Development of a multi-modal liquid biopsy for the early detection of lung cancer

Background Most lung cancer (LC) patients present with advanced, incurable disease. Early LC detection and surgery brings significant chance of cure.

In Manchester, we pioneered community-based Low Dose Computed Tomography (LDCT) screening in mobile scanners within socially deprived areas of Greater Manchester (GM) where LC is disproportionally high; 80% of detected cancers were stage I-II. In parallel, we collected blood samples. LDCT limitations are false-positives, overdiagnosis, high cost and radiation exposure.

A reliable liquid biopsy identifying early LCs could: increase compliance with, frequency of, and triage for screening; refine LDCT detection and potentially distinguish aggressive from indolent nodules.

We continue to collect blood samples in GM and Leeds community screening studies with comprehensive clinical follow-up, generating a unique biobank for biomarker research, including risk-matched, cancer negative controls.

Samples are already banked from 1,367 participants, 117 participants with a positive scan and 75 participants re-called for monitoring of indeterminate pulmonary nodules (IPNs). Within 4-years, samples from >250 LDCT-positive screening participants with stage I-II LC will be banked. Diagnosis of stage I LC using circulating tumour (ct)DNA alone is insufficiently sensitive.

Our pilot data suggests rare circulating cell (CTC) candidate number combined with our new ctDNA methylation assay could improve sensitivity. Aims To develop a robust, multi-modal blood test, sensitive and specific enough to detect stage I-II LCs.

To perform a pilot sub-study investigating relationships between liquid biopsy outputs and LDCT scan radiomics features.

Methods Our multidisciplinary team will apply High-Definition Single Cell analysis (multi-phenotype, rare cell enumeration and copy number profiles), ctDNA methylation (bisulphite-free preserving DNA fragments) and LUNG-CLiP ctDNA mutation analysis (most sensitive method published) to 400 participants’ blood samples blinded for outcome (200 stage I-II LC, 200 risk matched LDCT negative controls, 5-years cancer free).

We will determine and compare sensitivity and specificity of single vs multi-modality liquid biopsy outputs using sophisticated machine learning approaches.

Our exploratory radiomics sub-study will include serial scans of 50 cases with IPNs correlating liquid biopsy outputs and radiomics features associated with malignancy on follow up.

Utility of Results and Next Steps We will submit a second program proposal in year 2 to expand and power a study to define synergies between radiomics and liquid biopsy outputs.

If our multi-modal blood test performance is sufficient, we will design appropriate cohort studies to evaluate its utility to broaden screening eligibility, optimise screening interval, manage IPNs and reduce screening related harm (reducing false positives).