 0.57 | 0.36–0.91 | 45 | 0.93 | 0.50–1.72 | 35 | 0.70 | 0.39–1.28 | |||||
| 104.4–193.7 | 225 | 104 | 0.54 | 0.38–0.77 | 36 | 0.91 | 0.45–1.86 | 58 | 0.51 | 0.31–0.85 | 34 | 0.57 | 0.30–1.08 | 35 | 0.62 | 0.32–1.20 | |||||
| ≥193.8 | 225 | 84 | 0.38 | 0.25–0.57 | <0.001 | 21 | 0.38 | 0.15–0.95 | 0.366 | 35 | 0.26 | 0.14–0.47 | <0.001 | 30 | 0.49 | 0.25–0.99 | 0.062 | 30 | 0.45 | 0.20–1.00 | 0.039 |
| Log2ferritin | 900 | 455 | 0.80 | 0.72–0.88 | 115 | 0.88 | 0.73–1.07 | 236 | 0.73 | 0.64–0.84 | 154 | 0.87 | 0.74–1.03 | 148 | 0.80 | 0.67–0.96 | |||||
| Iron (µg/dl) | |||||||||||||||||||||
| ≤89 | 233 | 145 | Referent | 33 | Referent | 76 | Referent | 45 | Referent | 47 | Referent | ||||||||||
| 89–111 | 219 | 99 | 0.69 | 0.50–0.97 | 21 | 0.74 | 0.35–1.56 | 61 | 0.74 | 0.47–1.16 | 41 | 0.77 | 0.43–1.41 | 32 | 0.77 | 0.43–1.37 | |||||
| 112–140 | 227 | 119 | 0.80 | 0.58–1.09 | 33 | 1.09 | 0.57–2.11 | 54 | 0.61 | 0.38–0.97 | 41 | 0.81 | 0.46–1.42 | 34 | 0.76 | 0.43–1.36 | |||||
| ≥141 | 220 | 91 | 0.62 | 0.44–0.88 | 0.034 | 28 | 1.00 | 0.49–2.02 | 0.973 | 45 | 0.54 | 0.33–0.89 | 0.029 | 27 | 0.52 | 0.26–1.02 | 0.153 | 35 | 0.88 | 0.49–1.57 | 0.435 |
| Per 10 µg/dl | 899 | 454 | 0.97 | 0.94–1.00 | 115 | 1.01 | 0.95–1.07 | 236 | 0.95 | 0.91–1.00 | 154 | 0.96 | 0.91–1.02 | 148 | 1.00 | 0.95–1.06 | |||||
| TS (%) | |||||||||||||||||||||
| ≤23.35 | 225 | 136 | Referent | 28 | Referent | 73 | Referent | 42 | Referent | 49 | Referent | ||||||||||
| 23.35–29.69 | 225 | 117 | 0.78 | 0.56–1.08 | 25 | 1.08 | 0.51–2.29 | 70 | 0.82 | 0.53–1.27 | 49 | 0.84 | 0.45–1.54 | 34 | 0.75 | 0.43–1.31 | |||||
| 29.69–37.78 | 225 | 105 | 0.75 | 0.53–1.05 | 31 | 1.44 | 0.67–3.08 | 48 | 0.56 | 0.35–0.92 | 37 | 0.80 | 0.42–1.50 | 30 | 0.72 | 0.39–1.33 | |||||
| ≥37.78 | 224 | 95 | 0.60 | 0.43–0.86 | 0.021 | 30 | 1.26 | 0.59–2.67 | 0.576 | 45 | 0.47 | 0.29–0.78 | 0.01 | 26 | 0.48 | 0.25–0.94 | 0.073 | 35 | 0.67 | 0.36–1.24 | 0.179 |
| Per 10% | 899 | 453 | 0.87 | 0.78–0.97 | 114 | 1.05 | 0.85–1.31 | 236 | 0.78 | 0.67–0.92 | 154 | 0.88 | 0.73–1.07 | 148 | 0.91 | 0.76–1.10 | |||||
| TIBC (µg/dl) | |||||||||||||||||||||
| ≤341.77 | 226 | 111 | Referent | 32 | Referent | 54 | Referent | 35 | Referent | 40 | Referent | ||||||||||
| 341.78–371.8 | 230 | 79 | 0.72 | 0.50–1.04 | 25 | 0.71 | 0.35–1.42 | 44 | 0.87 | 0.52–1.45 | 33 | 0.97 | 0.53–1.78 | 27 | 0.73 | 0.38–1.41 | |||||
| 371.9–410.41 | 222 | 116 | 1.16 | 0.82–1.65 | 31 | 1.28 | 0.62–2.67 | 56 | 1.09 | 0.66–1.82 | 51 | 1.38 | 0.77–2.46 | 31 | 0.89 | 0.45–1.78 | |||||
| ≥410.42 | 221 | 149 | 1.65 | 1.16–2.36 | 0.002 | 28 | 0.96 | 0.48–1.96 | 0.957 | 82 | 2.01 | 1.21–3.32 | 0.007 | 35 | 0.96 | 0.50–1.85 | 0.474 | 51 | 1.45 | 0.77–2.75 | 0.156 |
| Per 50 µg/dl | 899 | 455 | 1.13 | 1.02–1.25 | 116 | 0.93 | 0.75–1.16 | 236 | 1.22 | 1.06–1.41 | 154 | 1.01 | 0.84–1.21 | 149 | 1.29 | 1.06–1.56 | |||||
Notes.
Site unknown, n = 104.
Histology unclassified, n = 153.
Matched for age at recruitment, sex, center, date of blood extraction and further adjusted for education (categorical), tobacco smoking (categorical) and heme iron intake (mg/2,000 kcal).
For almost two thirds of cases (289 out of 456) the diagnosis of cancer took place more than five years after blood collection; for about one quarter were (111 cases) this period of time was 2 to five years, and 56 cases where diagnosed within 2 years after blood drawing. The analysis of the effect of biomarkers of iron status according to of time since blood collection is shown in Table 4. Concerning ferritin it seems that the effect is stronger for cases diagnosed after a short period of blood collection; however the inverse association remains significant even for cases diagnosed after 5 years since blood draw. On the other hand, the inverse association with gastric cancer risk for iron and TS, and the positive association of TIBC remained statistically significant only for cases diagnosed within 2 years of blood collection.
Table 4. OR and 95% confidence interval for gastric cancer and iron metrics by time between blood draw and diagnosis for all gastric cancer cases.
| Time between blood draw and gastric cancer diagnosis (years) | ||||||
|---|---|---|---|---|---|---|
| <2 years (n = 56) | 2–5 years (n = 111) | >5 years (n = 289) | ||||
| OR | 95% CI | OR | 95% CI | OR | 95% CI | |
| Ferritin (ng/ml) | ||||||
| Log2 ferritin | 0.62 | 0.51–0.76 | 0.76 | 0.65–0.89 | 0.88 | 0.79–0.98 |
| Iron (µg/dl) | ||||||
| Per 10 µg/dl | 0.86 | 0.80–0.93 | 1.03 | 0.98–1.08 | 0.98 | 0.94–1.01 |
| TS (%) | ||||||
| Per 10% | 0.75 | 0.65–0.87 | 1.02 | 0.93–1.11 | 0.94 | 0.88–1.00 |
| TIBC (µg/dl) | ||||||
| Per 50 µg/dl | 1.24 | 0.99–1.55 | 1.10 | 0.92–1.30 | 1.10 | 0.98–1.22 |
Measurement of plasma pepsinogen 1 was available for a subset of 363 subjects (123 cases and 240 controls), among which only 19 had levels below 22 μg/the cut-off used for diagnosis of chronic atrophic gastritis (CAG) (21). Although concentrations of ferritin were consistently lower in subjects with CAG, there were too few individuals to assess the effect of interaction between CAG and iron biomarkers on gastric cancer risk. However, we explored the potential effect of physiological features associated with CAG by assessing the effect of body iron status indicators taking into account the pepsinogen 1 levels within a subset of 363 individuals with measured plasmatic pepsinogen 1. Adjusting for pepsinogen 1 levels did not affect risk estimates for gastric cancer risk for any of the indicators of body iron status (data not shown).
Finally, we further analysed possible interaction of iron biomarkers with relevant variables, such as Hp infection, plasmatic levels of vitamin C, and self-reported alcohol intake, tobacco smoking, sex, BMI, and waist-to-hip ratio, but interactions were not statistically significant (data not shown).
Discussion
The results of this study represent the largest prospective analysis of the association of serum iron biomarkers with risk of gastric adenocarcinoma in European Populations. In this analysis, higher levels of ferritin, iron and TS, and lower levels of TIBC were inversely associated with risk of gastric adenocarcinoma. Moreover, these associations seem to be restricted to NCGC. It is important to mention that the differences between gastric cancer subgroups could be due to the number of cases we had available for each group. In fact, NCGC was the group with the biggest number of cases and this could have influenced the associations found for this site. Moreover, NCGC and GCC have different etiological factors so it is expected to see some differential effects when analysing both sub-sites. Also, the same pattern between gastric cancer sites has been shown in studies studying on meat and heme iron intake (5).
Overall, our findings are consistent with those from previous studies, and seem to suggest an inverse association between body iron status and gastric cancer risk, with ferritin – considered as the most specific biomarker for body iron stores – showing the strongest association amongst all iron indices, reaching a 27% reduction in risk for each doubling of the concentration of ferritin for NCGC. In a study within a cohort of Finnish men, the authors found limited evidence for the association of iron metrics with all gastric cancers (15); nonetheless, both serum ferritin and serum iron had an inverse, or possibly “u”-shaped relationships with this malignancy in the NCGC site, although associations were not statistically significant. Four other prospective studies have examined the association between iron biomarkers and all gastric cancers combined, although the predominant sub-site was NCGC. Two nested case-control studies in Japanese populations (16,17) assessed ferritin levels and found an inverse association (statistically significant) with gastric cancer, in both studies the risk of gastric cancers was lower for higher levels of transferrin but the association was not significant. Another Finnish cohort (18) reported significant association of gastric cancer risk with lower levels of transferrin and ferritin and higher levels of TIBC in men but not in women. On the contrary, one American cohort (12) reported a borderline statistically significant positive association between gastric cancer and TS for women, but no association in men.
There is definitive evidence that iron overload induces stress and DNA damage, which can enhance carcinogenic risk (7;22;23). However, other evidence suggests that iron deficiency and anaemia could also lead to increased levels of oxidative stress and DNA damage, increasing the carcinogenic risk, especially in the gastrointestinal tract. Iron deficiency may negatively influence several iron-dependent metabolic functions that are associated with genome protection and maintenance (e.g., immune responses against cancer-initiated cells, metabolism of toxic compounds, and redox regulation of DNA biosynthesis and repair) (24).
It has been suggested that the inverse association between ferritin and gastric cancer risk could be at least in part explained by achlorhydria or hypochlorhydria, which decreases the absorption of non-heme iron in the gastric mucosa (16). This condition often accompanies chronic atrophic gastritis, which is strongly associated to gastric cancer. In our study very few subjects had been diagnosed with atrophic gastritis, but ferritin levels were lower in those subjects, both among cases and controls. However, our sample size was too small to assess a potential interaction between CAG and ferritin levels on the risk of gastric cancer. We indirectly assessed the potential effect of gastritis and atrophy using the plasmatic levels of pepsinogen 1 as a proxy, but adjusting for pepsinogen 1 levels did not materially affect the association of ferritin (and other markers) with gastric cancer risk. Therefore, our results do not support the hypothesis that ferritin is a marker of CAG, the actual factor associated with gastric cancer risk as was reported in the above mentioned Japanese study (16). Another possible explanation for the inverse association between ferritin and gastric cancer is that low serum pre-diagnosis ferritin is due to bleeding from early undetected lesions of gastric cancer. If this was the case, one may expect that the effect is higher when measurement of ferritin and the diagnosis of gastric cancer are close in time than when serum ferritin was measured in samples collected long before the occurrence of clinical cancer. According to our results in table 4, the association is weakened for cases with more than 5 years between blood collection and diagnosis, and is stronger for the subsets of cases diagnosed within two years after blood collection, supporting the hypothesis that undetected blood loss could contribute, at least in part, to the inverse association between ferritin and gastric cancer risk. However, such an effect was not observed in previous studies in Japanese cohorts (16, 17) where the association was found in subjects whose serum collection preceded gastric cancer by up to 15 years.
In our study, four biological markers were used for the assessment of body iron status. Furthermore, we also used available information from the EPIC cohort study concerning iron intake and meat consumption. We found no correlations between iron indices and dietary iron or meat, except for a weak correlation of ferritin with processed meat. One could argue about the precision of laboratory measurements and the quality of the dietary questionnaires used in the study, however, we have used reliable techniques in our determinations and validated food frequency questionnaires. A possible explanation could be that iron has a tightly regulated absorption which is affected by its form and other nutrients; hence, higher iron intake is not necessarily linked to high body iron stores (25,26) and this could explain the contrast between the associations of studies on body iron status and iron intake with gastric cancer risk.
Hp infection is capable of inducing haemorrhagic gastritis leading to iron loss; furthermore Hp infection may also induce gastric atrophy, which reduces gastric acidity and ascorbic acid levels leading to poor absorption of iron, and sequester iron from the host for growth (15). Very few cases were negative for Hp serology and therefore the interaction between iron markers and Hp infection was not formally addressed; however, it must be noted that results did not change significantly when analyses were restricted to infected individuals (data not shown). Moreover, in recent analysis we have shown that eventually all NCGC cases have been previously infected, suggesting that Hp infection is a necessary cause of NCGC (27). Therefore, it would be very difficult to examine whether there is an interaction of body iron status with Hp infection on risk of NCGC.
Our study has several strengths. It is the largest study researching biomarkers of iron status and gastric cancer risk, and it is nested in a prospective study with long-term follow-up which allowed the use of plasma retrieved previous to diagnosis, which reduces the possibility that cancer affected the associations we observed in this study. Also, it enabled us to assess diet prior to diagnoses using use data from food frequency questionnaires. Iron measurements performed by our laboratory experts were determined by reliable techniques.
The major limitation of our study is that the determination of body iron status was based on a single set of measurements. Preferably we should have had several measurements at regular intervals up to the development of cancer, since timing of blood collection and specific individual conditions could alter some of these measurements. Other limitations were the few subjects with information on pepsinogen 1 levels, which limited further research on the possible link between gastric atrophy and gastric cancer.
In summary, our results do not support the hypothesis that increased body iron status is associated with increased risk of gastric cancer. Instead, it seems that higher risk of cancer of the stomach, mainly in the non-cardia region, is associated with lower body iron status. However, alternative explanations for this association cannot be ruled out. It is possible that the inverse association of serum ferritin (and other markers) with gastric cancer risk is an indirect one: lower ferritin levels could be a consequence of bleeding associated with early gastric cancer lesions, and probably also of achlorhydria or hypochlrhydria induced by chronic gastritis, a risk factor of gastric cancer. Further investigation is needed to clarify the interplay of these factors and iron status in gastric carcinogenesis.
Novelty & Impact.
Although it seems biologically plausible for iron to be associated with gastric cancer (GC), the evidence is rather heterogeneous and insufficient. This study represent the largest prospective analysis of the association of serum iron biomarkers with risk of gastric adenocarcinoma in European Populations. Our results have shown a statistically significant inverse association between ferritin and transferrin saturation and GC. Additional analysis suggests that time since diagnosis and pepsinogen levels could modify these findings.
Funding
This study was supported by the Europe Against Cancer Programme of the European Commission (SANCO); German Cancer Aid; German Cancer Research Centre; German Federal Ministry of Education and Research; Danish Cancer Society; the Dutch Cancer Registry; World Cancer Research Foundation ref. 2011/428. AGAUR, Generalitat de Catalunya (exp. 2014 SGR 726). The Health Research Funds (FIS Exp PI11/1486) and RD12/0036/0018. CIBER Epidemiología y Salud Pública (CIBERESP), Spain; Spanish Regional Governments of Andalusia, Asturias, Basque Country, Murcia (No 6236) and Navarra and the Catalan Institute of Oncology (ICO-IDIBELL); Cancer Research UK; Medical Research Council, UK; Hellenic Health Foundation; Italian Association for Research on Cancer; Italian National Research Council; Dutch Ministry of Public Health, Welfare and Sports; Dutch Ministry of Health; Dutch Prevention Funds; LK Research Funds; Dutch ZON (Zorg Onderzoek Nederland); World Cancer Research Fund (WCRF); Statistics Netherlands; Swedish Cancer Society; Swedish Scientific Council; Regional Government of Skane and Västerbotten, Sweden; Research Council of Norway; Nordforsk Centre of Excellence programme in Food, Nutrition and Health (Helga), Associazione Italiana per la Ricerca sul Cancro-AIRC- (Italy).
Footnotes
Conflicts of Interest: None of the authors have a conflict of interest.
Reference List
- (1).de MC, Ferlay J, Franceschi S, Vignat J, Bray F, Forman D, Plummer M. Global burden of cancers attributable to infections in 2008: a review and synthetic analysis. Lancet Oncol. 2012 Jun;13(6):607–15. doi: 10.1016/S1470-2045(12)70137-7. [DOI] [PubMed] [Google Scholar]
- (2).IARC Monographs on the Evaluation of Carcinogenic Risks to Humans, Volume 83, Tobacco Smoke and Involuntary Smoking. Lyon: E100 IARC Press; 2012. [PMC free article] [PubMed] [Google Scholar]
- (3).World Cancer Research Fund / American Institute for Cancer Research. Food, Nutrition, Physical Activity, and the Prevention of Cancer: a Global Perspective. Washington DC: AICR; 2007. [Google Scholar]
- (4).Gonzalez CA, Jakszyn P, Pera G, Agudo A, Bingham S, Palli D, Ferrari P, Boeing H, Del Giudice G, Plebani M, Carneiro F, et al. Meat intake and risk of stomach and esophageal adenocarcinoma within the European Prospective Investigation Into Cancer and Nutrition (EPIC) J Natl Cancer Inst. 2006 Mar 1;98(5):345–54. doi: 10.1093/jnci/djj071. [DOI] [PubMed] [Google Scholar]
- (5).Jakszyn P, Agudo A, Lujan-Barroso L, Bueno-de-Mesquita HB, Jenab M, Navarro C, Palli D, Boeing H, Manjer J, Numans ME, Igali L, et al. Dietary intake of heme iron and risk of gastric cancer in the European prospective investigation into cancer and nutrition study. Int J Cancer. 2012 Jun 1;130(11):2654–63. doi: 10.1002/ijc.26263. [DOI] [PubMed] [Google Scholar]
- (6).Hughes R, Cross AJ, Pollock JR, Bingham S. Dose-dependent effect of dietary meat on endogenous colonic N-nitrosation. Carcinogenesis. 2001 Jan;22(1):199–202. doi: 10.1093/carcin/22.1.199. [DOI] [PubMed] [Google Scholar]
- (7).Toyokuni S. Role of iron in carcinogenesis: cancer as a ferrotoxic disease. Cancer Sci. 2009 Jan;100(1):9–16. doi: 10.1111/j.1349-7006.2008.01001.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- (8).Agudo A, Bonet C, Sala N, Munoz X, Aranda N, Fonseca-Nunes A, Clavel-Chapelon F, Boutron-Ruault MC, Vineis P, Panico S, Palli D, et al. Hemochromatosis (HFE) gene mutations and risk of gastric cancer in the European Prospective Investigation into Cancer and Nutrition (EPIC) study. Carcinogenesis. 2013 Jun;34(6):1244–50. doi: 10.1093/carcin/bgt045. [DOI] [PMC free article] [PubMed] [Google Scholar]
- (9).Huang X. Iron overload and its association with cancer risk in humans: evidence for iron as a carcinogenic metal. Mutat Res. 2003 Dec 10;533(1–2):153–71. doi: 10.1016/j.mrfmmm.2003.08.023. [DOI] [PubMed] [Google Scholar]
- (10).Andrews NC, Schmidt PJ. Iron homeostasis. Annu Rev Physiol. 2007;69:69–85. doi: 10.1146/annurev.physiol.69.031905.164337. [DOI] [PubMed] [Google Scholar]
- (11).Garry PJ, Hunt WC, Baumgartner RN. Effects of iron intake on iron stores in elderly men and women: longitudinal and cross-sectional results. J Am Coll Nutr. 2000 Apr;19(2):262–9. doi: 10.1080/07315724.2000.10718925. [DOI] [PubMed] [Google Scholar]
- (12).Herrinton LJ, Friedman GD, Baer D, Selby JV. Transferrin saturation and risk of cancer. Am J Epidemiol. 1995 Oct 1;142(7):692–8. doi: 10.1093/oxfordjournals.aje.a117698. [DOI] [PubMed] [Google Scholar]
- (13).Richie JP, Jr, Kleinman W, Marina P, Abraham P, Wynder EL, Muscat JE. Blood iron, glutathione, and micronutrient levels and the risk of oral cancer. Nutr Cancer. 2008;60(4):474–82. doi: 10.1080/01635580801956477. [DOI] [PMC free article] [PubMed] [Google Scholar]
- (14).Fonseca-Nunes A, Jakszyn P, Agudo A. Iron and cancer risk--a systematic review and meta-analysis of the epidemiological evidence. Cancer Epidemiol Biomarkers Prev. 2014 Jan;23(1):12–31. doi: 10.1158/1055-9965.EPI-13-0733. [DOI] [PubMed] [Google Scholar]
- (15).Cook MB, Kamangar F, Weinstein SJ, Albanes D, Virtamo J, Taylor PR, Abnet CC, Wood RJ, Petty G, Cross AJ, Dawsey SM. Iron in relation to gastric cancer in the Alpha-tocopherol, Beta-carotene Cancer Prevention Study. Cancer Epidemiol Biomarkers Prev. 2012 Nov;21(11):2033–42. doi: 10.1158/1055-9965.EPI-12-0799. [DOI] [PMC free article] [PubMed] [Google Scholar]
- (16).Akiba S, Neriishi K, Blot WJ, Kabuto M, Stevens RG, Kato H, Land CE. Serum ferritin and stomach cancer risk among a Japanese population. Cancer. 1991 Mar 15;67(6):1707–12. doi: 10.1002/1097-0142(19910315)67:6<1707::aid-cncr2820670638>3.0.co;2-c. [DOI] [PubMed] [Google Scholar]
- (17).Nomura A, Chyou PH, Stemmermann GN. Association of serum ferritin levels with the risk of stomach cancer. Cancer Epidemiol Biomarkers Prev. 1992 Nov;1(7):547–50. [PubMed] [Google Scholar]
- (18).Knekt P, Reunanen A, Takkunen H, Aromaa A, Heliövaara M, Hakulinen T. Body iron stores and risk of cancer. Int J Cancer. 1994;56:379–82. doi: 10.1002/ijc.2910560315. [DOI] [PubMed] [Google Scholar]
- (19).Riboli E, Hunt KJ, Slimani N, Ferrari P, Norat T, Fahey M, Charrondiere UR, Hemon B, Casagrande C, Vignat J, Overvad K, et al. European Prospective Investigation into Cancer and Nutrition (EPIC): study populations and data collection. Public Health Nutr. 2002 Dec;5(6B):1113–24. doi: 10.1079/PHN2002394. [DOI] [PubMed] [Google Scholar]
- (20).Gomez F, Simo JM, Camps J, Cliville X, Bertran N, Ferre N, Bofill C, Joven J. Evaluation of a particle-enhanced turbidimetric immunoassay for the measurement of ferritin: application to patients participating in an autologous blood transfusion program. Clin Biochem. 2000 Apr;33(3):191–6. doi: 10.1016/s0009-9120(00)00064-3. [DOI] [PubMed] [Google Scholar]
- (21).Megraud F, Buissonniere A, Lujan BL, Agudo A, Duell EJ, Boutron-Ruault MC, Clavel-Chapelon F, Palli D, Krogh V, Mattiello A, Tumino R, et al. Helicobacter pylori infection assessed by ELISA and by immunoblot and noncardia gastric cancer risk in a prospective study: the Eurgast-EPIC project. Ann Oncol. 2012 May;23(5):1320–4. doi: 10.1093/annonc/mdr384. [DOI] [PubMed] [Google Scholar]
- (22).Nelson RL. Iron and colorectal cancer risk: human studies. Nutr Rev. 2001 May;59(5):140–8. doi: 10.1111/j.1753-4887.2001.tb07002.x. [DOI] [PubMed] [Google Scholar]
- (23).Pra D, Franke SI, Giulian R, Yoneama ML, Dias JF, Erdtmann B, Henriques JA. Genotoxicity and mutagenicity of iron and copper in mice. Biometals. 2008 Jun;21(3):289–97. doi: 10.1007/s10534-007-9118-3. [DOI] [PubMed] [Google Scholar]
- (24).Pra D, Rech Franke SI, Pegas Henriques JA, Fenech M. A possible link between iron deficiency and gastrointestinal carcinogenesis. Nutr Cancer. 2009;61(4):415–26. doi: 10.1080/01635580902803701. [DOI] [PubMed] [Google Scholar]
- (25).Lund EK, Wharf SG, Fairweather-Tait SJ, Johnson IT. Increases in the concentrations of available iron in response to dietary iron supplementation are associated with changes in crypt cell proliferation in rat large intestine. J Nutr. 1998 Feb;128(2):175–9. doi: 10.1093/jn/128.2.175. [DOI] [PubMed] [Google Scholar]
- (26).Kato I, Dnistrian AM, Schwartz M, Toniolo P, Koenig K, Shore RE, Zeleniuch-Jacquotte A, Akhmedkhanov A, Riboli E. Iron intake, body iron stores and colorectal cancer risk in women: a nested case-control study. Int J Cancer. 1999 Mar 1;80(5):693–8. doi: 10.1002/(sici)1097-0215(19990301)80:5<693::aid-ijc11>3.0.co;2-g. [DOI] [PubMed] [Google Scholar]
- (27).González CA, Megraud F, Buissonniere A, Lujan Barroso L, Agudo A, Duell EJ, et al. Helicobacter pylori infection assessed by ELISA and by immunoblot and noncardia gastric cancer risk in a prospective study: the Eurgast-EPIC project. Ann Oncol. 2012;23:1320–4. doi: 10.1093/annonc/mdr384. [DOI] [PubMed] [Google Scholar]