Plasma-derived ctDNA provides easily accessible minimally invasive material for molecular testing. There is increasing evidence that underlying genotypes including microsatellite instability (MSI-H), tumor mutation burden, PD-L1 expression or amplification and other molecular changes can be associated with distinct response patterns to iCPI. We isolated ctDNA from plasma samples collected from patients with advanced cancers before starting therapy with iCPI and tested them using a targeted NGS panel (Tempus xF) at average depths of 20,000x(raw reads)/5,000x(unique reads) to detect single nucleotide variants, insertions, deletions in 105 genes; copy number amplifications in 6 genes; copy number deletions in 2 genes; and gene rearrangements (translocations) in 7 genes spanning ~0.3 Mb of genomic space. Variants of unknown significance were excluded. MSI-H status was also tested. Results were compared to molecular testing of archival tumor tissue to assess agreement with ctDNA, and to treatment outcomes. Twenty patients with advanced cancers (breast 5, colorectal 2, cholangiocarcinoma 2, uveal melanoma 2, other 9) dispositioned to start therapy with iCPI (PD-L1/PD1 antibody, n = 18; LAG3 antibody, n = 2) were enrolled. Of 20 patients, 19 had ctDNA results, which passed quality control. The rates of complete detection of all molecular alterations present in archival tumor tissue was 47% (n = 9) and of partial (key driver alterations) detection 32% (n = 6) with a total complete or partial detection rate of 79%. Individual detection rates in ctDNA compared to tissue per patient ranged from 0%-100% (median, 78% in 19 patients). Seventeen patients had a total of 38 alterations (TP53, n = 14; NF1, n = 5; PIK3CA, n = 2; KRAS, n = 2; ESR1, n = 2; FBXW7, n = 2; others [including PD-L1 amplification], n = 11) in ctDNA not reported in archival tissue. Patients with lower amounts of ctDNA (<median variant allele frequency) had longer median time-to-treatment failure (TTF) on iCPI (3.6 vs. 1.9 months, P = 0.018) and longer overall survival (OS; not reached vs. 4.7 months, P = 0.015) compared to patients with higher amounts of ctDNA. In conclusion, targeted NGS testing of pretreatment ctDNA samples demonstrated a high rate of detection for driver alterations previously noted in archival tumor tissue. Higher quantity of ctDNA was associated with shorter TTF on iCPI and shorter OS compared to lower quantity of ctDNA.

VIEW THE PUBLICATION