Exosomal biomarkers: A novel frontier in the diagnosis of gastrointestinal cancers

Table 1 Exosome formation and composition

Aspect	Details	Ref.
Origin	Derived from endosomal structures through endocytosis	[10]
	Invaginated endosomes form from the plasma membrane
Content sorting & packaging	Tightly regulated process of active sorting and packaging of diverse contents	[11,12]
Diverse contents	Lipids, proteins, DNA, mRNA, miRNAs (18-25 nt), lncRNA (> 200 nt), circRNA	[11,13,14]
Protein components	Rab family proteins, sorting - associated proteins, tetraspanins, HSPs, integrins, vacuolar proteins. Co-existence depends on specific proteins and their functions	[15]
ESCRT - dependent biogenesis	ESCRT-0 recognizes ubiquitinated cargo, starting the pathway	[16-19]
	Tsg101 in ESCRT-I forms a complex with ubiquitinated cargo, activating ESCRT-II
	ESCRT-II leads to ESCRT - III formation etc.
	ESCRT-III recruits deubiquitination machinery, packages cargo, promotes vesicle budding. Inward buds mature into MVBs
	ESCRT-III is degraded by an ATPase, regulated by Rab family proteins
Tetraspanins’ role	Transmembrane proteins that induce membrane - curved structures for vesicle formation	[20]
HSPs’ role	Mediate protein distribution in ILVs (exosome precursors) and include cytoskeleton proteins	[21,22]
RNA - related features	Enrichment of miRNAs with 3’-end nucleotide additions and 5’-terminal oligopyrimidine	[23-26]
	Specific RNA modifications during exosome formation enrich specific RNA species
	RNAs can induce genetic and epigenetic modifications in recipient cells
Lipid structure	Cholesterol, phospholipids, glycerophospholipids, sphingolipids, ceramides form a stable bilayer membrane for exosomes	[27]

miRNA: MicroRNA; lncRNA: Long noncoding RNA; circRNA: Circular RNA; HSPs: Heat shock proteins; ESCRT: Endosomal sorting complex required for transport; MVB: Multivesicular body; ILV: Intraluminal vesicle; Tsg101: Tumor susceptibility gene 101.

Table 2 Composition of exosome

Biomolecule	Function	Ref.
Hsc70, Hsp70, Hsp60, Hsp90 (HSPs)	Mediate protein distribution in ILVs (exosome precursors) and the inclusion of cytoskeleton proteins	[21,22,31]
GTPase Rab, flotillins, annexins, ARF6	Participate in exosome release and membrane fusion	[32,33]
Major histocompatibility complex class II molecules	Function not detailed in this text, likely related to immune response	[34]
Programmed cell death 6-interacting proteins	Play a role in programmed cell death	[35]
Tsg101 proteins	Are involved in the sorting and transportation of exosomes	[22]
CD9, CD63, CD81 (tetraspanin family transmembrane proteins)	Function not fully described, likely related to exosome structure and function	[35]
Transforming growth factor-β, apoptosis - related factor ligands (in tumor cell secreted exosomes)	Associated with tumor - specific processes	[36]
Dipeptidyl peptidase IV, matrix metallopeptidase 9	Participate in extracellular matrix remodeling, related to tumor invasion and metastasis	[37]
Epithelial cell adhesion molecule, epidermal growth factor receptor, survivin, insulin-like growth factor 1 receptor	Can be used as biomarkers for clinical diagnosis and prognosis	[38]
miRNA	Intercellular transport via exosomes may critically regulate gene expression and protein translation	[36,39-44]
	Short sequence motifs (EXOmotifs) guide miRNA into exosomes
	hnRNPA2B1 selectively binds miRNA, recognizing EXOmotifs and controlling its encapsulation within exosomes
	RNA binding proteins and RNA sequence motifs contribute to its selective sorting into exosomes
	Encapsulated in exosomes, RNA binding proteins protect miRNA from hydrolytic degradation, enabling it to exercise effects through cell-to-cell communication
circRNA	Part of the diverse ncRNA species in exosomes; specific functions not elaborated in the given text	[39,40]
lncRNA	Part of the diverse ncRNA species in exosomes; specific functions not elaborated in the given text	[39,40]
EXOmotifs	Guide miRNA into exosomes	[42]
hnRNPA2B1	Selectively binds exosomal miRNAs, recognizes EXOmotifs, and controls their encapsulation within exosomes	[42]
RNA binding proteins	Contribute to the selective sorting of miRNA into exosomes and protect RNAs (including miRNA) from hydrolytic degradation when encapsulated in exosomes	[36,43]
Cholesterol	Involved in exosome formation and release; contributes to the overall lipid composition affecting exosome properties, though its concentration may differ from parent cells	[32,45-47]
Sphingolipids (incl. Sphingomyelin)	Involved in exosome formation and release; sphingomyelin has a higher concentration in exosomes compared to parent cells	[32,45-47]
Phosphatidylcholine	Participates in exosome formation and release; present at a reduced content in exosomes relative to parent cells	[32,45-47]
Phosphatidylethanolamine	Involved in exosome formation and release; shows an observable enrichment in exosomes	[32,45-47]
Ceramide	Plays a role in exosome formation and release	[32,45,46]
Glycerophospholipids	Contribute to exosome formation and release	[32,45,46]
Lipid rafts	Implicated in exosome formation and help facilitate the secretion of specific molecules into the extracellular space; contribute to exosome’s permeability and circulating stability, making exosomes suitable for drug delivery	[45,48,27]

HSPs: Heat shock proteins; ILV: Intraluminal vesicle; miRNA: MicroRNA; lncRNA: Long noncoding RNA; circRNA: Circular RNA; Tsg101: Tumor susceptibility gene 101; ARF6: ADP ribosylation factor 6; hnRNPA2B1: Heterogeneous nuclear ribonucleoprotein A2B1.

Table 3 Comparison of exosome isolation methods: Advantages and disadvantages

Isolation method	Advantages	Disadvantages	Ref.
Ultracentrifugation	Considered the gold standard; convenient and cost - effective; can be combined with density - gradient mechanisms to achieve high - purity exosome yield and aid in morphological identification	Co-purifies lipoproteins and protein aggregates alongside EVs; combined with density - gradient mechanisms may result in lower yield and longer processing time	[53-59]
Ultrafiltration	Simple, faster procedure, no need for specialized equipment; refined method can achieve higher capture efficiency of different - sized exosomes compared to ultracentrifugation	Low recovery rate due to protein contamination and potential exosome damage during filtration; filter clogging can occur	[55,57,60-62]
Polymer - based precipitation separation (using PEG)	Simple and scalable	Pellets may be contaminated with other particles, large aggregates, and associated proteins, potentially affecting subsequent analysis	[58,63]
Microfluidic technology	Promising for rapid, efficient exosome isolation; can achieve high recovery rate in a short time	Not suitable for large - volume sample separation due to handling limitations	[49,64]
Other techniques (chromatography, hydrostatic filtration dialysis, size - exclusion chromatography, lipid - based separation, immunoaffinity - based methods)	Allow for seamless integration with clinical diagnostics, broadening potential clinical applications of exosomes	N/A	[49]

EVs: Extracellular vesicles; PEG: Polyethylene glycol; N/A: Not applicable.

Table 4 Comparison of exosome detection methods: Advantages and disadvantages

Detection method	Advantages	Disadvantages	Ref.
Nanoparticle tracking analysis	Simple; can determine both particle size and concentration; can detect vesicles in the 10-1000 nm diameter range, covering the typical exosome size range of 50-150 nm	N/A	[66,67]
Dynamic light scattering	Provides information on relative particle size; can calculate absolute size distribution when microvesicle concentration is known; accurate for samples with exosomes of one specific size	Larger particles may hinder detection of smaller particles in samples with various particle sizes	[68-70]
Enzyme-linked immunosorbent assay	A plate - based biochemical diagnostic tool for detecting and quantifying ligands, antibodies, and hormones; can assay exosomal membrane proteins and other marker proteins	Time - consuming (several hours to detect exosomes); requires a relatively large sample volume; low sensitivity for exosome detection	[71]
Colorimetric detection	User - friendly operation and straightforward signal readout for point-of-care testing; can be partitioned into AuNP-based assays (using AuNPs as signal transducers/amplifiers with high extinction coefficient and distance - dependent optical properties) and enzyme-H2O2-TMB-based assays (using enzymes to catalyze TMB solution for color signals)	Generally provides binary or semi-quantitative results	[72-74]
Fluorescent detection (including fluorescence spectrophotometry)	High sensitivity and excellent selectivity; can provide insights into exosome origins; can be divided into direct (specific recognition between exosome surface antigens and fluorescent - labeled aptamers or antibodies) and indirect (exosomes triggering fluorescence recovery) modes; can monitor exosome dynamics in real-time	N/A	[73,75-79]
Transmission electron microscopy	High imaging resolution (< 1 nm), well-suited for visualizing nanoparticles and assessing exosome morphology and heterogeneity	Fixation and dehydration steps in sample preparation may affect microvesicle morphology and size distribution	[71,80]
Cryogenic transmission electron microscopy	Eliminates potential effects on exosomes during sample preparation	N/A	[81]
Atomic force microscopy	Allows for sub-nanometer resolution imaging; can simultaneously measure exosome size distribution and map mechanical properties with nanometer accuracy; useful for quantifying and detecting exosome abundance, structure, biomechanics, etc. in tumor samples	N/A	[68,82-84]
Microfluidics (including integration with SPR technology and electrochemical detection)	Decreased reagent consumption, minimized contamination, reduced analysis times, increased throughput, and ease of integration and automation; enhanced exploration of exosome physicochemical and biochemical attributes at the microscale; can be integrated with SPR technology for multiparametric profiling of exosomes; electrochemical detection methods can be rapid and sensitive	Integration with SPR technology requires bulky, intricate instrumentation and is prone to severe interferences	[68,72,85-90]

SPR: Surface plasmon resonance; N/A: Not applicable.

Table 5 Most promising exosomal biomarkers with diagnostic significance in gastrointestinal cancer

GI cancer type	Exosome origin	Candidates’ biomarker	Clinical samples	Supporting evidence	Ref.
Colorectal cancer	Serum	Hsa-circ-0004771	179 patients; 45 healthy donors	AUC of 0.86, 0.88 to differentiate stage I/II CRC patients and CRC patients from healthy controls	[137]
	Plasma	Epcam-CD63	59 cancer patients; 20 healthy donors	AUC of 0.96	[138]
	Plasma	CD147		AUC of 0.932, P < 0.001	[139]
	Serum	LncRNA UCA1+, circRNA HIPK3		AUC of 0.900, P < 0.0001	[140]
	Plasma	miR-96-5p and miR-149	102 CRC patients	Significantly decreased in CRC tumor exosomes, and was significantly normalized after surgery	[98]
	Serum	miR-99b-5p and miR-150-5p	169 CRC patients, 155 healthy donors, and 20 benign disease patients	The AUC of miR-99b-5p was 0.628 (32.1% sensitivity and 90.8% specificity), the AUC of miR-150-5p was 0.707 (75.2% sensitivity and 58.8% specificity)	[99]
	Plasma	miRNA-27a and miRNA-130a	369 peripheral blood samples	The AUC of miR-27a (miR-130a) was 0.773 (0.742) in the training phase, 0.82 (0.787) in the validation phase	[100]
	Serum	miR-23a and miR-301a	12 CRC patients and 8 healthy donors	AUC values for miR-23a and miR-301a were 0.900 and 0.840, respectively	[101]
	Plasma	Let-7b-3p, miR-139-3p, and miR-145-3p	15 colon cancer patients and 10 healthy donors	Their combination (AUC = 0.927) showed an advantage in identifying CRC patients	[102]
	Serum	miR-126, miR-1290, miR-23a, and miR-940		Showed high diagnostic values to differentiate CRC patients at TNM stage I from healthy controls	[103]
Esophageal cancer	Serum	miR-652-5p	93 OSCC patients and 93 healthy individuals	AUC of 0.901	[106]
	Plasma	miR-93-5p	83 ESCC patients and 83 healthy individuals	The expression level in the plasma of ESCC patients being 1.39 times higher than that of the control population (P = 0.035)	[107]
	Serum	miRNA-182	125 ESCC patients and 60 healthy individuals	AUC = 0.837, 95%CI: 0.776-0.887	[108]
	Serum	UCA1, POU3F3, ESCCAL-1 and PEG10	313 ESCC patients and 313 control individuals without ESCC history	The AUC for UCA1, POU3F3, ESCCAL-1 and PEG10 were 0.733, 0.717, 0.676, 0.648, respectively	[109]
	Plasma	NR_039819, NR_036133, NR_003353, ENST00000442416.1, and ENST00000416100.1	295 ESCC patients, 43 esophagitis patients, and 49 healthy volunteers	The combined diagnostic value of these five lncRNAs revealed an AUC of 0.9995 (P < 0.001)	[110]
		CircRNA has-circ-0001946 and has-circ-0043603	3 pairs of ESCC frozen tumor and non-tumor tissues	The AUC, sensitivity and specificity of hsa_circ_0001946 was 0.894, 92.80%, of hsa_circ_0043603 was 0.836, 64.92%	[111]
Gastric cancer	Serum	Lnc HOTTIP	126 GC patients; 120 healthy donors	AUC of 0.827	[141]
	Serum	miR-19b-3p and miR-106a-5p	130 GC patients and 130 healthy donors	Levels of exosomes of patients with GC were markedly overexpressed compared to healthy donors	[114]
	Serum	miR10b-5p, miR132-3p, miR185-5p, miR195-5p, miR-20a3p, and miR296-5p	The training (49 gastric cancer vs 47 NCs) and validation phases (154 gastric cancer vs 120 NCs)	AUC were 0.764 and 0.702 for the training and validation phases, respectively	[115]
	Serum	miR-10a-5p, miR-19b-3p, miR-215-5p, and miR-18a-5p	A pair of 43 primary adenocarcinoma GC tissue samples with corresponding adjacent non-malignant counterparts	AUC of 0.801, 0.721, 0.780 and 0.736, respectively	[116]
	Serum	LncRNA PCSK2-2:1	29 healthy people and 63 gastric cancer patients	AUC of 0.896	[117]
	Plasma	LncRNA SLC2A12-10:1	60 GC patients and 60 age-matched healthy controls	The area under the ROC curve was 0.776	[118]
	Plasma	LncUEGC1 and lncUEGC2	Five healthy individuals and ten stages I GC patients and from culture media of four human primary stomach epithelial cells and four GCCs	LncUEGC1 exhibited AUC values of 0.8760 and 0.8406 in discriminating EGC patients from healthy individuals	[119]
	Plasma	Hsa_circ_0065149		Low expression levels in GC tissues were significantly associated with the tumor diameter (P = 0.034) and perineural invasion (P = 0.037)	[120]
	Serum	CircSHKBP1	72 paired GC tissues and normal tissues	The expression of circSHKBP1 was 2.31-fold higher in GC tissues on average than in normal tissues	[121]
	Serum	miR-15b-3p	108 GC patients; 108 healthy donors	AUC of 0.820; specificity of 80.6%; sensitivity of 74.1%	[142]
Hepatocellular and biliary cancer	Plasma	AFP; GPC3; ALB; APOH; FABP1; FGB; FGG; AHSG; RBP4; TF mRNA	36 HCC patients; 26 cirrhosis	AUC of 0.87; sensitivity of 93.8%; specificity of 74.5%	[143]
	Serum	miR-21		Higher in HCC patients	[144]
	Serum	CEA; GPC-3, and PD-L1	12 HCC patients; 12 hepatitis B; 6 healthy donors	Higher in HCC patients	[74]
	Plasma	miRNA-122, miRNA-21, and miRNA-96	50 patients with HCC and 50 patients with hepatic cirrhosis and 50 healthy volunteers	AUC of 0.924; 95%CI; sensitivity 82%, specificity 92% to discriminating HCC from the cirrhosis group	[123]
	Serum	miR-10b-5p	28 healthy individuals, 60 with chronic liver disease, and 90 with HCC	AUC of 0.934	[124]
	Plasma	miR-21-5p and miR-92a-3p	20 healthy individuals, 38 with liver cirrhosis, and 48 with HCC	MiR-21-5p was up-regulated and miR-92a-3p was down-regulated, and after incorporating AFP, the AUC was 0.85	[125]
	Serum	miR-4661-5p	15 normal subjects, 20 with CH, 10 with LC, 18 with HCC (Edmonson grade 1), and 45 with moderate to poorly differentiated HCC	AUC of 0.917 diagnose HCC in all stages, AUC of 0.923 in early stage	[126]
	Plasma	RN7SL1, SNHG1, ZFAS1, and LINC01359	57 plasma cell-free RNA transcriptome and 20 exosomal RNA transcriptomes	RN7SL1 discriminated HCC samples from negative controls (AUC = 0.87; 95%CI: 0.817-0.920)	[127]
	Plasma	miR-96-5p, miR-151a-5p, miR-191-5p, and miR-4732-3p	5 CCA patients and 4 GBC patients before and after surgery, 40 healthy individuals, 45 more CCA patients and 24 more GBC patients to validate	AUC of 0.733, 0.7639, 0.5417, and 0.6544, respectively	[128]
Pancreatic cancer	Plasma	miRNA-10b	3 PDAC patients; 3 CP patients; 3 healthy donors	Higher in PDAC patients	[145]
	Plasma	miRNA-10b	PDAC patients; CP patients and healthy donors	Higher in PDAC patients	[146]
	Mouse plasma samples	miR-3970-5p	9 healthy donors; 9 pancreatic intraepithelial neoplasia patients; 9 PDAC patients	Accuracy of 65%	[147]
	Serum	EpCAM, Glypican1		90% accuracy for pancreatic cancer or normal pancreatic epithelial cell lines; 87 and 90% predictive accuracy for healthy control and EPC individual samples	[148]
	Plasma	miR-4525, miR-451a and miR-21	55 patients with PDAC and 20 healthy volunteers	The exosomal levels from PDAC patients were significantly higher than those from healthy volunteers	[130]
	Serum	miR-1246, miR-4306, and miR-4644	131 pancreatic cancer patients and 89 controls	Significantly increased in 83% of pancreatic cancer patients compared to healthy controls, P < 0.05	[131]
	Plasma	LINC01268, LINC02802, AC124854.1, AL132657.1	78 pancreatic cancer patients and 70 healthy controls	The ROC analysis revealed AUC values of 0.8421, 0.6544, 0.7190, and 0.6231 for LINC01268, LINC02802, AC124854.1, and AL132657.1, respectively	[132]
	Plasma	Long-stranded RNAs (FGA, KRT19, HIST1H2BK, ITIH2, MARCH2, CLDN1, MAL2 and TIMP1)	Samples from 284 patients with PDAC, 100 patients with chronic pancreatitis and 117 healthy controls	Diagnosed PDAC with 0.949 AUC to identify stage I/II tumors	[133]
	Serum	Small nucleolar RNAs: WASF2, ARF6, SNORA74A, and SNORA25	27 pancreatic cancer patients and 13 controls	The AUCs of WASF2, ARF6, SNORA74A, and SNORA25 in serum from patients in the early stages of pancreatic cancer (stages 0, I, and IIA) were > 0.9	[134]
	Serum	Glypican1	190 pancreatic cancer patients and 131 controls	Sensitivity of 100%; specificity of 100%; positive predictive value of 100%; negative predictive value of 100%; AUC of 1.0	[135]

GI: Gastrointestinal; AUC: Area under the receiver operating characteristic curve; CRC: Colorectal cancer; OSCC: Oesophageal squamous cell carcinoma; ESCC: Esophageal squamous cell carcinoma; CI: Confidence interval; UCA1: Urothelial carcinoma associated 1; POU3F3: POU class 3 homeobox 3; ESCCAL-1: ESCC-associated lncRNA-1; PEG10: Paternal expressed gene 10; GC: Gastric cancer; NC: Normal control; ROC: Receiver operating characteristic; EGC: Early stage gastric cancer; GCC: Gastric cancer cell; AFP: Alpha fetoprotein; GPC3: Glypican-3; ALB: Albumin; APOH: Apolipoprotein H; FABP1: Fatty acid binding protein 1; FGB: Fibrinogen beta chain; FGG: Fibrinogen gamma chain; AHSG: Alpha 2-HS glycoprotein; RBP4: RNA binding protein 4; TF: Transferrin; CEA: Carcinoembryonic antigen; PD-L1: Programmed death-ligand 1; HCC: Hepatocellular carcinoma; CH: Chronic hepatitis; LC: Liver cirrhosis; CCA: Cholangiocarcinoma; GBC: Gallbladder carcinoma; PDAC: Pancreatic ductal adenocarcinoma; CP: Chronic pancreatitis; EPC: Early pancreatic cancer; FGA: Fibrinogen alpha chain; KRT19: Keratin 19; HIST1H2BK: Histone cluster 1 H2B family member K; ITIH2: Inter-alpha-trypsin inhibitor heavy chain 2; MARCH2: Membrane associated ring-CH-type finger 2; CLDN1: Claudin 1; MAL2: Mal T-cell differentiation protein 2; TIMP1: Tissue inhibitor of metalloproteinases 1; WASF2: WAS protein family member 2; ARF6: ADP ribosylation factor 6; SNORA74A: Small nucleolar RNA H/ACA Box 74A; SNORA25: Small nucleolar RNA H/ACA box 25; TNM: Tumor, lymph node, metastasis.

Table 6 Limitations for the use of exosomal biomarkers

Limitation aspect	Details
Extraction and isolation	Techniques are not widespread and have limitations. Different isolation methods produce distinct EV subpopulations with variations in miRNAs, proteins, diameters, and functions. Purity of exosomes is a major concern as it impacts diagnostic accuracy
Standardization	Lack of standardized operating procedures and data - handling methods. Current liquid biopsy clinical guidelines lack uniform and robust evidence. Variations in sensitivity and specificity occur due to different detection techniques/assays
Large scale clinical validation	Sample sizes in relevant studies are insufficient, data are limited, and validation periods are often too short to draw definitive conclusions. Most studies focus only on specificity and sensitivity of exosome detection systems, neglecting other crucial aspects like consistency, reproducibility, accuracy, reference ranges, and minimum detection limits

EV: Extracellular vesicle; miRNA: MicroRNA.