Predicting therapy response by analysis of metastasis founder cells: emerging perspectives for personalized tumor therapy

ABSTRACT

Introduction

Circulating tumor and disseminated cancer cells can be detected after surgical removal of the primary tumor in non-metastatic patients. Despite being considered the prime targets of adjuvant therapy, they have not been implemented into clinical decision making yet.

Areas covered

Here we review the recent progress in understanding the biology of circulating tumor cells and disseminated cancer cells as well as the technical challenges associated with quantification, isolation, and preclinical model development. We highlight the first examples of clinical studies in which metastasis founder cells have been used as surrogate markers in adjuvant cancer patients and address the current hurdles in implementing these in routine clinical application. Finally, we provide a perspective on how the combination of technologies to detect and isolate metastasis founders, single-cell multi-omics, development of preclinical models, and their drug responses in specific niches could improve personalized adjuvant treatment strategies.

Expert opinion

The specific target cells of adjuvant cancer treatment are metastasis founder cells that remain in the body of the patients after surgical removal of the primary tumor. We, therefore, believe that the success of adjuvant therapies will be improved by implementing circulating and disseminated cancer cells in future clinical decision making.

KEYWORDS:

• Circulating tumor cells
• disseminated cancer cells
• metastasis founder cells

1. Introduction

Precision oncology defines the strategy to match individual cancer patients with the best available therapy. The drug selection is based on molecular, cellular, or functional criteria that are thought to determine treatment responses. In metastasized patients, therapies targeting driver mutations that confer strong growth advantage, e.g., BRAF mutation in melanoma, or EGFR mutation in lung cancer patients initially shows strong efficacy, high response rates, and significant reduction of tumor mass. However, unfortunately in most patients relapses occur [1–]. Most targeted approaches in metastatic patients therefore successfully prolong patient survival by eliminating a predominant cancer cell clone, while therapy-induced selection pressure finally leads to acquisition of resistance mechanisms or growth advantage of rare subclones within the heterogeneous pool of cancer cells [4].

Consequently, the clinical success of personalized treatments is anticipated to be greater in early therapeutic settings when only a few cancer cells need to be eliminated. For instance, targeting minimal residual disease by adjuvant therapies to prevent metastatic relapses. In recent years, we have learned that only a scarce population of cancer cells within a primary tumor (PT) has the potential to survive all steps along the complex invasion–metastasis cascade and succeed in finally colonizing in distant organs [5]. So-called circulating tumor cells (CTCs) in blood as well as disseminated cancer cells (DCCs) which can be detected in distant organs of non-metastasized cancer patients, have already overcome many hurdles in the metastatic cascade and their presence can be correlated with poor outcome in most cancers [6,7]. Despite this knowledge and the fact that two-thirds of cancer deaths are attributed to metastasis [8], patient stratification for precision oncology in adjuvant therapy setting currently still relies on molecular information from PT cells, cells that in most patients have been removed by surgery. To successfully translate state of the art precision medicine approaches to adjuvant therapy, strategies have to be developed that include analysis of the earliest metastatic seeds, i.e., single CTC and DCC, optimally to prevent colonization in distant organs and outgrowth of metastases. However, the integration of single-cell analysis in personalized adjuvant treatment concepts is still hindered due to a range of clinical, biological, and technical challenges such as (i) routine tissue availability for quantification and isolation of single DCC and translation of multi-omics precision medicine tools to single cells; (ii) using CTCs/DCCs or their derivates as surrogate markers in adjuvant clinical studies; and (iii) establishment of preclinical CTC/DCC models for functional assays and drug testing.

Here, we review recent works to provide a better understanding of metastasis founder cell biology with direct implications for improving precision medicine in the adjuvant therapy setting. Finally, we offer a perspective on how PT-based personalized therapy concepts could be adapted to rare CTC-/DCC-based approaches for tailored treatment strategies targeting metastasis founder cells.

2. Identification of metastasis founder cells

Detection and quantification of DCCs have first been reported in the 1980s and 1990s [9,10] and their presence has been associated with reduced disease-free or overall survival in many different epithelial cancers [6]. Even with most advanced high-resolution imaging technologies, it is currently not possible to detect single DCC within organs. As a result, studies on DCCs have been mainly conducted in accessible mesenchymal tissues, such as bone marrow (BM) or lymph nodes (LN). In these organs, DCCs can be detected by specific staining against epithelial antigens, i.e., Epithelial Cell Adhesion Molecule (EpCAM, CD326) or cytokeratins, in 10–60% of patients with 1–10 cells per 2 million bone marrow cells [6]. Unfortunately, neither automated nor FDA approved systems currently exist for the detection of DCCs. Although several tools have been described [11], the consensus protocol for detection of DCCs still relies on cytology staining and manual cell isolation by micromanipulation hampering its integration in routine diagnostic workflows [12].

In contrast, CTCs can be accessed by a minimally invasive blood draw that can be taken serially and has led to the use of the term ‘liquid biopsy’. Although not as progressed along the metastatic cascade as DCCs, during the last decade CTCs have been extensively investigated. Semiautomated workflows have been established to reliably detect CTCs in patients with epithelial cancers [13,14], which in the case of the CellSearch system has led to FDA approval for quantification of CTCs in breast, colorectal, and prostate cancers [15]. CTC quantification could be applied as a valuable marker in metastatic patients for risk stratification and monitoring of therapeutic efficacy with a detection threshold of 2–5 CTCs per 7.5 ml blood across cancer types [16–19]. Although the presence of CTCs in early cancer, typically taken before surgical removal of the PT, is also correlated with reduced progression-free and overall survival, their abundance is even lower compared to patients with metastases [20]. For instance, in metastatic stage IV breast cancer, greater than or equal to 5 CTCs/7.5 ml blood can be detected in 50% of patients [21], while only in 20% of stage I–III patients more than a single CTC/7.5 ml blood can be found [22].

3. Molecular characterization of metastasis founder cells

Since not all early CTC or DCC positive patients are doomed to develop metastases, it will be important to identify and fully characterize the fraction of actual metastasis founder cells. While cancer stem cells (CSC) are well described as a small subset of cells responsible for sustaining tumorigenesis by self-renewal and differentiation in the context of primary tumors [23] less is known about stem-like phenotypes of CTCs or DCCs, since it is still unknown whether single cells or clusters predominantly initiate metastases and to what extent other cell types within the metastatic microenvironment influence the process. To that end, molecular characterization of single CTC/DCC can provide important insights. To date, molecular analysis of early CTCs/DCCs has been mainly focused on genomic heterogeneity and evolution of cancer progression. In light of the data showing that most DCCs present fewer and discordant genomic abnormalities compared to their matched PT, it is fair to presume that they seed early during PT development and evolve at distant organs independently [24–28]. Several studies have shown that even actionable targets are disparate in early CTCs/DCCs and PT. In esophageal cancer, HER2 expression in DCCs has been shown to significantly differ from matched PT, and interestingly, HER2 gain on single DCCs, but not in PT, correlated with a high risk for early death [26,29]. In melanoma, the presence of BRAFV600E mutation was inconsistent between PT and LN-derived DCCs [28]. Discordance between DCCs and matched PT has also been reported for the expression of estrogen receptor as well as the presence of PI3K mutations in breast cancer [30–32]. With regard to HER2 expression, discordance between PT has been observed for both DCCs and CTCs in non-metastatic breast cancer patients [33–35] indicating that adjuvant therapies could be improved by implementation of routine CTC/DCC-based stratification.

Currently, the majority of the studies investigating molecular features of single CTC/DCC focus on the analysis of either genomic abnormalities or gene expression data [36]. As reported in recent tumor tissue-based precision medicine studies multi-omics approaches such as combined genomics and transcriptomics analysis can more precisely predict individual patient’s responses to therapies [37–39]. However, despite progress in establishing multi-omics approaches for single cells [40], considerable effort is still needed to translate these technologies and especially their integrative bioinformatics data analysis into comprehensive diagnostic workflows for rare single cells, such as CTCs and DCCs.

4. Metastasis founder cells as surrogate markers for adjuvant treatment success

In addition to the discussed technological hurdles, improved and especially expedited assessment of clinical outcomes for novel personalized treatment strategies tailored to metastasis founder cells are needed. Currently, adjuvant studies based on classical clinical endpoints such as survival or disease progression can take decades depending on cancer type. First studies corroborate the utility of CTCs as well as DCCs as surrogate markers for assessment of adjuvant therapy success. Gruber et al. repetitively quantified DCCs in BM of breast cancer patients during surgery for up to 24 months after PT removal. They observed that the majority of initially DCC-positive patients became DCC-negative in the course of adjuvant treatment and that persistence of DCCs after treatment was associated with disease recurrence [41]. Xenidis et al. quantified CK-19 mRNA-positive CTC by RT-PCR in blood samples of patients with early breast cancer before and after adjuvant chemotherapy. 51% of initially DCC-positive patients became DCC-negative after adjuvant chemotherapy and those patients that remained or subsequently became DCC-positive post-chemotherapy had an increased risk for clinical relapse [42]. These promising data indicate that CTCs/DCCs have the potential to be used as a tool for assessing the efficacy of adjuvant therapies, to shorten the duration of clinical trials for testing new drugs and for early detection of non-responsive patient populations. Furthermore, better stratification of patients benefiting from adjuvant therapies is important to avoid overtreatment and unnecessary therapy-related side effects. In this context, in a retrospective study the success of adjuvant radiotherapy was associated with CTC positivity in breast cancer [43]. In line with this study, the detection of DCCs in BM of patients with early breast cancer was associated with the success of regional radiation and systemic hormone therapy for locoregional relapse [44,45].

Moreover, few trials have shown that secondary adjuvant therapies can be guided by quantification or even by treatment relevant information of CTCs/DCCs. Geogoulias et al. quantified chemotherapy-resistant CTCs in stage I–III breast cancer patients and showed that subsequent trastuzumab treatment can eliminate CTCs and reduce the risk for disease recurrence [46]. In a second study, repetitive quantification of DCCs in BM of early breast cancer patients provided evidence that secondary adjuvant therapy can be guided by DCC quantification [47]. In the future, additional studies are needed to link the presence of CTCs/DCCs or better their molecular features to the treatment outcome of patients with adjuvant therapies and to develop tailored strategies for specific eradication of these metastatic progenitors.

5. Preclinical models representing patient-derived metastasis founder cells

A major barrier for a better understanding of the biology of metastasis founder cells and the development of improved adjuvant treatments constitutes the lack of functional data from representative preclinical models. Beyond the correlation between DCC detection and the clinical outcome of patients, there is still only little functional evidence as to whether DCCs bear metastasis-inducing capacity in general, and if so which DCC populations in particular. Consequently, experimental data addressing seeding, metastatic colonization and drug target/pathway identification rely mainly on mouse models and/or human cell lines derived from PT or metastases that genetically and presumably also functionally differ from human DCCs [25,48,49]. Reports addressing specific model generation from patient-derived DCCs are rare, given the cultivation and expansion of DCCs is challenging on account of their low abundance and the difficulties associated with isolation of viable DCCs. In the case of BM, short-term culture of cytokeratin positive cells from early-stage prostate, breast, colon, and kidney cancer patients has been demonstrated [50,51]. Only selective immortalization of micrometastatic carcinoma cells using integration of SV40 DNA has enabled long-term cultivation of BM-derived DCCs from prostate, breast, lung, and colon cancer patients without overt metastases [52]. In addition, our group was recently able to successfully develop stable DCC-derived models from LN of non-metastatic melanoma patients at early stages of metastatic progression using a combination of in vitro spheroid culture and transplantation into NSG mice [28]. Interestingly, model generation was possible only if DCCs were derived from small colonies, while not possible for LN with single DCC [28].

Efforts in generating DCC-derived models might benefit from recent technical developments made in the field of liquid biopsy through optimizing the isolation and culturing conditions of CTCs from metastatic patients. Within the last decades a range of methods have been developed to specifically isolate CTCs [53] and to cultivate these rare cells by optimized in vitro culture conditions such as spheroid and organoid cultures, or by in vivo transplantation into immunodeficient mice [54–58]. To date, the generation of CTC-derived models has been reported for melanoma, breast, colorectal, lung, gastroesophageal, pancreatic, and prostate cancer patients (Table 1) [54–69]. CTC models from metastatic melanoma patients that recapitulate the patient’s disease and responses to therapies have been developed, highlighting the power of such models as potential ‘avatars’ for testing personalized treatments in the future [60]. Furthermore, microfluidic devices have been shown to facilitate CTC expansion from metastatic patients as in vitro spheres, thus allowing rapid testing of CTC models for drug sensitivity by panels of single drugs and drug combinations [58,70]. However, generation of CTC-derived models from early-stage patients mirroring the target cells of adjuvant therapy has only been shown to be successful in short-term cultures, presumably owing to their even lower abundance [71].

Table 1. Overview of studies reporting established CTC-derived preclinical models from advanced patients

Type of cancer	Type of culture	Enrichment method	Expansion Technology	Number of permanent CTC models	Success rate (models per cases)	Reference
Breast cancer	in vivo	RosetteSep/flow cytometry sorted	Transplantation in femoral medullar cavity	N/A	3/110 (3%)	Baccelli et al., 2013 [54]
Breast cancer	in vitro	Flow cytometry sorted	2D cell culture	3	3/8 (37.5%)	Zhang et al., 2013 [61]
SCLC	in vivo	RosetteSep	s.c. injection with matrigel	4	4/6 (67%)	Hodgkinson et al., 2014 [57]
Prostate cancer	in vitro	RosetteSep	Organoid culture	1	1/17 (6%)	Gao et al., 2014 [56]
Breast cancer	in vitro	iChip	Sphere culture	6	6/36 (17%)	Yu et al., 2014 [58]
Colon cancer	in vitro	RosetteSep	Sphere culture	1	1/71 (1%)	Cayrefourcq et al., 2015 [55]
Melanoma	in vivo	RosetteSep	s.c. injection with matrigel	6	6/47 (13%)	Girotti et al., 2016 [60]
NSCLC	in vivo	RosetteSep	s.c. injection with matrigel	1	1/1 (100%)	Morrow et al., 2016 [62]
Colon cancer	in vitro	RosetteSep	3D sphere culture	2	3/16 (19%)	Grillet et al., 2017 [63]
SCLC	in vivo	CTC-iChip RosetteSep/Ficoll	s.c. injection with matrigel	16 1	16/42 (35%) 1/4 (25%)	Drapkin et al., 2018 [65]
Breast cancer	in vivo	Gradient centrifugation	s.c. injection with matrigel	1	1/32 (3%)	Pereira-Veiga et al., 2019 [64]
Breast cancer	in vivo	Flow cytometry sorted	Intracardiac injection	1	1/3 (33%)	Vishnoi et al., 2019 [66]
Prostate cancer	in vivo	RosetteSep	Intercapsular fat pad with matrigel	1	1/22 (5%)	Faugeroux et al., 2020 [67]
Breast cancer	in vitro	RosetteSep	2D cell culture	1	1/50 (2%)	Koch et al., 2020 [68]
Gastroesophageal cancer	in vitro	RosetteSep	2D and 3D sphere culture	2	2/23 (9%)	Brungs et al., 2020 [59]
Pancreatic cancer	in vitro	Labyrinth microfluidic device	2D cell culture	3	3/10 (30%)	Rivera-Baez et al., 2020 [69]

6. Emerging technologies to expand and model patient-derived metastasis founder cells

Increasing the number of available CTCs for analysis could significantly increase the likelihood for generating models from non-metastatic patients. Recently, it has been calculated that the probability to detect sufficient number of CTCs is increasing with analyzing larger blood volumes [72]. This hypothesis has been utilized by isolating CTCs from leukapheresis products instead of venous blood draw [67,73]. A small prospective study in 23 cancer patients showed an increase in CTC detection rate from 28% in peripheral blood to 72% in leukapheresis product [73]. To that end, technologies that use biomedical devices for intravenous CTC capture have been developed [74].

Additionally, a considerable amount of evidence indicates that the metastatic niche (e.g. hypoxia, immune cells, vasculature, and extracellular matrix) strongly influences the seeding capacity and phenotype of DCCs, and consequently their outgrowth as metastatic colonies [75]. Exciting recent developments in establishing in vitro or ex vivo systems mimicking human organs can recapitulate metastatic niche microenvironment for investigating DCC biology and drug responses, e.g., natural matrices, synthetic hydrogels, and scaffolds, iPS-based niches, organ on a chip technologies, tissue engineering, or tissue slices from different organs might serve as ideal systems in the future [76–84].

7. Conclusion

Detection of early DCCs and CTCs has been shown to correlate with systemic progression of most solid cancers and therefore at least some of these cells constitute founder of metastasis and target cells of adjuvant therapies. With state of the art technologies CTCs/DCCs can be isolated and their molecular analysis clearly indicates that information derived from these rare cells may potentially support the identification of novel and more therapeutically relevant targets. Their quantification and molecular information has been shown to serve as surrogate markers for adjuvant treatment and improve patient stratification. However, the hurdles in obtaining the respective tissues and lack of automated procedures to quantify, isolate, and analyze these rare cells and to develop representative preclinical models of metastatic founder cells has so far hindered their routine clinical application.

8. Expert opinion

The success of personalized cancer therapy will be markedly improved if the molecular, cellular, or functional information used for therapy selection would precisely mirror the features of cells responsible for disease progression. Therefore, it is essential for adjuvant therapy strategies to address the earliest seeds constituting rare metastasis founder cells rather than a heterogeneous pool of surgically removed PT cells only. Identification and more specifically targeting of metastasis founder cells would bear the potential to evolve the currently rather empirical concept of adjuvant treatment into a real precision medicine approach in order to more efficiently prevent metastatic relapses.

Early CTCs and DCCs have been shown to represent metastasis founder cells and their presence correlates with disease progression in most solid cancers. However, in order to implement early CTCs and DCCs in routine clinical decision-making practice there is yet much to be learned. What are the molecular and actionable features of CTCs/DCCs? How do these cells functionally respond to therapies within their metastatic niche microenvironment? Which of the detectable CTCs/DCCs bear metastasis-inducing capacity in patients? To answer these questions and translate bulk tumor tissue-based diagnostic tools into rare CTC/DCC approaches, improved methodologies for detection and isolation of, ideally viable, CTCs, and especially DCCs are needed. Here, much could be learned from the efforts and technological advancements in the field of liquid biopsy in metastasized patients which have to be transferred to earlier disease settings. Routine sampling of BM and LN should be integrated into automated workflows for detection and isolation of DCCs with standard diagnostic precision and accuracy (Figure 1).

Figure 1. Perspective for implementation of CTCs and DCCs in adjuvant precision medicine. Routine sampling of BM and LN should be performed and technologies for identification and isolation of CTCs/DCCs should be improved. Existing multi-omics technologies including genome, DNA methylome, transcriptome, small RNAs, and proteome data as well as corresponding bioinformatics analysis tools addressing single metastasis founder cells will support the discovery of improved biomarkers. In parallel, first preclinical CTC/DCC models should become available for discovery of novel adjuvant therapies and/or repurposing of approved and investigational drugs in the context of adjuvant treatment. Molecular and functional data should be linked to clinical information to develop and initiate innovative clinical trials

Abbreviations: PT: primary tumor, BM: bone marrow, LN: lymph node, CTC: circulating tumor cell, DCC: disseminated cancer cell

To better understand the biology of metastasis founder cells and to accurately predict therapy responses, the workflows for identification and isolation of early CTCs and DCCs should be combined with multi-omics data analysis. Single-cell technologies to obtain comprehensive information including genome, DNA methylome, transcriptome, small RNAs, and proteome data of single cells as well as corresponding bioinformatics analysis tools already exist, however they are yet to be applied to metastasis founder cells. This will help to identify the CTCs/DCCs that potentially colonize and discover specific biomarkers for patient stratification and activated pathways that can be therapeutically targeted. Additionally, much needed DCC- and early CTC-derived models could be potentially established using protocols developed for CTC expansion in the metastatic setting in combination with approaches to increase sample volume and thereby DCC/CTC number. As an alternative, considering recent advancements in the field of genome editing [85,86], one approach could also be to systematically introduce the earliest genetic, epigenetic and transcriptomic alterations identified by comprehensive molecular analysis of DCCs and early CTCs into normal nonmalignant cells by CRISPR-based technologies. In principle one could also reversely transform more advanced colonizing DCC models into their earlier and more immature counterparts by per se correcting specific molecular alternations which drive the transition of non-colonizing to colonizing DCCs by CRISPR technologies. Patient-derived or artificial preclinical models could then be subjected to high-throughput drug screens with follow-up hit validation in more complex systems representing progenitors in the context of their metastatic niche microenvironment (e.g., iPS niches, hydrogels, tissue slices), or alternatively perform small-to-medium scale drug screens directly in those complex multicellular systems for precision medicine and drug repurposing. Similar to recent studies on bulk tumor tissues [37,87–90], functional information could be combined with multi-omics data from single DCCs/CTCs to identify new drug targets and to predict responses to existing drugs. Ultimately, these findings have to be thoroughly examined and validated within innovatively designed clinical studies aimed at improving therapeutic strategies for adjuvant cancer patients in a truly precision medicine approach. If such comprehensive molecular and functional information could be linked to surrogate markers of adjuvant target cells in clinical settings, better stratification, improved personalized therapies and in case of resistance early therapy adjustment could be achieved. If these challenges can be tackled successfully, we anticipate that within a five year period sampling, identification and isolation of CTCs/DCCs will be improved, multi-omics data and first functional information for biomarker and drug discovery would be generated which can be examined within clinical trials in the long term (Figure 1). Although there is yet much work to be done, the necessary technologies seem to exist for tackling the hurdles that currently impede the implementation of metastasis founder cells into routine clinical workflows.

Article highlights

• Patient stratification for precision oncology in adjuvant therapy setting currently relies on molecular information from primary tumor

• Detection of circulating tumor cells (CTCs) and disseminated cancer cells (DCCs) predict poor outcome in non-metastasized cancer patients

• Actionable drug targets are disparate in early CTCs/DCCs and primary tumors

• Early CTCs as well as DCCs can be used as surrogate markers for assessment of adjuvant therapy success

• Protocols have been developed to establish preclinical models from metastatic CTCs that might pave the way for successful expansion of metastasis founder cells

• Linkage of multi-omics single cell data to CTCs/DCCs as surrogate markers in clinical settings might improve personalized therapy strategies for prevention of metastases development in future

Declaration of interest

The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

Reviewers disclosure

Peer reviewers on this manuscript have no relevant financial relationships or otherwise to disclose.

Acknowledgments

The authors thank S. Pausch, K. Weidele and S. Treitschke for help with the figure and table, and C. A. Klein (CAK) for critical reading of the manuscript. This work was supported from grants of the Deutsche Krebshilfe (70112504 Schwerpunktprogramm ‘Translationale Onkologie’, DETECT CTC subproject 6 to BP and CAK) and of the Bavarian ministry of economic affairs, energy and technology (AZ 20-3410.1-1-2 to CAK).

Additional information

Funding

The authors declare funding from Bavarian ministry of economic affairs, energy, and technology, [AZ 20-3410.1-1-2] and Deutsche Krebshilfe [70112504, DETECT CTC Subproject 6].