INTRODUCTION
Colorectal cancer (CRC) is the fourth most common cancer worldwide. It was newly diagnosed among 147,950 patients in the United States in 2020 and among 27,909 patients (54.4 patients per 100,000 people) in Korea based on national cancer statistics in 2018 [
1,
2]. Risk factors for the incidence of CRC include lifestyle-related factors such as high-fat diet, obesity, lack of physical activity, smoking, and alcohol consumption [
3]. Red meat consumption is also known to be a strong risk factor [
3], which can be considered in relation to the dietary iron level. Iron in red meat is present in the form of heme iron, and CRC risk increases according to heme iron intake [
4].
Iron, which has the ability to transfer unpaired electrons, is a key player in oxidation–reduction (redox) reactions. The oxidation states vary, and iron exhibits an oxidation state from −2 to 6. Because of the flexibility in accepting different oxidation states that allows iron to interact with various ligands, iron is essential for sustaining life [
5]. However, this ability of iron generates a large amount of hydroxyl radicals, which sometimes cause DNA damage and drive carcinogenesis [
5]. This suggests that iron overload in the human body is related to carcinogenesis development. In a study of 14,407 people who were observed for 10 years, 858 patients diagnosed with cancer had higher transferrin saturation and lower total iron-binding capacity than people without cancer; similar results have been reported in other studies [
6,
7].
In case of excessive levels of systemic iron, iron is sequestered and stored to prevent toxicity. Ferritin is a major iron storage protein that plays a critical role in the maintenance of systemic iron homeostasis [
8]. Therefore, serum ferritin levels generally decrease during iron deficiency and increase during iron overload [
8]. Based on these concepts, the positive correlation between cancer risk and serum ferritin level is a predictable hypothesis. However, in one meta-analysis, the serum ferritin level in CRC patients was lower than that in healthy patients, contrary to the expected result [
9]. There are only limited studies on the association between serum ferritin levels and CRC risk, especially in the Asian population.
This study aimed to analyze the association between CRC risk and serum ferritin levels in the Korean population using linkage data from the 2008 to 2012 Korea National Health and Nutrition Examination Survey (KNHANES) and the National Health Insurance Services (NHIS) claims database.
DISCUSSION
This is the largest cohort study to analyze the relationship between serum ferritin levels and CRC risk in the Korean population. Serum ferritin levels were inversely associated with CRC risk. In subgroup analysis, there were significant differences in patients with young age and without obesity, DM, and anemia when compared with the other groups. Particularly, in men, a higher ferritin level showed a lower HR, but there was no correlation in women; hence, the sex-based difference was clear.
A study on the relationship between iron storage status and CRC risk conducted in the 1980s reported an increasing tendency of CRC risk in high transferrin saturation, unlike the results of our long-term follow-up study, but the data were not significant and the number of patients with CRC development was small, with only 12 cases [
7]. The European Prospective Investigation into Cancer and Nutrition (EPIC)-Heidelberg study analyzed the relationship between iron status and cancer risk in various cancers. There were 256 CRC cases, and there was no difference in HR according to the quartile of serum ferritin levels [
14]. However, in contrast to the abovementioned studies, three nested case-control studies reported an inverse association between serum ferritin levels and CRC risk [
15–
17]. In particular, Cross et al. [
15] reported that CRC risk was inversely associated with serum ferritin level, serum iron level, and transferrin saturation, and all these markers were significant. On the contrary, although it is not CRC, in the EURGAST study which investigated stomach cancer risk, serum ferritin levels showed the strongest inverse relationship (HR of the fourth quartile to the first quartile 0.38; 95% CI, 0.25 to 0.57) among all markers of systemic iron status [
18]. These previous studies were conducted in Western countries, hence epidemiological evidence on the association between ferritin levels and CRC risk in Asians is lacking. To the best of our knowledge, this is the first large-scale nationwide cohort study to investigate this topic in Asians. The results of a meta-analysis of studies comparing CRC patients with normal subjects support our findings, although it was not an epidemiological study; in this report, CRC patients from Eastern countries had significantly lower serum ferritin levels than normal subjects [
9].
Two notable factors observed in the subgroup analysis of our study were age and sex. In terms of age, the significance of serum ferritin was prominent in those under 65 years of age but not in those aged 65 years or over. Elderly people are more prone to nutritional deficiency and chronic inflammation. Because these confounding factors can lead to an increase in the level of serum ferritin [
19], it seems that the relationship between ferritin levels and CRC risk is more prominent among younger subjects than among elderly individuals. According to the 2011 to 2016 United States statistics, the incidence of CRC decreased by 3.3% per year for those over 65 years of age but increased by 1% per year for those aged 50 to 64 and by 2% per year for those under 50 years of age [
2]. Therefore, considering the age-related differences in incidence patterns, active CRC screening is required if younger people have low serum ferritin levels. Second, there was a sex-based difference in the effect of ferritin on CRC risk. In men, the HR of CRC incidence related to ferritin tended to decrease from Q2 to Q4, and there was a clear difference between the HRs of Q4 and Q2. A difference in CRC risk according to ferritin level was also observed in women, but the difference was not significant. The reason seems to be related to iron intake in women, and as shown in
Table 2, the serum ferritin level in women is lower than that in men. The results of a dietary survey showed that, in general, many women had insufficient intake of red meat containing heme iron [
20]. In addition, many fertile women exhibited negative iron balance because the intake patterns of foods containing ingredients that impede iron absorption were higher than those of men, and there was also blood loss due to menstruation. These factors may cause complexity in analyzing the relationship between serum ferritin levels and CRC risk in women, unlike men. This may explain the sex-based difference shown in our study. This sex-based difference has previously been reported by Ekblom et al. [
16] where, similar to our study, the reduction of CRC risk with an increase in serum ferritin levels was significantly observed only in men. Therefore, young men with low serum ferritin levels should be carefully considered in active CRC screening tests.
There is strong evidence to suggest that dyslipidemia, hypertension, waist circumference, fasting glucose level, and metabolic syndrome are associated with increased CRC risk [
13,
21–
24]. In our study, serum ferritin levels in participants with metabolic syndrome showed a strong inverse correlation. Interestingly, the HR of participants without metabolic syndrome was also less than 0.6 (HR, 0.503; 95% CI, 0.286 to 0.886). This suggests that serum ferritin levels are associated with CRC risk, regardless of the presence or absence of metabolic syndrome. In addition, serum ferritin levels were significantly associated with CRC risk in participants with non-anemia and BMI of < 25 kg/m
2 but not in those with anemia and BMI of ≥ 25 kg/m
2.
Therefore, in view of these considerations, serum ferritin can assist in identifying people who need active CRC screening tests among healthy people who do not have risk factors commonly known to be related to CRC risk such as anemia, obesity, and metabolic syndrome.
Epidemiological evidence that excessive iron intake increases CRC risk has been reported historically [
4]. Most ingested iron is not absorbed and reaches the colorectum [
4]. Adenomatous polyposis coli deletion, which is found in most CRC cases, induces the intracellular accumulation of iron in the colorectal epithelium [
25]. This activates and enhances the Wnt pathway, a major oncogenic signaling pathway in CRC [
25]. As such, there is epidemiological and biological evidence for the association between iron intake and CRC risk, but in a large cohort study, there was no direct correlation between iron intake and systemic iron status [
17]. Moreover, contrary to studies showing that iron increases carcinogenic risk because iron is an important factor in genome protection, there are reports that iron deficiency induces oxidative stress and DNA damage, and this is particularly involved in tumorigenesis in gastrointestinal cancer [
26]. Therefore, there is still much to be clarified regarding the mechanism of the inverse association between ferritin levels and CRC risk identified in our study.
Aside from CRC pathogenesis in the cellular level of iron and ferritin, this inverse relationship between serum ferritin and CRC risk may be interpreted as a result of undetectable micro-bleeding, which is commonly observed in CRC. A previous study with 9,238 CRC patients reported that serum ferritin levels measured within 180 days of CRC diagnosis were low values [
27]. Serum ferritin levels measured around 1 year were normal in most cases [
27]. This suggests that factors such as micro-bleeding may affect the change of serum ferritin levels in CRC. However, contrary to this previous result, Kishida et al. [
28] reported no difference in serum iron levels in the early-stage CRC compared with those in healthy subjects. Therefore, there is insufficient evidence to explain the causal relationship between serum ferritin and CRC risk, although serum ferritin levels were inversely associated with CRC risk in our study.
Our study has some limitations. Ferritin levels can increase in acute inflammatory states [
8]. Therefore, serum C-reactive protein (CRP) levels can help differentiate acute infection or inflammatory disease, but we did not determine CRP values in the participants. However, because the KNHANES was conducted on healthy people, participants with acute infection may have been excluded. In addition, comprehensive iron status was not investigated, and periodic follow-up was not performed. All inspections and investigations in our study were carried out at the national administrative level; therefore, academic aspects of the inspection items were not sufficiently detailed. Second, micro-bleeding of undetectable early-stage CRC and consequent iron deficiency may be a confounding factor in our results. An analysis of the time interval from the measurement of serum ferritin level to the diagnosis of CRC may indirectly provide a clue to this question. However, evaluation of the HR according to time intervals from the time of registration to the time of diagnosis of CRC could not be performed because of limitations of the available data. Instead, in a previous cohort study on stomach cancer, where there is a risk of iron loss due to micro-bleeding, the risk of gastric cancer and serum ferritin levels was inversely related, regardless of the time interval [
18]. Another limitation of this study is that our main result is not sex-specific. Since there is a difference in serum ferritin levels according to sex, there may be controversy over applying our results equally to both men and women, although this result was adjusted for sex. To compensate for this limitation, we presented the result of subgroup analysis according to sex.
Despite these limitations, the present study has some strengths. It is the first large-scale cohort study conducted nationwide in an Asian population, and it is a cohort study with long-term follow-up. Contrary to the general results between excess iron intake and cancer risk, this study showed an inverse relationship between serum ferritin levels and CRC risk. In relation to CRC risk, it can be inferred that systemic iron status affects carcinogenesis differently from that of intraluminal iron. Additionally, the inverse relationship between ferritin levels and CRC risk was more prominent in young individuals and males, and a similar trend was found even in the absence of metabolic syndrome, commonly known to induce CRC risk. Therefore, in the case of healthy young men with low serum ferritin levels, more careful observation regarding the risk of CRC is required. To further clarify the significance of our findings, additional research is needed on how systemic iron status, including ferritin, affects colorectal carcinogenesis through interactions in the human body.