Liquid profiling for patients with advanced cancer is ready for clinical integration

Many studies have assessed the concordance of genomic alterations detected in major driver genes and treatment-relevant biomarkers between plasma and tissue and have found consistent sensitivities, ranging approximately between 70–90% across various solid tumors [5, 12‐14]. Discordance is mostly related to tumor heterogeneity and temporal dynamics, sampling biases and technical limitations. Tumors are often heterogeneous and the portion of the tumor sampled for tissue analysis may not fully represent the entire genetic landscape of the tumor, whereas ctDNA represents a mixture of DNA shed from various tumor regions. Moreover, tumor genomes evolve over time due to therapeutic pressure and clonal selection. The ctDNA shed into the bloodstream reflects the most recent state of the tumor, whereas tissue samples may have been obtained at an earlier stage of the disease and may not capture these changes. On the other hand, mutations derived from the hematopoietic system, which can lead to clonal expansions, can be picked up in cfDNA. Clonal hematopoiesis-related mutations can introduce specificity concerns in ctDNA analysis. Without careful validation and discrimination strategies, there is a risk of misinterpreting clonal hematopoiesis-related mutations as tumor-derived mutations, leading to false-positive results.

Taken together, tissue and ctDNA analysis each have strengths and limitations, and their findings may be complementary rather than identical. However, a robust detection agreement among key driver events as well as a comparable number of targetable alterations between ctDNA and tissue profiling [15‐17] has established the viability of ctDNA-based assays as a substitute for tissue-based testing.

Currently, the standard clinical application of genomic profiling of ctDNA is for treatment selection in the advanced cancer setting, meaning detecting alterations that can be matched to targeted therapies or identifying alterations that would be a contraindication for a particular therapy. As with tissue analysis, several scenarios may warrant the limited analysis of a single gene [18‐20] or using a cancer hotspot panel ([4, 21]; Table 1). However, the general direction of the treatment selection setting is moving toward employing larger comprehensive genomic profiling (CGP) panels to maximize the detection of therapeutic targets and to harvest the additional, valuable information such as mutations, copy numbers, fusions, tumor fraction in plasma (Fig. 1; [22, 23]). In fact, several ctDNA-based CGP tests have already received FDA approval/clearance for select indications (Table 2). Putting algorithmically estimated levels of plasma tumor fraction (TF) into context with the detected variant allele frequencies (VAF) of detected mutations can help with the interpretation of detected mutations and indicate whether they are derived from the germline, from subclones or even from the hematopoietic system (Fig. 2). For CGP panels, TF is usually estimated based on tumor aneuploidy measured as deviations in coverage across the genome. When TFs are below 5–10%, such an estimate is no longer informative, while SNVs can still reliably be detected down to 0.1% [16]. If the panel is large enough, also complex biomarkers such as microsatellite instability (MSI) status and blood tumor mutational burden (bTMB), which have significant relevance for immunotherapies, can be inferred [24]. The added insight that can only be obtained from CGP approaches is critical to downstream interpretation of ctDNA results and provides the best comprehensive overview of the patient’s sample (Table 1).

An increasing body of evidence supporting the usage, advantages and limitations of plasma-based CGP for guiding treatment decisions has been instrumental in updating clinical guidelines for routine ctDNA testing. This has led to terminologies and concepts such as “tissue-first”, “plasma-first” or matched tissue and liquid profiling, indicating in which clinical scenario it makes sense to perform initial testing on biopsy material, cfDNA from plasma, or both, respectively. Because each approach has its obvious advantages and disadvantages, it is difficult to design a universal testing strategy for patients with advanced cancer and, in the past, recommendations have often conflicted among authors of clinical guidelines [10, 25‐28] Recently, several of these professional multidisciplinary expert panels have reconvened to provide a general but thorough framework for assay and analyte selection for advanced cancer genotyping which are, for reasons of brevity, summarized here [8, 9]. Generally, it is recognized that ctDNA assays have demonstrated utility in the identification of actionable alterations to inform targeted treatment and may be used routinely for the management of advanced cancer patients, but the assay limitations must be considered. The “tissue-first” approach remains the gold standard for the majority of patients, especially as ctDNA assays are often limited in detecting important events of therapeutic relevance, such as fusions and SCNAs. As such, most guidelines stress the importance of the “tissue-first” approach and generally recommend “plasma-first” for most tumor types when tissue material is unavailable or inadequate [8]. However, a “plasma-first” approach may be performed when a quicker turnaround time is critical for a clinical decision, like in aggressive tumors such as advanced NSCLC [8]. Additionally, there are several scenarios in which ctDNA testing is preferred to standard tissue profiling, such as for the detection of ESR1 mutations in breast cancer and the detection of resistance-related mutations in NSCLC patients who previously received tyrosine kinase inhibitor (TKI) therapy [8]. However, negative ctDNA results can have different interpretations depending on the clinical context and the specific characteristics of the patient and the tumor. While it may suggest the absence of detectable tumor DNA in the bloodstream or a favorable response to treatment, it does not definitively rule out the presence of a mutations, in particular when the sample harbors low tumor content. Therefore, expert guidelines advise reflex tumor testing when non-informative results are obtained. There is also accumulating evidence that performing molecular profiling on both tissue and plasma in parallel significantly enhances the detection of actionable alterations, thus increasing the chance of being able to match patients to targeted therapies [4, 15]. In addition, joint tissue and liquid testing provides valuable complementary, technical and biological information that enables a more holistic interpretation and evaluation of the molecular results.