Background
Colorectal cancer (CRC) screening with faecal testing significantly impacts CRC mortality sufficiently to affect all-cause mortality [
1‐
8]. The quantitative faecal immunochemical test (FIT) is the most frequently used test [
9]. In the Danish CRC screening programme, participants with a positive FIT (faecal haemoglobin concentration (f-Hb) ≥100 ng Hb/mL buffer or ≥20 μg Hb/g faeces) are referred for colonoscopy. 37.0% of these colonoscopies reveal no neoplastic findings that explain the bleeding [
10]. This issue is not unique to the Danish screening programme, and the sensitivity of the FIT is often discussed [
11]. The presence of haemoglobin in the faeces without an obvious source of bleeding could indicate early-stage non-communicable chronic disease [
12]. Recent studies suggest that f-Hb is associated with several seemingly unrelated chronic conditions and causes of death, including cardiovascular disease (CVD), respiratory disease and/or neuropsychological disorders [
13‐
16]. However, these studies are limited by either the use of guaiac-based faecal occult blood test (gFOBT) instead of FIT, the lack of prescription medication associated with gastrointestinal (GI) bleeding as a covariate and/or the lack of individual-level adjustment for confounders. Therefore, we conducted a register-based study investigating the association between f-Hb and both all-cause mortality and cause-specific mortality in a population-based, FIT-tested cohort of participants in the Danish CRC screening programme.
Discussion
Our results indicate that f-Hb is associated with all-cause mortality and seemingly unrelated causes of death in a cohort of screening participants in the Danish national CRC screening program. The significance of our findings is underlined by a significant association relationship between f-Hb and all-cause mortality. For causes of death, this effect was especially notable with CVD and respiratory disease. However, the increased risk of dying from other cancers seems to plateau at f-Hb of ≥20.0 μg Hb/g faeces. In addition, an association between diabetes mortality and f-Hb was observed, although a low number of events and a wide CI reduce interpretability. Since diabetes is a common disease, it is possible that the cause of death in some diabetics may be registered to complications of the disease rather than diabetes itself [
29]. Less surprising was the observed relationship between f-Hb and CRC, which was particularly pronounced in the group with the highest f-Hb. However, the increase in aHR appeared to plateau in participants with f-Hb ≥20.0 μg Hb/g faeces, i.e. those just above the Danish positivity threshold. This should be interpreted cautiously as we have a low number of CRC deaths in each category, but one suggested explanation is that this stagnation in mortality around the positivity threshold shows the impact of CRC screening on the mortality curve.
The role of f-Hb in diseases with a systemic inflammatory component represents a growing area of research. A recent review summarizes the hypotheses for the observed patterns, suggesting that elevated f-Hb is an indicator of systemic inflammation and possibly a marker of susceptibility to non-communicable diseases [
12]. Recently, Scottish and Danish researchers presented findings that indicated a relationship between f-Hb (measured by gFOBT) and both all-cause mortality and seemingly unrelated causes of death in CRC screening populations with very long follow-up [
15,
16]. Korean researchers categorized a FIT-based screening population as either FIT positive or negative, concluding that FIT-positive participants had a higher risk of dying from non-CRC causes, circulatory disease and respiratory disease [
30]. Furthermore, in a study from Taiwan, there was a similar dose-response relationship between f-Hb concentration on all-cause mortality and CRC mortality in a cohort of CRC screening participants [
31].
Other studies have focused on the relationship between f-Hb and the likelihood of having an underlying non-communicable disease. Japanese researchers have demonstrated an association between FIT positivity and concentrations of glycated Hb in an apparently healthy population [
13]. Furthermore, a South Korean study showed increasing incidence rates of diabetes with an increasing number of positive FIT during a 6½-year follow-up period [
32]. In a large community-based study, researchers from Taiwan suggested a significantly increased risk of developing and dying from CVD with increasing f-Hb [
33]. Lastly, Moon et al. support these findings by showing an increased risk of myocardial infarction, ischemic stroke and all-cause mortality in the FIT-positive population [
34]. Other studies have investigated the association between f-Hb and the development of inflammatory diseases. One study presents a 16.0% higher risk of developing rheumatoid arthritis among the FIT positive in a large population of Korean screening participants [
35]. Moreover, a Taiwanese study on periodontal disease and f-Hb concentrations reported that participants with a concentration of ≥20.0 μg Hb/g faeces had a 33.0% higher risk of their periodontal disease deteriorating to severe when compared to those with f-Hb concentrations <20.0 μg Hb/g faeces [
36]. In another study, the authors found that having a positive FIT result was associated with an increased risk of having psoriasis. The authors highlighted the association between psoriasis and a number of conditions documented to be associated with elevated f-Hb such as metabolic syndrome, dyslipidaemia, obesity, diabetes and hypertension [
37]. Moreover, a Taiwanese study has shown an association between metabolic syndrome and elevated f-Hb concentrations [
38]. In extension, a Taiwanese study on air pollution and CRC reports that high concentrations of ambient fine particulate matter (known to cause inflammation) increase the risk of having a positive FIT by 11.0% [
39]. Scottish researchers found an association between the consumption of medication as a proxy marker for disease and f-Hb. Their findings suggest an increased risk of having heart diseases, diabetes, hypertension and depression if participants have a positive gFOBT result [
14]. Our results suggest that several characteristics such as higher age, male sex, lower education and lower income are more common among those with elevated f-Hb concentrations. This is consistent with existing literature [
40].
Considering the findings of these studies in the context of our results, there appears to be mounting evidence that f-Hb does indicate the presence of non-communicable diseases and since many of the reported associations have a systemic inflammatory component, the proposed relationship between inflammation and GI bleeding seems plausible. However, the mechanisms underlying these associations remain elusive. An explanation is that the use of medications known to cause GI bleeding, such as oral anticoagulants, aspirin, corticosteroids and non-steroidal anti-inflammatory drugs, could affect the associations presented in this study. However, the inclusion of these drug categories as a confounding factor in the multivariate analysis strengthens the interpretability of our results. This, combined with differing conclusions from recent meta-analyses, reduces our concern about the impact of these drugs on the overall interpretation [
41,
42]. Another potentially contributing factor to the elevated f-Hb could be diseases that cause bleeding in the GI tract. The immunochemical response of the FIT should in theory, not detect bleedings originating from the upper GI tract due to the digestion of the haemoglobin proteins. However, FIT-positive participants with a negative colonoscopy have been reported to have a higher risk of some upper GI cancers [
43]. While the scientific evidence on this topic is limited and findings incoherent, it cannot be ruled out that upper GI bleedings can affect the FIT measured f-Hb [
43,
44]. It is also possible that lower GI bleedings from sources such as haemorrhoids, diverticular diseases and inflammatory bowel disease may impact the reported f-Hb to some degree. While a relationship between false-positive FIT result (positive FIT and negative colonoscopy) and both inflammatory bowel disease and other non-neoplastic findings (such as anal fissures) has been reported, no association was found with diverticulosis and haemorrhoids in the same studies [
45,
46]. To mitigate this uncertainty, we accounted for the potential effect of both types of bleeding in our multivariate analyses.
The strengths of our study include a very large population of screening participants, extensive and individual-level adjustment for confounding effects and our ability to adjust for a numerical f-Hb. Limitations include the risk of misclassification in the registrations from the DRCD, which is affected by the input quality. We have addressed this by including contributing causes of death in the analyses and have no indications that misclassification could confound the results to any significant extent. Limited follow-up is another disadvantage of our study, but given the size of our population, the use of time-sensitive analyses and the coherence to other studies with longer follow-up, we do not believe that this affects the overall interpretation of our results. Data on over-the-counter medication is not available in Danish registers and represents another limitation. This may result in some underestimation of the effect medications such as aspirin have on the f-Hb. However, because we were able to adjust for the effect of prescription medication, the impact of this limitation is reduced.
Assuming f-Hb is an effective and practical biomarker for non-communicable diseases, several new pathways focusing on maximizing the diagnostic gain from CRC screening and diagnostic initiatives in other clinical areas could be established. For instance, a study from China suggests using f-Hb to predict complications and survival after R0 gastrectomy [
47].
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.