Faecal Haemoglobin level and mortality in colorectal cancer screening

Screening for early detection of colorectal cancer (CRC) have become an integrated part of health care systemsin many developed countries in recent years. Screening is conducted with the aim of reducing incidence andmortality by allowing for early diagnosis and treatment of cancers or by preventing onset by removingprecursor lesions. Most programs invites participants to submit a stool sample that are checked for the presenceof faecal haemoglobin (f-Hb) - an established biomarker for CRC. Previously, the guaiac faecal occult bloodtest (gFOBT) was the primary method for detecting f-Hb in screening. The test produced a positive or negativeresult. Today, the gFOBT has been replaced by the faecal Immunochemical Test (FIT) which produces aquantifiably interpretable result from which a positivity threshold can be established.

In the 1970s-1990s, three separate clinical trials created the scientific foundation of modern day CRC screening. The three trials, conducted in Minnesota (USA), Nottingham (UK) and on Funen (Denmark), allbrandomized a sizeable part of an adult population to undergo several rounds of biennial CRC screening. All three studies concluded that gFOBT-based screening have a preventive effect on CRC mortality. Years later, research groups in the US and in the UK, each presented findings suggesting a slightly reduced, but persistent, protective effect of screening on CRC mortality after 20 and 30 years of follow-up respectively.

When we started the work presented in this dissertation, the Danish CRC screening program had just completed its first round. The program appeared successful on several important parameters such as participation and detection of early cancers and high-risk adenomas. However, some problems did appear, one being the approximately 30% of participants with a negative colonoscopy. This large group of participants had faecal haemoglobin (f-Hb) levels above the positivity threshold (100 ng/mL) and no findings to explain the bleeding. This pattern appear to not be unique for screening in Denmark and researchers have begun considering other explanations than colorectal neoplasia as the source of the bleeding. Recently, a British study by Libby et al. presented findings that suggest an association between f-Hb and all-cause mortality (both including and excluding CRC deaths) in gFOBT-tested screening participants. Results also showed an association between f-Hb and a number of seemingly unrelated causes of death, such as cardiovascular disease and neuropsychological disease. Authors speculated that f-Hb could be used to indicate the presence of other serious and/or chronic conditions. While succeeding in creating initial support for their hypothesis, the study did, however, have some methodological limitations that needs to be addressed. This warrants additional studies on the subject.

Aim
The overall aim of this dissertation was to investigate the effects of screening for CRC and to investigate the proposed association between f-Hb and mortality outcomes. This divided into three studies where we wanted:

1. To investigate the role of CRC screening on both overall and cause-specific mortality in a pooled study of Danish and American data.

2. To investigate the association between f-Hb and both overall mortality and seemingly unrelated causes of death in a Danish gFOBT-tested screening population with more than 30 years of follow-up

3. To investigate the association between incrementally increasing f-Hb and mortality outcomes in a modern-day FIT-based Danish screening population.

Study 1. The long-term effect of colorectal cancer screening
In study 1, we pooled the data from the Danish and the US screening trials and enriched it with follow-up data. We then conducted an individual participant data meta-analysis comparing screening participants to the control population in terms of mortality. We found that CRC screening provides a sustained reduction in CRC mortality and a significant reduction in all-cause mortality (adjusted for compliance). We also observed no benefits from screening in the youngest group of women aged 50-59. 

Study 2. Faecal haemoglobin and mortality outcomes after >30 years of follow-up In study
2, we used enriched the Danish trial data from study 1 with register-data on education, income, cause of death and comorbidity and conducted a 33-year follow-up. We compared gFOBT positive to negative participants on a number of outcomes including all-cause mortality, CRC mortality and causes of death. We found an association between positive gFOBT result and increased risk of both all-cause mortality and several seemingly unrelated causes of death such as cardiovascular disease. The study did have a number of limitations as we were not able to adjust for the effects of prescription medication nor could we quantify the f-Hb levels to investigate potential dose-response relationships. 

Study 3. Quantified faecal haemoglobin and mortality outcomes in the Danish colorectal cancer screening population In study
3, we collected data on current screening participants in Denmark from several national registers. We introduced prescription medication as a covariate and FIT-levels as the exposure to address limitations of study 2. Participants were divided by FIT-level and compared in terms of mortality outcomes. Our results showed that even an incremental increase in FIT increased the risk of all-cause mortality – even after excluding CRC deaths. The same is true for different and seemingly CRC-unrelated causes of death. Interestingly, we observed a clear dose-response relationship between FIT and several of our outcomes underlining the association. 

Conclusions & future perspectives
In conclusion, our results suggests that CRC screening as a concept appear viable for sustained reductions in mortality, but the preventive benefit varies significantly by age and gender – something that is worth considering when designing future initiatives. Our results also suggest that elevated f-Hb levels does appear to be associated with an increased both all-cause mortality and several causes of death not usually connected to CRC. This supports the notion that f-Hb could indicate the presence of non-communicable, chronic conditions. We believe that f-Hb may one day become an established biomarker for non-CRC diseases, which could add important perspectives to CRC screening and create a more nuanced understanding of the complicated relationship between gut health and health outcomes. Due to the complex and multifactorial nature of this topic, more work is needed for it to have a clinical impact.