INVESTIGATING PLASMA MICRORNA AND PROTEIN MARKERS FOR EARLY DETECTION OF LUNG CANCER

Abstract

Background: Lung cancer (LC) is the leading cause of cancer-related death worldwide and in Canada. Current diagnostic strategies are limited by challenges such as poor accessibility, invasiveness, low sensitivity, and limited specificity. microRNAs (miRNAs, miRs) are small RNA molecules that regulate most physiological processes and make excellent biomarkers due to their disease specificity and stability in blood. Similarly, some proteins are also disease-specific and stable in plasma, making them promising targets for blood-based cancer detection.

Hypothesis: Plasma-derived miRNA and protein markers can be used to distinguish among patients with different subtypes of LC and patients with non-neoplastic lung disease (NNLD).

Aim 1: Quantitate the abundance of candidate markers in plasma samples from study cohort. Aim 2: Investigate associations between candidate marker abundance and clinicopathologic characteristics within LC subtype groups. Aim 3: Construct and validate a Fuzzy Inference System (FIS) classifier to discriminate among patients with different LC subtypes and non-neoplastic disease.

Methods: Plasma samples were obtained from 132 patients, including 41 with lung neuroendocrine neoplasms (NEN), 30 with lung adenocarcinoma (LUAD), 30 with lung squamous cell carcinoma (LUSC), and 31 with NNLD. Candidate miRNA and protein marker abundances were quantitated. The results were tested for associations with clinicopathological variables and used to develop a FIS classifier. Model performance was evaluated using an independent validation cohort to assess generalizability.

Results: Two circulating miRNAs and one protein marker differed significantly among diagnostic groups. miR-375 was elevated in lung NEN relative to LUAD, LUSC, and NNLD (KW: p < 0.0001). miR-205 was elevated in LUSC relative to LUAD and NNLD (KW: p = 0.003 and p = 0.002, respectively). Carcinoembryonic antigen (CEA) was elevated in NEN relative to LUSC and NNLD (KW: p = 0.002 and p < 0.0001, respectively). An FIS classifier distinguishing lung NEN from the remaining diagnostic groups achieved 90.7% accuracy in the training cohort and 95.3% accuracy in the independent validation cohort.

Significance: This project lays the foundation for establishing an inexpensive and minimally invasive liquid biopsy test to reduce time-to-diagnosis of LC subtypes and enable diagnosis for those with limited access to care.