Blood-based CRC screening platform delivers promising performance

An investigational blood-based test performs satisfactorily for the early detection of colorectal cancer (CRC), having met all prespecified primary endpoints in the large PREEMPT CRC trial.

The blood-based test was developed using case-control samples from patients with CRC and advanced adenomas and leveraged multiple technologies including genomics, proteomics, transcriptomics, and other omics, according to lead investigator Dr Aasma Shaukat from the New York University Grossman School of Medicine in New York, US.

The data, Shaukat continued, were then fed into a machine learning algorithm to identify the signatures that were associated with advanced colorectal neoplasia (ie, CRC and advanced adenoma). Clinical validation was performed for two classifiers—one was a genomics platform, and another was a multiomics platform that combines genomics with a protein classification model. All endpoints were powered at 96 percent or greater.

In the prespecified analysis, the genomics platform had a sensitivity of 79.2 percent (95 percent confidence interval [CI], 68.4–86.9) for detecting CRC, a specificity of 91.5 percent (95 percent CI, 91.2–91.9) for detecting nonadvanced colorectal neoplasia (non-ACN), a negative predictive value (NPV) of 90.8 percent (95 percent CI, 90.7–90.9) for non-ACN, and positive predictive value (PPV) of 15.5 percent (95 percent CI, 14.2–16.8) for ACN. [Shaukat A, et al, DDW 2024]

For detecting CRC by stage, the sensitivity was 57.1 percent for stage I, 100 percent for stage II, 82.4 percent for stage III, and 100 percent for stage IV.

“All CRCs were staged, except for one case that was detected by the blood test,” Shaukat noted.

Looking at the secondary endpoint, the sensitivity for detecting advanced adenomas was 12.5 percent (95 percent CI, 11.3–13.8), which did not meet the prespecified threshold and missed it by 0.2 percent, she added. The sensitivity for advanced adenomas with high-grade dysplasia or carcinoma in situ was 29.1 percent (95 percent CI, 21.4–38.2).

Results for the multiomics platform were similar. The sensitivity for detecting CRC was 77.8 percent, the specificity for detecting non-ACN was 93.7 percent (95 percent CI, 93.4–94.0), the NPV for non-ACN was 90.8 percent (95 percent CI, 90.7–91.0), and the PPV for ACN was 17.7 percent (95 percent CI, 16.1–19.4).

In a post hoc analysis, Shaukat and colleagues adjusted the specificity for detecting non-ACN to the Centers for Medicare and Medicaid Services (CMS) National Coverage Determination (NCD) cutoff of 90 percent, given that the genomics and multiomics platforms had different sensitivity and specificity, making it difficult to compare them directly.

For the genomics platform, there was a corresponding increase in the sensitivity for detecting advanced adenomas (14.8 percent) while maintaining sensitivity for CRC. For the multiomics platform, on the other hand, sensitivity increased both for detecting CRC (80.6 percent) and advanced adenomas (16.4 percent).

“While physicians have screening tools available for colorectal cancer, the associated inconvenience and discomfort of these methods deter many from getting recommended screenings,” according to Shaukat, adding that only 59 percent of eligible individuals aged 45 years and older in the US are up to date with CRC screening as of 2021.

“By providing a more convenient option, the blood-based test for the early detection of CRC has the potential to boost screening adherence rates and improve accessibility for people in all communities,” she said.

As the largest study to evaluate a blood-based CRC screening platform, PREEMPT CRC included 27,010 participants (mean age 58.1 years, 55.8 percent female, 73 percent White) who were at average risk of CRC; had no personal history of CRC, colorectal adenoma, or IBD; had no family history of CRC or hereditary gastrointestinal cancer syndromes; and were screen-eligible.

All participants underwent a blood draw and then a standard-of-care screening colonoscopy. Blood samples underwent comprehensive analysis, including histopathological evaluation, and the data were validated clinically. “This was all overseen by a data quality oversight committee,” Shaukat said.

PREEMPT CRC had several limitations. The study was not designed to assess the impact of the blood-based screening platform on CRC incidence and mortality, to compare adherence and effectiveness of different screening modalities, and to test screening interval.

“Work is currently ongoing to enhance the clinical performance of the test,” Shaukat said.