Current Applications of Liquid Biopsy in Gastrointestinal Cancer Disease—From Early Cancer Detection to Individualized Cancer Treatment

Simple Summary Gastrointestinal (GI) cancers are a common cancer, affecting both men and women, normally diagnosed through tissue biopsies in combination with imaging techniques and standardized biomarkers leading to patient selection for local or systemic therapies. Liquid biopsies (LBs)—due to their non-invasive nature as well as low risk—are the current focus of cancer research and could be a promising tool for early cancer detection and treatment surveillance, thus leading to better patient outcomes. In this review, we provide an overview of different types of LBs enabling early detection and monitoring of GI cancers and their clinical application. Abstract Worldwide, gastrointestinal (GI) cancers account for a significant amount of cancer-related mortality. Tests that allow an early diagnosis could lead to an improvement in patient survival. Liquid biopsies (LBs) due to their non-invasive nature as well as low risk are the current focus of cancer research and could be a promising tool for early cancer detection. LB involves the sampling of any biological fluid (e.g., blood, urine, saliva) to enrich and analyze the tumor’s biological material. LBs can detect tumor-associated components such as circulating tumor DNA (ctDNA), extracellular vesicles (EVs), and circulating tumor cells (CTCs). These components can reflect the status of the disease and can facilitate clinical decisions. LBs offer a unique and new way to assess cancers at all stages of treatment, from cancer screenings to prognosis to management of multidisciplinary therapies. In this review, we will provide insights into the current status of the various types of LBs enabling early detection and monitoring of GI cancers and their use in in vitro diagnostics.


Introduction
Gastrointestinal (GI) cancers are responsible for more cancer-related deaths than lung and breast cancer. Colorectal cancer (CRC) is the major type of GI cancer, with 1.9 million new cases diagnosed worldwide in 2020, making it after lung and breast cancer the third most common cancer of all organs. According to the International Agency for Research on Cancer, in the same year, 1.1 million new cases of gastric cancer, 900,000 new cases of liver cancer, 600,000 new cases of esophageal cancer, and 500,000 new cases of pancreatic cancer were diagnosed across the globe [1].
Although the prognosis of many GI cancers has improved over the past decades [2,3], a late cancer diagnosis is still the leading reason for cancer-related deaths among all

Circulating Tumor Cells (CTCs)
CTCs are tumor cells, shed from a primary tumor. They can enter the bloodstream or lymphatic system, potentially spreading into distant organs possibly leading to metastases [17,18]. Nevertheless, only a minority of CTCs become solid metastatic lesions because of a complex sequence of events needed, i.e., the detachment from the primary tumor, migration through the circulating blood, immune escape, and survival. It remains unclear how the detachment process from the primary tumor tissue takes place. Evidence supports the involvement of epithelial to mesenchymal transition, by which transformed Due to the crucial role of LB markers, our main focus is on research findings and clinical applications in gastrointestinal cancers.

Circulating Tumor Cells (CTCs)
CTCs are tumor cells, shed from a primary tumor. They can enter the bloodstream or lymphatic system, potentially spreading into distant organs possibly leading to metastases [17,18]. Nevertheless, only a minority of CTCs become solid metastatic lesions because of a complex sequence of events needed, i.e., the detachment from the primary tumor, migration through the circulating blood, immune escape, and survival. It remains unclear how the detachment process from the primary tumor tissue takes place. Evidence supports the involvement of epithelial to mesenchymal transition, by which transformed epithelial cells can acquire the ability to invade, resist apoptosis, and disseminate. This could be the main driver for the detachment of tumor cells from the primary tumor [19][20][21][22][23][24]. Other reports hypothesize that cells split into different clusters [25]. It is noteworthy that gastrointestinal cancers compared to breast cancer have lower numbers of CTCs in peripheral blood due to portal vein circulations and a steady 'first-pass effect' in the liver [26]. Therefore, portal vein blood might be a unique sample site to isolate CTCs from gastrointestinal cancers. It has already been shown that the number of enriched CTCs from portal vein blood is higher than in the systemic circulation [27,28]. Portal vein blood can be collected intraoperatively or even by endoscopic ultrasound (EUS)-guided sampling [29].
In addition to the number of CTCs, the analysis of physical (size, density, and electric charge) and biological (cell surface expression) properties could play a crucial role in future clinical use [30].

Isolation and Enrichment of CTCs
The isolation and enrichment of CTCs are technically challenging because of their low numbers and elimination by the body's immune system. The sensitivity of capture methods has improved in the last few years. First, based on cancer-specific characteristics, the cancer cells are separated from other blood components followed by enrichment procedures. There are two main techniques for isolating CTCs, one based on immunoaffinity properties, and the other based on biophysical properties. The immunoaffinity-based technology, including positive or negative selection assays, isolates CTCs with an antibody-immobilized inert surface combined with magnetic beads [31] (Figure 2a). The epithelial cell adhesion molecule (EpCAM) is a commonly used cell surface marker for positive CTC selection and an immunomagnetic assay called CELLSEARCH ® . To date, CELLSEARCH ® is the only CTC technology that has gained FDA approval enabling the direct visualization and quantification of CTCs and the identification of living cells without the need for cell lysis. The functionality of CELLSEARCH ® for the detection of CTCs in GI cancer was confirmed by different studies [32,33]. In another approach, CTCs from blood samples of patients with CRC were pre-enriched through binding to VAR2CSA protein-coupled magnetic beads, and finally the colon-related mRNA transcripts USH1C and CKMT1A were detected by RT-qPCR [34]. Microfluidic chips as an immunoaffinity technology allow the selection of CTCs from small volumes of fluid under laminar flow, eliminating the need for sample processing. In a study by Lim et al. [35], single CTCs and CTC clusters were captured on the membrane of a centrifugal microfluidic device, picked without fixation, and used for further molecular analysis. Researchers began to develop CTC isolation technologies based on the biophysical properties of CTCs to overcome the bias and narrow spectrum of immunoaffinity-based approaches for CTC isolation. These methods are characterized as label-free and isolate CTCs from the blood based on biophysical properties, such as density, size, deformability, and electrical charge [36]. In order to confirm that the enriched cells consist only of CTCs, the cells must undergo characterization. Currently, this is achieved by immunocytochemistry-based assays, including immunofluorescence and immunohistochemistry, and molecular approaches, including RT-qPCR, FISH, and next-generation sequencing [37]. Using GILUPI Cell Collector (CC), a novel in vivo CTC detection device, researchers reported overcoming the limitations of small blood sample volumes. However, the clinical relevance of the CTCs detected was inferior to the CTCs identified by Cell Search [38].

Circulating Tumor DNA (ctDNA)
A group of French scientists detected cfDNA fragments in the circulating plasma of patients with autoimmune disorders in 1948 [73]. Later it became clear that the release of cfDNA is not restricted to autoimmune disorders, but was also found amongst others in pregnant women [74], septic patients [75], people suffering from different types of cancers, and even in healthy individuals [76]. Numerous studies have been conducted to uncover the mechanisms by which DNA fragments are released from cells into the plasma or serum. Major mechanisms involve apoptosis, necrosis, phagocytosis, NETosis, or active secretion [77]. Basically, every cell and tissue type is able to release cfDNA into circulation. Consequently, based on cell and tissue-specific methylation patterns from comprehensive databases including the Cancer Genome Atlas (TCGA) an assignment of cfDNA to different origins became possible [78]. Accordingly, the major source of cfDNA in blood results from hematopoietic cells, followed by vascular endothelial cells (up to 10%). However, cfDNA from liver tissue can also frequently be detected at low levels in circulation (up to 1%) in healthy people [78]. CtDNA is a fraction of cfDNA that originates from primary tumors, metastases, or from CTCs. Additionally, some findings are suggestive of an active release of ctDNA from living tumor cells involving exosomes [79]. CtDNA is characterized by small 70-200 base pair fragments circulating freely within the blood [73] showing a major fragment size of around 170 base pairs, which corresponds to nucleosomal fragments resulting from apoptosis [80]. The half-life of ctDNA is very short ranging from 15 min to 2.5 h before it is finally cleared by the liver and/or kidneys which is a prerequisite for a precise biomarker [81]. Concentrations of ctDNA in the blood of patients with a malignant tumor are significantly increased compared to healthy individuals [82]; however, levels of released ctDNA significantly vary between different tumor types and tumor stages. Furthermore, the match of cancer-specific alterations in the genome of solid tumors to those of ctDNA is a major discriminator between ctDNA and physiological cell-free DNA at steady state [83][84][85].

Detection and Analysis of ctDNA
There are a number of technical challenges associated with the analysis of ctDNA. First, total cfDNA itself is present only at comparably low concentrations in the nanogram

Clinical Application/Relevance of CTCs
Several studies have observed the prognostic value of CTCs for overall survival (OS) in localized colorectal cancer (CRC) [39,40]. In a series of 287 patients, a group demonstrated that preoperative CTC detection with a ≥ 1 CTC/7.5 mL proved to be an independent prognostic marker, whereas another group with 519 patients stated no association after surgery [39,41]. In high-risk CRC patients requiring adjuvant chemotherapy, CTC detection in the blood was correlated with worse outcomes [41][42][43]. A meta-analysis containing 1847 patients (11 studies) indicated that the detection of CTCs in the peripheral blood with CELLSEARCH ® has predictive utility for patients with CRC [44]. VISNÚ-1, a multicentre, randomized phase III trial with 349 patients, indicated that the first-line FOLFOXIRIbevacizumab chemotherapy regimen significantly improved progression-free survival (PFS) in comparison to the FOLFOX-bevacizumab chemotherapy regimen in patients with metastatic CRC. This trial revealed that the CTC count might be a valuable non-invasive biomarker to aid in decision-making for patients undergoing intensive first-line therapy [45]. In a series of studies performed in metastatic CRC, patients with high baseline CTC count (≥3 CTCs/7.5 mL) could benefit from intensive chemotherapy regimens (four drugs), unlike patients with low CTC counts [44,[46][47][48]. The PRODIGE 17 trial, conducted in 106 untreated patients with advanced gastric and esophageal cancer, reported that dynamic changes in CTC counts between baseline and 28 days after treatment were significantly associated with PFS and OS and could help in tailoring treatment regimens to each individual patient [49]. The above-mentioned findings might therefore help in the development of individualized treatment approaches in gastrointestinal cancer patients. In Table 1, a broad application of CTCs and its clinical relevance in GICs is mentioned. A group of French scientists detected cfDNA fragments in the circulating plasma of patients with autoimmune disorders in 1948 [73]. Later it became clear that the release of cfDNA is not restricted to autoimmune disorders, but was also found amongst others in pregnant women [74], septic patients [75], people suffering from different types of cancers, and even in healthy individuals [76]. Numerous studies have been conducted to uncover the mechanisms by which DNA fragments are released from cells into the plasma or serum. Major mechanisms involve apoptosis, necrosis, phagocytosis, NETosis, or active secretion [77]. Basically, every cell and tissue type is able to release cfDNA into circulation. Consequently, based on cell and tissue-specific methylation patterns from comprehensive databases including the Cancer Genome Atlas (TCGA) an assignment of cfDNA to different origins became possible [78]. Accordingly, the major source of cfDNA in blood results from hematopoietic cells, followed by vascular endothelial cells (up to 10%). However, cfDNA from liver tissue can also frequently be detected at low levels in circulation (up to 1%) in healthy people [78]. CtDNA is a fraction of cfDNA that originates from primary tumors, metastases, or from CTCs. Additionally, some findings are suggestive of an active release of ctDNA from living tumor cells involving exosomes [79]. CtDNA is characterized by small 70-200 base pair fragments circulating freely within the blood [73] showing a major fragment size of around 170 base pairs, which corresponds to nucleosomal fragments resulting from apoptosis [80]. The half-life of ctDNA is very short ranging from 15 min to 2.5 h before it is finally cleared by the liver and/or kidneys which is a prerequisite for a precise biomarker [81]. Concentrations of ctDNA in the blood of patients with a malignant tumor are significantly increased compared to healthy individuals [82]; however, levels of released ctDNA significantly vary between different tumor types and tumor stages. Furthermore, the match of cancer-specific alterations in the genome of solid tumors to those of ctDNA is a major discriminator between ctDNA and physiological cell-free DNA at steady state [83][84][85].

Detection and Analysis of ctDNA
There are a number of technical challenges associated with the analysis of ctDNA. First, total cfDNA itself is present only at comparably low concentrations in the nanogram per ml range. Second, the fraction of ctDNA among total cfDNA in many cases is also relatively low. This becomes especially evident in the early stages of cancer development or for tumors that release only low amounts of DNA [76]. Therefore, an urgent need for enrichment approaches of ctDNA over total cfDNA still exists. In this context, it has been reported that ctDNA to some extent might be enriched in cfDNA fractions of smaller size (90-150 base pairs (bp)) separated from the major fraction of 170 bp fragments [86] (Figure 2b). Although a tendency to higher ctDNA content in the 90-150 bp fraction could be found, enrichment factors of twofold in more than 95% of cases were still not satisfying [86]. In principle, a plethora of downstream analyses has been established for the diagnosis, prognosis, monitoring, and prediction of treatment response for all major cancer diseases. Among them are single nucleotide polymorphism (SNP) as well as copy number variation (CNV) analyses. When sufficient amounts of ctDNA templates (at least one genome equivalent per assay) are available in patient specimens, targeted assays based on PCR, for example, represent powerful approaches. Such PCR-based techniques which are characterized by high sensitivity showed the ability to identify relevant alterations in genes including RAS, HER2/NEU, BRAF, MET, BRCA2, APC, TP53, ALK, ROS1, PTEN, and NF1. However, when ctDNA content drops below one tumor genome equivalent per assay multi-target approaches become mandatory. Multi-gene panels are well established to similarly test for hundreds of targets. However, when it comes to early detection of cancer for screening purposes ctDNA contents of less than 0.01% represent a serious technical challenge to robustly assure high sensitivity and specificity. Cancer-specific methylomes in ctDNA were therefore proposed as a promising alternative [87]. In combination with next-generation sequencing (NGS), thousands of differential hyper-/hypo-methylated regions (DMRs) have been determined for a variety of cancers in addition to chromosomal and copy number changes or point mutations [88,89]. Accordingly, complex signatures of DMRs, for example, allow for early diagnosis with higher reliability. In general, it has been suggested that even for only as low as 0.01% ctDNA content prediction of cancer disease should be possible with 100-1000 DMRs when sequencing coverage of at least 100-1000x is given [87].

Clinical Application/Relevance of ctDNA
Analysis of tumor-linked genetic alterations and DNA methylation profiling has been recognized as a method for detecting potential biomarkers for disease diagnosis and prognosis [90]. In a study by Tam et al. [91] the levels of TAC1 and SEPT9 methylation detected in postoperative sera of patients with CRC were independent predictors for tumor recurrence and unfavorable cancer-specific survival. Findings from several studies showed that high ctDNA levels combined with an increased number of mutations detected in the ctDNA were linked to poor survival and multi-site metastasis [92]. However, when the amount of ctDNA was <1 mutant template molecule per milliliter of plasma, tests fail to detect early-stage cancer [93]. Therefore, methylome analyses of cfDNA with thousands of cancer-specific DMRs might overcome such limitations. In 2019, the 'Galleri test' or the 'Galleri multicancer early detection (MCED) test' developed by GRAIL Inc. (Menlo Park, CA, USA) achieved Breakthrough Device designation. The Galleri test detects cancer-and tissue-specific alterations in the methylation patterns of cfDNA in a blood sample via NGS, which should allow early pan-cancer detection even for cancers with unknown primary. GRAIL's clinical trial includes three studies: the Circulating Cell-free Genome Atlas (CCGA) Study (Clinical Trial NCT02889978) [94], the STRIVE Study (Clinical Trial NCT03085888), and the SUMMIT Study (Clinical Trial NCT03934866). In these studies including 2482 cancer patients covering approximately 50 different cancer types, sensitivities and specificities were tested by using a target hybridization capture approach for the analyses of 100,000 differentially methylated regions [95]. In a predefined subset of 12 cancer types which comprised roughly 63% of all US cancer cases average sensitivity was 67.3% accumulated for stage I-III at a very high specificity of 99.3% [95]. Sensitivity for all 50 cancer types accumulated for stage I-III dropped to 43.9% with 18% sensitivity for stage I and 43% in stage II. Remarkably, sensitivities for pancreatic cancer were significantly higher even at early stages with 63% and 83% in stage I and II, respectively. Although high specificities might predestine this test for negative prediction in screening approaches, moderate sensitivities still require improvements, especially for early diagnosis. Another FDA-approved LB-based test is Epi proColon ® (Epigenomics AG, Berlin, Germany) targeting methylation changes used to screen for CRC. The test is based on a real-time PCR with a fluorescent hydrolysis probe and targets the methylation changes of the SEPT9 gene promoter in cfDNA isolated from plasma. To evaluate the clinical assessment, Epi proColon ® was involved in a prospective multicenter study (Clinical Trial NCT00855348) [96][97][98].
A ctDNA-guided approach has been used by numerous clinical trials (DYNAMIC, CIRCULATE-Japan, CIRCULATE-trial, CIRCULATE-PRODIGE, and IMPROVE-IT2), all focused towards precise adjuvant therapy for stage II colon cancer patients [99][100][101][102][103]. According to DYNMAIC-trial, a ctDNA-guided approach led to a reduction in the number of patients who received adjuvant therapy, and furthermore, ctDNA-positive patients appeared to benefit from adjuvant treatment. The other trials (CIRCULATE-trial, IMPROVE-IT2) mentioned above could help in decision-making before adjuvant treatment in stage II colon cancer. The outcome of these trials proposes that a survival benefit from adjuvant chemotherapy may be obtained in a well-defined subgroup of patients with stage II colon cancer-especially those with detectable ctDNA post-surgery. In the case of pancreatic cancer, ctDNA might be used as a marker for monitoring treatment efficacy and disease progression [104]. To prove the feasibility of a non-invasive detection in plasma, Liu et al. developed a pancreatic cancer detection assay (PANDA) for screening and validation of PDAC-specific DNA methylation in tissues and plasmas of PDAC patients [105]. In combination with age and CA19-9 plasma serum level, this assay showed encouraging results to discriminate PDAC plasma from non-malignant disease, showing its capability to be amended into a non-invasive diagnostics method for PDAC screening. In Table 2, a broad application of ctDNA/cfDNA and its clinical relevance in GICs is mentioned. Minimal invasive blood-based biomarker panel which could potentially be used as a diagnostic and screening tool in a select subset of high-risk populations [112] Pancreatic cancer 194 ctDNA in combination with exosomal DNA provides both predictive and prognostic information relevant to therapeutic stratification [113] Colorectal cancer 455 Reduced the usage of adjuvant chemotherapy [99] Colorectal cancer 250 Detection of residual disease [114] Locally advanced rectal cancer (LARC) 462 ctDNA analysis as a useful guide for adjuvant chemotherapy selection in LARC patients [115] Pancreatic cancer with liver metastasis 104 Use of circulating tumor DNA as an independent prognostic marker for advanced pancreatic cancer [116] Pancreatic cancer 135 ctDNA as an independent prognostic marker in advanced PDAC as well as an indicator of shorter disease-free survival in resected patients when detected after surgery [117] Pancreatic cancer 112 Increased ctDNA levels were a poor prognostic factor for survival. [118] Pancreatic cancer 259 Plasma cfDNA might provide a prognostic and diagnostic tool to assist surgical decision-making in PDAC patients [119]

Circulating Extracellular Vesicles (Tumor Exosomes)
Exosomes are a subpopulation of extracellular vesicles (EVs), ranging in size from 30-150 nm. They are derived from the endosomal pathway via the formation of late endosomes or multivesicular bodies (MVBs). As an important mediator of intracellular communication, exosomes transmit various biological molecules including proteins, lipids, and nucleic acids over distances within the protection of a lipidic bilayer-enclosed structure. Nearly all types of cells and all body fluids contain exosomes [124,125]. Cancer cells and other stromal cells in the tumor microenvironment (TME) also release exosomes and control tumor development through molecular exchanges mediated by exosomes [126,127]. Circulating extracellular vesicles (cEVs) are implied to be more stable in comparison to serological proteins as the lipidic bilayers defend the content from proteases and other enzymes [128].

Isolation of Tumor Exosomes
Exosomes can be isolated using various methodologies. Ultracentrifugation (UC) is the most widely used technique. Other techniques include differential centrifugation (DC), density gradient ultrafiltration (DG), size exclusion chromatography (SEC), precipitation, immunoaffinity capture based on the expression of endosomal surface proteins such as CD81, CD63, and CD9, and microfluidic-based assays [129] (Figure 2d).

Clinical Application/Relevance of Tumor Exosomes
Exosomes demonstrate significant advantages over other sources of LBs. First, exosomes exist in almost all body fluids and are characterized by highly stable lipidic bilayers. Second, living cells secret exosomes. They thus contain biological information from the parental cells and are more representative than cell-free DNA secreted during necrosis or apoptosis [130]. Third, exosomes express specific proteins such as CD63, ALIX, TS101, and HSP70, 20 which can be used as markers to discriminate exosomes from other vesicles making their identification clear and simple [131]. Fourth, as exosomes can present specific surface proteins from parental cells or target cells, they can help in the prediction of organ-specific metastasis [132]. Fifth, compared to CTCs, they can be isolated using classic methods such as ultracentrifugation [133].
Circulating exosomal PD-L1 was shown to contribute to immunosuppression, to reflect the immune status, and to better predict survival in patients with GICs, thereby making it a potential prognostic biomarker [134]. Additionally, EV proteins such as carcinoembryonic antigen-related cell adhesion molecules (CEACAMs), Tenascin C, Glypican-1, and ZIP-4 have been recognized as diagnostic biomarkers in GICs [135][136][137]. The EV-Glypican-1 (GPC-1) derived from plasma has been described as a potential marker of early pancreatic ductal adenocarcinoma (PDAC) with higher diagnostic accuracy than CA19-9 [137]. A clinical trial (NCT03032913) performed by Etienne BUSCAIL involved 20 PDAC patients and 20 non-cancer patients, whose blood samples were collected to detect CTCs and GPC1+ exosomes for diagnostic accuracy assessment of CTCs and Onco-exosome Quantification. Very recently, Lin et al. presented the development of a signature of four EV-proteins-monocyte marker CD14, Serpin A4 (a regulator of angiogenesis), CFP (a positive regulator of the complement system), and LBP (lipopolysaccharide binding protein)-as prognostic biomarkers in colorectal liver metastases (CRLM). Thereby, they used matching pre-and post-operative serum samples of patients undergoing CRLM surgery and finally validated the discovered proteins in three independent cohorts. Additionally, they showed that EV-bound CXCL7 could serve as a biomarker of early response in CRML patients undergoing systemic chemotherapy [138]. Furthermore, three cancerspecific phospholipids were found in a study that analyzed 20 dysregulated phospholipids in pancreatic cancer compared to controls. Among them, LysoPC 22:0 was linked with tumor stage, whereas CA19-9 and CA242 were associated with tumor diameter and positive lymph node count [139]. Finally, diagnostic accuracy for WASF2, ARF6 mRNAs, SNORA74A, and SNORA25snoRNAs in circulating exosomes was greater than for CA19-9 in discriminating PC patients from controls [140]. Even though a limited number of studies on the application of exosome-based drug delivery vectors in the treatment of GICs exist, some of them report intriguing advancements in the field. Using an in vitro model Pascucci L. et al. demonstrated the role of exosomal-mesenchymal stromal cells (exo-MSCs) in the packaging and delivery of active drugs, suggesting a possible option of using the MSCs as a warehouse to develop drugs with a better specificity [141]. Another study showed the application of milk-derived exosomes for the oral delivery of PAC in early-stage and advanced-stage pancreatic and other cancers. It also evaluated the anti-tumor potency of the milk-derived exosomes loaded with PAC [142]. Furthermore, studies have verified the link between cancer-derived exosomes and the modulation of immune response in pancreatic cancer and have also indicated the application of these cargo carriers in targeting pancreatic cancer cells, whether as anti-tumor drugs and other molecules, such as RNAi against mutant KRAS [143]. More applications and the clinical relevance of exosomes are summarized in Table 3.  Pancreatic cancer CEACAMs pancreatic fluid upregulated Exosome isolation is feasible from pancreatic duct fluid, and that exosomal proteins may be utilized to diagnose patients with PDAC. [135] Pancreatic cancer Tenascin C pancreatic fluid upregulated Exosome isolation is feasible from pancreatic duct fluid, and that exosomal proteins may be utilized to diagnose patients with PDAC. [135] Pancreatic cancer Glypcan-1 (GCP-1) serum upregulated GPC1 + crExos may serve as a potential non-invasive diagnostic and screening tool to detect early stages of pancreatic cancer to facilitate possible curative surgical therapy. [137] Pancreatic cancer ZIP-4 cell line upregulated Exosomal ZIP4 promotes cancer growth and is a novel diagnostic biomarker for pancreatic cancer [136] Pancreatic cancer DNA MAFs plasma upregulated Exosomal DNA in combination with ctDNA provides both predictive and prognostic information relevant to therapeutic stratification [113] HCC/Colongiocarcinoma Annexin V serum upregulated A novel non-invasive biomarker to assess the presence and possible extent of cancers in patients with advanced liver disease [150]

Tumor-Educated Blood Platelets (TEPs)
Platelets play a central role in blood coagulation and in the healing of wounds, and their relationship with cancer has been extensively investigated [156][157][158][159][160][161][162][163][164]. There are two major studies that indicated the involvement of platelets during tumor progression. These studies contributed to the development of the concept of tumor-educated platelets (TEPs). The first study by Trousseau (1868) observed spontaneous coagulation being common in cancerous patients, stating that circulating platelets were affected by cancer [165]. The second study (Billroth T, 1877) described '' thrombi filled with specific tumor elements" as part of metastasis, pointing out a direct interaction of cancer cells and platelets [166,167]. In recent years, several studies have focused on the impact of platelets during cancer progression, and some of them showed that platelet dysfunction and thrombotic disorders are important key factors in cancer development. By now it is well known that tumoreducated platelets (TEPs) are educated when they interact with the tumor cells in such a way as to lead to the detachment of biomolecules, such as proteins and RNA, tumor-specific splice events, and finally to megakaryocyte alteration [168]. Through this interaction, the RNA profile of blood platelets changes. This change has been used as an independent diagnostic marker for detecting TEPs in various solid tumors [169]. The RNA biomarkers of the directly transferred transcripts are EGFRvIII, PCA3, EML4-ALK, KRAS, EGFR, PIK3CA mutants, FOLH1, KLK2, KLK3, and NPY [170]. Here, we describe the isolation, detection, and clinical relevance of TEPsin GICs.

Isolation and Detection of Tumor-Educated Platelets
Tumor-educated platelets can be separated from peripheral blood by using a twostep centrifugation approach [171]. The first step separates the platelet-rich plasma (PRP) from the red blood and white blood cells, whereas the second centrifugation step yields the platelet pellet [172]. For the detection of the TEPs in the human plasma, the plasma pellets are dissolved in 1ml TRIZOL reagent. RNA is extracted from the plasma platelet pellet and the suboptimal quality of the platelet RNA is characterized by an absence of ribosomal 18S and 28S peaks and measured in a Bioanalyzer Picochip. Preferably, platelet mRNAs are measured by other methods, such as Fragment analyzer (Advanced Analytical Technologies, Ankeny, IA, USA) or Qubit Fluorometric Quantification (Thermo Fisher, Waltham, MA, USA) analysis (Figure 2c).

Clinical Application/Relevance of Tumor-Educated Blood Platelets
Platelet-related measures are considered important in anticipating long-term results in patients with GI cancer. A study distinguished 228 patients with localized and metastasized tumors from 55 healthy donors using the genetic profile of mRNA from TEPs. Additionally, mRNA sequencing of TEPs could accurately recognize MET or ERBB2-positive and mutant KRAS, EGFGR, or PIK3CA tumors. Moreover, TEPs have the ability to identify the location of primary tumors including colorectal cancer, non-small lung carcinoma, glioblastoma, pancreatic cancer, hepatobiliary cancer, and breast cancer [169]. Yang et al. showed that TIMP metallopeptidase inhibitor 1 (TIMP1) mRNA levels were higher in platelets from patients with CRC compared to those from healthy donors or patients with inflammatory bowel diseases, which could be another promising diagnostic signature [173]. A study in patients with hepatocellular carcinoma (HCC) conducted by Asghar et al. showed that the expression of TGF-β, NF-κβ, and VEGF was increased in TEPs of HCC patients compared to that from controls and thereby they suggested that these RNA based biomarkers could be used as a promising tool for early detection of HCC [174]. Furthermore, the alterations of platelet counts as a prognostic marker were studied in a clinical trial (NCT03717519) conducted by Corrado Pedrazzani. The study recruited 196 patients with synchronous colorectal liver metastases. In esophageal squamous cell carcinoma (ESCC), Ishibashi et al. performed a meta-analysis evaluating the prognostic values of platelet-related measures. The analysis revealed that a high platelet-to-lymphocyte ratio (PLR) was significantly correlated with poor OS [175]. A dual-center retrospective study performed by Yang et al.
showed a standardized indicator of platelet counts was used to forecast the prognosis of 586 CRC patients by using a development-validation cohort. In the development cohort, postoperative platelet count and postoperative/preoperative platelet ratio (PPR) were independent predictors of prognosis in CRC patients. In the validation cohort, the platelet/lymphocyte ratio and PPR were used to test the OS of CRC patients and showed the largest AUC in reviewing 1-year and 3-year OS (AUC: 0.663 and 0.673) [176]. Based on the studies above, we summarize that PLR and PPR could serve as reliable and economic indicators to evaluate the prognosis of GI cancer. Interestingly, many new approaches have been utilized to explore the clinical relevance of TEPs in GICs. We have summarized some of them in Table 4 below. TEPs have advantages over other blood-based sources due to their abundance in the blood, the ease with which they can be isolated, their high-quality RNA, and their ability to process RNA in response to foreign signals.

Conclusions
Due to the limitations of conventional tissue biopsies, there is an urgent need for new tumor biomarkers. As reviewed here, LBs have been well-evaluated and show promising results as an alternative clinical tool for the detection and treatment of gastrointestinal cancers, many of those currently being intensively investigated in various observational and interventional clinical trials (Tables 5 and 6). Multiple longitudinal biopsies allow for real-time monitoring of the tumor. This approach may facilitate the prediction of the possible treatment outcome and may help in choosing the optimal individualized therapeutic strategy. Therefore, the analysis of LB markers (TEPs, CTCs, ctDNA/cfDNA, and exosomes), in combination with modern imaging techniques and already existing protein markers might help to create an optimal clinical synergy that might be used as a standard procedure in the near future for early cancer detection and individualized cancer treatment.

Conflicts of Interest:
The authors declare no conflict of interest.