2.1 Study populations

From June 1, 2017 to May 31, 2022, a total of 250 HEV-ALF patients and 250 AHE patients were retrospectively screened from Suzhou Municipal Hospital, the First Affiliated Hospital of Zhejiang University School of Medicine, Linyi Traditional Hospital, the First People's Hospital of Yancheng City, the Second People's Hospital of Yancheng City and the Fifth People's Hospital of Wuxi. A total of 250 health controls (HCs) were all from the First People's Hospital of Yancheng City.

AHE inclusion criteria: (1) serum anti-HEV IgM positive, and/or more than two-fold increased titers in anti-HEV IgG; (2) HEV RNA positive with clinical symptoms, showing elevated liver enzymes and/or jaundice and/or nonspecific symptoms, fatigue, itching, and nausea included. Definition of HEV-ALF: (1) abnormal liver function (prothrombin activity ≤40% or international normalized ratio [INR] ≥1.5), jaundice and hepatic atrophy in 2 weeks; (2) stage 2 or 3 encephalopathy complicating end-stage disease; and (3) without chronic liver disease. Exclusion criteria: (1) co-infection with other hepatitis virus, and/or human immunodeficiency virus (HIV); (2) with other liver disease not caused by viral hepatitis; (3) incomplete data.^{20, 21}

Serum samples of AHE and HEV-ALF patients were collected when they were enrolled, and the samples were frozen at −80°C for testing. Clinical data of recruited patients were extracted from hospital medical records. Follow-up data collection was through medical records or telephone interviews with patients and their families. All patients or their families provided the informed consents.

2.2 Isolation of serum EVs

Serum samples were placed on ice and centrifuged at 4°C to remove sediment after they were thawed in the 25°C water bath. 500 μl serum supernatant was collected and added into qEV column (qEVoriginal/35 nm). More buffers were added when the final sample entered under the tube via the column top, and a 1.5 ml EV solution was obtained with high purity after collecting the void volume. Further, samples were added to a new ultrafiltration tube pretreated with 1 × PBS. The ultrafiltration tube was then centrifuged and a certain amount of 1 × PBS was added. Subsequently, the ultrafiltration inner tube was vortexed for 3 min for downstream experiments. We resuspended the isolated EV samples in phosphate buffer to obtain the concentrated EV samples with the volume of 200 μl. and placed them in cryopreservation tubes. Finally, the labeled samples were frozen at −80°C.

2.3 Transmission electron microscopy (TEM)

On the copper wire, 5 μl EV samples were dropped. The samples were incubated at room temperature for 5 min, and after incubation, we carefully absorbed the excess liquid with filter paper. Afterwards, we added a drop of 2% uranium dioxyacetate to the copper web. The samples were incubated at room temperature for 1 min, and we also used filter paper to absorb the excess liquid. The samples were dried at room temperature for 20 min, and at last, we observed the samples under the TEM (HT-7700, Hitachi).

2.4 Nanoparticle tracking analysis (NTA)

We thawed the frozen samples in the water bath at 25°C and next placed them on ice.

After diluted with 1 × PBS (1: 800), the samples can be directly used for NTA under the nano-particle tracking analyzer (ZetaVIEW S/N 20-602, PARTICLE METRIX).

2.5 Western blot

EV samples were resuspended using 1 × PBS, and lysed by adding RIPA lysate of equal volume. EV proteins were separated by SDS-PAGE gel and then transferred to PVDF membrane. PVDF membrane was further blocked in the skim milk for 1 h after transfer, and it was next incubated with primary antibody overnight at 4°C. Secondary antibody was added after washing to remove unbound primary antibody, and the membrane was incubated for 30 min at 25°C. The antibodies used were following: CD9 (Abcam, ab92726), CD63 (Abclonal, A5271), TSG101 (Abcam, ab125011), APOA 1 (Santa Cruz, sc-376818), APOB (Novus Biologicals 20737), and ALB (Santa Cruz, sc-271605).

2.6 Detection of anti-HEV antibodies and HEV RNA

In all patients, anti-HEV IgM and IgG antibody presence was determined using the HEV enzyme-linked immunosorbent assay (ELISA) kit (Wantai). The optical density (OD) >1.1 was considered as positive, while ≤1.1 was considered negative. Real-time reverse transcription-quantitative PCR (RT-qPCR) was used for the detection of HEV RNA. Total viral RNA from the samples was extracted using the viral nucleic acid extraction kit (Aikang).

2.7 ELISA

The levels of EV-derived ASS1 were detected using the ELISA kit (MSKBIO, No. 201908). The concentrated EVs were diluted 1:500 using 1 × PBS. We used 100 μl RIPA lysate to precipitate the EV samples on ice for 30 min. The samples were then diluted 1:3 with 1 × PBS. The blank control and standard substance were added to a microplate, and they were coated with ASS1 antibody. 100 μl diluted samples were added to the microplate. The plate was incubated for 1 h at 37°C, after which we discarded the liquid in the microplate and dried the plate. Then, 100 μl liquid A was added and incubated for 1 h at 37°C. After washing for 3 times, liquid B of equal volume was added and incubated for 30 min. The plate was also washed for 3 times. Next, 100 μl substrate solution was added. The plate was incubated in the dark at 37°C for 20 min, and 50 µl stop solution was added to terminate the reaction. At last, the plates were read at 450 nm wavelength with a microplate reader.

2.8 Statistical analysis

In this study, SPSS 26.0, GraphPad Prism 9 and MedCalc were used for statistical analysis. If the data is normally distributed, mean ± standard deviation was used, and t-test was performed to compare between two groups. One-way ANOVA was conducted among three groups, and LSD method was used for pairwise comparison among three groups. If not, median (quartile) was used, and Mann–Whitney U test was performed between two groups. Kruskal–Wallis H test was conducted among three groups, and Bonferroni method was used for pairwise comparison among three groups. The relationship between serum EV-derived ASS1 levels (Y) and HEV-related parameters (X) was analyzed using the linear regression. To determine the independent risk factors for the occurrence of HEV-ALF, we conducted the univariate and multivariate logistic regression analyses. Area under the curve (AUC) with 95% confidence interval was calculated by MedCalc, and the sensitivity and specificity were also given. Decision curve analysis (DCA) was used by the R rmda package to evaluate the benefit of serum EV-derived ASS1 levels in assisting decision-making at different threshold probabilities. Besides, we analyzed all data for orthogonal partial least squares discriminant analysis (OPLS-DA) through SIMCA (Version 14.1.0.2047, 64-bit, MKS Umetrics). A p < 0.05 was statistically significant.

3.1 Characteristics of isolated serum EVs

Under TEM, EVs isolated from serum were found to be rounded or round-like in shape with a diameter of 40–100 nm, which had a complete envelope and clear background (Figure 1A). NTA was used to measure the particle size and distribution. It was revealed that the particle size was concentrated between 65.9 and 249.3 nm, and the maximum distribution peak was 157.6 nm (Figure 1B). Through western blot, the expression of several marker proteins was shown, including CD9, CD63, and TSG101 (Figure 1C), while the expression of APOA 1, APOB, and ALB was not detected (Figure 1D).

3.2 Expression of serum EV-derived ASS1

The serum EV-derived ASS1 levels were detected in 250 HCs, 250 AHE patients and 250 HEV-ALF patients using ELISA. It was found that the levels of serum EV-derived ASS1 in the HEV-ALF group were significantly higher than those in the AHE group [HEV-ALF: 915.28 (691.96–1186.58) ng/L vs. AHE: 643.09 (457.68–841.68) ng/L, p < 0.001]. The serum EV-derived ASS1 levels in the AHE group were significantly higher than those in the HCs group [AHE: 643.09 (457.68–841.68) ng/L vs. HCs: 359.96 (269.92–492.90) ng/L, p < 0.001; Figure 2A]. The difference in the serum EV-derived ASS1 expression among three groups showed that the higher serum EV-derived ASS1 levels, the more severe the condition of patients with HEV infection, indicating the HEV-ALF occurrence.

3.3 Relationship between serum EV-derived ASS1 levels and dynamic changes in HEV-ALF patients

To clarify the relationship between serum EV-derived ASS1 levels and the condition of HEV-ALF patients, the correlation between serum EV-derived ASS1 levels and HEV-related parameters was assessed. The serum EV-derived ASS1 levels positively correlated with the levels of TBIL (r = 0.219, p < 0.001; Figure 2D), while neither significant difference between serum EV-derived ASS1 levels and ALT levels was discovered (r = 0.112, p = 0.078; Figure 2B), nor significant difference between EV-derived ASS1 levels and AST levels (r = 0.102, p = 0.108; Figure 2C).

Organ failure often occurs in HEV-ALF patients during the disease period. According to the number of failed organs, HEV-ALF patients were divided into the number 2 group (n = 150), number 3 group (n = 59) and number >3 group (n = 41). The levels of serum EV-derived ASS1 in the number 2 group were significantly lower than those in the number 3 group [Number 2: 835.02 (646.83–985.58) ng/L vs. Number 3: 997.94 (756.29–1221.75) ng/L, p < 0.001], and the levels of serum EV-derived ASS1 in the number 3 group were significantly lower than those in the number >3 group [Number 3: 997.94 (756.29–1221.75) ng/L vs. Number >3: 1593.55 (981.92–1966.08) ng/L, p < 0.001; Figure 2E].

Furthermore, based on the dynamic changes of HEV-ALF patients' condition, we divided HEV-ALF patients into the improvement group (n = 131), fluctuation group (n = 86) and deterioration group (n = 33). It was discovered that the levels of serum EV-derived ASS1 in the improvement group were significantly lower than those in the fluctuation group [Improvement: 849.69 (623.99–1050.75) ng/L vs. Fluctuation: 943.12(743.26–1219.18) ng/L, p = 0.013], and the levels of serum EV-derived ASS1 in the fluctuation group were significantly lower than those in the deterioration group [Fluctuation: 943.12 (743.26–1219.18) ng/L vs. Deterioration: 1305.33 (906.00–1816.94) ng/L, p = 0.004; Figure 2F].

3.4 Serum EV-derived ASS1 level is an independent risk factor for the HEV-ALF occurrence

We described the clinical characteristics of 250 HCs, 250 AHE patients and 250 HEV-ALF patients through Table 1. The mean age of the HEV-ALF group was 56.50 ± 11.58 years, that of the AHE group was 53.68 ± 12.39 years, and that of the HCs group was 54.70 ± 11.67 (p = 0.028). There was no statistically significant difference in gender among the three groups (p > 0.05). In the HEV-ALF group, 61.20% of patients were male, 54.80% in the AHE group were male, and 51.60% in the HCs group were male (p > 0.05). Besides, no significant difference existed in BMI among the three groups (p > 0.05).

All patients in the HCs, AHE and HEV-ALF groups underwent laboratory tests. When comparing the levels of laboratory parameters among the three groups, we identified significant differences in serum EV-derived ASS1, WBC, RDW, CRP, PLT, PT, ALT, AST, GGT, TP, ALB, TBIL, CR, TG, TCH, CHE, TT3, TT4, TSH, AFP, and FER levels (all p < 0.05). In contrast, there existed no significant difference in UREA level (all p > 0.05).

Univariate and multivariate logistic regression analyses were performed to determine risk factors of the HEV-ALF occurrence (Table 2). Univariate logistic regression analysis showed that age and the levels of serum EV-derived ASS1, PLT, PT, ALT, AST, TP, ALB, TBIL, TCH, CHE, TSH, and FER were significantly different between the AHE group and the HEV-ALF group (all p < 0.05). Moreover, multivariate logistic regression analysis suggested that serum EV-derived ASS1, TP, TBIL, CHE, TSH, and FER were independent risk factors for the occurrence of HEV-ALF (all p < 0.05).

In addition, OPLS-DA was carried out to further rank and evaluate the prediction ability of these factors. The results suggested these factors can unambiguously distinguish patients in the two groups (Figure 3A,B). The serum EV-derived ASS1 level was considered as the main indicator, with the second highest predictive capability only to the level of TBIL (Figure 3C,D).

3.5 Prediction ability of serum EV-derived ASS1 levels for the occurrence of HEV-ALF

The ability of serum EV-derived ASS1 levels to predict the occurrence of HEV-ALF was evaluated. According to ROC curve analysis, the AUC of serum EV-derived ASS1 levels to predict the HEV-ALF occurrence was 0.728 (0.684–0.772), and the sensitivity and specificity were 72.80% and 64.80%, respectively (Figure 4A). Additionally, DCA results found that the levels of serum EV-derived ASS1 had a high decision-making ability in predicting the occurrence of HEV-ALF (Figure 4B).

3.6 Expression of serum EV-derived ASS1 in the elderly with HEV-ALF

If infected with HEV, pregnant women, the elderly and people with underlying liver diseases are more likely be severe. In our study, patients with other liver diseases have been excluded and no pregnant women were found, therefore the levels of serum EV-derived ASS1 of the elderly were explored. Patients with HEV-ALF were divided into the age <60 group (n = 147) and age ≥60 group (n = 103) depending on whether they were elderly or not. The levels of serum EV-derived ASS1 in the age ≥60 group were higher than those in the age <60 group, but no significant difference was found (Age ≥60: 943.31 (727.94–1367.38) ng/L vs. Age <60: 896.22 (688.97–1094.48) ng/L, p = 0.102; Figure 5).