Table 2.

Most relevant studies published on the role of EVs and their cargo for HCC diagnosis, emphasizing percentage of patients in early stages of disease.

Study	EVs cargo target	Number of Patients,	Early stage HCC	Validated maker used for comparison	Main Findings (Sensitivity/Specificity, AUC)
Rui et al. (84)	miR-425-5p, let-7d-5p, miR-122-5p	124 HCC, 46 non-tumor donors	52.4% Stages I-II	N/A	Training cohort: AUC for HCC miR-425-5p, let-7d-5p, miR-122-5p, 3 miR signature: 0.65, 0.68, 0.83, 0.95. AUC for early HCC: 0.70, 0.76, 0.84, 0.92
Yao et al. (85)	exosomal lncRNA H19-204, THEMIS2-211 and PRKACA-202	168 HCC, 101 normal controls	64.8% TMN I-II	AFP	Combined score (H19-204 + LncRNA THEMIS2-211+ PRKACA-202) had AUC, Se, Sp for HCC in the validation set: 0.88, 70.9, 94.6%; LncRNA THEMIS2-211 for aerly stage HCC had AUC, Se, Sp of 0.81, 82.8, 70.8%
Chen et al. (86)	miR-34a	60 HCC, 60 healthy controls	48.3% TMN I-II	AFP	Sensitivity and specificity of serum exosomes miR-34a, AFP and their combined detection for the diagnosis of HCC were 78.3 and 51.7%, 61.7 and 98.3%, 68.33 and 93.33%, respectively; exo-miR-34a + AFP AUC 0.85.
Guo et al. (87)	exo_circ_0006602	87 HCC, 30 healthy controls	100% early stage	AFP	All Stages HCC, sensitivity 0.77; specificity 0.933; AUC 0.907. In cases of TNM stage I, the AUC value was 0.9564, Se 93.3%, Sp 89.7%
von Felden et al. (88)	3 small RNA clusters signature - smRC (unannotated)	105 HCC (), 85 CLD, 19 healthy controls	100% BCLC 0-A	AFP	3-smRC signature for early stage HCC has an AUC 0.87, 86% sensitivity, 91% specificity, 89% positive predictive value, significantly better than AFP + US.
Kim et al. (89)	onco-lncRNAs, DLEU2, HOTTIP, MALAT1, and SNHG1	72 HCC, 21 chronic hepatitis, 25 liver cirrhosis, 21 healthy controls	100% Very early stage, mUICC I	AFP	For all-stage HCC diagnosis, the combination of AFP and EV-MALAT1 had AUC 0.91 (HCC vs. nontumor); EV-MALAT1 and EV-SNHG1 had AUC 0.88 (HCC vs. CH/LC); For Very early HCC (mUICC stage I) EVDLEU2 and EV-MALAT1, and EV-HOTTIP and EVMALAT1 combinations had best AUC of 0.92 (HCC vs. nontumor) whereas combination EV-MALAT1 and EV-SNHG1 had AUC 0.98 for (HCC vs. CH/LC)
Lyu et al. (90)	hsa_circ_0070396	111 HCC, 50 chronic hepatitis, 58 liver cirrhosis, 54 healthy controls	28% TNM I-II	AFP	AUC, Se, Sp for HCC vs. healthy controls 0.857, 62.16, 98.15%; HCC vs. chronic hepatitis 0.774, 76.58, 68%; HCC vs. liver cirrhosis 0.66, 46.85, 81.03%
Hao et al. (91)	miR-320a	104 HCC; 55 CLD; 50 healthy controls	37.8% TNM I-II	AFP	HCC vs. Healthy controls (Se/Sp, AUC): 77.9%/80%, 0.86. AUC value for serum exosomal miR-320a was 0.860 with a sensitivity; HCC vs. CLD (Se/Sp, AUC): 71.6%/81.8%, 0.83
Cui et al. (92)	LDHC	112 HCC, 100 healthy controls	44.6% Stages I-II	N/A	AUC of exosomal LDHC in distinguishing early-stage HCC patients from healthy controls 0.9451, sensitivity and specificity 88.2 and 93.3%
Wang et al. (93)	miR-122, miR-21, miR-96	50 HCC, 50 LC, 50 healthy controls	20% TNM I	AFP	miRNA panel had high accuracy in discriminating HCC from the cirrhosis group (AUC 0.924; sensitivity 82%, specificity 92%)
Sorop et al. (94)	miR-21-5p, miR-92a-3p	48 HCC (), LC 38, healthy controls 20	87.5% BCLC A	AFP	AUC of combined miR/AFP score for HCC diagnosis 0.85
Cho et al. (95)	miR-25-3p, miR-140-3p, miR-423-3p, miR-1269a, miR-4,661-5p, and miR-4,746-5p	147 HCC, 42 liver cirrhosis, 46 chronic hepatitis, 41 normal controls, in 3 cohorts: screening, test and validation	31.3% Stages I-II	AFP (>20 ng/mL)	mUICC I and II vs. CH and LC (Se/Sp, AUC): miR-4,661-5p + miR-4,746-5p: 86.6%/90.4%, 0.95; mUICC I vs. CH and LC (Se/Sp, AUC): miR-4,661-5p + miR-4,746-5p: 86.6%/93.1%, 0.95
Lu et al. (96)	lncRNAs: ENSG00000248932.1, ENST00000440688.1, ENST00000457302.2	200 HCC (), 200 CLD, 200 healthy controls	49%/34% TNM I/II	AFP	Three lncRNAs: AUC 0.96/0.53 in training/validation cohorts; Three lncRNAs vs. AFP: 0.97/0.87 in training/validation cohorts
Li et al. (97)	lncRNAs	71 HCC, 37 CLD, 94 healthy controls	63% early-stage HCC	AFP (>10 ng/mL)	Support vector machine model (Se/Sp, AUC): Training set: 84%/94%, 0.95; Validation set: 89%/91%, 0.98; Testing set: 85%/95%, 0.96; Among patients with AFP measurements (60 HCC, 17 with benign liver lesions): Support vector machine model: AUC, 0.95; AFP alone: 0.83 (p = 0.037)
El Gwad et al. (98)	lncRNA-RP11-513I15.6, miR-1,262 and RAB11A	60 HCC, 42 CLD, 18 healthy controls	90% BCLC 0/A	AFP	All stages HCC Combined (RP11-513I15.6 + miRNA 1,262 + AFP) Se/Sp/PPV/NPV/Accuracy: 100%/76.7%/81.1%/100%/88.3%; Early stage HCC: 100%/76.7%/79.4%/100%/ 87.7%
Pu et al. (99)	miR-144-3p, miR-21-5p	24 HCC; 16 CLD, 17 healthy controls	38% BCLC stage 0	AFP	miR-144-3p: AUC 0.75; miR-21-5p: AUC 0.44; miR-144-3p + miR-21-5p: AUC 0.78; AFP: AUC 0.63 (p > 0.05 vs. all miR groups)
Wang et al. (100)	miR-122, miR-148a and miR-1,246	68 HCC; 53 LC; 50 CLD; 64 healthy controls	74% TNM I-II	AFP	Early stage HCC vs. cirrhosis: miR-122: AUC 0.80; miR-148a: AUC 0.86; miR-1,246: AUC 0.76 AFP: AUC 0.67; miR-122 + miR-148a + AFP: AUC 0.95
Xu et al. (101)	hnRNPH1 mRNA	88 HCC; 135 CLD, 68 healthy controls	16% TNM I-II	AFP (>20 ng/mL)	HCC vs. chronic hepatitis: hnRNPH1 alone: 85.2%/76.5%, AUC 0.87; AFP alone: 69.3%/87.9%, AUC 0.79; hnRNPH1 + AFP: AUC 0.89 (p < 0.05 vs. AFP); HCC vs. cirrhosis: hnRNPH1 alone: 86.4%/54.0%, AUC 0.65; AFP alone: 46.6%/88.3%, AUC 0.67 hnRNPH1 + AFP: AUC 0.75 (p < 0.05 vs. AFP)
Xu et al. (102)	ENSG00000258332.1 and LINC00635 (expression relative to GADPH)	115 HCC; 241 CLD, 120 healthy controls	26% TNM I-II	AFP (>20 ng/mL)	Training cohort: HCC vs. chronic hepatitis (Se/Sp, AUC): ENSG00000258332.1: 71.6%/83.4%, 0.72; LINC00635: 76.2%/77.7%, 0.75; AFP: 54.7%/75.3%, 0.67; All 3 markers: 83.6%/87.7%, 0.89 (p < 0.05 vs. AFP); Validation cohort: HCC vs. chronic hepatitis (Se/Sp, AUC): ENSG00000258332.1: 73.5%/80.5%, 0.72; LINC00635: 79.6%/75.2%, AUC, 0.73; AFP: 52.5%/74.1%, 0.63; All 3 markers 84.5%/85.3%, 0.89 (p < 0.05 vs. AFP)