Implementing the ESMO recommendations for the use of circulating tumor DNA (ctDNA) assays in routine clinical application/diagnostics

Introduction

The liquid biopsy (LB) is a highly advanced, minimally invasive approach that facilitates the detection and analysis of tumor cells or their genetic material in a patient’s body fluids, significantly improving clinical decision-making in the field of oncology. This dynamic molecular diagnostic field is marked by rapid advancements and offers a wealth of innovative approaches and technologies.

In particular, the detection of ctDNA from plasma has provided a plethora of compelling evidence for their clinical effectiveness and utility. Consequently, numerous international specialty societies and working groups have already advocated for the use of ctDNA assays [1], [2], [3], [4], [5], [6]. A groundbreaking development in this field recently occurred when the European Society for Medical Oncology (ESMO) recommended ctDNA assays as both a complement and an alternative to traditional genotyping for specific tumor indications [7], highlighting their central role in modern oncological practice. Despite the widespread endorsements however, the integration of LB into routine care has been slow, primarily due to lacking preanalytical and analytical harmonization. This delay can be attributed to methodological challenges such as the absence of standardized workflows and analytical sensitivity concerns, conceptual hurdles including doubts about appropriate indications for LB and its clinical utility, as well as operational uncertainties like the absence of reimbursement [8]. Implementing strategies that meet the diagnostic standardization needs is imperative to mitigate uncertainties in sample handling, processing, and result reporting [8].

The main objective of this mini review is to provide easily understandable and practical guidelines for the optimal use of ctDNA assays to molecular oncology laboratories and healthcare professionals. To achieve this goal, we have summarized the recommendations from ESMO and supplemented them with relevant insights from extensive literature research to ensure that the guidelines remain current, and evidence based. Our approach includes addressing frequently asked questions (see Table 1), synthesizing information from ESMO’s recommendations, and considering the latest findings from relevant literature sources.

Q1. For which scenarios/indications does the European Society for Medical Oncology (ESMO) recommend ctDNA tests?

The analysis of ctDNA unlocks a wide range of potential applications (see Figure 1; [9]). Based on the evidence of clinical validity, validated and adequately sensitive ctDNA tests can be used in routine practice, provided that their limitations are understood and considered ([7]; also see Q4). Specifically, ESMO currently recommends molecular profiling (= liquid profiling, LP), particularly for the identification of genetic aberrations linked to specific therapies and as a consequence facilitate treatment decision. While tissue-based testing is still considered the preferred diagnostic method and the gold standard for many cancer patients, plasma-based ctDNA assays are the preferred option for initial diagnosis in therapy-naive cases or in situations of tumor progression under the following circumstances:

1. When a tissue biopsy is not possible or insufficient for medical or practical reasons, or

2. When the timely delivery of results is clinically significant, especially in aggressive tumors [7, 10]. Tumor-specific clinical recommendations can be found in Table 2.

Figure 1:

A schematic timeline illustrating the hypothetical journey of a patient with non-small cell lung cancer (NSCLC), commencing with surgery or initial treatment, followed by disease relapse and subsequent therapies (adapted from [9]). The potential roles of liquid biopsies throughout the patient’s treatment course, alongside the anticipated variations in ctDNA levels are indicated. Initially, the patient presents with a sole primary tumor site, until the emergence of multiple metastases and distinct clones, each denoted by different colors. Liquid biopsy applications endorsed by the ESMO [7] are marked with an asterisk (*).

Additionally, in certain indications (e.g., lung tumors), ctDNA testing can be performed in parallel or as a complement to tissue testing [2, 5]. If there is limited tissue available, parallel ctDNA analysis can save material for the analysis of tissue-based biomarkers (e.g., PD-L1, MSI, detection of fusions/amplifications) [2]. Furthermore, in cases where tissue testing yields negative or inconclusive results, further ctDNA analysis can be conducted [2, 5, 11].

Other applications of ctDNA analysis (i.e., detection of minimal residual disease (MRD), monitoring, population screening; see Figure 1) are still being studied by the ESMO. The reason for this is that, despite promising studies, the clinical benefit of ctDNA analyses for these applications has not yet been sufficiently established through prospective studies [7]. The status is summarized as follows:

Detection of minimal residual disease (MRD): It is clinically established that the prognosis is significantly improved, and the risk of relapse is substantially reduced when no ctDNA is detected in the blood after therapy with curative intent [7, 8, 12]. Ongoing prospective studies aim to validate the clinical benefits of ctDNA analyses for minimal residual disease (MRD) detection, with potential implications for improved outcomes and safe therapy de-escalation [7]. It is worth pointing out that nearly simultaneously with the ESMO recommendations [7], a naturally not yet considered Phase II study involving patients with colon carcinoma (Stage II) was completed. In this study, therapy de-escalation guided by ctDNA resulted in approximately 50 % of patients being spared unnecessary chemotherapy, without adversely affecting recurrence-free survival [13].

Monitoring: Depending on the tumor type and stage, a relapse can be detected up to 12 months earlier using ctDNA assays than with conventional imaging techniques [5, 7, 8, 12]. Early intervention, believed to enhance the chances of cure [12], is a focus of ongoing prospective studies aiming to demonstrate that therapy adjustment based on ctDNA monitoring improves outcomes [6, 7]. In this context, ESMO highlights a promising Phase III study (PADA-1), in which early therapy change based on rising ctDNA levels (ESR1) led to a doubling of the median progression-free survival time in breast cancer patients [7, 14].

Population screening: The use of ctDNA assays as a method for early tumor detection is logical due to the minimally invasive nature of sampling, especially when combined with other methods, but it is not yet recommended [7, 12]. Being aware of the low concentration of ctDNA amidst a significant background of “normal” cfDNA, screening options require not only accuracy of detection, but also need highly sensitive and specific methods [12]. Additionally, the low prevalence of tumor diseases and their resulting low positive predictive value further complicate early detection efforts [8, 12, 15]. For example, 98 % of lung cancers occur in individuals aged 50 and above, with an incidence rate of 1 in 600. Considering a hypothetical ctDNA assay specificity of 99 %, this would result in six false positives for every true positive, causing significant anxiety and expense [12, 15, 16]. However, the potential benefit of a high negative predictive value in screening warrants further discussion. Data from large studies assessing ctDNA assays as population screening tools for multiple tumor types are still expected, but at present population screening cannot be considered a validated application for ctDNA assays [7].

Q2. Which markers/genes should be tested?

The ESMO Scale for Clinical Actionability of Molecular Targets (ESCAT) classifies the “actionability” of specific genomic alterations (i.e., how well genomic alterations match the efficacy of drugs). ESCAT Tier I means that the match between a genomic alteration and a drug has been shown to lead to clinically validated and significant improvements in clinical outcomes. ESCAT Tier I is synonymous with clinical benefit and should, therefore, determine treatment decisions in daily practice [7, 17]. ESCAT Tier II refers to genomic alterations that are expected to define a patient population that could benefit from targeted therapy, but additional data are required to confirm this [7, 17].

In this context, a total of 21 markers for ctDNA testing are currently recommended in 13 tumor entities (see Table 2; [7]). The number varies by entity, with a maximum of 11 markers for non-small cell lung carcinoma (NSCLC) and only one marker for soft tissue sarcomas (see Table 2). These markers can be further divided into point mutations (SNVs) or small insertions/deletions (ATM, BRAF, BRCA1/2, EGFR, ESR1, FGFR, IDH1, KRAS, NRAS, PALB2, PIK3CA, PTEN), fusions (FGFR2, FGFR3, NTRK 1/2/3), amplifications (MET, ERBB2), or markers where all the mutation types mentioned above may be relevant (ALK, ERBB2, MET, RET, ROS1). Microsatellite instability (MSI) is a predictive biomarker for immune checkpoint inhibitor therapy and refers to the number of mutations in repeated microsatellite regions that have accumulated due to “mismatch repair deficit” [7, 18]. A tumor with high MSI usually responds better to therapy with immune checkpoint inhibitors [7, 18, 19].

Q3. What panel size/method should be used for ctDNA analysis?

The utilization of well-established ctDNA methods, validated using standardized reference materials (e.g., for Limit of Detection (LOD), Limit of Blank (LOB), technical accuracy, and precision), is indispensable for their integration into routine clinical practice [5, 7, 8, 12]. Moreover, external proficiency testing and collaboration among laboratories are essential for ensuring quality assurance [8, 12].

There is already an increasing variety of methods for ctDNA detection with different panel sizes available (see Table 3; [20]). Typically, these methods require a compromise between sensitivity, genome coverage, and associated costs [1, 3, 21, 22]. This raises the crucial question of which methods best align with the current recommendations. Depending on specific indications, both tissue-based and ctDNA tests generally recommend testing a small number of markers (with ESCAT Tier I and II) ([7, 17]; see Table 2; see Q2). There is currently no “one-size-fits-all” ctDNA approach that would fit all patients or tumor entities [5]. In clinical practice, the choice of method and panel size (RT-PCR, digital PCR, and NGS assays) should be determined by availability, reimbursement status, and the number of actionable genetic aberrations of ESCAT Tier I (see Table 2) in a tumor-specific context [7]. Therefore, the ctDNA approach must be adapted according to the patient and tumor entity (see Table 3).

One important factor to consider is the detection limit of the ctDNA assay. The tumor fraction in plasma, quantified as variant allele frequency in percentages (VAF), correlates with both tumor burden and stage, although it is often notably low (<1 % VAF) – particularly in early-stage disease [21, 23, 24]. The sensitivity of existing tests and methods varies significantly (refer to Table 3), and currently, there is no specified threshold for the limit of detection of ctDNA deemed “adequate” [7]. The German Federal Medical Council (Bundesärztekammer) has recently addressed this gap, stipulating as part of the RiliBäk guidelines that ctDNA assays from body fluids must be capable of detecting a VAF of at least 0.5 % [25]. Likewise, in the context of NGS based assays, detection of ctDNA variants with a VAF of less than 1 % should not rely solely on increasing sequencing coverage/depth. Instead, it should involve the integration of unique molecular identifiers (UMIs) and the implementation of additional algorithms for error suppression [25].

Another important aspect is the type of aberration to be detected. For example, single nucleotide variants (SNVs), deletions, insertions, and deletions/insertions can be well covered by smaller hotspot mutation panels (see Table 3). Other types of therapeutically relevant aberrations, such as MSI or fusions/amplifications (see Table 2), are more challenging to detect from plasma, despite promising approaches (e.g., [18, 26]; see also Table 3), and should, if possible, be determined from tissue [7].

Large (external) NGS LB panels/services offered by commercial companies such as Guardant360^® CDx (Guardant Health, Redwood City, USA) or FoundationOne^® Liquid CDx (Foundation Medicine, Cambridge, MA), cover dozens to hundreds of genes and can also detect aberrations like MSI or fusions/amplifications, albeit with lower sensitivity compared to tissue [18, 26, 27]. Besides the cost (ranging between approximately $5,800 and $6,800; [28]), a key concern against the use of these large NGS panels is that for only a relatively small portion of all possible tumor-associated genetic deviations a targeted therapy is available and recommended [7, 17, 21, 29]; see Table 2). Therefore, the utilization of large NGS panels must be carefully evaluated in relation to the expected potential additional benefit for the patient [5, 17]. In addition, the outsourcing of analytics to external commercial providers is associated with a loss of the profession in both the laboratory diagnostic and clinical setting.

Tumor-informed approaches, which necessitate the development of a patient-tailored LP panel based on prior tumor tissue genotyping, pose a challenge in adhering to the recommended applications of ctDNA assays due to time constraints and the prerequisite of a prior biopsy ([7, 21]; see also Q1).

Finally, to briefly address the problem of the current IVDR, this form of diagnostics, which is at least currently predominantly offered as RUO by manufacturers, is only available from laboratories that are either ISO 15189 accredited or comply with valid national regulations i.e. RiliBäk, Germany (Art. 5.5c, IVDR 2017), validated and may be used for diagnostics in humans. This will have to be checked by country-specific authorities such as AGES in Austria.

Q4. What limitations of ctDNA analysis should be considered?

The use of ctDNA assays in clinical practice requires an understanding and consideration of the limitations of the test:

1. ctDNA assays generally have high specificity but lower sensitivity, which can result in false-negative or false-positive results [7, 15].

2. When interpreting ctDNA test results, it must be considered that the amount of ctDNA in the plasma sample may not be sufficient to detect various genetic variants [7].

3. If no mutations can be detected using an ctDNA assay, it does not completely rule out the possibility of a tumor disease, especially due to other mutations or genetic aberrations [7].

4. In the case of a previously diagnosed tumor disease, a tissue biopsy is required if no mutations are detectable in the target regions or if the ctDNA assay is considered “non-informative” [7, 10].

5. Detecting ctDNA in the plasma poses challenges in cases of brain tumors, primarily attributed to the presence of the blood-brain barrier [5, 7]. In such cases, cerebrospinal fluid can be considered as an alternative sample for ctDNA analysis [5].

6. LBs may have limitations in detecting specific aberrations (see also Q3). Somatic copy-number variations (CNVs) should only be determined using ctDNA assays when tissue analysis is not possible [7].

7. Clonal hematopoiesis of indeterminate potential (CHIP) mutations can lead to false-positive results but can potentially be ruled out through leukocyte analysis (see also Q5).

Q5. How to deal with clonal hematopoiesis of indeterminate potential (CHIP) mutations?

Clonal hematopoiesis of indeterminate potential (CHIP) involves the accumulation of somatic mutations in hematopoietic stem cells, leading to the clonal spread of mutations in blood cells. CHIP is associated with the natural aging process in both healthy individuals and cancer patients. The VAFs of CHIP variants typically range from around 0.1–1 % and often overlap with the range of ctDNA variants. As a result, these variants represent important natural biological confounders that could potentially lead to inappropriate therapeutic decisions [5, 7, 23, 30]. To distinguish CHIP variants from “true” ctDNA variants, cfDNA (plasma) and genomic DNA extracted from a samples buffy coat can be analyzed separately and then compared [1, 5, 23, 31, 32]. Whether genomic DNA from buffy coats should be analyzed to correct for CHIP-related variants is still a subject of debate, primarily due to the additional costs [5, 21]. However, it is recommended to routinely collect buffy coats to have material available for subsequent CHIP analysis if needed [7].

CHIP mutations are more frequent in certain genes: 15 typical CHIP genes include DNMT3A, TET2, ASXL1, PPM1D, TP53, JAK2, RUNX1, SF3B1, SRSF2, IDH1, IDH2, U2AF1, CBL, ATM, and CHEK2, with about 50 % of CHIP mutations found in the DNMT3A, TET2, and ASXL1 genes [30, 31]. For many mutations, it is therefore unlikely from the outset that they are false positives due to CHIP, even if no additional buffy coat/CHIP analysis is performed [7]. Examples include VHL, SPOP mutations in kidney and prostate cancer, EGFR mutations in lung cancer, and PIK3CA and ESR1 mutations in breast cancer. KRAS mutations can occur as CHIP mutations, but they are so rare that they can be neglected for treatment decisions in lung and colon cancer patients [7, 11]. In this context, ESMO recommends additional buffy coat/CHIP analysis when the genes to be tested:

1. Are known to frequently carry CHIP mutations (e.g., TP53).

2. Or involve clinically relevant (“actionable”) tumor suppressor genes (e.g., DNA repair genes) [7].

In the context of the recent ESMO recommendations [7], this definition specifically applies to genes such as ATM, BRCA1/2, PALB2, and PTEN (see Table 2). As a result, additional CHIP alongside ctDNA analysis is recommended for prostate cancer (ATM, BRCA1/2, PALB2, and PTEN), breast cancer (BRCA1/2), and ovarian cancer (BRCA1/2) entities. For BRCA1/2 and PALB2, it should be noted that pathogenic germline mutations should be confirmed or excluded through tissue testing if detected by ctDNA assays [7]. In breast cancer, testing for BRCA1/2 germline mutations is recommended independently of the LP results. This makes the application of ctDNA assays for the mentioned genes in prostate, ovarian, and breast cancer more complex.

Q6. What pre-analytical steps are necessary for ctDNA analysis?

To ensure a valid test result, proper sample collection and pre-analytics are essential. The main challenges in the pre-analytical phase usually involve the varying physiological states of the patient (see also Q7), the low ratio of ctDNA to the total amount of circulating free DNA (cfDNA), the possibility of sample contamination by DNA released during the lysis of leukocytes, and the unstable nature of DNA and its implications for storage and transport [1, 7]. While there is not a unified approach, some recommendations have been made by professional societies (e.g., [1, 2, 7]), which are summarized below:

A consensus already exists for the use of cell-stabilizing blood collection tubes (such as Streck cfDNA BCT or PAXgene Blood ccfDNA tubes) [5, 19, 32]. Cell-stabilizing tubes, inhibit the lysis of leukocytes and extend the shelf life of blood samples for ctDNA analysis by up to 14 days and allow for logistics, including shipping to another laboratory [1, 2, 5, 7, 33, 34]. Ideally however, plasma preparation should be performed within three days of blood collection [5, 34]. Blood samples in cell-stabilizing tubes should be transported and stored at room temperature. Improper transport temperatures can be identified by an enlarged buffy coat layer or hemolysis [35]. Hemolytic samples should not be analyzed due to the increased risk of amplified background noise in genomic DNA [36]. The presence of icterus or lipemia could potentially affect PCR steps and, thus, distort results [36] and should be documented accordingly.

Plasma preparation should involve double centrifugation at room temperature with gentle deceleration at each step. In the first step, centrifugation is carried out at 1600×g for 10 min [8, 15, 32, 37]. Subsequently, the plasma fraction (supernatant) is transferred to an empty tube without disturbing the buffy coat (it is recommended to leave approximately 500 µL of residual plasma). The buffy coat can be separated at this step to have material available for subsequent CHIP analysis if necessary (see also Q5). In the next step, a second centrifugation of the plasma is performed at 3,000 to 16,000×g for at least 10 min [34, 37, 38]. The supernatant is again transferred to a new tube without disturbing the cell layer (approximately 300 µL of residual plasma should remain). The obtained plasma can be used directly for cfDNA extraction or stored at temperatures between −70 °C and −100 °C [1, 15, 19, 36]. Neither plasma nor cfDNA samples should be subjected to repeated freeze-thaw cycles [1, 8]. Various methods for DNA extraction from plasma are currently available, with the QIAamp circulating nucleic acid kit (QIAGEN, Hilden, Germany) considered the gold standard [8].

Q7. At what time points should ctDNA analysis be conducted?

The timing of blood collection for ctDNA analysis should be carefully chosen based on the clinical question at hand, as various factors can influence the release of cfDNA and thus the detectable VAF (e.g., administered therapies, inflammatory processes, surgeries, exercise; [1, 6, 7, 32, 39]). Notably, the VAF represents the percentage of ctDNA within (healthy) cfDNA and is inversely proportional to the total cfDNA in a blood sample. Therefore, methodologies capable of reporting the absolute number of mutant molecules, expressed as copies per mL of plasma, may provide a more accurate reflection of the actual disease status [32, 40].

Currently, ESMO provides limited guidelines regarding the timing of blood collection. Under the current recommended indications (see Q1), ctDNA analysis for genotyping is intended to be performed at the time of initial diagnosis before therapy initiation or at relapse [7]. Furthermore, blood collection for genotyping during active therapy should be avoided to minimize false-negative results [7].

The use of ctDNA assays for monitoring or recurrence control is not yet recommended by ESMO (see also Q1; [7]). For minimal residual disease (MRD) detection (although the application itself is not yet recommended) ctDNA analysis should take place at least one to two weeks after surgery [7, 38].

Regardless, a common strategy is to collect blood for ctDNA analysis at the time of the initial diagnosis (before therapy initiation or before surgery) to establish a baseline for monitoring, and after each therapy cycle to facilitate effective treatment monitoring [6, 14]. In addition to regular follow-up examinations for recurrence control (e.g., once per quarter), blood collection can help detect a possible disease recurrence early (see [16]). Depending on the clinical context (monitoring, MRD, or recurrence control), it may be useful to increase the volume of blood to be analyzed (and thus sensitivity) compared to genotyping, up to 60 mL [1, 32], in contrast to the common 20 mL [5].

Q8. What elements should be included in the report of an ctDNA assay?

In addition to the standard elements of a clinical report, the findings report of a ctDNA analysis should include the following information according to ESMO [7]:

1. Pre-analytical variables: The date of sample collection and treatment exposure (on/off treatment) at the time of collection should be considered.

2. Result: Cases in which a variant cannot be detected should be labeled as “non-informative” or “not detected” rather than “negative.” It should also highlight the possibility of discordant results between tissue tests and ctDNA tests.

3. Potential germline variants: Pathogenic variants associated with genetically inherited cancer susceptibility, where the allele frequency suggests a potential germline background, should be separately indicated. Patient consent should be obtained before investigating the germline origin of a variant.

4. Potential CHIP variants: Variants frequently associated with CHIP (as discussed in Q5) should be highlighted to indicate their possible non-tumor origin. If subsequent testing to exclude CHIP is possible, it should be mentioned in the report.

5. Variant allele frequencies: Variant allele frequencies should be provided as they can provide information about potential germline origin, clonal relatedness of variants in the same panel, and the potential for false-positive results.

6. Target regions covered/mutation types of the assay: The target regions of the ctDNA test and the types of mutations (e.g., SNVs (point mutations), small insertions/deletions, amplifications, fusions, MSI, TMB (Tumor Mutational Burden), etc.) that the test can detect should be stated.

7. Detection limit for different types of variants: The detection limit for each type of variant should be provided. In cases where the plasma DNA used is the limiting factor, the indicated sensitivity should be adjusted and emphasized.

Summary and outlook

The aim of this mini review was to help catalyze the integration of liquid biopsy (LB) into routine care by offering practical, clear, and user-friendly clinical guidance for molecular oncology laboratories and healthcare professionals regarding the optimal use of circulating tumor DNA (ctDNA) assays. Recent recommendations suggest the use of LB for the purpose of LP in scenarios where tissue testing is impractical or when immediate results are vital. While current recommendations tacitly favor smaller-sized liquid biopsy panels, the selection of panel size and analysis techniques should be tailored based on patient-specific, tumor-specific, and logistical considerations. Recognizing and navigating the inherent limitations and challenges of these assays, such as Clonal Hematopoiesis of Indeterminate Potential (CHIP), is crucial for the nuanced application of these assays in clinical practice. In conclusion, a thoughtful and adaptable approach to incorporating these assays into clinical practice will be essential for realizing their full potential in advancing personalized and timely cancer care.