Status of liquid profiling in precision oncology – the need for integrative diagnostics for successful implementation into standard care

Introduction

Although the presence of cell-free DNA (cfDNA) was described as early as 1948 by Mandel and Metais [1], it still took almost fifty years for its potential value in the management of cancer patients to be recognized by the scientific community [2, 3]. The term liquid biopsy in use today, coined by Pantel and Alix-Panabières [4], comprises analysis of circulating tumor cells (CTC), circulating tumor DNA (ctDNA), miRNAs, exosomens, and tumor-derived platelets. Of these biomarkers, ctDNA analysis, which can most appropriately be described as “liquid profiling”, promises the broadest application in the treatment of oncologic patients. It can be used as a minimally invasive alternative to tissue biopsy to determine the genetic tumor profile. Unlike tissue biopsy, it provides information on all tumor sites that shed their DNA into the circulation, allowing real-time tracking of genetic tumor evolution. Accordingly, the benefit of liquid profiling has been demonstrated for several clinical applications, including (i) detection of minimal residual disease and prognostic assessment [5], [6], [7], [8], (ii) earlier detection of recurrence compared to standard of care imaging techniques [9], [10], [11], (iii) selection of appropriate targeted therapeutics based on the genetic tumor profile as so-called companion diagnostics [12], [13], [14], [15], (iv) treatment monitoring and detection of resistance mechanisms [10, 16], [17], [18], [19], [20], [21], and (v) guidance of adjuvant therapy [5], [6], [7]. Despite all these promising scientific results, liquid profiling has not yet been implemented into standard care to the extent one might expect. This is most likely due to the technical challenges of analyzing ctDNA, resulting in heterogeneous study results, disagreement among and reluctance by clinicians, and a limited availability of reimbursement options [20, 22]. The technical challenges are due to the inherent characteristics of ctDNA, including its highly fragmented nature with a mean size of 167 bp and its low fraction of total cfDNA, sometimes less than 0.01% [20, 23]. These characteristics of ctDNA must be taken into account during the preanalytical and analytical phases in order to ensure reliable test results. During the preanalytical workflow, which includes all steps from venipuncture to ctDNA analysis, special care must be taken to prevent degradation of cfDNA or leukocyte lysis causing further dilution of the ctDNA fraction. As for the analytical part, highly sensitive methods specifically designed for liquid profiling should be used, such as digital approaches (digital PCR, BEAMing) or next generation sequencing (NGS) with molecular barcoding. Finally, the lack of standardization and harmonization of preanalytical and analytical workflows is reported as a major barrier to implementation in standard care [24]. Nevertheless, there are initial reports demonstrating the applicability and suitability of liquid profiling (LP) in a routine set-up [12, 20, 25], [26], [27].

In this manuscript, an overview of the current state of implementation of liquid profiling in the standard care of cancer patients’ management at the University Hospital Mannheim (UMM), Medical Faculty Mannheim, University of Heidelberg is provided. Clinical cases are presented for each application to highlight both - value and limitations - of ctDNA testing. Finally, the need for and benefits of an integrative diagnostic approach are discussed.

Current status and perspectives: liquid profiling in standard care

In 2016, the use of liquid profiling as alternative to tissue based-testing in cases where biopsies are either not available or not feasible was first included into an S3 guideline in Germany for detection of RAS mutation status in colorectal cancer (CRC) patients [28]. At that time, the UMM was the first hospital in Germany to establish liquid profiling for this indication and to offer it as part of standard care without an option for reimbursement. Shortly thereafter, the Institute of Clinical Chemistry of UMM was the first to receive ISO 15189 accreditation for several ctDNA assays and to be designated as a reference institute for ctDNA testing by the Reference Institute for Bioanalytics (RfB). Recently, plasma-based RAS and BRAF mutation analysis data for CRC patients obtained as part of standard care between 2016 and 2021 were evaluated and results regarding diagnostic performance, requesting behavior, and integration in clinical decision-making were reported [20]. However, within the last two years, the diagnostic liquid profiling portfolio has expanded significantly. Likewise, the indications as well as requests for liquid profiling have increased. This can be explained by both a broader diagnostic spectrum and an integration of liquid profiling into clinical workflows.

The following paragraph will focus on the impact of liquid profiling for the diagnostic process in cancer treatment, as depicted in the guidelines, and on benefits for both patients and clinicians. The focus will be on benefits and limitations associated with the available diagnostic options of ctDNA testing, how these can be optimally used, and what further changes are necessary for successful implementation into routine care on a broad scale.

Guideline-based management of cancer patients as part of clinical care

According to guidelines, follow-up of cancer patients includes regular clinical examinations, determination of blood-based protein tumor markers, standard imaging if necessary with evaluation of tumor load according to standardized criteria such as RECIST [29], in each case at predefined intervals. This is regularly combined with standardized first-line treatment protocols depending on tumor type and stage, usually based on surgery, radiotherapy and chemotherapy in adjuvant or neoadjuvant setting. In case of relapse or progressive disease (PD), molecular tumor profiling is required for further therapy stratification. This will require additional tissue biopsies, which carry the risk of complications such as bleeding and infection [30]. If this is not feasible, archived tumor tissue can be used for genetic analysis. However, this may not reflect the current genetic tumor profile as it does not take into account intra- and inter-tumor heterogeneity. Therefore, this approach bears the risk of suboptimal treatment decisions. This is supported by evidence from autopsy studies showing a high degree of tumor heterogeneity from multiple biopsy samples [31].

Liquid profiling for companion diagnostics – implementation of small panels in standard care

Introduction of liquid profiling yields the potential to overcome the above-mentioned challenges by enabling real-time assessment of cumulative tumor mutational status. It is therefore not surprising that liquid profiling was first introduced into standard care in colorectal cancer (CRC) and non-small cell lung cancer (NSCLC) patients, for whom current guidelines include targeted therapies that require specific mutational testing prior to treatment. The first FDA-approved assay for this purpose was for EGFR mutation testing in NSCLC patients [32], and the first real-world data of liquid profiling were reported in the context of companion diagnostics for tyrosine kinase inhibitor (TKI) selection in these patients [12, 26]. For the areas described, targeted assays have mainly been developed using either PCR-based enrichment methods or digital methods - the latter having the highest sensitivity. These methods allow the analysis of some, less known, partly therapeutically relevant hot-spot mutations with high analytical sensitivity.

In CRC, the prerequisite for administration of anti-EGFR antibodies such as cetuximab or panitumumab according to the guidelines is the presence of wild-type RAS status [28]. Based on recently reported clinical experience from 2016 to 2021 [20], liquid profiling performed as part of standard care achieved a high level of overall agreement with tissue-based testing (91.7%), underscoring the reliability of this diagnostic approach. As a result, liquid profiling gained an increasing acceptance by clinicians, which over time translated into higher demand and inclusion of ctDNA test results in clinical decisions. Interestingly, RAS mutations occurring as a resistance mechanism to anti-EGFR therapy were detected with a very low variant allele frequency (VAF) of <0.1% in 20% of cases. This highlights the subclonal nature of emerging resistance mechanisms, which require highly sensitive and tissue-sample independent methods such as ctDNA analysis that can also account for spatial heterogeneity.

To illustrate the value of liquid profiling for detection of known driver and/or resistance-causing mutations, a case of a 56-year-old male CRC patient is presented. In this particular case, illustrated in Figure 1, cetuximab treatment was given in combination with irinotecan after a RAS wild-type status was revealed by liquid profiling. This treatment showed stable metastatic disease for liver, lung and bone. After 273 days of treatment, newly emerging KRAS mutations were detected by liquid profiling, indicating resistance to cetuximab. Based on this finding, targeted therapy was discontinued. Shortly thereafter, a CT scan performed to rule out pulmonary embolism revealed multimetastatic progression with progression of bone and lung lesions, lymphangiosis carcinomatosa, and pericardial tumoral affection. Unfortunately, the patient deceased 98 days later.

Figure 1:

Liquid profiling used for companion diagnostics and detection of resistance.

A 56 year-old CRC patient was monitored by LP to detect emerging RAS sequence variation with anti-EGFR therapy. LP positivity correlated with PD detected by imaging.

This case highlights the value of liquid profiling for treatment stratification based on genetic findings. For this application, liquid profiling is currently at the threshold of implementation in Germany, with a few specialized centers, primary in the university setting, offering this test in routine clinical practice. However, further dissemination is mainly hampered by the limited reimbursement options, which are restricted to EGFR T790 M in NSCLC and PIK3CA detection in breast cancer patients in Germany (EBM 19462 and 19,460). To overcome this obstacle, the inclusion of liquid profiling in the respective drug information is necessary. However, based on the liquid profiling assay, treatment resistance was evident in this case, but it was not possible to identify alternative genetic targets based on the digital assay used for this patient.

Liquid profiling for companion diagnostics – integration of NGS-based approaches as part of molecular tumor board

In contrast to targeted approaches, where at most a small panel of sequence variations can be analyzed in parallel, NGS-based approaches facilitate the simultaneous detection of a large number of genetic alterations in several genes. Based on this development, it becomes technically feasible to target multiple potentially relevant mutations rather than focusing on a few common alterations. However, this is accompanied with a vast amount of complex data that need to be evaluated in a clinical context. To meet this demand, molecular tumor boards (MTB) have been established at university hospitals [33, 34]. Here, these sophisticated diagnostic tests are performed and evaluated by a multidisciplinary team. For the interpretation of NGS tests, a bioinformatics pipeline is necessary to enable quality assessment, alignment, and variant calling. Identified variants must be annotated, and appropriate recommendations must be provided for each individual patient. Typically, patients for whom there are no other guideline-based therapy options are referred to MTB. Recently, this has demonstrated that targeted off-label therapies based on NGS liquid profiling test results can help improve progression-free and overall survival [15, 35].

These approaches can be illustrated by the example case of a 37 year-old female patient who suffered from metachronous pulmonary, hepatic, and osseous metastasized estrogen-receptor positive, HER2-negative breast cancer (Figure 2). After long-term guideline-guided treatment, the patient was eventually enrolled in a clinical trial. Due to adverse side effects, the patient was presented at the molecular tumor board to identify alternative genetically stratified therapy options. Panel-based NGS of tissue obtained from biopsy of a liver metastasis identified a class four PIK3CA missense mutation that can be targeted by alpelisib. However, this drug is no longer available in Germany. The also identified FGFR1 amplification is known to refer resistance to endocrine therapies. Liquid profiling identified the identical sequence variations of PIK3CA with a high VAF, proving the presence of ctDNA in the circulation. The identified copy number variation (CNV) of FGFR1 could not be confirmed by liquid profiling because it was not included in the NGS panel used for ctDNA analysis. However, additional amplifications could be revealed by liquid profiling, with MET CNV representing a potential molecular target. The urgent need for therapeutic adjustment was reinforced by the identification of progression detected on CT a short time later.

Figure 2:

Liquid profiling for identification of molecular druggable targets.

A 37 year-old breast cancer patient presented for re-evaluation of therapy due to adverse side effects. To identify a drug target as an alternative therapeutic option, a liver biopsy was performed and a PIK3CA variation was identified indicating a potential response to alpelisib. Concurrent LP identified MET and EGFR amplification in addition to the PIK3CA variation, the former indicating a potential response to tepotinib.

This case highlights the value of NGS-based liquid profiling approaches for identification of novel targetable molecular alterations that can be used directly for selection of matching drugs. However, the broader spectrum of genetic alterations covered in large panels has the disadvantages that a higher degree of coverage by the mean of ultra-deep sequencing is required, resulting in much higher costs and lower sensitivity compared with digital approaches [24]. The need for expensive instrumentation, bioinformatics support, lack of harmonization, and lack of appropriate software tools for annotation and drug selection, as well as the lack of reimbursement options, are responsible for the missing dissemination of NGS-based liquid profiling within Germany and other European countries. In addition, the currently commercially available assays are not CE-certified and thus each assay has to be validated as a laboratory-developed test. Hence, this approach is currently primarily restricted to very few MTBs and is often not considered in clinical decision making. Due to cost, lower sensitivity, the variety of included targets, and the missing possibility to customize NGS panels to patient- and/or indication-specific needs, the use of NGS-based liquid profiling is restricted to situations where treatment adjustment is required, such as relapse or PD.

Liquid profiling for monitoring of cancer patients – on the way to personalized diagnostics

After identification of tumor specific genetic alterations (either from tissue biopsy or from NGS-based liquid profiling), these can be used as personalized molecular tumor markers for monitoring response to treatment. However, oncogenic driver mutations or variations under selection pressure are of limited suitability for this purpose, as a resistant subclone could develop under targeted therapy that would not be detected by this approach. Thus, at least one additional truncating variation present in all subclones should be selected for personalized follow-up. Because these truncating mutations are patient- and tumor-specific and often not covered by commercially available liquid profiling assays, at least not in small, cost-effective panels, the design and establishment of individualized, laboratory-developed assays becomes necessary. Digital technologies are best suited for such an approach, as they can be easily customized and are referred as the gold standard for liquid profiling in terms of sensitivity.

First, somatic tumor variation is confirmed by Sanger sequencing and presence in germline is excluded. Second, after optimization of designed ddPCR, the limit of detection and quantification are determined by serial dilutions. Third, the validated assay is used as a personalized tumor marker for monitoring of disease.

To illustrate the use of liquid profiling for personalized follow-up, the case of a 51-year-old woman with metachronous hepatic metastasized breast cancer is described. Three years after initial diagnosis, the patient had a newly occurring solitary liver metastasis that was surgically removed. Tissue-based genetic testing revealed a TP53 deletion among other sequence variations. This was selected for the establishment of a personalized digital droplet PCR (ddPCR) assay. Figure 3 shows that the identified variation was somatic, as it could not be detected in germline DNA and was therefore suitable as a target for an individualized assay. After design of specific primers and probes, the optimal annealing temperature was assessed and limit of detection and quantification were determined during assay validation. Subsequently, the assay could be used for monitoring. As shown in Figure 4, the patient had stable disease thereafter, as indicated by imaging and ddPCR findings.

Figure 3:

Process to establish a liquid profiling personalized assay: TP53 deletion.

Figure 4:

Liquid profiling laboratory developed test (LDT) for personalized follow-up: TP53 deletion.

After validation of a personalized assays, this was used for monitoring of response to therapy of this 51 year-old female breast cancer patient. During follow-up imaging revealed stable disease, and LP confirmed this by absence of detectable ctDNA.

This case demonstrates that liquid profiling can be used as a personalized molecular tumor marker for monitoring treatment response. To ensure the required sensitivity for the detection of micrometastatic lesions and to optimize cost-effectiveness, a ddPCR approach is currently one of the most appropriate options. However, these tests need to be validated as laboratory-developed tests and thus require a high validation effort. Another obvious limitation is the lack of suitable time points for follow-up of patients, e.g., whether a predefined time interval in line with imaging is appropriate or should be adjusted.

Integrative evaluation of liquid profiling and imaging findings enabling precision diagnostics

Based on the cases presented above, the benefits of liquid profiling during the oncology treatment process have been demonstrated. Yet, there are situations in which liquid profiling can provide false negative results. These include situations in which ctDNA is not shed into the circulation, such as in case of solitary brain metastases or in cases of pleural affection [9, 20]. In addition, liquid profiling based on genetic alterations does not provide topologic information. This can be supplemented by protein tumor markers or, more precisely, by imaging. Imaging techniques include CT, but also more sensitive modalities such as dedicated liver MRI or PET/MRI, which provide precise topologic information. Thus, localized, advanced and disseminated disease can be distinguished, allowing further stratification of treatment. However, molecular targets cannot be identified by imaging, and the procedures may be associated with radiation exposure in some cases and relevant costs in others.

Because the benefits of liquid profiling and imaging are complementary, integrative diagnostics represent a promising approach to address the mutual limitations.

Summary and outlook

In conclusion, the benefit of liquid profiling has been demonstrated in various studies for several clinical scenarios. To date, clinical utility has only been proven for companion diagnostics. Due to the lack of large-scale prospective clinical trials, the lack of harmonized pre-analytical and analytical workflows, and the limited reimbursement options, liquid profiling has not yet been fully implemented in standard care and is currently at the threshold of implementation in clinical workflows in Germany. However, initial real-world data and the experience reported in this manuscript underscore the feasibility and benefit of liquid profiling in the management of cancer patients, not only for companion diagnostics but also for monitoring of response to treatment. The potential for optimization identified in the case reports can be exploited through better interdisciplinary collaboration with imaging experts. This allows all diagnostic aspects to be considered in a patient-specific context and can thus help to foster truly integrated decision-making processes towards precision diagnostics.