Liquid biopsy investigates circulating tumor cells (CTCs) and/or cell-free nucleic acids in the peripheral blood of cancer patients (Fig. 1) and is considered one of the most advanced non-invasive diagnostic systems with which to obtain key molecular information relevant to clinical decisions and the practice of precision medicine (1-5). Diagnostic actions include, but are not limited to, early diagnosis, staging, prognosis, the prediction of therapeutic responses, and follow-up during therapeutic intervention (5-13). Historically, the applications of liquid biopsy for the characterization of cancer patients have been focused on CTCs (1). Looking for CTCs in peripheral blood has generated a very large number of reports focusing on diagnosis, prognosis and therapeutic management (6). The downstream characterization of CTCs, including the identification of possible therapeutic targets (e.g., mutations or other traits of aggressiveness) in this peculiar tumor cell subset not only has had a great impact on diagnosis and prognostication, but also has an impact on clinical protocols, charting the route to precision medicine (14-16). In this respect, an excellent example is colorectal cancer (CRC), one of the most frequent malignancies worldwide (17). As is known, the transformation of normal colonic epithelium into CRC is punctuated by the progressive accumulation of acquired genetic and epigenetic alterations deeply altering morphological parameters, cell growth potential and differentiation, and shutting down apoptosis. Recent basic and clinical research on CTCs in patients with CRC has underlined that the molecular detection of CTCs in peripheral blood is feasible, and their phenotypic characterization drives therapeutic protocols for tailored clinical interventions (18). Moreover, the real-time monitoring of CTCs in patients with CRC has been extensively applied for a better mechanistic understanding of the factors determining clinical outcome and the efficacy of therapeutic treatment, as well as the stability of therapeutic effects over time (18-21).

In addition to CTCs, the formal demonstration that free nucleic acids are present (although short-lived) in biological fluids (plasma being investigated by most authors), has led to the development of a large wealth of studies aimed at circulating DNA and RNA (22-24). This strategy, similar to CTC detection, allows for non-invasive diagnosis, and at the same time it represents a convenient method for directly interrogating tumor aberrations, addressing tumor heterogeneity and metastatic dissemination across multiple, longitudinally collected clinical specimens (6,7). On the other hand, this approach suffers from important drawbacks, i.e., the fragmentation of circulating free DNA (cfDNA), the instability of RNA, low analyte concentrations, and the confounding, variable presence of DNA and RNA from normal tissues and mutated cells from the hematopoietic compartment (clonal hematopoiesis) (25). Limitations notwithstanding, the analysis of cfDNA has successfully identified mutations arising in, and the pathognomonicity of, tumor DNA, while the analysis of circulating free RNA (cfRNA) has been mostly focused on miRNA patterns strongly associated with neoplastic transformation/progression (22). Examples of the detection of tumor cfDNA are presented in Table I (18-20,26-52), while examples of the detection of circulating miRNAs are presented in Table II (53-104).

In order to identify specific DNA mutations and quantify miRNA levels in plasma and other body fluids of cancer patients, several types of technologies for DNA/RNA analysis have been proposed. For cfDNA analysis, the golden standards are possibly quantitative PCR (qPCR) and digital PCR; however, several additional technologies have been proposed (Table I), such as polymerase chain reaction-single strand conformation polymorphism (PCR-SSCP) analysis (27), multiplex digital PCR (dPCR), allele-specific qPCR (18,39), whole genome sequencing (WGS) (28), cancer personalized profiling deep sequencing (Capp-Seq) (30), methylation-specific PCR (31,37,44), the Discrimination of Rare EpiAlleles by Melt qPCR (DREAMing) (33), bidirectional pyrophosphorolysis-activated polymerization (bi-PAP) real-time PCR (40) and tagged-amplicon deep sequencing (TAm-Seq) (42). For miRNA analysis, qPCR and reverse transcription (RT)-PCR (53-58), NGS RNA sequencing (63), miRNA microarray analysis (60) and digital PCR (65) are the most commonly used technologies (Table II).

A common step, and under many respects a complication of all the above-mentioned technologies, is the need to amplify the minute amounts of target analytes by an enzymatic reaction with DNA modifying enzymes, most often Taq polymerase and its derivatives. Biosensing platforms hold great promise for the simple and rapid detection of cfDNA and cfRNA (135), since they skip this time-consuming, analyte-dependent, PCR amplification step. Novel PCR-free biosensing approaches are able to detect KRAS and BRAF mutations in the serum of patients with lung cancer and melanoma (136).

Digital PCR (137) is based on the limiting dilution of DNA, and single molecule detection to identify and quantify the target mutated DNA in a given sample (138,139). This experimental approach is very useful for the identification of rare variants and in non-invasive diagnosis on peripheral blood, since only a small concentration of template is required for the analysis. Next generation sequencing (NGS) is a high throughput DNA sequencing technology which allows for the analysis, in a single reaction, a large variety of different DNA aberrations across multigene panels (140,141), although comprehensiveness may somewhat detract from sensitivity. Different commercial NGS platforms are available, such as Genome Analyzer and HiSeq 2000 (Illumina), HeliScope (Helicos BioSciences), SOLiD and Ion Torrent (Life Technologies), Roche/454 (Roche). In these instruments, templates, primers or polymerase enzymes are immobilized on a solid support or on microbeads before sequencing, allowing the process of millions of microreactions carried out in parallel on each spatially distinct template.

However, as already pointed out, several challenges are related to liquid biopsy, the most important of which is the amount of target molecules to be detected and quantified. As far as cfDNA detection is concerned, these target molecules are so diluted by normal DNA that existing sequencing methods, such as Sanger sequencing, were not considered sufficiently sensitive to detect tumor-associated DNA mutation. As shown in Table I, the most commonly used approach was based on mutation-specific PCR, a technology proven to exhibit sufficient specificity and sensitivity allowing for the detection of the weak tumor signal present in the patient's circulation. This technology may be associated with important drawbacks when the quantification of miRNAs is considered, suffering from biases in the template-to-product ratios of the amplified target sequences (141). In addition, differential RT efficiency on different miRNA targets may also introduce variability when miRNA patterning is considered. Once again, PCR-free detection strategies are of great interest (142,143).

Liquid biopsy is a complex strategy requiring pre-analytical steps, post-analytical optimization, and the careful selection of optimal analytes for specific biological queries. In vivo model systems may be very useful in addressing and isolating these numerous individual variables (that are both technical and biological), and validate complex multi-step approaches. It is surprising, in this respect, that only few reports are available focusing on the use of animal models. For example, Garcia-Olmo et al directly compared the tumor ctDNA concentration and the number of circulating cancer cells in rats with xenograft tumors during the spread of CRC (144). Of note, they found that high ctDNA levels preceded the presence of CTCs. Rago et al (145) developed an elegant and highly sensitive qPCR test to quantify ctDNA by targeting LINE-1 in mouse xenografts, demonstrating that this experimental system enables the monitoring of systemic tumor burden and close examination of the therapeutic management on a variety of animal tumor models. These studies demonstrate the importance of ctDNA and how it intertwines with CTCs. In a more recent study, Thierry et al (146) evaluated the relative quantitative contributions of non-tumor, tumor and mutated ctDNA, as well as ctDNA integrity, in an animal model. In this case, they found differences between patients with CRC and nude mice xenografted with human colon cell lines, suggesting that further research is necessary to validate in vivo model systems based on mice xenografted with tumor cell lines.

As for miRNAs, different independent studies have firmly demonstrated that miRNAs released into the circulation by tumor xenografts are distinct from 'background' mouse miRNAs. This is a key point, since pre-existing miRNAs present in mouse body fluids may be a powerful confounding parameter, possibly altering conclusions and implications of any circulating miRNA signature. In this respect, the use of laboratory mouse strains has the advantage that it sets a 'background' mouse miRNA pattern that is stable and easily quantifiable. Mitchell et al demonstrated that several miRNAs originating from xenografted human prostate cancer cells are present in the circulation (one of the most interesting being miR-141), and are readily measured in plasma, allowing a clear distinction between tumor-xenografted mice and controls (147). Selth et al (148) performed global miRNA profiling and identified a set of miRNAs exhibiting significantly altered serum levels in transgenic mice bearing prostate adenocarcinoma tumors. Among the most interesting miRNAs, they focused their attention on miR-141 and miR-375. Waters et al observed a complex miRNA dysregulation in the circulation of athymic nude mice subcutaneously injected with MDA-MB-231 cells. Some miRNAs (such as miR-10b) were undetectable in the circulation, some others (miR-195 and miR-497) were significantly decreased, the miR-221 content was not altered, and a positive correlation was observed between miR-497 and miR-195. That study highlighted the distinct roles of miRNAs in the circulation and in disease dissemination and progression, all of which may be candidates as molecular targets for diagnosis, as well as for systemic therapy (149). More recently, Greystoke et al developed a robust protocol that allowed for the specific profiling of human tumor miRNAs in microliters of tail vein plasma (150). In a recent study, Gasparello et al presented the analysis of KRAS variants and the content of miR-141, miR-221 and miR-222 in mice xenografted with colon cancer cell lines (151). These results support the existence of multiple, finely tuned (non-housekeeping) control gateways that selectively regulate the release/accumulation of distinct ctDNA and miRNA species in culture and tumor xenograft models (Fig. 2).

ctDNA and miRNAs find application in a variety of clinical cancer settings.

Despite the fact that the majority of the analytical technologies are based on PCR and RT-PCR (see the Technologies section above and Tables I and II), PCR-free methods have attracted great interest in biomedicine. In fact, several articles have been published dealing with PCR-free methods for the detection of point mutations. In addition to the already cited limitation of PCR-based approaches, the need for repeated steps involving heating and cooling is an important limitation of all the PCR-based technologies, particularly when the PCR steps for the amplification of nucleic acids are associated with procedures performed in microfluidic-based devices (143,161-164). Several alternative isothermal-amplification methods (which do not require thermal cycling) have been developed to overcome this limitation, including nucleic-acid-sequence-based polymerization (NASBA), loop-mediated amplification (LAMP), helicase-dependent amplification (HAD), rolling-circle amplification (RCA), recombinase-polymerase amplification (RPA) and multiple-displacement amplification (MDA) (142). Recently, isothermal circular-strand-displacement polymerization (ICSPD) has emerged as a novel and promising method for nucleic-acid amplification and detection (163). Finally, promising opportunities are offered by selected isothermal amplification approaches that are based on a simple design of the amplification process and can be integrated in microfluidic devices (142). Such methods allow for the minimization of potential sample contaminations and minimize the sample volume required for the analysis (143). In this respect, the direct detection of point mutations in non-PCR-amplified human genomic DNA has been recently demonstrated by surface plasmon resonance imaging (SPR-I). Attomolar concentrations of target genomic DNA have been detected, demonstrating the ultra-sensitivity of the new method and its potential application in several biomedical fields, including liquid biopsy methods (164).

Tables III (165-173) and IV (174-182) summarize patents and patent applications related to the development of liquid biopsy protocols in cancer diagnosis. It is of interest to go through the claims of these patents, as they reflect the consideration given to liquid biopsy by a large part of the scientific community. Several patents build on the concept that mutations of tumor-associated genes (present in cfDNA from body fluids) and/or miRNA profiles are prognostic (associated with outcome), and/or predictive (associated with susceptibility to specific treatments). Examples of patents related to cfDNA and miRNAs are numerous and have steadily increased over the years. In fact, CTC counts, and molecular signatures originating from or associated with CTCs have been shown to be associated with conventional cancer molecular genotyping in tissues.

For instance, in EP2426217A1 (166) a method is described for detecting cell free nucleic acids, preferably cfDNA in a body fluid sample from an individual or a patient. A general claim, present in many other similar patent applications, relates to a method that comprises the step of accurately and sensitively determining the concentration of cell free nucleic acids in the sample and/or the index of integrity of said cell free nucleic acid and/or the determination of the presence of genetic polymorphisms [such as known single nucleotide polymorphisms (SNPs) or mutations]. The invention encompasses also a method to discriminate body fluid individuals where cfDNA are highly released. The majority of the approaches described in Table III have been validated on a variety of body fluids (urine, saliva, serum, plasma, bone marrow, lymphatic fluid, lacrimal fluid, serous fluid, peritoneal fluid, pleural fluid, ductal fluid from breast, gastric juice, or pancreatic juice) and cancers [breast cancer, CRC, periampullary cancer, melanoma, prostate cancer, gastric cancer, leukemia/lymphoma, renal cell carcinoma, hepatocellular carcinoma (HCC), neural-derived tumor, head and neck cancer, lung cancer, or sarcoma]. In addition to SNPs and tumor-associated mutations (presented in EP2426217A1, US7718364B2, WO2016168844A1 and US20160053301A1) (166,168,171,173), another interesting marker is DNA methylation. In US7718364B2 (168), a method is provided for assessing allelic losses and the hypermethylation of genes in the CpG tumor promoter region on specific chromosomal regions in patients suffering from melanoma, neuroblastoma, breast, colorectal and prostate cancer. The method is based on the evidence that free DNA and hyper-methylation of genes in the CpG tumor promoter region may be identified in the bone marrow, serum, plasma and tumor tissue samples of cancer patients. Table III lists examples of patents and patent applications focusing on cfDNA analysis in the body fluids of cancer patients.

As far as miRNAs are concerned, US8216784B2 (175) and EP2806273B1 (176) deal with cancer-derived microvesicle-associated miRNAs as a diagnostic marker for the detection of cancer. The method is based on the analysis of one or more miRNAs selected from a group comprising miR-21, miR-141, miR-200a, miR-200b, miR-200c, miR-203, miR-205 and miR-214. The method can be applied to a variety of biologicals, including milk, blood, serum, plasma, ascites, cyst fluid, pleural fluid, peritoneal fluid, cerebral spinal fluid, tears, urine, saliva, sputum, or combinations thereof. EP3011058A1 (178) is an example of patents focusing on total miRNA analysis. In addition, in this case, a shortlist of candidate miRNAs is provided, such as the combination of miR-365, miR-425, miR-143, miR-133a, miR-15a and miR-18a. The alteration in the level of the miRNA in the test sample (such as serum, plasma, or whole blood), relative to the level of a corresponding miRNA in a control sample, is indicative of the subject either having, or being at risk of developing, cancer, or the response of a subject to any treatment for the cancer. Further examples of miRNAs identified as cancer biomarkers are shown in Table IV.

Liquid biopsy is implemented (and is technologically tested) in several ongoing clinical trials (Tables V and VI) (183-204). For instance, NCT02639832 (183) is focused on the presence of tumor-derived CTCs or ctDNA using an investigational medical device known as the LiquidBiopsy. Using the LiquidBiopsy platform, recovered cells or DNA can also be investigated to obtain genetic information that may be useful to physicians for treating and understanding disease. The LiquidBiopsy device is able to purify the tiny numbers of tumor cells or ctDNA in blood. Even if a tumor is too small to be found by other means, such as an X-ray, it is possible that ctDNA or CTCs may be found in the blood. Genetic information can then be recovered from these cells or DNA to identify genetic alterations that are related to abnormal growth in a tumor. The proponents claim that this will potentially allow researchers to study tumor cells or tumor DNA from a blood sample instead of a biopsy sample, and may influence cancer diagnosis, treatment and drug selection in the future. In NCT02784639 (195) a method is employed which simultaneously allows the determination of three parameters: The specific quantification of tumor-derived ccfDNA, the ccfDNA fragmentation index and SNP or point mutation detection. The evaluation and validation of the method will be performed by determining the KRAS/BRAF mutational status prior to anti-EGFR therapy in patients with CRC. The protocol will detect the six most frequent KRAS mutations in CRC (G12D, G12V, G13D, G12S, G12C and G12A) and BRAF V600E. The goal of this multicenter prospective study is to validate, and ultimately translate in routine clinical practice, the use of plasma analysis of cfDNA for the determination of KRAS mutation status in patients with CRC. Table V lists several examples of ongoing clinical trials based on the analysis of cfDNA in breast, lung, colorectal, stomach, gastric and pancreatic cancers, HCC, non-Hodgkin lymphoma and melanoma.

As regards clinical trials related to miRNAs, one example is NCT02928627 (198), which is focused on miRNAs, such as miR-221 and miR-222, that are both dysregulated in the tumor tissues in approximately 80% of patients with HCC. The aim of that study was to evaluate whether these two miRNAs are overexpressed not only in tumor tissues, but also in the blood of cancer patients. An association between tumor tissue and blood levels will also be evaluated. Table VI lists additional examples of ongoing clinical trials in breast, esophageal and prostate cancer, HCC, lymphoma and leukemias.

The interest in non-invasive tumor diagnosis by liquid biopsy is demonstrated by several international projects supported by public funding. A few selected examples (LIQBIOPSENS, CANCER-ID, PRECISE, BILOBA and ULTRAPLACAD) are reported in the following sections irrespectively of their success in terms of delivered products (manuscripts, validated protocols and platforms, or patents).

The liquid biopsy of cancer is mainly based on the analysis of CTCs and/or cell-free nucleic acids in the peripheral blood of cancer patients, as well as in other body fluids suitable for diagnostic assessment. Among these, cerebrospinal fluid for tumors of the central nervous system, saliva for tumors affecting the head and neck, pleural effusion in the case of respiratory tract cancers and urine for urinary tract cancers. At present, liquid biopsy should be considered one of the most advanced non-invasive diagnostic systems suitable for performing key clinically relevant actions possibly leading to precision medicine. Historically, the applications of liquid biopsy for the characterization of cancer patients have been focused on CTCs. More recently, this analysis has been extended to cfDNA and miRNAs, demonstrated to be associated with cancer, with potential applications in early diagnosis, staging, prognosis, prediction of therapeutic responses, therapeutic outcome, and follow-up during therapeutic intervention. Liquid biopsy measures alterations in gene structure, regulation and expression that are the hallmarks of cancer. These analytes include nucleotide variants, promoter methylation, copy number variations of specific genes, chromosomal rearrangements, mutations affecting transcription, splicing and RNA maturation, translational efficiency and differences in miRNA signatures. The analysis of all these parameters can be approached by a variety of technological platforms. Moreover, the great interest of liquid biopsy is that this approach avoids certain key issues associated with invasive surgical biopsy. These include, but are not limited to: i) A static representation of the tumor pathology strictly limited to the tumor tissue sampling; ii) ethical and practical issues preventing repeated tissue biopsy; iii) tumor heterogeneity, particularly during progression and metastatic dissemination (making multiple sampling necessary); iv) easier patient monitoring by non-invasive analytical procedures (i.e., liquid biopsy). Therefore, despite the fact that the liquid biopsy approach suffers from important drawbacks (fragmentation of cfDNA, instability of RNA, low yield of isolated samples to be analyzed and variable presence of normal DNA and RNA) this approach is generally deemed of great interest for future applications, patent development and clinical trials. These will ultimately verify the potential of liquid biopsy in cancer. It is not expected that liquid biopsy will replace surgical biopsy; however, it will probably complement within a few years the information routinely obtained by excisional, incisional, surgical and needle biopsies, becoming a tool of choice for the dynamic monitoring of patients during clinical treatment and during the long-term surveillance of their health status.