Real-World Data on Liquid Biopsy Use in Non-Small Cell Lung Cancer in the Community Setting

Nonsmall cell lung cancer (NSCLC), the most common type of lung cancer, remains an important contributor to global burden of disease and leading cause of cancer mortality.^[1] The introduction of precision medicine with targeted therapies directed toward driver genetic aberrations has changed the treatment paradigm in NSCLC and has suggested possible benefits small observational studies, when compared with standard chemotherapy.^[2–4] The challenge remains in effectively extracting tumor DNA and expeditiously detecting actionable aberrations (AA) to guide treatment decisions.

Tissue biopsy is considered the gold standard for harvesting tumor DNA. Recent advancements in cancer research have introduced liquid biopsy as a surrogate modality.^[5–8] Liquid biopsy describes the extraction of circulating tumor DNA (ctDNA) from patients' plasma and analyzing it using different techniques, namely, next-generation sequencing (NGS), to identify AA that impart tumorigeneses or install resistance to therapy. Liquid biopsy has been shown to facilitate diagnosis, prognosis, disease monitoring, and targeted therapy in several studies.^[9–12] The TRACERx study^[6] detected cancerous ctDNA in plasma as early as 6 months before cancer diagnosis, making liquid biopsy a potential screening modality in patients who are yet to show clinical or radiologic signs of disease. Furthermore, as a minimally invasive modality, liquid biopsy carries substantially lower risk to patients when compared with tissue biopsy, and liquid biopsy has potential to serve as a surrogate option for AA detection when tissue biopsy is contraindicated or not performed.^[8,11]

Conversely, liquid biopsy has important disadvantages, namely test sensitivity that reaches 85% at most^[12] and cost. CtDNA is found in rare concentrations in the plasma and false-negative results are prevalent, particularly when tumor burden is low. Liquid biopsy carries high cost and variable insurance coverage, requiring careful patient selection.

In this retrospective data analysis, we included 279 patients with NSCLC who had liquid biopsy ordered between November 2014 and July 2019, as part of routine clinical management. The study was approved by Mount Sinai Medical Center institutional review board committee. Plasma NGS was performed using the Guardant360 platform, and tissue NGS was performed using two different vendors, Caris Life Sciences and Foundation Medicine, chosen at the discretion of the physician. Both modalities targeted the following AAs, fusions and translocations: EGFR, MET, BRCA1, BRCA2, KRAS, RET, ERBB2, BRAF, ROS1, ALK, and NTRK2.

A 10-mL blood samples were collected in Streck tubes and shipped to Guardant Health. The plasma was isolated by centrifugation of blood by 1600g for 10 minutes at 48°C. Using the QIAamp circulating nucleic acid kit (Qiagen), ctDNA was then extracted, concentrated, and size selected using Agencourt AMPure XP beads (Beckman Coulter), and quantified by Qubit fluorometer (Life Technologies).

After isolation and oligo-nucleotide indexing, ctDNA digital sequencing library was performed. The library was amplified and enriched for target genes. After sequencing, algorithmic reconstruction of the digitized sequencing signals were used to reconstruct and subsequently stored in a secure database for analysis.

Simple descriptive statistics was used to describe findings. The results of the genetic aberrations were reported in the form of a bar graph. Microsoft Excel for Mac, version 16.42, was used for statistical calculation.

Over a period of 5 years, 337 plasma samples taken from 279 patients were sent for ctDNA NGS testing. The median age at diagnosis was 73 years (range 36–93 y, SD 10.4 y), 51% were men and 49% were women. The majority were White or Caucasian (80% versus 8% Black or African American versus 12% Multiracial or unknown race) and had a history of smoking (79%).

Excluding synonymous mutations and variants of unknown significance, 705 somatic aberrations were detected in 239 of 337 (71%) of plasma samples and in 201 of 279 (72%) of patients. Of the aberrations, 254 of 705 (36%) were identified as therapeutically actionable. These were detected in 139 of 239 (58%) of samples, taken from 106 of 201 (53%) of patients.

An overall 106 patients had 254 AA identified by plasma NGS testing through liquid biopsy. Commonly detected AA were EGFR (n = 127, 50%), KRAS (n = 61, 24%), BRAF (n = 24, 9.5%), MET (n = 23, 9%), RET (n = 5, 1.9%), BRCA1 (n = 5, 1.9%), BRCA2 (n = 4, 1.6%), ERBB2 (n = 4, 1.6%), and ALK (n = 1, 0.4%). None of the patients had aberrations in ROS1 or NTRK2. One patient had microsatellite instability-high cells (see Figure 1).

An overall 62 of 106 of patients had tissue biopsy performed in addition to their liquid biopsy, with 49 AA detected in 45 (73%) of the patients. (see Figure 2). Commonly detected AA were EGFR (n = 28, 57%), KRAS (n = 10, 20%), BRCA2 (n = 3, 6%), MET (n = 3, 6%), BRAF (n = 2, 4%), BRCA1 (n = 1, 2%), RET (n = 1, 2%) and ROS1 (n = 1, 2%) (see Figure 3).

The three commonest co-occurring aberrations detected on liquid biopsy were EGFR and BRAF (n = 14), EGFR and MET (n = 13), and MET and BRAF (n = 10). The commonest mutually exclusive aberrations were KRAS and EGFR. In 44 of 106 (42%) of patients, tissue NGS was not performed because of insufficient tissue or technically difficult biopsy and any precision treatment was determined by their liquid biopsy NGS results.

We evaluated whether consistent AAs were found on liquid and tissue biopsies in same patients who underwent both modalities (Figure 3). Concordant AAs were found in 39 of 45 (87%) of patients who underwent both tissues and plasma testing and was demonstrated in EGFR (n = 25, 64%), KRAS (n = 11, 28%), BRAF (n = 2, 5%), and MET (n = 1, 3%). In the remaining 6 of 45 (13%) of patients, tissue NGS was discordant from their plasma NGS results.

In this retrospective study, we present real-world data from the community setting, assessing AA detection rates and genomic patterns of ctDNA between liquid biopsy and tissue biopsy genotyping modalities. Our study shows that integrating liquid biopsy into the clinical practice of patients with NSCLC increased AA detection rate in 44 patients who had no tissue NGS testing and no other means of assessing the genomic makeup of their tumor. In those patients, plasma sampling via liquid biopsy offered additional actionable targets and increased their potential therapeutic arsenal. Neither treatment assignment nor survival analysis was evaluated in this study; however, a literature review shows no consensus on survival benefits upon routine use of broad-based genomic sequencing in large populations of patients with NSCLC in the community.^[13–15] The reasons for the lack of consensus are multifactorial, ranging from difficulties in identifying AA in fairly low tumor burden, the scarcity of robust response to treatments, access to treatment and to clinical trials, and variable insurance coverage.

In congruency with literature, our study demonstrated that EGFR and KRAS were the most commonly identified AAs on both plasma and tissue NGS testing. Mutations in EGFR identified by liquid biopsy testing were most frequently co-occurring with aberrations in BRAF and MET and were mutually exclusive with KRAS. These data are consistent with previously published literature.^[14–17] An overall concordance of 87% was observed in AA detection rate when comparing between plasma and tissue NGS.

This study is limited by its retrospective nature. The information regarding tissue NGS and mortality was captured through review of the electronic medical records, which may be incomplete. The liquid biopsy information was obtained from the Guardant360 database. Detection bias may be introduced if providers choose to order liquid biopsies in selected cases depending on specific characteristics (eg, insurance coverage and nonsmoking status). Evaluating whether clinicians had access to the genomic NGS data and offered patients informed therapy was not routinely documented, limiting the understanding of clinician decision-making with broad-based genomic-sequencing results.

Although tissue biopsy remains the gold standard in cancer management, our experience advocates for using liquid biopsy alongside tissue biopsy, as means of increasing AA detection rate and enhancing the delivery of precision medicine to patients with NSCLC. Certainly, more powered studies are needed to assess whether incremental benefit exists between tissue NGS and liquid biopsy; however, our results cautiously display a role for using liquid biopsy as part of routine clinical management.

The authors thank Leylah Dubrowsky, PhD (Guardant Health, Genetics Department) for assistance with data analysis and interpretation. She did not receive compensation for her role in this study.