Evaluation of a New Blood-based Test to Detect Colorectal Cancer and Its Precursors

1. BACKGROUND Colorectal Cancer (CRC) is one of the leading causes of cancer in developed regions around the world and its incidence is rising(1,2). Many nations or regions have introduced CRC population screening programmes to detect cancerous lesions before symptoms arise or to detect precursor lesions whose eradication may prevent CRC (3). The Faecal Immunochemical Test (FIT) is used to screen for CRC in many nations and regions around the world. Its advantages include ease of use, the ability to be performed at home and to be sent by mail to a laboratory, and its cost-efficiency (4). Moreover, these characteristics enable population-wide screening without the need for all participants to undergo a colonoscopy. However, a downside of the FIT is its performance. In the Netherlands, where a FIT cut-off concentration of 47 mcg Hb/g faeces is used, the FIT produces a negative result in 15% of people who have CRC (5). For the detection of CRC and Advanced Adenomas (AA), referred to as Advanced Neoplasia (AN), the FIT falsely provides a negative result in up to 70% of people with AN. In addition, more than half of the people with a positive FIT result in the Netherlands do not have any (pre)cancerous lesions at colonoscopy. One way to improve the efficiency of CRC screening is to introduce a second test to FIT-positive screening participants, to further select individuals for a colonoscopy. Ideally, this test should have a high sensitivity since a false negative result would lead to denying a colonoscopy to a FIT-positive individual with a (pre)cancerous lesion. In addition, the specificity of this test should be high enough to substantially reduce the number of FIT-positive individuals without AN undergoing colonoscopy. The CELTiC panel is a blood-based test comprising of four mRNA markers (LGALS4, CEACAM6, TSPAN8, and COL1A2) (6). In a sample of 128 individuals the panel reached an are under the curve (AUC) of 0.82 in discriminating individuals with AN (n = 92; AA n = 25, CRC n = 67) from individuals with a positive FIT result but a normal colonoscopy (n = 36). The CELTiC panel could potentially function as a second test for individuals with a positive FIT result and reduce the number of unnecessary colonoscopies (7). Furthermore, by simultaneously lowering the cut-off concentration of the FIT and introducing the CELTiC panel, screening programs may potentially detect more individuals with AN without increasing the number of colonoscopies performed.
2. QUALITY ASSURANCE This cross-sectional diagnostic test study will examine the CELTiC panel in a population of 800 FIT-positive individuals in Italy and the Netherlands. Sites in both nations will adhere to the same protocol, including blood sample collection, processing, and data collection. Monitoring will be put in place at both sites to ascertain adequate data collection and storage. Laboratory personnel and endoscopy staff will be blinded to the results of the colonoscopy and lab analysis respectively.
3. OBJECTIVES Primary Objective: to estimate the sensitivity of the CELTiC panel for detecting AN at a cut-off with an expected sensitivity of 90%. Secondary Objective(s) include: (i) to estimate the specificity of the CELTiC panel for detecting AN at a cut-off with an expected sensitivity of 90%; (ii) to estimate the positive predictive value of the CELTiC panel for detecting advanced neoplasia at a pre-specified threshold with an expected sensitivity of 90%; (iii) to estimate the specificity of the CELTiC panel for detecting advanced neoplasia and CRC at an observed sensitivity of 90%; (iv) to estimate the positive predictive value of the CELTiC panel for detecting advanced neoplasia and CRC at on observed sensitivity of 90%; (v) to estimate the Area under the receiver operator characteristic curve (c-statistic); (vi) to estimate the association between CELTiC score and FIT concentration.
4. SAMPLE SIZE

The CRC screening programs of Italy and the Netherlands differ in multiple aspects: in Italy, individuals between 50 and 69 years old are screening every two years with FIT set at a threshold of 20 ug Hb/g feces. In contrast, the Netherlands screens individuals between 55 and 75 years old every two years with FIT set at a threshold of 47 ug Hb/g feces. Therefore, if we were to analyse the results from a combination of these two populations, the results may not be externally valid; the population would consist of more individuals with a relatively high FIT results compared to any other FIT-population in the world. Since FIT is an important risk factor for AN, this may induce selection bias.

Therefore, we have opted to perform two analyses: one with all individuals recruited in Italy (FIT >= 20), and one with all individuals with a FIT of 47 ug Hb/g or higher (both in the Netherlands and Italy). We will aim to estimate a cut-off of CELTiC with a sensitivity of 90% for detecting AN. In addition, we are interested in an estimate with a 95% confidence interval that is not wider than 5% at either side and, for practical reasons, to split the recruiting between Italy in the Netherlands in a 3:2 ratio. Given the positive predictive value of FIT in both screening programs at their respective cut-offs, we will need in total 800 participants for this trial, 347 of whom need to be diagnosed with AN.