Clinical Applications of Circulating Tumor DNA, Circulating Tumor Cells, and Exosomes as Liquid Biopsy-Based Tumor Biomarkers

Along with improved knowledge of cancer biology and biotechnical progress, the diagnostic approaches have evolved from tissue biopsies to liquid biopsies. As they provide a minimally invasive tumor detection, liquid biopsies allow early diagnosis and serial assessments of tumor progression. Discovery and use of circulating tumor markers circulating tumor DNA (ctDNA), circulating tumor cells (CTC), and exosomes have largely expanded the possibility of early diagnosis of cancer, patient stratification, as well as developing a personalized treatment. Based on these circulomes, liquid biopsies can be developed, but each type of liquid biopsies has its own merits and limitations. While ctDNA-based methods represent the most advanced techniques, sensitivity improvement is expected given the rarity of ctDNA in circulation. As intact cancer cells, CTC provide information on cancer cells. However, current CTC capturing procedures are still lack of efficiency. Exosomes are abundant, but they are highly heterogeneous and there is a lack of specific markers for identification. Future efforts are needed to improve operational parameters and clinical performance of each method. Prior to a broad use in clinical settings, it is crucial to standardize the procedure for the specific liquid biopsy method and validate the test with adequate specificity and sensitivity for clinical applications.


INTRODUCTION
Targeted therapies and immunotherapies have achieved a great success in treating cancer patients compared to conventional therapies such as chemotherapies. However, the treatment outcome largely depends on expression level of respective targets and immune checkpoints, which are usually assessed with a tissue biopsy. While tissue biopsies represent the gold standard for cancer diagnosis, frequent tissue biopsies are often impractical to perform due to invasiveness of the procedure and the risk of disease spreading it may incur [1,2]. Moreover, the characteristics openaccess of cancer tissue can evolve during disease progression, the evidence acquired from tissue biopsies is, therefore, a limited snapshot of specific lesion at a specific time [3]. This limitation leads to incomplete information [4], possibly resulting in errors for tumor diagnosis and/or treatment decision [5]. In contrast to tissue biopsies, liquid biopsies usually use blood specimens that are much easier to collect and can be used to enrich blood-derived circulomes such as circulating tumor DNA (ctDNA), circulating tumor cells (CTC), and exosomes. These circulomes have been found to be related to tumorgenesis and thereby they could be important tools for tumor identification and treatment follow-up. ctDNA is part of cell free DNA (cfDNA). In 1989, cfDNA carrying neoplastic characteristics mutations was reported in plasma of cancer patients [6]. These circulating DNA are actually ctDNA. Since then, tumor-derived alterations, including specific gene mutations, epigenetic alterations, and copy number variations (CNV) have been broadly observed in ctDNA of cancer patients. Several commercial ctDNA detection kits have been approved for cancer diagnosis, such as cobas EGFR mutation test v2 (Roche) and Epi proColon (Epigenomics), both of which are elaborately discussed in ctDNA section [7][8][9]. Similarly, CTC are also shed by tumors and have been described in patients of several cancer types, mainly breast, colorectal, and prostate cancers. CTC count of ≥ 5 CTC in 7.5 mL of blood has been demonstrated as a reliable prognostic tool for these cancers. CTC cluster was reported to have higher metastatic potential [10]. Now CTC detection has been incorporated into National Comprehensive Cancer Network (NCCN) clinical practice guideline (2017, v3) and American Joint Committee on Cancer (AJCC) staging manual (2018, 8th edition) for metastatic breast cancer diagnosis [11]. Unlike ctDNA and CTC, exosomes are abundant in biological fluids. They also closely correlate with tumors and are elevated in cancer patients compared to healthy controls. Moreover, exosomal proteins are significantly higher in cancer patients at advanced stages than those at early stages. The correlation between exosomal proteins and clinical outcomes could serve as predictive and prognostic biomarkers as protein levels of exosomes changed during and after chemotherapy [12]. Additionally, tumor derived exosomes (TDE) are able to induce oncogenic transformation of normal cells [13]. Thus, ctDNA, CTC and exosomes could serve as potential biomarkers for various clinical applications. This review focuses on the isolation procedures and the clinical applications of these circulating biomarkers for cancer diagnosis, tumor progression monitoring, patient selection, and evaluation of treatment effectiveness.
ctDNA Cancer cells have long been known for harboring mutations of genes essential for cell growth control. For example, it was reported that in breast and colon cancers, almost every cancer cell contains approximately 80 mutated genes on average [14]. The ctDNA are tumor-derived DNA fragments found in blood. The typical length of ctDNA ranges from 80 to 200 bps and peaks at 160-180 bps, which is a bit shorter than majority of cfDNA [15]. While mechanisms of ctDNA formation are not fully understood, apoptosis and necrosis of cancer cells are considered as major sources of ctDNA [16]. Active secretion of ctDNA by the cells in tumor tissue can be another mechanism [17]. The abundance of ctDNA was estimated to be less than 0.01% of total circulating DNA or cfDNA, equivalent to less than 10 ng/mL of plasma [18,19]. Due to the rarity of ctDNA in blood, efficient enrichment and sensitive detection are necessary for many applications of ctDNA. A variety of nucleic acid isolation kits are commercially available for fast and convenient isolation of cfDNA. Two approaches, PCR-based and NGS-based methods, are primarily available for detection of gene mutations in ctDNA. While the former is usually used to detect small number of gene mutations, the latter is employed for detection of gene mutations of a large panel. Nowadays, with advent of sensitive digital polymerase chain reaction (PCR) technology, one can quantify as little as 0.001% ~ 0.01% mutated alleles  [20]. In addition to these PCR-based approaches, a recently available technology named single molecular array (SIMOA) may provide another tool for ctDNA detection. It is based on direct detection of single molecules of DNA on magnetic beads. Briefly, ctDNA samples are subjected to denature to become single-strand DNAs, which are subsequently captured by magnetic beads through hybridization with specific probes (complementary sequences) attached to the beads. After hybridization with biotin-labeled probes, the complexes are incubated with streptavidin-ß-galactosidase for detection. Although SIMOA technology is usually used for protein analysis, it is supersensitive in DNA detection with a reported limit of detection of 0.07 fM [21], providing an attractive alternative to the methods that rely on DNA amplification. Along with discovery of tumor-driven gene mutations such as EGFR, BRAF, KRAS, and P53, ctDNA analysis has been extensively explored for cancer diagnosis and clinical response monitoring [22,23]. It was reported that a panel of 16 genes, including but not limited to EGFR, BRAF, KRAS, and NRAS, can be collectively used to detect eight common cancer types [24]. In a study of metastatic melanoma patients treated with immunotherapy, mutations of genes BRAF, NRAS, TERT and ALK were found in ctDNA samples from five of ten (50%) patients. Moreover, three patients with disease progression exhibited an increase in ctDNA levels [25]. In another study of breast cancer [26], ctDNA levels were reported to correlate with degree of tumor burden better than CA15-3, a typical protein biomarker used for monitoring breast cancer. ctDNA also displayed a better performance in monitoring the treatment response in this study. In addition to its potential applications in cancer diagnosis and treatment monitoring, ctDNA analysis of tumor-driven genes mutations can be used to predict drug resistance and cancer recurrence. For instance, dynamic ctDNA profiling of genes TP53, PIK3CA, mTOR, and Pten in HER2-positive breast cancer patients could be used to identify those resistant to anti-HER1/2 tyrosine kinase inhibitor therapy with sensitivity of 85.7% [27]. Additionally, simultaneous analysis of KRAS, TP53, PIK3CA and APC gene mutations in ctDNA samples from colorectal cancer patients could highly predict the recurrence of cancer within one year [28]. In colorectal cancer patients after surgical resection, postoperative gene mutations of KRAS, APC and P53 detected in ctDNA samples indicated a high risk of recurrence, especially in patients without the adjuvant chemotherapy treatment (79% versus 9.8% at median follow-up of 27 months) [29]. All of these results indicated that ctDNA could serve as a promising biomarker for patient selection, treatment evaluation and recurrence monitoring, as well as drug resistance prediction. Clinical uses of ctDNA analyses were reported in various studies and some of them are summarized in Table 1.
Currently, there are two ctDNA-based tests approved by Food and Drug Administration (FDA). The first one was approved in 2016 as a companion diagnostic for detection of EGFR mutation in patients with non-small cell lung cancer (NSCLC) (Cobas ® EGFR Mutation Test v2, Roche). This test is to select the patients who would benefit from erlotinib and osimertinib treatment [7,8]. It is based on an optimized quantitative PCR technology using pre-amplification to improve the detection efficiency of mutated alleles.
The clinical specificity and sensitivity of this test in NSCLC were 97.9% and 72.1%, respectively. Compared with tissue biopsy-based assay, the cobas® EGFR Mutation Test v1, the concordance between ctDNA and tissue biopsy analyses was 91% [30]. Another approved diagnostic test is Epi proColon (Epigenomics). This test is used for colorectal cancer diagnosis and it is based on the detection of ctDNA methylation [31]. Investigation of tumor epigenetics, including methylation and histone post-transcriptional modifications, represent important applications of ctDNA analysis [32]. In recent years, several studies have been conducted to analyze ctDNA methylation patterns for cancer screening, early diagnosis, and follow-up of the disease progression [33,34]. Some examples in this application are shown in Table 1 They are used to detect four major types of gene alterations, including point mutation, insertions/deletions, copy number changes, and gene fusions. Normally offered in the Clinical Laboratory Improvement Amendments (CLIA)-regulated and College of American Pathology (CAP)-accredited laboratories, these tests have shown good performance. As an example, Guardant360 ® exhibits clinical sensitivity of 85% and specificity of 99.9% in advanced stage solid tumors, and the analytical sensitivity and specificity of this test were 100% and 99.99%, respectively [9]. For further information about these LDT, readers can refer to review article [35]. While ctDNA analysis has shown a great potential in various clinical applications, it exhibits several challenges. First, it is difficult to obtain good quantity of ctDNA as its content in peripheral blood is extremely low (< 0.01% of 10 ng/mL plasma). Moreover, half-lives of ctDNA are reported to be 16 minutes to several hours [28], which are short and may partially explain low abundance of ctDNA. Second, ctDNA detection platforms can vary from genome-wide analysis to a single gene interrogation. These analyses exhibit different detection limits (0.1-1.0% for allele-specific PCR, 0.01-2.0% for NGS, and 0.01% for digital PCR and cancer personalized profiling by deep sequencing (CAPP-Seq)) [36]. Therefore, analysis results can vary with different platform technologies, rendering evaluation of results difficult among different laboratories that employ different platforms. Another challenge lies in poor concordance between liquid biopsies and tissue biopsies in certain cases. As an example, the concordance of EGFR L858R and 19del mutation between ctDNA-and tissue biopsy-based analyses in NSCLC patients was 91%, while it was only 61% for T790M mutation [37]. This might be attributable to different disease stages, as high concordance between liquid and tissue biopsies of PIK3CA mutation status was observed in advanced breast cancer patients, but a poor concordance was reported in patients at early-stage [38]. Combined analyses of other biomarkers with ctDNA samples could be used to improve the clinical value of ctDNA-based liquid biopsies. An example was provided by Cohen et al, who reported that through a combined analysis of eight protein biomarkers and ctDNA markers, 8 different cancer types could be identified with specificity of >99%, and sensitivity of 69% to 98%, depending on cancer type [24]. Additionally, a combined analysis of DNA from both exosomes and CTC might be an alternative to obtain more reliable results in clinical applications [39].

CTC
CTC are tumor cells that detached from primary and metastatic tumor lesions and enter into blood circulation. Since their discovery in 1869, CTC have been widely identified in patients with breast cancer [40], non-small cell lung cancer [41], prostate cancer [42], colon cancer [43], and pancreatic cancer [44]. As intact cells derived from tumor, CTC can provide valuable information on tumor composition, heterogeneity, invasiveness, and drug resistance. As early as in 2004, it was found in a clinical study conducted in breast cancer patients that the number of CTC was inversely correlated to progression-free survival (PFS) and overall survival (OS) of the patients receiving an anti-cancer treatment [45]. Similar findings were also reported in patients with prostate cancer and patients with colorectal cancer (CRC) [46,47]. In patients with castration-resistant prostate cancer, Danila et al reported that the 2-year OS was 46% in the patients with low level of CTC and 2% in the patients with high level of CTC, indicating a prognostic value of CTC [48]. More studies on the clinical applications of CTC enumeration are listed in Table 2.
In addition to CTC enumeration, genetic and protein analyses of CTC also offer a potential value for cancer diagnosis and prognosis. The commonly studied tumor-driven gene mutations were reported in CTC. For example, EGFR and PIK3CA mutations were observed in CTC of NSCLC and breast cancer patients, respectively [49,50]. Moreover, the presence of EGFR mutation was found to be associated with shorter PFS in NSCLC patients and  Table 3. The studies demonstrated a significant value of CTC in cancer diagnosis, prognosis and treatment response evaluation. Indeed, in patients with different cancers, CTC showed distinct pattern of CNV, which was consistent with results of tissue biopsy analysis from metastatic lesions in the same patients [51]. Similar results were also observed for DNA methylation in CTC. As reported by Chimonidou et al [52], the methylation patterns of tumor suppressor and metastasis suppressor genes in CTC were found remarkably distinct in patients versus healthy individuals, and in metastatic patients versus operable patients. All of these findings indicated that CTC could serve not only as a diagnostic biomarker, but also as a prognostic biomarker as they provided valuable information for tracking cancer metastasis. Protein analysis can be performed on CTC in clinical studies as well. A study conducted in patients with castration-resistance prostate cancer revealed androgen receptor variant 7 (Arv7)positive CTC as specific predictive biomarker for prostate cancer [53]. In another study, expression of prostate-specific antigen (PSA) and prostate-specific membrane antigen (PSMA) was analyzed in CTC [54]. It was found that these two proteins could represent different status of androgen receptor (AR) activation. Moreover, the elevated number of "AR-on" CTC (CTC with PSA + /PSMAphenotype) was associated with reduction in OS in patients. Recently, the expression of immune checkpoint inhibitors (e.g. PD-L1) and other therapeutic targets such as HER2 was also found in CTC [55]. The expression analysis of these therapeutic targets in CTC showed important clinical value in responsiveness and drug resistance evaluation. As an example, a study in NSCLC patients reported that the number of CTC with higher PD-L1 inversely correlated to clinical response. In other words, the more CTC expressing high level of PD-L1 in the patients, the stronger resistance to anti-PD1 antibody treatment [55]. The prevalence of CTC is approximately 5 to 200 CTC in 7.5 mL of blood. Given this rarity, CTC enrichment is critical for its clinical application. There are three strategies employed for CTC enrichment, including immuno-affinity-based capture, isolation by size of epithelial tumor cells (ISET) [56], and microfluidics-based methods [57]. Immunoaffinitybased capture approaches are most widely used, and microfluidics-based methods are usually used in combination with immunoaffinity-based capture [58]. The comparison of these enrichment methods is summarized in Table 4. The interested readers could also refer to several reviews for more information [59,60]. The most widely used platform for CTC enrichment is an epithelial adhesion molecule (EpCAM)-based immunoaffinity capture assay such as CellSearch ® (Veridex). EpCAM is an epithelial cell-specific marker ubiquitously expressed on normal epithelial cells and cancer cells, but not on blood cells, whereas CD45 is a leukocyte-specific marker. Cytokeratins (CK) are also epithelial cell-specific markers and highly expressed on cancer cells. Thus, CTC are defined as CD45 -CK + EpCAM + cells by immunofluorence staining. The cut-off threshold in this method for distinguishing positive from negative results is 5 CTC in 7.5 mL of blood. This method was approved by FDA in 2004 for prognosis of metastatic breast cancer [45].

Arv7
Castrationresistance PC Arv7-positive CTC were considered as a specific biomarker for evaluation of treatment outcomes [52] PSA and PSMA PC The expression of PSA and PSMA in CTC could represent the status of estrogen receptor activation, and reflect the response of hormone therapy HER2 BC CTC-HER2-postive patients showed longer PFS than CTC-HER2-negative patients, but no difference was observed between these patients for OS [101] Arv7: androgen receptor variant 7; BC: breast cancer; CNV: copy number variation; HNC: head and neck cancer; NSCLC: non-small cell lung cancer; PC: prostate cancer; PSA: Prostate-specific antigen; PSMA: prostate-specific membrane antigen; PAR2: proteinase-activated receptor-2; NSCLC: non-small cell lung cancer; Table 3. Summary of studies based on CTC analysis in cancer patients (Cont'd).

Target
Cancer type

Observations
Ref.

PAR2 Glioma
Exosomes positive for PAR2 mediate the hypoxia-induced angiogenesis. [74] PD-L1 NSCLC the number of CTC with higher PD-L1 in patients showed negative correlation to the clinical response [54] Arv7: androgen receptor variant 7; BC: breast cancer; CNV: copy number variation; HNC: head and neck cancer; NSCLC: non-small cell lung cancer; PC: prostate cancer; PSA: Prostate-specific antigen; PSMA: prostate-specific membrane antigen; PAR2: proteinase-activated receptor-2; NSCLC: non-small cell lung cancer;  [66]. Thus, the use of CellSearch ® has resulted in low sensitivity (5-30%) of CTC detection [67], as it detects epithelial CTC only. These findings indicate that CTC in peripheral blood are highly heterogeneous, and thereby it is inadequate to identify and characterize CTC based on EpCAM marker only. Despite encouraging advancements have been achieved in the last decades, several barriers still exist, which prevent wide-spread clinical applications of CTC. First, the rarity of CTC poses a significant challenge to CTC utilization, warranting enrichment of CTC with high purity and quality. Second, CTC are highly heterogeneous and have no universal markers. The identification of CK + EpCAM + CD45cells as CTC is imprecise as quite a few CTC neither express CK nor EpCAM. Finally, CTC numeration is strongly dependent on the blood volume, enrichment method, and photographic or image processing system [68]. Distinct cutoffs were reported among different methods for distinguishing positive from negative samples (5 CTC/7.5 mL for CellSearch ® , 50 CTC/mL for ISET, versus 14 CTC/mL for CTC-chip) [69,70], rendering analytical results from different methods difficult to compare. Thus, the future efforts will be needed to improve current methodologies for expanded use of CTC in the following aspects: 1) Development of more useful isolation method that may combine different isolation mechanisms (e.g. Immunoaffinity-based capture and microfluidics); 2) Establishment of standard isolation method that yields as many CTC as possible without losing their heterogeneity; 3) Identification of more specific markers to CTC, and with more defined criteria to objectively identify CTC; 4) Confirmation of clinical value of CTC as a diagnostic, prognostic or predictive biomarker through large-scale clinical trials. Advances in these aspects would enable the better value of CTC as biomarkers for tumor management.

Exosomes
Exosomes are small lipid bilayer vesicles of 30 to 200 nanometers in diameter [71]. Released from almost every type of eukaryocytes through a sequential process, they are enriched with multiple cargoes of cellular origin, including lipids, proteins, and nucleic acids. Since the initial discovery in sheep reticulocytes in 1983 [72], exosomes have been considered as powerful shuttles for transporting biofunctional cargoes among cells, and have been implicated in both physiological and pathological processes. Studies have indicated that tumor-derived exosomes (TDE) participate in cancer development. As reported by Roccaro et al, the exosomes released from bone marrow-derived mesenchymal cells promoted the multiple myeloma (MM) development in animal models [73]. Similar result was observed in breast cancer cells that TDE could induce oncogenic transformation of normal cells [74]. This could be attributable to the involvement of exosomes in various processes that facilitated tumor progression, for example, angiogenesis, EMT, and drug resistance. Angiogenesis is possibly mediated by activation of the PAR2 signaling, an established angiogenic pathway, by exosomes [75]. The EMT regulators such as TGF-ß, ß-catenin, and tumor necrosis factor alpha (TNFα) are widely found in exosomes and they enhance migratory and invasive capacity of cancer cells [76]. Drug resistance is mediated by TDE possibly via several mechanisms, such as encapsulating and exporting the drugs from cancer cells [77], transferring multi-drug resistance (MDR)-associated proteins into cancer cells, and binding to the drugs and thereby blocking them [78]. Due to easy access from biological fluids and given their roles in tumor progression and drug resistance, exosomes have been considered as attractive tools for cancer diagnosis and treatment evaluation. A markedly increased release of exosomes was observed in the serum of lung adenocarcinoma patients compared to non-cancer subjects (2.85 mg/ mL versus 0.77 mg/mL) [79,80]. Both exosomal proteins and nucleic acids have been explored as biomarkers for clinical use. In a study of pancreatic adenocarcinoma, author revealed that the quantities of glypican-1-positive exosomes could be used to reliably distinguish the adenocarcinoma patients from non-cancer patients and healthy subjects [81]. The messenger RNA of EGFRv III, a common tumorigenic mutation known as an inframe deletion of exons 2-7 in the coding sequence of EGFR, was found in the exosomes of glioblastoma patients [82]. There are many other cargoes reported in exosomes, which are listed in Table 5. One type of cargoes of exosomes is microRNA (miRNA). Contents of miRNA can be different in exosomes from cancer patients versus healthy subjects. In the aforementioned study of lung adenocarcinoma, the mean miRNA concentration was significantly higher in patients than in control subjects (158.6 ng/mL versus 68.1 ng/mL) [79]. Besides, it was found that various specific miRNA could discriminate cancer patients from healthy subjects. For example, 8 species of miRNA were uniformly elevated in the patients with advancedstage ovarian tumor, while they remained at much lower levels in benign patients and even absent in healthy controls [83]. Recently, PD-L1 was found to be specifically packaged into exosomes of prostate cancer cells and to be involved in inhibiting T cell activation and promoting cancer cell growth [84]. Moreover, another clinical study of melanoma showed that exosomal PD-L1 was exclusively related to pembrozumab resistance, and the level of exosomal PD-L1 is much higher in non-responders than that in responders [85]. This finding suggested that exosomal checkpoint inhibitor levels could serve as a potentially useful surrogate for predicting the clinical response in immunotherapy. In summary, exosomes are multifunctional entities that play important roles in tumor progression and drug resistance. However, high quality exosomes are difficult to prepare as they are highly heterogeneous in terms of size, cargo content, and cellular origin. Five types of methods have been developed, which include differential centrifugation, size exclusion chromatography, immune-capture, polyethylene glycol precipitation, and microfluidic-based methods. These methods are based on the distinct features of density, size, surface proteins, and hydrophobicity of exosomes, respectively. The principles, applications and commercial products, as well as advantages and disadvantages of each method are provided in Table 6, and in a review article [86]. Despite availability of various methods for exosomes isolation, there is no single perfect method, and therefore methods are usually used in combination. As different laboratories use different methods and/or different markers for cancer identification, the results are hardly comparable among laboratories. Thus, further efforts are needed to standardize the exosomesisolating method and to identify common markers for use.

Comparison of three circulating tumor markers
Circulating tumor markers ctDNA, CTC, and exosomes have largely expanded the possibility of early diagnosis of cancer, patient stratification, as well as developing a personalized treatment. Based on these circulomes liquid biopsies can be developed, but each type of liquid biopsies has its own merits and limitations. Among the three circulomes, ctDNAbased methods represent the most advanced techniques. Several commercial products or LDT platforms have been launched. As it detects gene mutations with the sensitivity of 0.01%, a ctDNA test is broadly used for patient selection, medication guidance, and recurrence monitoring. However, it is difficult to distinguish real CNV from an operational error or aging-associated clonal hematopoietic mutations of indeterminate potential [87]. Moreover, PCR-based ctDNA detection is of low throughput and a sequencing-based method is of high cost. CTC is also an approved biomarker for prognosis of breast cancer. By enumeration, CTC is approved as adjunctive diagnosis for breast cancer staging. As intact cancer cells, they can be subjected to single cell analysis at both nucleic acid and protein levels. Furthermore, CTC can be cultivated in vitro for other analyses. However, CTC capture and analysis are costly. In addition, ambiguous correlation between CTC and tumor progression and treatment outcome might complicate cancer diagnosis and prognosis [88]. Exosomes are abundant, which makes them easier to obtain. Moreover, extensive studies have revealed that the miRNA enclosed in exosomes are differentially  [78,109] cIAP: cellular inhibitor of apoptosis protein; CRC: colorectal cancer; ERG: ETS transcription factor; NSCLC: non-small cell lung cancer; OS: overall survival; PCA3: prostate cancer associated 3; TACSTD2: tumor-associated calcium-signal transducer 2; TMPRSS2: trans-membrane serine protease 2; XIAP: X-linked inhibitor of apoptosis protein  [111,112] cIAP: cellular inhibitor of apoptosis protein; CRC: colorectal cancer; ERG: ETS transcription factor; NSCLC: non-small cell lung cancer; OS: overall survival; PCA3: prostate cancer associated 3; TACSTD2: tumor-associated calcium-signal transducer 2; TMPRSS2: trans-membrane serine protease 2; XIAP: X-linked inhibitor of apoptosis protein Half-life 16 minutes to several hours 1-2 hours 2 minutes to 6 hours in mice [89,90], and no report in human.

Sensitivity of analytical method
0.01%

Types of detection CNVs, point mutation, methylation
Change of CTC numbers Single cell sequencing The presence and abundance of specific miRNA or proteins; present in cancer patients versus healthy controls, which holds a great potential for cancer diagnosis. However, exosomes are highly heterogeneous and there is a lack of specific markers for identification. Thus, technologies remain to be further developed before one can effectively apply exosomes in cancer diagnosis. For further comparison among these three types of circulomes, readers can refer to Table 7.

Summary and future perspectives
These circulomes and the associated liquid biopsy methods elaborated above provide promising supplemental tools, and in some cases, the tool alternative to tissue biopsies. Although remarkable advances have been made in last few decades, clinical applications of the liquid biopsies is still challenging due to the rarity and difficulty in enrichment of these circulomes. Future efforts are needed to improve clinical performance of each method.
For ctDNA analysis, efforts should be directed to develop high throughput digital PCR instrument with a further improved signal-noise ratio. For CTC analysis, an urgent need would be to develop an immune-affinity enrichment protocol based on multiple antibodies against both epithelial and mesenchymal markers. For exosomes, the focus of efforts should be to identify cancer tissue origin of exosomes and to validate the correlation between exosomes and cancers in large clinical studies. It should be noted that except approved methods, the isolation procedures of circulomes and design of clinical trials vary significantly in different laboratories and clinical centers, resulting in poor comparability. Thus, prior to a broad use in clinical settings, it is crucial to standardize the procedure for the specific liquid biopsy method and validate the test with adequate specificity and sensitivity for clinical applications. Progresses in liquid biopsy technology would yield significant benefit for cancer diagnosis, patient selection, and treatment effect monitoring.