Isolation and Characterization of Circulating Biomarkers to Predict Patient Outcomes in Late-Stage Non-Small Cell Lung Cancer

Abstract

This work hypothesizes that the molecular characterization of extracellular vesicles (EVs) and circulating tumor cells (CTCs), biomarkers found in peripheral blood draws, can be used to monitor, and predict, treatment efficacy and patient outcomes in lung cancer. First, microfluidic EV isolation technologies were developed, and EV protein characterization assays were adapted for use with these technologies. Isolation strategies included (1) Annexin V-phosphatidylserine binding to capture tumor-derived EVs on the device surface and (2) isolating natural killer (NK) cell-derived EVs by first capturing NK cells followed by on-chip EV biogenesis. Western blot protein analyses were optimized for these technologies to verify the presence of EV-specific proteins (CD9, FLOT1, HLA-C) along with cell type-specific proteins (CD56). Finally, the DICE device was developed to both isolate EVs and characterize select EV-Proteins (CD9, Vimentin, EGFR) on-chip. Building off the developed EV characterization methods, EGFR mutations were detected longitudinally in both EV-RNA and EV-Protein from 10 metastatic NSCLC patients. For these patients, identifying the presence of sensitizing (exon 19 del, L858R) and resistance (T790M) EGFR mutations informs sensitivity to tyrosine kinase inhibitors. We demonstrated the presence of exon 19 del and L858R mutations within EV-Protein, marking the first study that demonstrates the presence of these mutations in patient-derived EVs. At the EV-RNA level, exon 19 del mutations were detected in 88% (n=7/8) patients and an increase in mutation burden mirrored disease progression and a decrease mirrored stable disease in 100% (n=5/5) of patients and in 86% (n=12/14) samples. We additionally profiled 2 patients for L858R and T790M mutations, however the detection was more modest at 60% (n=6/10) and 30% (n=3/10) samples, respectively. As such, EV-RNA exon 19 del mutation burden has the potential to inform treatment decisions in a subpopulation of metastatic NSCLC patients. In a cohort of 26 stage III NSCLC patients, CTCs were isolated from all 26 patients using the microfluidic graphene oxide (GO) chip at six timepoints during radiation and immunotherapy therapies. Significantly, it was found that having a decrease in CTCs of less than 75% between pre-treatment and week 4 of radiation therapy is predictive of significantly shorter progression free survival time, 7 months vs 21 months stable monitoring time (p=0.005, log-rank test). Additionally, evaluation of PD-L1 expression on the CTCs demonstrated that having a higher proportion of PD-L1 CTCs before starting treatment was a potential indicator of metastatic potential (p=0.057, log-rank test). Finally, microarray mRNA analysis demonstrated that CTCs develop a more aggressive, proliferative phenotype during radiation treatment. Finally, to assess the heterogeneity of EV biogenesis for applications in NK-EV therapeutics, the droplet microfluidic CellMag-CARWash system was adapted to isolate single NK-92MI cells bound with anti-CD56 Dynabeads. It was found that the CellMag-CARWash isolates cells that have 3+ Dynabeads attached, with an overall efficiency of 58% ± 7 (n=4), while calculations indicate that 1.8 beads should be needed to isolate cells. From a mixed cell population, the CellMag-CARWash isolated NK cells with 95% ± 2 (n=4) purity, achieving an isolation efficiency of 42% ± 14 (n=3). Prolonged droplet stability was demonstrated, and cell viability is 50% after 24 hours in droplets. Taken together, these technological advancements represent necessary developments to move liquid biopsies from the lab to the clinic. These novel isolation and characterization strategies will need to continue to be tested in pilot cohorts and validated in larger cohorts.