Abstract
BACKGROUND: Existing staging approaches fall short in precisely forecasting the likelihood of recurrence and survival outcomes among patients undergoing surgery for early-stage breast cancer (EBC). Our study hypothesized that multivariate long non-coding RNA (lncRNA) expression profiles, when systematically integrated into a composite model, may synergistically refine postoperative risk categorization and enhance prognostic forecasting precision in this patient cohort. METHODS: For the discovery set, lncRNA expression profiling associated with breast cancer progression was discovered by analyzing the differential expression profiles in three paired primary breast cancer tumor tissues and liver metastases. We found 12 distinctially expressed lncRNAs. A total of 400 patients were consecutively recruited and randomized to either training group or validation group. We first confirmed the expression of these lncRNAs using qRT-PCR. Subsequently, employing the LASSO Cox regression model with five lncRNA features as covariates, we constructed a five-lncRNA signature. We then validated this signature in an independent cohort to assess its prognostic and predictive capabilities in disease-free survival (DFS) duration. RESULTS: We constructed a classifier using the LASSO model, incorporating five specific lncRNAs: CBR3-AS1, HNF4A-AS1, LINC00622, LINC00993 and LINC00342. Utilizing this tool, we successfully stratified patients into two distinct categories: high- and low-risk groups. Significant differences were observed in both DFS and overall survival (OS) between the two groups. Within the initial patient cohort, significant differences of 5-year DFS was observed across high- and low-risk group (61.1% vs. 92.2%, HR 6.3, 95% CI 3.5-11.6; P < 0.001). The 5-year DFS rate was 72.9% and 85.4% for high- and low-risk group respectively in validation cohort (HR 2.6, 95% CI: 1.5-4.5; P = 0.001). The 5-lncRNA signature emerged as an independent prognostic indicator, demonstrating superior prognostic value compared to conventional clinicopathological risk factors. CONCLUSIONS: The integrated model combining 5-lncRNA molecular signature with clinical parameters demonstrates significant prognostic stratification capacity and therapeutic decision-making value in EBC management. It may help patients consult and personalize disease management.