Background
Colorectal cancer (CRC) is a major health concern worldwide, ranking third in males and second in females, with over 1.2 million new cancer cases and an estimated 608,700 deaths in 2008 [
1]. Survival of colorectal cancer is inversely related to the stage at diagnosis. Five-year survival rates range from more than 90% for stage I to less than 10% for stage IV CRC [
2]. Therefore, detection and removal of the tumour in an early, or preferably, a premalignant stage is vital [
3]. Since CRC often only becomes symptomatic when progressed to an advanced state, secondary prevention through screening is an important instrument for reducing death from CRC [
4].
CRC as a disease lends itself well for screening as it has a high prevalence and a well-defined precursor lesion (i.e., adenoma) with a long dwell-time, providing an excellent window of opportunity for detection and resection of a lesion before becoming symptomatic. Indeed, incidence and mortality rates of CRC have declined in countries where screening has been introduced [
5‐
7]. Recently a nationwide screening programme was also implemented in the Netherlands, using a faecal immunochemical test (FIT). Men and women aged 55 to 75 years are invited every 2 years, which amounts to a total number of 2.2 million individuals per year in the Netherlands being invited to participate in the program. Of the estimated 7% FIT-positive screenees, approximately half will have advanced adenomas or carcinomas at colonoscopy, equal to over 45.000 individuals annually [
8]. As these patients carry an increased risk to develop metachronous advanced lesions in the future [
9‐
12] it is standard practice to enrol these individuals in a colonoscopy-based surveillance program [
13‐
15].
There are several downsides, though, to the current approach for managing the cancer risk in the surveillance population. Firstly, evidence of the impact of current surveillance strategies on the ultimate endpoint CRC-related mortality is very limited. The effect of surveillance has primarily been evaluated for intermediate endpoints, i.e., the yield of (advanced) adenomas upon surveillance colonoscopy [
12,
16]. However, adenomas are very common with a prevalence of 18-35% reported in screening series [
17,
18], whilst only up to 5% of these adenomas will eventually progress to malignancy [
19]. This suggests that focussing on adenoma yield as a primary endpoint represents actual overdiagnosis. Moreover, the technical advancements in colonoscopy-equipment and the recent emphasis on quality assurance have resulted in increasingly more and often smaller adenomas being detected. These small lesions are more likely to remain stable over time [
20,
21]. As a result, the surveillance population will expand even further, putting the colonoscopy capacity and health care budgets under pressure [
22,
23]. Finally, it will expose post-polypectomy patients to a burdensome and risky procedure that has not proven to be effective for this population.
For these reasons there is a demand for a surveillance tool that is easy to apply, well-tolerated and accurate in identifying high risk adenomas, as to reserve colonoscopy only for those individuals that are most likely to benefit.
Today, despite its limitations, colonoscopy is still the only test used for surveillance of patients after removal of polyps and CRC. The FIT, which detects small amounts of human haemoglobin in the faeces, has been proposed as a method for surveillance [
24]. While simple and increasing the likelihood of neoplasia being present when positive, its sensitivity is relatively low due to the fact that not all CRCs, and especially not all advanced adenomas bleed [
25,
26]. Yet, it was shown that the diagnosis of CRC and advanced adenomas was made 25 and 24 months (median) earlier, respectively, when offering a yearly FIT in the interval between colonoscopies and shortening the interval in case of a positive test. This indicates that FIT could be used to detect missed or rapidly developing lesions in surveillance programs [
24].
In contrast to tests detecting blood, tests based on molecular markers derived from the neoplastic cells in the colon have the potential to be more accurate. CRCs are known to acquire discriminating epigenetic and genetic changes as they develop and progress, which form the basis of stool based DNA testing. Recently, a multitarget stool DNA test (Cologuard®, Exact Sciences, Madison, WI, USA) combining a molecular assay for hypermethylated promoter CpG islands (
NDRG4 and
BMP3) and mutant
KRAS with an immunoassay for human haemoglobine has been reported on [
27]. Sensitivities for the detection of CRC were 92.3% with Cologuard® and 73.8% with FIT (p =0.015). Also, Cologuard® detected significantly more advanced adenomas than the FIT (69.2% versus 46.2%, p = 0.004) and significantly more sessile serrated polyps measuring 1 cm or more (42.4% versus 5.1%). Specificities of Cologuard® and FIT were 86.6 and 94.9%, respectively. The results of this study have led to FDA approval of Cologuard® in 2014.
We hypothesise that Cologuard®- or FIT-based surveillance is a cost-effective first-line surveillance strategy to select individuals that need colonoscopy for diagnosis confirmation and therapeutic removal.
Discussion
In this cross-sectional observational cohort MOCCAS study, the performance of a multitarget stool DNA test (Cologuard®) and FITs will be assessed as an alternative for surveillance colonoscopy. To this end, the results of the stool based molecular tests and FITs will be compared to the findings of the (routine) colonoscopy in a surveillance population. These data will then be used as input for model-based analyses. By simulating different surveillance strategies varying in testing frequency and rules for referral to the screening population, the model will predict outcomes such as cancer incidence and mortality, resource utilisation and cost-effectiveness.
Multiple modalities are available in the prevention of CRC. Colonoscopy is a one-staged method in which lesions are detected and removed simultaneously. Other methods, such as CT-colonography, faecal immunochemical tests and multitarget stool DNA tests, are two-staged and are used as triage for colonoscopy. The two-staged methods are associated with lower sensitivities, but are generally advantageous in terms of participation rates, risks and costs [
27,
28,
38]. When choosing an optimal strategy it should be emphasised that the weight of these factors is different for surveillance compared to screening. Where screening targets the “healthy” population, surveillance is aimed at a narrow high risk group with a higher positivity rate for colonoscopy. Sensitivity of the method is therefore most important. The accuracy of molecular tests might approach colonoscopy in the detection of (high risk) colorectal polyps and –carcinomas when the test is performed more frequently than colonoscopy. Participation in these high risk groups is generally high, as a result of a patient’s awareness of his/her risk.
The optimal model-predicted surveillance strategy, as identified in the current study, will be evaluated in clinical practice through a randomised study. Besides investigating a potential alternative to surveillance colonoscopy, this study will generate new insights in the molecular profiles of precancerous lesions by relating the expression of both diagnostic biomarkers and progression biomarkers to the results of Cologuard®. Moreover, because whole stool samples, as well as FIT samples are collected, a large biobank is established that provides extensive opportunities for future research.
Acknowledgements
The authors thank Exact Sciences (Madison, WI, USA) for assisting in the logistics and analysis concerning their Cologuard®.
We thank Sentinel (Milan, Italy) for providing the FOB Gold® infrastructure necessary to perform this study.