EUnetHTA assessment of stool DNA testing for early detection of colorectal cancer has been published

In late July 2019, the report of the project OTJA10, titled “Stool DNA testing for early detection of colorectal cancer” was published on the website of the European Network for HTA (EUnetHTA).

The project was executed by Austrian Public Health Institute (GOEG), Slovenian Institute of Public Health (NIJZ), Agency for Medicinal Products and Medical Devices of the Republic of Slovenia (JAZMP), Private University of Health Sciences and Medical Informatics and Technology (UMIT) in Tyrol, Austria.

The dedicated reviewers come from the National Institute for Health and Care Excellence (NICE), Italian Agency for Regional Health Services (AGENAS), Danish Social & Health Services and Labor Market (DEFACTUM) and Basque Office for HTA (OSTEBA).

Deoxyribonucleic acid (DNA) stool testing for the early detection of colorectal cancer (CRC) is a non-invasive technology that supplements the established stool tests [e.g., fecal immunochemical test (FIT) or guaiac (based) fecal occult blood test (gFOBT)] for CRC detection with the stool-based analysis of tumor DNA.

The currently CE-marked stool DNA tests in Europe are ColoAlert® (PharmGenomics) and Cologuard® (Exact Sciences). However, only ColoAlert® is sold on the European market. Both technologies are to be used solely for screening as well as prevention purposes in public health. The expected benefit of DNA stool testing is having a non-invasive screening test that is superior to gFOBT and FIT in terms of test accuracy and comparable in terms of patient compliance and can replace those tests in the screening pathway.

The assessment was based on the systematic literature search performed in Medline, Cochrane Library, and EMBASE in August 2018. Clinical trial registries (ClinicalTrials.gov, World Health Organization (WHO) International Clinical Trials Registry Platform and the EU Clinical Trials Register) were assessed for registered ongoing clinical trials or observational studies; an initial search was completed by an update search in March 2019. In addition, a hand search (in reference lists of relevant studies), as well as an internet search, were performed.

Results - clinical effectiveness:

Regarding the outcomes on test accuracy, two studies were identified that investigated Cologuard® (Imperiale et al.], Brenner et al.) and one that investigated ColoAlert® (Dollinger et al.). Regarding outcomes on mortality, morbidity, health-related, and non-health-related QoL, no (primary) studies were identified. Five patient interviews were done for this assessment, which inform outcomes on patient satisfaction as well as some QoL aspects. For the outcomes on patient satisfaction, an additional five surveys were identified [Matthew V. Abola TFF et al., Schroy PC, 3rd et al., Berger BM et al., Calderwood AH et al., Schroy PC et al.], but only one of which investigated one of the currently available tests [Matthew V. Abola TFF et al.] (Cologuard®). With regard to the comparators defined in PICO, none of the included studies investigated (flexible) sigmoidoscopy, SEPTIN9 test or CTC compared with stool DNA testing.

• One study by Johnson et al. compared colonoscopy results in an unblinded and a blinded group with a positive DNA test. More total adenomatous/sessile serrated polyps (70% versus 53%, p=0.013) and advanced neoplasms (28% versus 21%, p=0.27) were detected in the unblinded than in the blinded group. Among polyps detected, flat or slightly raised lesions in the right colon were proportionately more frequent with unblinded (40%) than with blinded examinations (9%; p=0.0017). Median withdrawal time was 19 min in the unblinded group compared with 13 min in the blinded group (p<0.001). The study authors concluded that knowledge of a positive DNA test result had a beneficial impact on the diagnostic yield and quality of subsequent colonoscopy
• No primary studies were identified regarding the expected effect of DNA stool tests on CRC or overall mortality. The benefits of the screening strategies can be transformed and interpreted on an individual level. That is, when compared to No Screening, one 50-year-old individual is expected to gain, on average, 144 life days with the COL strategy, 141 life days with Cologuard, 133 life days with the FIT, and 131 life days with the ColoAlert strategy
• The comparative effectiveness of modeling the five CRC screening strategies yielded the following results regarding morbidity outcomes: per 1000 screened individuals, COL, Cologuard, FIT and ColoAlert averted 62, 52, 45 and 44 CRC cases, respectively
• When asked for the preferred strategies for routine CRC screening, 45% of patients in the study by Schroy et al. preferred DNA testing, 32% gFOBT, and 15% colonoscopy, with 8% expressing no preference. In the survey of patients without CRC-screening experience (Schroy et al.), 51% preferred colonoscopy as a screening option, versus 28% for DNA testing and 18% for gFOBT. Subjects who preferred colonoscopy rated accuracy as the most influential test feature. For those who preferred DNA, concerns about the amount of discomfort were rated highest, and for those who preferred gFOBT, frequency of testing was rated highest

Results - safety:

• None of the primary studies reported adverse events of DNA or the other included stool tests. No studies were found that investigated (if and) how the frequency or severity of harms change over time or in different settings. None of the primary studies distinguished susceptible patient groups that are more likely to be harmed through the use of DNA stool tests. No primary studies reported (or have been found on) user-dependent harms of DNA stool tests. In addition, no studies were found that directly investigated consequences of false positive or false negative test results from the viewpoint of patient safety (e.g. delayed treatment or overtreatment). No study reported on incidental findings of DNA stool testing (because the test only gives either a ‘negative’ or a ‘positive’ finding)

Results – benefit-harm analysis:

• Based on the base-case analysis with a screening adherence of 100% in all screening strategies and considering the total number of colonoscopies as a measure of burden for individuals, on the one hand, life-years gained (LYG) and CRC deaths averted as measures of benefits on the other hand, the 3-yearly ColoAlert strategy was dominated by biennial FIT (i.e. provided less health benefit at higher burden). To gain one additional LY with the 3-yearly Cologuard strategy compared with the biennial FIT, there is an expected incremental burden of an additional 19 colonoscopies. To gain one LY with 10-yearly colonoscopy compared with the Cologuard strategy, there is an expected incremental burden of an additional 167 colonoscopies. To avoid one CRC-related death with 3-yearly Cologuard compared with the biennial FIT, there is an expected incremental burden of additional 208 colonoscopies. To avoid one CRC-related death with 10-yearly colonoscopy compared with 3-yearly Cologuard, there is an expected incremental burden of additional 1235 colonoscopies
• Moving from No Screening to FIT would result in an average benefit of 133 additional life days for the screenee and an average burden of approximately 1 expected additional colonoscopies during their lifetime. Moving from FIT to Cologuard would result in an additional benefit of 8 life days and an additional average burden of approximately 0.4 colonoscopies. Finally, moving from Cologuard to COL would result in a further benefit of 3 life days and an additional average burden of approximately 1.5 colonoscopies
• Considering the tradeoff between the potential psychological harm of positive test results versus the benefit of CRC death averted, all stool tests were dominated
• The systematic sensitivity analyses showed that model-predicted benefit–harm results were particularly sensitive to screening participation and/or adherence rates. At a 20% adherence rate of 10-yearly colonoscopy and a 30% adherence rate of stool test screening, COL is dominated in the benefit–harm analysis considering the tradeoff between LYG and colonoscopies. Increasing the adherence rates of stool tests led to not only increasing LYG but also increasing numbers of colonoscopies. Nevertheless, COL is dominated on the whole range of adherence rates

Conclusions:

• Stool DNA testing with Cologuard® had higher sensitivity for the detection of CRC and advanced adenoma compared with the FIT, but lower specificity. However, these results depend to a degree on the exact type of FIT used
• Stool DNA testing with ColoAlert® (although referring to a former version of the product) had higher sensitivity for the detection of CRC and adenoma compared with gFOBT, but lower specificity. The combined test failure rate of all three stool tests in this study was higher than of FIT (as seen in the study by Imperiale et al.). There was no direct evidence available for the test accuracy for only advanced adenoma and no information on the exact proportion of test failures in the DNA assay alone compared to the other stool tests
• Overall certainty of the evidence was moderate to high for Cologuard® results (two studies, both referring to the same Cologuard® study population) and low to very low for ColoAlert® results (one study)
• Based on the decision-analytic modeling, CRC screening with a 10-yearly colonoscopy screening strategy was more effective than all other investigated screening strategies when considering longterm screenee-/patient-relevant outcomes, such as remaining life expectancy, CRC mortality, and risk of developing CRC, and when assuming 100% adherence rates. The 3-yearly ColoAlert® screening strategy, the 3-yearly Cologuard® screening strategy, and the biennial FIT screening were less effective than 10-yearly colonoscopy screening, but also led to less burden resulting from colonoscopies. The ColoAlert® strategy yielded fewer LYGs and caused more additional colonoscopies, compared with a ‘no screening’ strategy, than biennial FIT and, therefore, was dominated
• Stool DNA testing shows a promising benefit–harm balance when comparing different screening strategies, although this only currently refers to Cologuard®. By contrast, ColoAlert® is the only stool DNA test currently sold in Europe and, moreover, is available at a lower cost than Cologuard®. A high degree of uncertainty surrounds the evidence on ColoAlert®

The full report in English can be found here.

Subscribe to our newsletter delivered every second week not to miss important reimbursement information.