Exosomal miRNAs have drawn increasing attention in the field of liquid biopsy owing to their high degree of cancer specificity and stability. This study aimed to establish an exosome-based liquid biopsy signature to predict molecular residual disease (MRD) in stage II and III colorectal cancer (CRC).

We analyzed 175 postoperative serum samples from stage II and III CRC patients from 2 independent institutions, performed quantitative reverse-transcription polymerase chain reaction, and developed and validated a risk-stratification model using logistic regression.

We first evaluated the expression of the 8 miRNAs identified in our previous tissue-based study in exosomes, and found that 5 of 8 were significantly downregulated in the recurrent group. A panel of 5 miRNAs robustly predicted recurrence after surgery in a training cohort (area under the curve [AUC]: 0.83). Subgroup analysis of the adjuvant chemotherapy (ACT) naïve group demonstrated that recurrence rates were 3.1% (1/32) in the low-risk group and 64.7% (11/17) in the high-risk group. We successfully validated the performance of this exosomal miRNA panel in an independent testing cohort (AUC: 0.77). Furthermore, when we developed our risk model, named Clinical Liquid Biopsy via Exosomes for Assessment of Molecular Residual Disease in Colorectal Cancer Score (CLEAR score), by adding tumor factor and stage to the panel, predictive accuracy was dramatically improved (AUC: 0.84).

We have developed an exosome-based liquid biopsy signature that enables robust detection of MRD in patients with stage II-III CRC. These findings could help facilitate more informed clinical decision-making for ACT.

NCT06654622.