Summary of findings
The studies reviewed included a variety of designs such as blinded control studies, cross-sectional studies, and cohort studies. The study settings included hospitals and outpatient centers, as well as private homes for participants who were asked to mail in stool collection cards. Typically, either a FIT or a MT-sDNA test was collected from a participant before his or her colonoscopy to compare results between the 2 screening methods.
Eleven studies (totaling 44,470 participants) evaluated FITs and/or MT-sDNA, testing for detection of CRC. Accuracy (correct screening result), sensitivity (true positives), and specificity (true negatives) were reported by all studies. Eleven studies provided probability values (P values). In addition, many of the studies addressed stool collection methods, patient participation, test comfort/usability, and cost-effectiveness. This literature review also includes 2 systematic reviews.
Studies took place in the United States, Canada, Italy, the Netherlands, and Spain, and participants were selected using similar criteria for age (40-85 years, with a median age range of 60-65 years) and degree of risk. An average risk level was defined as being asymptomatic and without familial risk (having a first-degree relative with CRC) and applied to 10 studies (24,259 of the participants). Four studies included participants at higher risk for CRC (defined as having a first-degree relative with CRC or the participant having had a previous neoplasm). Some studies accepted patients who had colonoscopies before participation in the study. Colonoscopy results varied and included both negative and positive results for neoplastic changes, depending on the focus of the particular study (ie, testing center proficiency or diagnostic sensitivity/specificity of a specific test).
All studies included quantitative design, but some mixed-method studies provided additional qualitative data regarding the participant’s experience and the screening product’s ease of use. Measures used in data analysis included Chi square (χ2) with a confidence interval (CI) of 95%, multivariate analysis, positive predictive value, and linear regression models.
Evidence related to FITs
Seven studies investigated the sensitivity and specificity of the FIT for detecting advanced neoplasia (AN) and CRC.6-12 Variations existed in the collection method of samples (eg, single fecal samples vs multiple samples). A prospective cohort study with 2959 participants at average risk for CRC was conducted for a period of 7 years, during which the FIT was completed 4 times.5 The authors found a high detection rate of AN in every round. Another study included 2 cohorts with different risk levels (average vs high risk) and found no significant difference in FIT results between the 2 groups.6 Considering both cohorts, overall FIT accuracy was high (88.4%-91.7%; P =.051), and the difference between the groups was not statistically significant. Similar findings occurred in 3 additional studies reviewed.7-9
The FIT was found to be moderately successful at screening for adenomas and sessile-serrated adenomas/polyps (SSA/Ps) but even better at detecting CRC.7 FIT was best at detecting CRC vs adenomas and SSA/Ps. The proximal vs distal and left-side vs right-side location of lesions also were found to affect FIT sensitivity.7,10 This finding was supported in a later study, which also noted that FIT sensitivity and specificity were affected by the morphology of the lesion detected: flat and nonhemorrhagic lesions (SSA/Ps) were less likely to be detected than adenomas.11
The systematic review and meta-analysis found FITs to have high levels of accuracy in detecting CRC.12 They also determined the FITs had high levels of specificity (94%) and moderately high levels of sensitivity (79%) in detecting CRC (depending on cutoff values for Hgb used by the particular FIT manufacturer).
Several studies included qualitative data in addition to a quantitative approach. These areas focused on practical issues of patient comfort and usability, patient participation in screening, and sample analyzation methods (eg, manual vs automated).6,8,10,12
Evidence related to MT-sDNA tests
The MT-sDNA test was used to assess colorectal neoplasms and was found to have a CRC detection rate of 85%.13 The detection rate of advanced adenomas (AAs) was not as high as those found in a later study,14 but detection rates were found to be improved with an increase in adenoma size. The location of the nonmetastatic neoplasm did not affect detection rate. The clinical performance of the MT-sDNA test was the focus of a study including 1003 individuals with average CRC risk.14 This case-controlled study occurred in multiple test centers where specimens were analyzed to measure for 5 different DNA markers. Specific cutoff levels were used to determine either positive or negative scores for detecting CRC, AAs, and SSA/Ps. Sensitivity for CRC was high (98%) regardless of cancer stage level; specificity was 90%. Detection of AA with high-grade dysplasia had 83% sensitivity. The authors’ conclusions identified MT-sDNA as an effective tool in early detection of CRC.14