关键词: 肿瘤, 中等长度非编码RNA, 高通量测序, 分子标志物, 精准诊疗

Abstract: Objective Precision surgery and individualized cancer treatment increasingly rely on molecular biomarkers that inform tumor classification, recurrence risk, prognosis, and therapeutic response. Mid-sized non-coding RNAs （mncRNAs） participate in ribosome biogenesis, RNA processing, and selective translation and have been associated with malignant progression and treatment response in multiple cancers, suggesting their potential as a source of novel tumor biomarkers. However, the complex secondary structures, nucleotide modifications, and diverse terminal configurations of mncRNAs make their detection highly susceptible to library-preparation bias, limiting systematic biomarker discovery. To address these challenges, we developed a random-primer DNA-adapter ligation sequencing method and compared it with two commonly used library-preparation strategies to evaluate its technical performance for mncRNA biomarker discovery in cancer. Methods HeLa cells were used as a tumor cell model and profiled using RNA-adapter ligation, poly（A）-tailing/template-switching, and the random-primer DNA-adapter ligation methods developed in this study. In the random-primer methods, an RNA adapter was ligated to the RNA 3′ end, followed by reverse transcription using a random-sequence-containing RT primer and SuperScript IV, and subsequent ligation of a DNA adapter to the cDNA. The resulting datasets were compared in terms of full-length RNA recovery, RNA length distribution, and RNA biotype composition. Results The three strategies exhibited distinct technical biases. RNA-adapter ligation yielded the lowest proportion of full-length RNA reads. Poly（A）-tailing/template-switching improved full-length RNA recovery to some extent but predominantly generated shorter RNA fragments. In contrast, the random-primer methods achieved the highest proportion of full-length RNA reads （approximately 40%） and substantially enhanced the detection of tRNA-related sequences. All three approaches detected diverse classes of mncRNAs, including rRNAs, snoRNAs, tRNAs, snRNAs, and miRNAs, while showing clear preferences for specific RNA biotypes, indicating that library preparation can directly influence the spectrum of candidate tumor biomarkers identified. Conclusion The random-primer methods developed in this study showed advantages in recovering full-length RNAs and achieving broader coverage of tRNAs/tsRNAs, providing a new technical option for discovery-oriented studies of structurally complex and extensively modified mncRNAs. The method was validated using HeLa cells and should be further evaluated in surgical tumor specimens and blood-based samples from different solid tumors to establish its cross-cancer applicability and its ability to identify biomarkers associated with molecular stratification, recurrence, prognosis, and therapeutic response.

Key words: cancer, mid-sized non-coding RNA, high-throughput sequencing, molecular biomarker, precision medicine