Abstract 1063: IsomiR relative proportion (IRP) within miRNA family as an efficient biomarker for noninvasive colorectal cancer detection

• Published in: Cancer research (Chicago, Ill.) Vol. 84; no. 6_Supplement; p. 1063
• Main Authors: Hu, Yeting, Lu, Wei, Liu, Chengcheng, Dai, Xiaozhuan, Xu, Jiasheng, Kong, Xiangxing, Zhu, Yunfeng, Ye, Yao, Chen, Lihao, An, Xiaolong, Wu, Fan, Ma, Shiqing, Wang, Nana, Xiao, Qian, Liu, Jun, Ding, Kefeng
• Format: Journal Article
• Language: English
• Published: 22-03-2024
• ISSN: 1538-7445; 1538-7445
• DOI: 10.1158/1538-7445.AM2024-1063

Background: Cell free miRNA has been widely used for disease diagnosis, particularly in cancers. The latest research has also demonstrated that isomiRs (isoform miRNAs) can serve as superior markers, displaying enhanced diagnostic performance in cancer diagnosis. However, the low detection sensitivities of miRNA/isomiRs in previous methods in plasma have limited their broad implementations for screening purpose. In this study, we propose a new isomiRs feature of isomiR relative proportion (IRP), which was validated with high performance in Colorectal Cancer (CRC) early diagnosis.
Methods: A total of 1036 participants were enrolled from the National Colorectal Cancer Cohort (NCRCC), and 836 participants were eligible for plasma collection, including 238 controls, 373 CRC patients, 41 inflammatory bowel disease (IBD), 5 intramucosal carcinoma (Tis) and 179 advanced precancerous lesions (APL). Total circulating small RNA was extracted, followed by library preparation using UMI-based approach. Subsequently, small RNA sequencing was conducted using DNB-seq, and bioinformatic analysis was performed thereafter. The samples were randomly assigned into training (49%), test (30%) and validation (21%) sets. Models based on isomiR relative proportion (IRP), isomiR relative expression (IRE) or both features were constructed by beam search and random forest algorithms in training and test sets, and results were verified in validation set.
Results: Differential IRP or IRE features were identified between cancer and controls. IRP identified 132 specific differential expressed isomiRs. Models constructed using either IRP or IRE features showed high performance (both AUC > 0.9). However, there was improvement in IRP feature model in the validation set (sensitivity of IRP: 0.923, sensitivity of IRE: 0.859) compared with IRE model. We then found that there may be complementarities between IRP and IRE features, and the combination model (34 IRP+IRE features) improved cancer detection performance (AUC: 0.953). Finally, the risk scores predicted by this combination model increased gradually across CRC progression (R=0.921), and the sensitivity of APL was 0.626.
Conclusion: We improved the small-RNA sequencing method and developed an IRP method, and present that IRP as a new feature not only demonstrate the ability of CRC early detection, but also revealed the CRC progressing process. Besides, the combination of IRP and IRE improved the performance of diagnosis. These results highlight the potential value of the IRP feature in early cancer detection.
Citation Format: Yeting Hu, Wei Lu, Chengcheng Liu, Xiaozhuan Dai, Jiasheng Xu, Xiangxing Kong, Yunfeng Zhu, Yao Ye, Lihao Chen, Xiaolong An, Fan Wu, Shiqing Ma, Nana Wang, Qian Xiao, Jun Liu, Kefeng Ding. IsomiR relative proportion (IRP) within miRNA family as an efficient biomarker for noninvasive colorectal cancer detection [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 1063.