Liquid biopsy testing is emerging as a valuable, minimally invasive diagnostic technique that overcomes the constraints of tissue availability. With shorter turnaround times and less pre-analytic processing than traditional tissue analysis, these assays offer a more streamlined path from sample to result. Because they capture genomic information across multiple tumor sites over time, liquid biopsy tests can also provide a more comprehensive, systemic view of a patient's cancer than a single tissue biopsy.
In a recent webinar supported by Illumina, Gang Zheng shared how the Mayo Clinic laboratory approaches liquid biopsy. He also presented data from a recent implementation of a liquid biopsy assay to enable comprehensive genomic profiling (CGP).
Key takeaways were:
An introduction to liquid biopsy and how it’s used at Mayo Clinic
Justification for bringing a large panel liquid biopsy research assay in house
Assay performance across key genomic variants
Comparative performance versus reference assays
Approaches to clonal hematopoiesis detection and filtering
Dr. Gang Zheng is Co-Director of Mayo Clinical Genomics Laboratory and Professor of the Department of Laboratory Medicine and Pathology.
Circulating tumor DNA (ctDNA) testing in oncology
Circulating cell-free DNA (cfDNA) is released into the bloodstream primarily through apoptosis and necrosis, with most cfDNA in healthy individuals originating from hematopoietic cells, particularly white blood cells.
Where cancer is present, cfDNA will contain some ctDNA fragments, which are typically 150–200 base pairs in length, corresponding to nucleosome-sized DNA fragments. Importantly, the proportion of ctDNA within total cfDNA can vary substantially, ranging from extremely low levels to relatively high tumor fractions depending on disease burden, tumor biology, and treatment status.
Because ctDNA has a short half-life measured in hours, it enables near real-time monitoring of tumor dynamics and treatment response.
The presentation highlighted the broad clinical utility of ctDNA for CGP. Current assays can detect single nucleotide variants (SNVs), small insertions and deletions, gene fusions, rearrangements, and copy number variations (CNVs), alongside biomarkers such as microsatellite instability (MSI) and tumor mutational burden (TMB).
ctDNA analysis can be used in different stages of cancer diagnostics, from early tumor detection to genomic analysis that guides treatment. During treatment, it can detect and monitor residual disease and treatment response.
“Overall, ctDNA testing provides a dynamic and real-time view of tumor biology supporting more informed clinical decision-making,” said Zheng.
What are the challenges with ctDNA testing?
In many patients, the ctDNA fraction represents less than 1 percent of total cfDNA, particularly in early-stage disease or tumors with low shedding rates. This presents a major analytical hurdle because the intrinsic error rate of conventional NGS platforms falls within a similar range – approximately 0.1 to 1 percent. As a result, distinguishing true low-frequency tumor variants from background sequencing noise can be technically difficult and may affect assay sensitivity.
Furthermore, the biology of ctDNA contributes to a key analytical challenge: clonal hematopoiesis can introduce mutations that may confound ctDNA results. For Zheng and the Mayo Clinic team, these biological and technical constraints formed a major aspect of the clinical feasibility study for in-house ctDNA testing.
Why implement CGP?
Larger sequencing panels increase the likelihood of identifying rare or emerging biomarkers that may be overlooked by smaller targeted assays. This broader approach also helps “future proof” testing strategies by reducing the need for repeat analyses as new clinically relevant biomarkers are identified.
Comprehensive profiling is increasingly central to precision oncology, supporting targeted therapy selection, off-label treatment opportunities, and enrollment in biomarker-driven clinical trials. Zheng emphasized that ctDNA assays may also provide a more complete picture of tumor heterogeneity because circulating DNA can originate from multiple tumor sites throughout the body.
“Adopting a ctDNA assay is not just about generating more data,” Zheng said. “It’s about enabling more informed and timely clinical decisions.”
The Clinic implemented the TruSight™ Oncology (TSO) 500 v2 assay* and carried out analytical validation testing.
Clinical feasibility testing
Reproducibility
For within-run precision testing, eight samples were analyzed in triplicate during the same sequencing run. Between-run reproducibility was assessed using approximately 14 samples tested across multiple independent runs. According to Zheng, the assay demonstrated “high concordance across all variant types,” including SNVs, insertions and deletions, CNVs, gene fusions, and MSI and TMB classification (Table 1).
Table 1. Precision
Limit of detection
Using reference mutation samples, the team found that assay performance remained strong at DNA inputs of 10 ng or greater, with high detection rates across SNVs, indels, CNVs, fusions, and MSI. TMB analysis required higher input levels for optimal performance.
At 10 ng input, the assay consistently detected variants at allele frequencies between 0.5 and 1 percent, while increasing input to 15 ng improved sensitivity further to approximately 0.25 to 0.5 percent for most variant classes. Zheng noted that TMB detection was more dependent on DNA quantity, with high TMB status detectable at roughly a 0.5 percent ctDNA fraction when using 15 ng input.
Accuracy
The team evaluated how well the TSO 500 V2 assay agreed with established reference methods: a commercial 500-plus CGP panel and an in-house 33-gene panel. Overall sensitivity, accuracy, and positive predictive value (PPV) was greater than 99 percent for SNVs, deletions and insertions, gene fusions, and MSI (Table 2).
TMBs showed slightly lower accuracy than other markers, and the scores were noticeably lower for CNVs. This is because CNVs are quantitative signals inferred from the reading depths, which require normalization and bioinformatics modeling, making them more sensitive to variability across different samples, especially at the low tumor fractions.
“Overall, these results show that the TSO 500 V2 assay can provide highly accurate and reliable detection across multiple variant classes,” said Zheng.
Table 2. Accuracy
MSI classification testing also supported the assay's ability to distinguish MSI status with high confidence.
Blood-based TMB estimation
ctDNA testing can provide a non-invasive estimate of TMB to help inform immunotherapy decisions – particularly checkpoint inhibitor treatment. However, there are important challenges.
First, low ctDNA fractions lead to under-detection of mutations. Second, the tumor DNA shedding varies significantly depending on tumor type stage and treatment. And third, clonal hematopoiesis can introduce false positives.
To address this issue, the team performed cross-platform calibration of TMB and aligned clinical cutoffs, which confirmed an optimal ratio of 32 mutations per megabase for the TSO 500 V2 platform.
“At this cutoff, the assay achieved 95 percent sensitivity, 98 percent specificity, and 97 percent overall accuracy,” reported Zhang.
Clinical actionability
Zheng presented data comparing a small targeted panel with the expanded TSO 500 V2 ctDNA panel in a clinical validation cohort. The larger panel identified more than five times as many pathogenic or likely pathogenic variants – increasing the proportion of patients with at least one clinically relevant finding from approximately 58 percent to 96 percent. The median number of pathogenic variants detected per patient also rose substantially, from two to six.
Using AMP/ASCO/CAP classification criteria, the expanded panel markedly increased the detection of clinically actionable tier I and tier II alterations, rising from 130 findings with the smaller panel to 480 with CGP. Even when focusing only on tier I variants, the expanded assay showed a clear advantage, largely driven by improved identification of tumors with high tumor mutational burden.
Overall, Zheng said, broader ctDNA profiling “significantly improves the likelihood of identifying pathogenic alterations,” potentially expanding treatment options and clinical trial eligibility for patients.
Dealing with clonal hematopoiesis
Despite the strong analytical performance of large ctDNA CGP panels, Zheng identified clonal hematopoiesis as a persistent challenge in liquid biopsy interpretation. Clonal hematopoiesis arises from somatic mutations in hematopoietic cells rather than tumor cells, and its prevalence increases substantially with age. Commonly affected genes include DNMT3A, TET2, and ASXL1, all of which may appear in ctDNA analyses despite not being tumor-derived.
This issue is clinically significant because some clonal hematopoiesis-associated genes are potentially actionable. Misclassifying these alterations as tumor mutations could lead to inappropriate treatment decisions.
To address this growing challenge, Zheng described the development of a high-sensitivity clonal hematopoiesis panel at Mayo Clinic. It aims to improve identification of clonal hematopoiesis-derived mutations and reduce false-positive ctDNA findings by distinguishing hematopoietic alterations from true tumor-derived variants. The NGS-based assay targets 37 clonal hematopoiesis-associated genes, and is designed to detect SNVs and indels at variant allele frequencies as low as approximately 0.25 percent.
Zheng outlined several strategies for managing clonal hematopoiesis in ctDNA testing, each with distinct trade-offs in accuracy, turnaround time, and cost. In a sequential approach, ctDNA is analyzed first, with reflex testing of matched blood cells performed only when potential CH-associated variants are identified; however, this reactive workflow can delay results. A parallel approach, which Mayo Clinic plans to adopt, sequences ctDNA and matched white blood cells simultaneously, allowing more accurate discrimination between tumor-derived and hematopoietic variants, although it requires additional resources.
A third option involves bioinformatics-based algorithms that infer the likelihood that variants originate from clonal hematopoiesis without additional sequencing. While this strategy is more scalable, Zheng noted that it may be less accurate than approaches incorporating matched blood-cell analysis.
Overall, Zheng said, the optimal strategy depends on the clinical context and the intended use of testing.
Summary
ctDNA testing offers a minimally invasive, real-time approach to tumor genomic profiling with clear clinical advantages. Mayo Clinic’s experience showed that while smaller targeted panels are feasible and clinically valuable, expanding to CGP substantially improves the detection of actionable alterations and clinically relevant biomarkers.
He noted that the large CGP panel demonstrated robust analytical performance across multiple variant classes, supporting its use in routine clinical testing. However, Zheng emphasized that clonal hematopoiesis remains a major interpretive challenge because hematopoietic mutations can introduce biologically confounding signals into ctDNA analyses.
Ultimately, Zheng said, “the value of ctDNA lies not only in generating more data, but in accurate interpretation to support better clinical decision-making.”
*For Research Use Only. Not for use in diagnostic procedures.
Newsletters
Receive the latest pathologist news, personalities, education, and career development – weekly to your inbox.
