Treatment of Non–Small-Cell Lung Cancer Based on Circulating Cell-Free DNA and Impact of Variation Allele Frequency

Abstract Background Next-generation sequencing of circulating cell-free DNA (cfDNA) can identify sensitizing and resistance mutations in non–small-cell lung cancer (NSCLC). cfDNA is helpful when tissue is insufficient for genomic testing or repeat biopsy is not feasible or poses unacceptable risk. Here we report the experience of cfDNA testing at the time of diagnosis and how this intervention can help avoid further invasive interventions, how it can be used to determine initiation of therapy, and how variation allele frequency of the somatic alteration affects response to subsequent treatment. Patients and Methods This is a single-institution retrospective study of patients with advanced NSCLC who had cfDNA from plasma tested using the Guardant360 panel, which identifies somatic genomic alterations by massive parallel sequencing of target genes. An institutional Clinical Laboratory Improvement Amendments tissue panel using fluorescence in situ hybridization (for MET, RET, ROS1, and ALK) and next-generation sequencing for selected genes was used for tissue analysis. Actionable mutations are those with US Food and Drug Administration–approved targeted therapies (EGFR, ALK, ROS, BRAF, NTRK fusions) or therapies soon to be approved (RET fusions and MET amplifications, or MET exon 14 skipping mutation). Results A total of 163 blood samples from 143 patients were evaluated, 82 at diagnosis and 81 at disease progression. A total of 94 cases had tissue and cfDNA testing performed within 12 weeks of each other. Seventy-six (81%) of 94 cases were concordant, of which 22 cases were concordantly positive and 54 concordantly negative. Eighteen (19%) of 94 cases were discordant, of which 11 had negative blood and positive tissue results, and 7 had positive blood and negative tissue results. cfDNA testing had a sensitivity of 67% (95% confidence interval [CI], 51%, 83%), specificity of 89% (95% CI, 81%, 97%), negative predictive value of 83% (95% CI, 74%, 92%), and positive predictive value of 76% (95% CI, 60%, 91%). Nineteen (21%) of 82 cfDNA samples analyzed at diagnosis had actionable mutations identified (4 EGFR exon 19 deletion, 2 EGFR exon 21 L858R, 2 EGFR L861Q, 1 L861R, 4 EML4-ALK fusion, 2 CD74-ROS1 fusion, 2 MET exon 14 skipping mutation, 2 KIF5B-RET fusion). Of the 82 patients with cfDNA testing performed at the time of diagnosis, 8 patients (10%) initiated targeted therapy on the basis of cfDNA results only, with 6 patients experiencing partial response, 1 patient complete response, and 1 patient stable disease. The response rate for patients who initiated targeted therapies on the basis of cfDNA only at diagnosis was 88%. Variant allele frequency had no impact on response. Conclusions Initiation of targeted therapy for advanced NSCLC was feasible based only on identification of actionable mutations by cfDNA testing in 9% of the cases for which tissue diagnosis could not be obtained. Actionable targets were identified by cfDNA in 20% of the samples sent at diagnosis. A substantial number of patients benefited from cfDNA testing at initial diagnosis because it identified actionable mutations that led to appropriate targeted treatments.