A novel aptamer-based non-invasive test for lung cancer: A proof-of-concept.

Abstract

8033 Background: Lung cancer (LC) is the leading cause of cancer-related mortality, primarily due to late-stage diagnoses. Low-dose computed tomography (LDCT) screening lowers mortality rates by detecting LC at earlier stages, but the program is limited by high costs, capacity constraints, and low compliance. We describe the early-phase development of a test based on the APTASHAPE technology, designed as a cost-effective, scalable tool to pre-qualify individuals for LDCT screening. This technique uses RNA aptamers to analyze protein composition in lung cancer patients, identifying cancer-specific protein fingerprints across all stages. Variations in aptamer ratios reflect plasma protein composition, profiled through next-generation sequencing and machine learning. Methods: A discovery cohort of 24 LC patients (stage I+II, n=12; stage III+IV, n=12) and 24 individuals initially referred on suspicion of LC but ultimately diagnosed as non-cancer cases were analyzed. Additionally, a test cohort of 48 LC patients (stage I+II, n=24; stage III+IV, n=24) and 48 non-LC cases were analyzed. In four rounds of Systematic Evolution of Ligands by EXponential Enrichment (SELEX), a library of 10 15 2’-fluoro-protected RNA aptamers was incubated with a pool of plasma prepared from the LC patients in the discovery cohort to facilitate binding to the plasma proteins. Non-binders were removed, and bound aptamers were amplified by PCR. Following SELEX, linear regression identified the aptamers capturing LC-specific protein signatures. The selected aptamers were then applied to the test cohort and their ability to differentiate between LC and non-LC cases was evaluated using principal component analysis and receiver operating characteristic (ROC) curve. Results: In the discovery cohort, statistical analysis identified 13 aptamers whose binding to plasma proteins formed a cancer-specific fingerprint, able to discriminate participants with lung cancer from those without. We used this profile to predict LC in the test cohort and obtained an area under the curve (AUC) of 0.74 (95% confidence interval (CI) 0.62-0.87). Importantly, the discriminatory ability was equally effective for stage I+II and stage III+IV (AUC=0.71 (95% CI 0.58-0.84) and AUC=0.74 (95% CI 0.61-0.86), respectively). Conclusions: We present a proof-of-concept for a promising, cost-effective, and scalable technique for pre-qualifying individuals for LDCT screening. While still in the earliest stage of development, we anticipate that expanding the study population will improve the machine learning algorithm and markedly increase the AUC value. Importantly, this approach holds significant promise in detecting early-stage lung cancer -an area where blood-based technologies usually face substantial limitations. Ongoing optimizations aim to enhance its performance.