Journal: Cancer prevention research (Philadelphia, Pa.)
This study evaluated a noninvasive multimodal stool RNA test (mm-stRNA) that combines five human mRNA biomarkers with a fecal immunochemical test (FIT) via a machine learning–based algorithm to detect colorectal cancer (CRC) and advanced precancerous lesions (APLs).
In the eAArly DETECT multicenter U.S. study, 265 participants (34 CRC, 68 APL, 163 controls) provided stool samples. RNA was extracted from stabilized stool for quantification of the five mRNA markers, and FIT was run at two hemoglobin thresholds (5 and 17 µg Hb/g). The algorithm was developed in a stratified split-sample design and then locked and applied to the entire cohort.
Key performance:
- mm-stRNA test:
- CRC sensitivity: 97.1%
- APL sensitivity: 83.8%
- Specificity: 95.7%
- FIT alone:
- At 5 µg Hb/g: CRC sensitivity 76.5%, APL sensitivity 45.6%, specificity 84.0%
- At 17 µg Hb/g: CRC sensitivity 70.6%, APL sensitivity 36.8%, specificity 90.8%
The multimodal RNA+FIT approach showed substantially higher sensitivity for both CRC and APLs, with high specificity, compared with FIT alone. The authors emphasize that these promising results require confirmation in an independent, prospective cohort before clinical adoption.