Development of a diagnostic model using the circulating long noncoding RNAs LINC00857 and KLHDC7B-DT in lung...

Lung adenocarcinoma (LUAD), a deadly malignancy, lacks clinically validated and reliable biomarkers. Long noncoding RNAs (lncRNAs) are involved in various physiological and pathological cancer processes. However, no clear molecular diagnostic markers has been identified in LUAD. Therefore, this study aims to identify novel LUAD-associated lncRNAs and develop a robust, non-invasive diagnostic model to improve the early identification of this disease.

We included a total of 646 patients, who were divided into a training set (n=388), a validation set (n=258). We here obtained LUAD-related lncRNAs from The Cancer Genome Atlas (TCGA) database and analyzed them by machine learning, including least absolute shrinkage and selection operator (LASSO) regression, support vector machine-recursive feature elimination (SVM-RFE), and random forest (RF), in conjunction with weighted gene co-expression network analysis (WGCNA), to identify differential lncRNAs associated with LUAD. Binary logistic regression model and receiver operating characteristic (ROC) curves were used to assess the diagnostic performance of the characterized genes. In addition, the diagnostic performance of characterized lncRNAs was compared with carcinoembryonic antigen (CEA) in LUAD plasma.

We successfully identified two characterized lncRNAs LINC00857 and Kelch domain containing 7B divergent transcript (KLHDC7B-DT) integrated into a lncRNA diagnostic model. The model performed superiorly in distinguishing LUADs from controls in several different cohorts, particularly in early stage I/II cancer. Furthermore, LINC00857 and KLHDC7B-DT showed better diagnostic efficacy than the existing clinical serum marker CEA.

This study demonstrated the potential of LINC00857 and KLHDC7B-DT as noninvasive biomarkers for the early detection of LUAD.