A Multicenter Study to Assess EGFR Mutational Status in Plasma: Focus on an Optimized Workflow for Liquid Biopsy in a Clinical Setting

Abstract

A multicenter study was performed to determine an optimal workflow for liquid biopsy in a clinical setting. In total, 549 plasma samples from 234 non-small cell lung cancer (NSCLC) patients were collected. Epidermal Growth Factor Receptor (EGFR) circulating cell-free tumor DNA (ctDNA) mutational analysis was performed using digital droplet PCR (ddPCR). The influence of (pre-) analytical variables on ctDNA analysis was investigated. Sensitivity of ctDNA analysis was influenced by an interplay between increased plasma volume (p < 0.001) and short transit time (p = 0.018). Multistep, high-speed centrifugation both increased plasma generation (p < 0.001) and reduced genomic DNA (gDNA) contamination. Longer transit time increased the risk of hemolysis (p < 0.001) and low temperatures were shown to have a negative effect. Metastatic sites were found to be strongly associated with ctDNA detection (p < 0.001), as well as allele frequency (p = 0.034). Activating mutations were detected in a higher concentration and allele frequency compared to the T790M mutation (p = 0.003, and p = 0.002, respectively). Optimization of (pre-) analytical variables is key to successful ctDNA analysis. Sufficient plasma volumes without hemolysis or gDNA contamination can be achieved by using multistep, high-speed centrifugation, coupled with short transit time and temperature regulation. Metastatic site location influenced ctDNA detection. Finally, ctDNA levels might have further value in detecting resistance mechanisms.

Keywords: EGFR; ctDNA; ddPCR; liquid biopsy; non-small cell lung cancer (NSCLC).

Figure 1

Influence of metastases on detectable EGFR mutated ctDNA levels. (A) distribution of concentration (copies/mL); (B) distribution of allele frequency (%); Br: brain metastases; ET: extrathoracic metastases; AF (%): percentage of allele frequency; +: present; −: absent; median with 95% confidence interval (CI); * p < 0.05.

Figure 2

Comparison of ctDNA (activating mutations vs. T790M mutation) (n = 43). (A,B) concentration (copies/mL) of EGFR activating mutations (ex19del, L858R, L861Q, and G719X) versus T790M mutation detected in samples taken at radiological progression (n = 29) (A) and during follow up (n = 14) (B), respectively. The lowest detectable concentration was 190 and 40 copies/mL, with a median of 1950 and 360 copies/mL, respectively; (C,D) distribution of the allele frequencies detected in samples taken at radiological progression (n = 29) (C) and during follow up (n = 14), respectively. The lowest allele frequency detected was 0.0985 and 0.0012%, with a median of 2.530 and 0.0107%, respectively; * p < 0.05; ** p < 0.001.

Figure 3

Kinetics of mutated ctDNA. (A) initial increase and decrease until undetectable levels of mutated ctDNA after osimertinib therapy. A spike of the activating mutation was detected from week 13 until week 17. At week 15, brain metastases with penetration of the blood brain barrier were detected via brain magnetic resonance imaging (MRI); (B) the T790M mutation follows the pattern of the activating mutation throughout several treatment schemes, however, at lower allele frequencies; (C,D) the activating mutation was detected at very low allele frequencies. A blood sample in a few weeks’ time resulted in the detection of increased allele frequency of this mutation together with the T790M resistance mutation. This corresponded with progressive disease to first- and second-generation EGFR TKI therapy respectively; (E,F) increasing and extremely high AFs of the activating mutations were detected, respectively. Due to progressive disease under first generation EGFR TKI therapy, tissue biopsies were performed at the last time point, which confirmed the absence of the T790M mutation; AF (%): percentage of allele frequency; PFS: progression-free survival in weeks; : brain MRI.