Detecting actionable mutations from matched plasma-based versus tissue next-generation sequencing in advanced non-small cell lung cancer: a retrospective single centre analysis on site

Abstract

Background: Liquid biopsies (LB) are used increasingly to detect actionable mutations in patients newly diagnosed with advanced non-small cell lung cancer (aNSCLC), though tissue biopsies (TB) still remain the gold standard. The value of systematically combining LB and TB next-generation sequencing (NGS) for genomic profiling in these patients remains controversial.

Methods: This single-centre retrospective study included 102 matched TB and LB samples collected from aNSCLC patients at diagnosis. Four circulating free DNA (cfDNA)-based NGS assays (1-4) were compared on site for performance and concordance with TB to detect ESMO Scale for Clinical Actionability of molecular Targets (ESCAT) I/II. Additionally, cfDNA droplet digital PCR methylation (ddPCR-met) testing estimated the tumour fraction to refine the interpretation of wild-type (WT) results.

Results: Out of 102 patients, 13% had stage IIIB disease, and 11% presented with brain-only metastases. Adenocarcinoma was the predominant subtype (84%). Ninety LB samples yielded interpretable results across the four assays. Positive percent agreement with TB ranged from 56% (assay 2) to 79% (assay 4), with high concordance, particularly for single-nucleotide variants (SNVs). Hybrid capture-based assays (3 and 4) detected eight and seven gene fusions, respectively, while amplicon-based assays (1 and 2) detected only two each. Assay 3 only identified 12 MET amplifications, five of which were confirmed by fluorescence in situ hybridisation (FISH) but were missed by TB-based NGS. Five out of six negative cfDNA samples with ddPCR-met testing were WT across all assays. The plasma-first approach added incremental value, up to 21% (assay 3). Amplicon-based assays were faster and required less input of DNA for analysis. Patients with stage IIIB or brain-only metastases were significantly more likely to have negative/low levels of cfDNA ddPCR-met.

Conclusions: LB-based NGS demonstrated high concordance with TB in newly diagnosed aNSCLC, particularly for detection of SNV. Hybrid capture assays showed superior performance in identifying gene fusions and MET amplifications. The incremental value of a plasma-first strategy was limited in this real-life study. Thus, LB-based NGS on site should be seen as a complementary tool to TB-based NGS or an alternative when tissue samples are unavailable. Additionally, cfDNA methylation analysis enhances diagnostic accuracy in specific cases.

Keywords: Actionable genomic alterations; CtDNA; Liquid biopsy; Methylation; NGS; Non-small cell lung carcinoma; Precision medicine.

Fig. 1

Flowchart depicting all assays used on patients’ samples (TB and LB samples)

Fig. 2

Comparison between ESCAT I/II alterations detected in TB and matched LB samples using amplicon-based cfDNA assays. Assay 1 (n = 102): (a) Venn diagram depicting the proportion of actionable alterations detected in TB and LB, (b) Bar plot representation of alterations distribution in TB only, LB only and both (concordance), (c) Frequency of activable alterations detected by reflex testing using TB or LB first. Assay 2 (n = 98): (d) Venn diagram depicting proportion of actionable alterations detected in TB and LB, (e) Bar plot representation of alterations distribution in TB only, LB only and both (concordance), (f) Frequency of activable alterations detected by reflex testing using first TB or LB

Fig. 3

Comparison between ESCAT I/II alterations detected with TB and matched LB samples using hybrid capture-based cfDNA assays. Assay 3 (n = 102): (a) Venn diagram depicting the proportion of actionable alterations detected in TB and LB, (b) Bar plot representation of alterations distribution in TB only, LB only and both (concordance), (c) Frequency of activable alterations detected by reflex testing using TB or LB first. Assay 4 (n = 90): (d) Venn diagram depicting the proportion of actionable alterations detected in TB and LB, (e) Bar plot representation of the distribution of alterations in TB only, LB only and both (concordance), (f) Frequency of activable alterations detected by reflex testing using first TB or LB

Fig. 4

Comparison between ESCAT I/II alterations detected in plasma using hybrid capture-based assays and amplicon-based assays: (a) Correlation matrix depicting pairwise relationships between the assays for ESCAT I/II mutations. The strength and direction of correlations are represented by the colour intensity and hue, respectively, according to the provided scale bar as well as the numerical value provided on the left. (b) Bar plot representation of alterations detected specifically in one, two or three assays for both amplicon and hybrid capture-based methods. (c) Correlation of VAF for ESCATI/II mutations across all panels. The coefficient of determination (R) as well as p-value are shown

Fig. 5

(a) Comparison of the detection of ESCAT I/II mutations across all panels and tissue testing. Baseline characteristics as well as the results from ddPCR for both of the two markers (HOXB4 and MROH6) are shown. Concordance of panels with tissue testing is shown in percent to the left. Tissue was used as reference and percent of patients with detected mutations are shown for each panel. (b) Oncoprint of MET amplifications. Baseline characteristics as well as results from ddPCR for both of the two markers (HOXB4 and MROH6) are shown. FISH neg: Sample was negative for MET amplification in tissue FISH testing. FISH NI: FISH sample was non informative in tissue MET amplification testing