The value of ctDNA tumor fraction in interpreting negative liquid biopsy results

Study findings

The results presented in Table 1 illustrate the central finding of Rolfo et al.^[11]: the interpretation of negative LBx results depends critically on the amount of ctDNA in plasma, as measured by the ctDNA TF. When results are interpreted without stratification, the concordance between LBx and TBx is modest, with a positive percent agreement (PPA) of 63% and a negative predictive value (NPV) of 66%. These metrics highlight the inherent limitations of LBx when considered in isolation, and they help explain why regulatory agencies currently recommend reflex tissue testing in the case of negative findings.

Introducing ctDNA TF as a stratification biomarker could fundamentally shift this paradigm. In samples with high TF (≥ 1%), the sensitivity and reliability of LBx improve dramatically, with PPA and NPV reaching 98% and 97%, respectively. In these cases, negative LBx results can be regarded as “informative negatives”, allowing clinicians to exclude actionable driver mutations and initiate therapy without waiting for tissue confirmation. This has immediate clinical significance, as it accelerates treatment initiation, particularly in patients with advanced NSCLC, where delays are associated with poorer survival outcomes^[12].

Conversely, low-TF samples present a higher clinical risk. When TF is < 1%, the probability of false-negative results increases substantially. Among NSCLC patients in this subgroup, more than half of the negative LBx cases were subsequently found to harbor actionable mutations upon reflex TBx. These findings underscore the necessity of tissue confirmation in low-TF contexts, as failure to do so could result in missed targetable alterations, such as EGFR, Erb-B2 receptor tyrosine kinase 2 (ERBB2), ERBB2 or ALK, all of which have established targeted therapies that improve survival^[13].

Data from the reflex testing cohort further illustrate the consequences of relying only on LBx. Among 505 patients with negative LBx who underwent TBx, 37% were found to carry actionable driver mutations, most commonly KRAS, EGFR, ERBB2 and ALK. Without reflex TBx, nearly four out of ten patients in the low-TF subgroup could have been denied access to potentially life-saving targeted therapies. Although reflex TBx delayed treatment initiation (median 4.7 weeks vs. 1.8 weeks for immediate therapy), it significantly increased the likelihood of receiving a matched targeted therapy (17% vs. 4%) and translated into meaningful survival benefits, with a real-world progression-free survival (rwPFS) of 15.9 months and overall survival (rwOS) of 45.8 months among patients who ultimately received therapy guided by tissue-based molecular profiling^[11].

Clinical relevance

Taken together, these results support a two-tiered clinical strategy for interpreting negative LBx results. High-TF negative results can be trusted and should prompt timely initiation of systemic therapy, avoiding unnecessary invasive procedures. By contrast, low-TF negative results remain indeterminate and require tissue confirmation to ensure no actionable mutations are missed. This stratified approach exemplifies precision oncology, aligning not only therapy with molecular alterations but also diagnostic intensity with the likelihood of obtaining clinically relevant information^[14].

The implications of this framework extend beyond NSCLC. In colorectal cancer, accurate exclusion of Rat sarcoma protein (RAS) and BRAF mutations is critical for guiding anti-EGFR therapy^[15]. Similarly, in breast and prostate cancers, where emerging biomarkers increasingly influence treatment decisions, high-TF negatives may suffice to guide therapy, whereas low-TF negatives necessitate further investigation^[16].

Importantly, the clinical interpretation of TF- stratified negative results may need to be contextualized according to the class of genomic alteration being assessed. As highlighted by Rolfo et al.^[11], current evidence for TF-based interpretation is strongest for short variants and rearrangements, which were the focus of their analysis. For other classes of genomic alterations, such as copy-number alterations or high-level amplifications, detection is generally more dependent on tumor DNA fraction, suggesting that TF thresholds may need to be interpreted in an alteration-specific manner in clinical practice.

The clinical significance of these findings is considerable. For patients with high ctDNA TF and negative results, clinicians could be confident that the absence of detectable alterations reflects the tumor’s genomic profile, enabling timely initiation of immunotherapy or chemo-immunotherapy^[3]. Accelerating time to treatment is particularly critical in NSCLC, where even short delays can reduce the likelihood of patients receiving targeted therapy^[16]. Conversely, patients with low TF require a different clinical pathway, since negative results in this context cannot be taken at face value, and reflex to TBx becomes essential. Rolfo et al.^[11] demonstrated that more than one-third of patients in the low-TF subgroup benefited from tissue-based detection of actionable mutations, and timely confirmatory testing significantly increased the uptake of targeted therapies, leading to longer progression-free and overall survival.

An additional consideration is that ctDNA TF is not a static parameter and may change over time in response to treatment or disease evolution. In patients with initially high TF, a subsequent decline following systemic therapy may reflect biological response rather than technical failure of LBx^[17]. In this context, a negative LBx result should be interpreted cautiously, taking into account prior TF levels and treatment history. Rather than representing a false-negative result, declining TF may indicate effective tumor suppression, whereas re-emergence or increase in TF on serial testing may signal molecular progression before radiographic evidence^[18]. Although prospective data are still limited, emerging real-world experience suggests that longitudinal assessment of TF may add clinical value by contextualizing serial negative results and informing decisions on surveillance, treatment continuation, or the need for repeat TBx. Importantly, this dynamic interpretation reinforces the role of TF as a contextual biomarker, complementing other clinical and radiologic parameters.

Beyond patient outcomes, ctDNA TF also has important implications for healthcare systems. TBx is invasive, costly, and often associated with complications such as pneumothorax or bleeding, in addition to patient anxiety and discomfort^[19]. Identifying patients who truly require tissue confirmation could reduce unnecessary procedures, optimize resource allocation, and lower costs. For patients with high TF, reflex biopsy may be avoided, whereas for those with low TF, prioritizing tissue confirmation ensures that no opportunities for targeted therapy are missed^[11].

The strengths of the study by Rolfo et al.^[11] reinforce the credibility of these conclusions. The analysis included over 3,800 paired samples from multiple tumor types, validated in a nationwide clinicogenomic database, providing robust statistical power and reflecting real-world practice. The TF estimation methodology was innovative, incorporating aneuploidy patterns, variant allele frequencies and multiomic signatures, while systematically excluding germline and clonal hematopoiesis confounders^[11].

These strengths also point toward broader implications for precision oncology. With prospective validation, ctDNA TF could be integrated into diagnostic algorithms and even clinical guidelines such as those from the National Comprehensive Cancer Network (NCCN) or European Society for Medical Oncology (ESMO). Workflows might be structured so that patients with a positive LBx receive immediate targeted therapy, patients with negative LBx and high TF proceed with immunotherapy, and patients with a negative LBx and low TF undergo reflex to TBx. Such algorithms could streamline clinical practice, reduce variability among clinicians, and standardize care pathways across institutions. Moreover, by reducing unnecessary biopsies, TF-guided interpretation has the potential to lower costs and improve patient experience, aligning precision oncology with both clinical and economic priorities.

Limitations of the study

While ctDNA TF represents a promising interpretive biomarker, several limitations must be considered when translating it into clinical practice. First, the retrospective nature of the data collection introduces potential biases that may not fully capture the complexity of patient selection, sample handling and sequencing conditions in real-world settings. Without prospective validation, the generalizability of these findings remains constrained. Another important limitation concerns the molecular scope of the analysis. The reliability of ctDNA TF has been primarily demonstrated for short variants and rearrangements, which are the most common targetable alterations in cancers such as NSCLC and colorectal cancer. However, copy-number alterations, which often require a higher tumor DNA fraction for accurate detection^[20], may not be as reliably assessed using the current approach. This restricts the applicability of the conclusions to a subset of genomic alterations and suggests that further refinement is needed before TF can be applied across all variant types.

In addition, the methodology for TF estimation is proprietary to Foundation Medicine and optimized for their comprehensive genomic profiling platform. It remains unclear whether comparable accuracy and predictive value can be achieved with other commercial assays or laboratory-developed tests. This platform dependence raises questions regarding reproducibility, standardization and cross-laboratory validation.

Beyond these study-specific limitations, important research blind spots remain, including the lack of standardized ctDNA TF thresholds across assays and tumor types, limited prospective validation of TF-guided clinical algorithms, and uncertainty regarding the interpretation of low TF in specific clinical contexts, such as indolent disease or low-volume metastatic burden. Addressing these research blind spots will be essential for broader clinical adoption of TF-stratified LBx interpretation.

Despite these limitations, ctDNA TF offers a relatively simple and quantitative biomarker that could fundamentally improve the interpretation of negative LBx results. In a field where rapid action is crucial and false reassurance can be harmful, ctDNA TF provides the contextual information that clinicians have long needed.