Abstract
BACKGROUND: Blood-based liquid biopsies are poised to revolutionize brain cancer diagnosis and monitoring. Recent studies have shown that circulating tumor DNA (ctDNA) can be detected in blood via droplet digital PCR (ddPCR), but due to low ctDNA burden (generally <0.5%) and difficulty multiplexing, timely detection and precise quantification of clinically relevant ctDNA changes using ddPCR remains challenging. To solve these problems, we have developed tail-ligated dsDNA recombinase polymerase amplification sequencing (TLDRseq), a sequencing-based assay using (1) targeted PCR, (2) intra-dsDNA 5’-to-3’ tail ligation, (3) isothermal hairpin amplification, (4) Oxford Nanopore sequencing, and (5) informatic error correction. We designed TLDRseq with the goal of developing an easy-to-perform, low error rate, multiplexable assay that is easily adoptable by clinical/academic labs even in low-resource settings. METHODS: We established a biobank of 165 serial blood samples for analysis from 22 pediatric patients with primary brain tumors (4-24 timepoints and 2-8 targets/patient) undergoing treatment. Serial samples from a cohort of five patients with known H3K27M+TP53/BRAF disease (n=32/64 samples/targets) were sequenced and compared to prior ddPCR results. RESULTS: TLDRseq generates ultra-deep (>100,000 reads/target) and highly sensitive results (75% for all targets (n=48/64) and 100% per timepoint). Results strongly correlated with ddPCR (r=.9957; P (two tailed) <.0001), while offering an exceptionally low limit of detection for H3.3K27M (~0.004% LOD). As an example of multiplex tumor evolution tracking, a patient with H3K27M/BRAFV600E DMG showed consistently positive BRAFV600E signal and negative H3K27M signal in blood. After targeted dabrafenib therapy BRAF signal was eventually lost (later confirmed lost at autopsy), while H3K27M signal spiked ~1-month prior to radiographic progression. Ongoing work will complete cohort analysis of 83 unique mutations over time with clinical/radiographic and paired tumor sequencing comparisons. CONCLUSIONS: This work represents exciting progress towards routine, low-cost, rapid, sensitive detection, and precise quantification of ctDNA in brain tumor plasma, and will increase our understanding of disease evolution under treatment.