Lung Plasma Genotyping: Comprehensive cfDNA Versus Tissue in Advanced NSCLC

The diagnostic and clinical question

In advanced (metastatic) non–small cell lung cancer (NSCLC), identifying actionable genomic drivers—such as EGFR mutations, ALK and ROS1 rearrangements, BRAF mutations, MET alterations, RET fusions, and others—is a prerequisite for selecting patients who may be eligible for targeted therapy drug classes rather than empiric chemotherapy or immunotherapy alone. Historically, this genotyping has relied on tumor tissue. But tissue has real-world limitations: biopsies can be insufficient for molecular analysis, the procedure carries risk, and the turnaround from biopsy to a complete genomic report can be long. These problems raised a practical question: can genotyping performed on a simple blood draw reliably detect the same drivers as tissue, quickly enough and accurately enough to guide the choice of therapy class?

This article examines the evidence behind comprehensive plasma cell-free DNA (cfDNA) genotyping in advanced NSCLC, focusing on the NILE study, analytic and clinical validation of the Guardant360 assay, and the resulting "plasma-first versus tissue-first" framing for molecular workup.

What the analyte and assay are, and what a result can establish

The analyte in plasma genotyping is cell-free DNA (cfDNA)—short fragments of DNA circulating in the bloodstream. A fraction of this cfDNA in a cancer patient is tumor-derived (circulating tumor DNA, or ctDNA), shed by tumor cells. Because tumor-derived fragments carry the tumor's somatic alterations, sequencing cfDNA can, in principle, reveal the same driver mutations and fusions found in tissue, without the need for a tissue sample.

Guardant360 is a next-generation sequencing assay that interrogates cfDNA across a panel of cancer-associated genes, detecting point mutations, insertions/deletions, copy number changes, and selected fusions [2]. Because the tumor-derived fraction can be very small, the assay is engineered to detect variants at very low allele frequencies—its reported limit of detection reaches approximately 0.02–0.04% variant allele fraction for certain alteration types [2].

What a plasma result can and cannot establish is central to interpreting it correctly:

• A positive result (an actionable alteration detected in plasma) is highly likely to reflect a true tumor alteration, given the assay's very high specificity (see below), and can inform eligibility for a corresponding targeted-therapy class.
• A negative result does not rule out an alteration. If a tumor sheds little ctDNA into the blood, a driver present in the tissue may not be detectable in plasma. This is why a negative or uninformative plasma test generally prompts reflex to tissue testing rather than ending the workup.

In short, plasma genotyping is best understood as a sensitive "rule-in" tool whose negative results require confirmation.

The evidence

Analytic and clinical validation (Odegaard 2018)

Before concordance with tissue can be interpreted, the assay itself must be shown to measure what it claims to measure. Odegaard and colleagues reported the analytic and clinical validation of Guardant360, demonstrating high analytic accuracy with positive percent agreement (PPA) and negative percent agreement (NPA) exceeding 99%, alongside the low limit of detection noted above (~0.02–0.04% allele fraction) [2]. These metrics establish that the platform can reliably detect low-abundance variants and that, when it calls a variant, that call is analytically trustworthy. Analytic validation of this kind underpins—but is distinct from—clinical concordance with an independent tissue reference.

Head-to-head with tissue (NILE, Leighl 2019)

The NILE (Noninvasive versus Invasive Lung Evaluation) study directly compared comprehensive cfDNA genotyping with standard-of-care tissue genotyping in patients with newly diagnosed metastatic NSCLC [1]. The study was designed to test whether plasma was noninferior to tissue for detecting guideline-relevant biomarkers.

Key findings:

• Comprehensive cfDNA genotyping detected a guideline-recommended biomarker in a proportion of patients that met the prespecified noninferiority threshold relative to tissue [1].
• Plasma identified a guideline biomarker in roughly 80% of the cases in which any such biomarker was found by either method, demonstrating substantial detection capability from blood alone [1].
• When comparing specific actionable alterations, the concordance (positive predictive value) of plasma relative to tissue exceeded 98.2% for the biomarkers assessed—meaning that when plasma called an actionable driver, tissue overwhelmingly agreed [1].
• Plasma testing turnaround was faster than tissue testing in the study, reflecting the practical advantage of a blood draw over biopsy-dependent workflows [1].

The combined message of NILE is that a comprehensive plasma assay was noninferior to tissue for driver detection in this setting, with very high agreement when plasma was positive—but, consistent with the biology of ctDNA shedding, plasma did not detect every biomarker found in tissue [1].

Where this stands: validated/cleared versus investigational, and open questions

Platform validation. Guardant360 has documented analytic and clinical validation with high agreement metrics and a well-characterized limit of detection [2], and comprehensive plasma genotyping in advanced NSCLC is supported by prospective comparative evidence from NILE [1]. Molecular-testing frameworks for NSCLC incorporate plasma genotyping as an option within the diagnostic algorithm, reflecting its accepted role in the workup [3].

The plasma-first versus tissue-first framing. The evidence supports two complementary workflow strategies rather than a single mandated one:

• A plasma-first approach can shorten time to a result and may be particularly useful when tissue is scarce, when a patient cannot safely undergo repeat biopsy, or when rapid results are needed to inform therapy-class eligibility. Because plasma positivity is highly concordant with tissue [1], an actionable positive can support decision-making without waiting for tissue.
• Because a negative plasma result does not exclude a driver, a tissue-first (or reflex-to-tissue) strategy remains important, and uninformative plasma testing should be followed by tissue genotyping [1,3]. In practice, many workflows are hybrid: plasma and tissue are complementary, not mutually exclusive.

Limitations and open questions.

• Sensitivity depends on ctDNA shedding. Tumors that shed little ctDNA produce false-negative plasma results; the ~80% any-biomarker detection in NILE underscores that plasma alone will miss a meaningful minority of cases [1].
• Negative predictive value is the central caveat. The strength of plasma genotyping lies in rule-in performance; a negative result is not definitive.
• Scope of detected alterations. Some alteration classes—particularly certain fusions and copy-number events—can be more challenging to detect in cfDNA than in tissue, so the specific gene and alteration type influence reliability.
• Clonal hematopoiesis. cfDNA can contain variants arising from blood cells rather than the tumor; interpretation requires awareness that not every detected variant is tumor-derived. (Readers should weigh this general principle when interpreting any cfDNA report.)

The evidence suggests that comprehensive plasma cfDNA genotyping is a validated, clinically useful tool for driver detection in advanced NSCLC—noninferior to tissue in a prospective comparison, with very high concordance when positive—provided its key limitation is respected: a negative plasma result does not substitute for tissue analysis. None of the above constitutes treatment advice; detection of a driver establishes only eligibility for consideration of a corresponding targeted-therapy class, and all therapeutic decisions remain the domain of the treating clinical team.

References

1. Leighl NB, et al. Clinical Utility of Comprehensive Cell-free DNA Analysis to Identify Genomic Biomarkers in Patients with Newly Diagnosed Metastatic Non–Small Cell Lung Cancer (NILE). Clin Cancer Res. 2019. PMID 30988079; DOI 10.1158/1078-0432.CCR-19-0624.
2. Odegaard JI, et al. Validation of a Plasma-Based Comprehensive Cancer Genotyping Assay Utilizing Orthogonal Tissue- and Plasma-Based Methodologies (Guardant360 analytic and clinical validation). Clin Cancer Res. 2018. PMID 29691297; DOI 10.1158/1078-0432.CCR-17-3831.
3. Non-Small Cell Lung Cancer molecular testing algorithm / plasma genotyping framing (guideline notes). 2024.