AKT-eNOS-NO pathway mediates spontaneous portosystemic shunts and therapeutic efficacy of SC79 in cirrhosis

Abstract

BACKGROUND

Spontaneous portosystemic shunts (SPSS) are common in patients with cirrhosis and portal hypertension; however, their prognostic significance and underlying mechanisms remain poorly defined.

AIM

To investigate the prognostic impact of SPSS in hepatitis B virus (HBV)-related cirrhosis and to explore the molecular mechanisms underlying its development.

METHODS

We retrospectively analyzed a nationwide multicenter cohort of patients with HBV-related cirrhosis from 2017 to 2021. Patients were stratified by imaging findings into SPSS and non-SPSS groups, and outcomes, including hepatic decompensation events and mortality, were compared between groups. Proteomic profiling of liver tissue and metabolomic profiling of serum were performed to identify differentially expressed molecules and enriched pathways. Mechanistic validation was conducted in a rat model of cirrhosis with SPSS.

RESULTS

Patients with SPSS exhibited more severe hepatic dysfunction and had higher rates of hepatic decompensation, including esophagogastric variceal bleeding, portal vein thrombosis, hepatic encephalopathy, ascites, and hepatocellular carcinoma, as well as higher mortality, than patients without SPSS. Multi-omics analyses identified 100 differentially expressed proteins and 54 metabolites enriched in 11 signaling pathways, notably the AKT-eNOS-nitric oxide (NO) axis, a key regulator of angiogenesis. In vivo, pharmacologic activation of AKT with SC79 increased hepatic p-AKT and p-eNOS levels, elevated NO levels, reduced the incidence of SPSS, attenuated cirrhosis severity, and decreased the expression of angiogenesis-related markers.

CONCLUSION

SPSS is associated with poor prognosis in HBV-related cirrhosis. Dysregulation of the AKT-eNOS-NO pathway may contribute to SPSS formation and represents a potential therapeutic target.

Key Words: Liver cirrhosis; Portal hypertension; Portosystemic shunt; Spontaneous; Prognosis; Proto-oncogene proteins c-AKT

Core Tip: Spontaneous portosystemic shunts (SPSS) are strongly associated with adverse outcomes in hepatitis B virus-related cirrhosis, including more frequent decompensation and higher mortality. By integrating multicenter clinical data with proteomic, metabolomic, and animal studies, this work identifies dysregulation of the AKT-eNOS-nitric oxide pathway as a potential mechanism underlying SPSS formation. Pharmacologic activation of AKT with SC79 reduced the incidence of SPSS and alleviated cirrhosis severity in vivo, suggesting that this pathway may represent a promising therapeutic target in cirrhosis with portal hypertension.

INTRODUCTION

Cirrhosis is the 11^th leading cause of death worldwide, accounting for approximately one million deaths annually[1]. The most common causes of cirrhosis-related mortality include refractory ascites, gastrointestinal bleeding, and hepatic encephalopathy (HE), all of which are recognized major complications of portal hypertension[2]. In addition, portal hypertension frequently leads to the development of spontaneous portosystemic shunts (SPSS), which are abnormal vascular channels connecting the portal venous system to the systemic circulation; the prevalence of SPSS in patients with cirrhosis ranges from 34.0% to 63.5%[3,4]. Through these shunts, portal venous blood bypasses the liver and enters the systemic circulation directly. SPSS was previously thought to reduce portal pressure to some extent, thereby lowering the risk of variceal bleeding and ascites[5,6]. However, recent evidence indicates that SPSS may exacerbate hepatic dysfunction, increase the incidence of decompensation events, and raise mortality by diverting blood flow away from the liver[7-9].

Given the evidence of potentially harmful sequelae of SPSS, understanding the mechanisms underlying its formation is crucial for timely intervention[10]. According to current understanding, increased portal pressure mechanically induces the passive reopening of embryonic vascular channels linking the portal and systemic venous systems, leading to SPSS formation[11]. However, recent findings also implicate vasoactive mediators in this process[12,13]. Using rat models, Fernandez et al[14] demonstrated that SPSS formation is driven by VEGF-dependent angiogenesis and is markedly suppressed by inhibition of the VEGF/VEGFR2 signaling pathway. In addition, hypoxia, oxidative stress, inflammation, and shear stress have all been shown to induce visceral VEGF overexpression and angiogenesis in portal hypertensive rats and patients[15]. Studies have shown that VEGF plays a central role in early-phase neovascularization, whereas platelet-derived growth factor contributes to vessel stabilization[16]. Similarly, placental growth factor stimulates endothelial and smooth muscle cell proliferation, thereby promoting collateral vessel formation[16]. However, current understanding of SPSS formation is based largely on animal models and focused analyses of selected angiogenic factors. Therefore, the molecular pathways driving SPSS in human cirrhotic tissues have not been systematically and comprehensively characterized.

To address this knowledge gap, we conducted a multicenter cohort study to investigate the prognostic significance of SPSS in patients with cirrhosis by comparing the rates of decompensation and mortality between patients with and without SPSS. To explore the mechanisms underlying SPSS formation, we performed, for the first time, integrated proteomic and metabolomic analyses of human liver tissue and serum samples from patients with cirrhosis. These analyses identified the AKT-eNOS-nitric oxide (NO) signaling pathway as a candidate pathway involved in SPSS formation. To validate these findings, we established a rat model of cirrhosis to determine whether treatment with the AKT phosphorylation activator SC79 could reduce the incidence of SPSS and ameliorate cirrhosis severity.

MATERIALS AND METHODS

Study population and ethics

This multicenter retrospective cohort study was conducted from January 2017 to December 2021 in patients diagnosed with hepatitis B virus (HBV)-related cirrhosis at five tertiary hospitals in China. The presence of SPSS was confirmed by contrast-enhanced abdominal computed tomography or magnetic resonance imaging. Patients were stratified into SPSS and non-SPSS groups, and data on clinical characteristics, laboratory test results, and follow-up evaluations were collected. The study protocol complied with the Declaration of Helsinki and was approved by the Ethics Committee of Mengchao Hepatobiliary Hospital, Fujian Medical University (Approval No. 2024_109_01). Written informed consent for participation was obtained from all participants.

Inclusion and exclusion criteria

Inclusion criteria: (1) Confirmed diagnosis of HBV-related cirrhosis based on histology, laboratory tests, or imaging; (2) Age > 18 years; and (3) Written informed consent.

Exclusion criteria: (1) Hepatocellular carcinoma (HCC) beyond the Milan criteria; (2) Presence of other malignancies; (3) A history of splenic artery embolization, splenectomy, hepatectomy, liver transplantation, or surgical shunt placement; (4) Prior transjugular intrahepatic portosystemic shunt; (5) Psychiatric or neurological disorders; and (6) Incomplete clinical or follow-up data.

The multi-omics analysis cohort was selected based on the same inclusion and exclusion criteria as the overall study cohort, with the additional exclusion of all patients with HCC to minimize potential confounding.

Proteomic analysis

Paraffin-embedded liver tissue samples were obtained from patients with cirrhosis undergoing liver transplantation (SPSS group, n = 5; non-SPSS group, n = 5), and proteomic profiling was performed by Jingjie PTM BioLab Co., Ltd. (Hangzhou, China). Briefly, the tissues were deparaffinized and lysed in four volumes of buffer containing sodium dodecyl sulfate (10 g/L) and protease inhibitor cocktail (1 × working concentration) by ultrasonication, followed by centrifugation at 12000 × g for 10 minutes at 4 °C. After protein concentrations were determined using the BCA assay, equal amounts of protein were reduced, alkylated, digested, and dissolved in mobile phase A for separation on an EASY-nLC 1200 UHPLC system. Mass spectrometry was then performed in data-independent acquisition mode with a high-energy collisional dissociation energy of 27. Differentially expressed proteins were defined by a variable importance in projection (VIP) score of > 1, an absolute log₂ fold change (FC) of > 1, raw P < 0.05, and a false discovery rate-adjusted P value < 0.05 after Benjamini-Hochberg correction.

Metabolomic analysis

Metabolomic profiling of serum samples from 80 patients (SPSS group, n = 50; non-SPSS group, n = 30) was performed by Panomix Biomedical Tech Co., Ltd. (Suzhou, China). The collected serum samples were first thawed at 4 °C, mixed with methanol, vortexed, centrifuged, dried, and reconstituted in 150 μL of methanol (800 mL/L) containing 2-chloro-L-phenylalanine (4 mg/L). Separation was performed on a Waters ACQUITY UHPLC system, and metabolites were detected using a Thermo Q Exactive mass spectrometer in both positive and negative electrospray ionization modes. The criteria for differential metabolites were identical to those applied in the proteomic analysis.

Integrated omics analysis

The identified differential proteins and metabolites were subjected to Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis. Proteins and metabolites mapped to the same pathways were annotated and visualized. Shared pathways between the proteomic and metabolomic datasets were identified using Venn diagrams, and selected pathways were further visualized using bubble plots.

Animal models and experimental groups

Male Sprague-Dawley rats (body weight, approximately 250 g; Laboratory Animal Center, Hangzhou Medical College) were housed at 25 °C under a 12-hour light/dark cycle with ad libitum access to food and water. The animal handling protocol was approved by the Laboratory Animal Welfare and Ethics Committee of Mengchao Hepatobiliary Hospital, Fujian Medical University (Approval No. MCHH-AEC-2023-12). Cirrhosis was induced by intraperitoneal injection of carbon tetrachloride (CCl₄, 200 mL/L in olive oil; 3 mL/kg) twice weekly for 16 weeks; the injection site was alternated between the left and right lower quadrants[17]. Successful establishment of the cirrhosis model was defined by macroscopic nodularity and histologic pseudolobule formation.

In the first part of the experiment, 40 rats underwent cecal venography at week 16 and were stratified into SPSS and non-SPSS groups based on the findings. In the second experiment, 39 rats were allocated equally to three groups: The normal control (sham) group, the cirrhosis control group (administered only CCl₄ as described above), and the cirrhosis + AKT activator group (administered CCl₄ as described above plus SC79 at 0.4 mg/kg intraperitoneally[18], once daily for 8 weeks starting from week 8; MedChemExpress). When CCl₄ and SC79 were administered on the same day, an interval of at least 8 hours was maintained between the two treatments.

Cecal venography and SPSS assessment

Under anesthesia with pentobarbital sodium (20 mg/kg), a lower midline abdominal incision was made to expose the cecal vein, which was then punctured with a 26G needle for injection of iodixanol contrast (1 mL/second; total volume, 1-2 mL). Images were acquired using a PHILIPS Azurion 7M20 system. The diagnosis of SPSS was made independently by two experienced interventional radiologists, and any discrepancies were resolved by consensus.

Histology and immunohistochemistry

For histologic examination, rat liver samples were fixed in 10% neutral formalin, paraffin-embedded, sectioned, and stained with hematoxylin and eosin, Sirius Red, or Masson’s trichrome. For immunohistochemistry, antigen retrieval was performed using citrate buffer, followed by blocking with normal horse serum (100 mL/L). The sections were incubated overnight at 4 °C with primary antibodies against AKT (1:200, Cell Signaling Technology), p-AKT (1:200, Cell Signaling Technology), CD31 (1:1000, Servicebio), CD34 (1:2000, Servicebio), and VEGFR2 (1:1500, Immunoway). For negative controls, the primary antibodies were replaced with phosphate-buffered saline. Signals were developed with 3,3-diaminobenzidine and counterstained with hematoxylin.

Western blotting

Liver tissue was lysed in RIPA buffer containing protease inhibitors and PMSF (Beyotime, China), and protein concentrations were quantified using the BCA assay. Equal amounts of protein were separated by SDS-PAGE and transferred to nitrocellulose membranes. The membranes were incubated overnight at 4 °C with primary antibodies against AKT (1:1000, Cell Signaling Technology, United States), p-AKT (Ser 473, 1:1000, Cell Signaling Technology, United States), eNOS (1:1000, Proteintech, United States), p-eNOS (Thr 495, 1:1000, Cell Signaling Technology, United States), and β-actin (1:40000, Abcam, United Kingdom), followed by incubation with HRP-conjugated secondary antibodies for 1 hour at room temperature. Signals were detected using enhanced chemiluminescence.

NO measurement

Liver tissue was homogenized by freeze-thaw cycles and then centrifuged. NO concentration was measured using a Griess assay kit according to the manufacturer’s instructions (Beyotime, China), with a standard curve (0-100 μM) generated at 540 nm. The results were normalized to protein concentration.

Serum biochemistry

Blood was collected from the tail vein without an anticoagulant, centrifuged to obtain serum, and analyzed for liver and renal function using an automated clinical analyzer (Hitachi, Japan).

Statistical analysis

Categorical variables were expressed as n (%) and compared using the χ² test or Fisher’s exact test. Continuous variables with a normal distribution were expressed as the mean ± SD and compared using the independent-samples t test, whereas nonnormally distributed variables were expressed as the median (interquartile range) and compared using the Mann-Whitney U test. Intergroup comparisons were performed using one-way ANOVA or the Kruskal-Wallis test with Bonferroni correction. Survival was analyzed using the Kaplan-Meier method and Cox regression. Quantitative analysis of Masson’s trichrome staining, Sirius red staining, and immunohistochemistry was performed using ImageJ. Statistical analyses were conducted using R and GraphPad Prism 10. A two-sided P value < 0.05 was considered statistically significant.

RESULTS

Higher incidence of decompensation and mortality in cirrhotic patients with SPSS

Study population and baseline characteristics: A total of 709 cirrhotic patients were enrolled. Imaging studies revealed that 418 patients (59.0%) had SPSS, whereas 291 (41.0%) did not. In the SPSS group, 197 patients (47.1%) had solitary shunts, whereas 221 (52.9%) had multiple shunts. The predominant types of solitary shunts were left gastric vein-(hemi)azygos vein shunts (51.8%, 102/197), splenorenal shunts (20.8%, 41/197), recanalized umbilical veins (18.3%, 36/197), and gastrorenal shunts (Figure 1A-D). Regarding baseline demographic and clinical characteristics, no significant between-group difference was observed in sex distribution, although patients in the SPSS group were older (P < 0.05; Supplementary Table 1). Hematologic tests showed significantly lower leukocyte, hemoglobin, and platelet counts in the SPSS group than in the non-SPSS group (all P < 0.05). The SPSS group showed significant worsening of the coagulation parameters, namely, international normalized ratio (INR), prothrombin time, and prothrombin activity, as well as the liver function indicators of albumin, total bilirubin, Child-Pugh score and classification, and Model for End-stage Liver Disease (MELD) score (Figure 1E-G and Supplementary Table 1, all P < 0.05). Although there were no intergroup differences in the main portal vein diameter, the SPSS group had significantly greater diameters of the left portal vein branch, splenic vein, and superior mesenteric vein, but smaller diameters of the right portal vein branch (Supplementary Table 1, all P < 0.05). Compared to the non-SPSS group, the SPSS group also exhibited higher baseline rates of the decompensation events, namely, esophagogastric varices (EGV), esophagogastric variceal bleeding (EGVB), portal vein thrombosis (PVT), HE, ascites, and HCC (Supplementary Table 1, all P < 0.05).

Figure 1 Higher incidence of decompensation and mortality in cirrhotic patients with spontaneous portosystemic shunts. A: Representative imaging of splenorenal shunt; B: Representative imaging of gastrorenal shunt; C: Representative imaging of umbilical vein recanalization; D: Representative imaging of left gastric-azygos/hemiazygos vein shunt; E: Model for End-stage Liver Disease score in the spontaneous portosystemic shunts (SPSS) and non-SPSS groups at baseline; F: Child-Pugh score in the SPSS and non-SPSS groups at baseline; G: Child-Pugh grade in the SPSS and non-SPSS groups at baseline; H: Cumulative incidence of esophagogastric variceal bleeding during follow-up; I: Cumulative incidence of portal vein thrombosis during follow-up; J: Cumulative incidence of hepatic encephalopathy during follow-up; K: Cumulative incidence of ascites during follow-up; L: Cumulative incidence of hepatocellular carcinoma during follow-up; M: Cumulative incidence of mortality during follow-up. MELD: Model for End-stage Liver Disease; EGVB: Esophagogastric variceal bleeding; SPSS: Spontaneous portosystemic shunts.

Follow-up and survival analysis: The median follow-up duration was 31.3 months. The two groups were compared for the cumulative incidences of mortality and several decompensation events, including EGVB, PVT, HE, ascites, and HCC. The cumulative mortality incidence was 5.2% in the non-SPSS group and 23.2% in the SPSS group. In addition, the cumulative incidences of the decompensation events EGVB, PVT, HE, ascites, and HCC were 3.1%, 2.1%, 3.8%, 14.1%, and 7.2% in the non-SPSS group and 17.9%, 12.7%, 32.8%, 50.2%, and 13.4% in the SPSS group, respectively. The between-group differences in the cumulative incidences of decompensation events and mortality were statistically significant (Figure 1H-M; all P < 0.05).

Univariate Cox regression analysis was performed to identify factors associated with the incidence of the decompensation events EGVB, PVT, HE, ascites, HCC, and mortality during follow-up. Variables that were statistically significant in the univariate analysis were subsequently entered into a multivariate Cox regression model to identify independent predictors. As summarized in Supplementary Table 2, the multivariate analysis confirmed that SPSS was an independent predictor of mortality and of all investigated decompensation events during follow-up.

Proteomic profiling identifies potential mechanisms underlying SPSS formation

Clinical features and data quality control: Liver tissues from 10 patients with cirrhosis (5 with SPSS and 5 without SPSS) were subjected to proteomic analysis. Compared with the non-SPSS group, the SPSS group exhibited significantly worse liver function, as reflected by higher INR (1.61 ± 0.26 vs 1.18 ± 0.16), Child-Pugh score (9.00 ± 1.87 vs 6.00 ± 0.71), and MELD score (13.60 ± 2.41 vs 9.00 ± 2.12; Supplementary Table 3, all P < 0.05). Histopathologic examination revealed characteristic features of cirrhosis in the SPSS group, including large regenerative nodules, dilation of periportal arterial branches, and extensive fibrous septa (collagen area, 25.6% ± 2.8%, which was 1.8-fold higher than that in the non-SPSS group; Figure 2A and B). Quality control analyses demonstrated high data reliability, with Pearson correlation coefficients > 0.8 and consistent protein intensity distributions (Figure 2C and D).

Figure 2 Proteomic profiling reveals mechanisms underlying spontaneous portosystemic shunts formation. A: Representative histological changes in patients undergoing proteomic analysis; B: Quantification of Masson-stained collagen (data are presented as mean ± SD, n = 5 per group); C: Pearson correlation analysis of proteomic data; D: Distribution of protein intensity values; E: Number of differential proteins between the spontaneous portosystemic shunts (SPSS) and non-SPSS groups; F: Volcano plot of differential proteins; G: Kyoto Encyclopedia of Genes and Genomes pathway enrichment analysis of differential proteins. ^aP < 0.01. H&E: Hematoxylin and eosin; SPSS: Spontaneous portosystemic shunts.

Bioinformatic analysis: Proteomic profiling identified 100 significantly differentially expressed proteins (log₂ FC > 1, P < 0.05), including 63 upregulated and 37 downregulated proteins in the SPSS group (Figure 2E). Key altered proteins included the drug-metabolizing enzymes CYP3A5 and GBA3, as well as the signaling regulators TLR3 and RGS10 (Figure 2F). KEGG pathway enrichment analysis highlighted three major altered pathways (Figure 2G): Disrupted hepatic metabolism, including bile secretion and primary bile acid biosynthesis; abnormal drug metabolism, including cytochrome P450-related pathways and vascular remodeling pathways, particularly VEGF signaling.

Metabolomic profiling reveals dysregulated pathways associated with SPSS

Clinical characteristics and quality control: Metabolomic analysis was performed on serum samples from 80 patients with cirrhosis (50 with SPSS and 30 without SPSS). Patients with SPSS had significantly worse liver function and portal hypertension-related parameters, including higher INR [1.37 (1.21-1.66) vs 1.05 (1.00-1.22)], Child-Pugh score [8.50 (8.00-10.00) vs 5.50 (5.00-6.00)], and MELD score [14.00 (11.00-18.00) vs 7.00 (7.00-10.00), Supplementary Table 4, all P < 0.05]. Notably, the splenic vein diameter was greater in SPSS patients [10.21 (8.79-11.71) mm vs 9.08 (7.73-10.04) mm, P = 0.009], as was the baseline incidence of decompensation events (including EGV, EGVB, HE, and ascites, all P < 0.05).

QC analysis revealed tight clustering of QC samples in both positive and negative ion modes, with more than 70% of metabolic features showing relative standard deviations < 30%, indicating high data quality. Orthogonal partial least squares discriminant analysis models successfully distinguished the SPSS and non-SPSS groups, and permutation testing confirmed model reliability. Taken together, these findings indicate that the metabolic profiles of the two groups were distinct (Figure 3A-D).

Figure 3 Metabolomic profiling identifies dysregulated pathways associated with spontaneous portosystemic shunts. A and B: Permutation tests in positive and negative ion modes; C and D: Orthogonal Partial Least Squares Discriminant Analysis score plots in positive and negative ion modes; E: Number of differential metabolites between spontaneous portosystemic shunts (SPSS) and non-SPSS groups; F: Volcano plot of differential metabolites; G: Kyoto Encyclopedia of Genes and Genomes pathway enrichment analysis. OPLS-DA: Orthogonal Partial Least Squares Discriminant Analysis; FC: Fold change; SPSS: Spontaneous portosystemic shunts.

Bioinformatic analysis of metabolomic data: Fifty-four metabolites showed significant alterations (VIP score > 1.0, P < 0.05), of which 42 were upregulated and 12 were downregulated (Figure 3E). Notable changes included increases in amino acid-related metabolites such as L-erythrulose and N2-gamma-glutamylglutamine, along with decreases in the nucleotide metabolites thymine and guanine, collectively suggesting disruption of amino acid metabolism and suppression of nucleotide biosynthesis (Figure 3F). KEGG enrichment analysis identified 20 significantly perturbed pathways (Figure 3G), primarily involving liver metabolism, particularly cysteine and methionine metabolism and primary bile acid biosynthesis; vascular function regulation, including cGMP-PKG signaling; and energy metabolism, including central carbon metabolism. The involvement of the cGMP-PKG pathway complements the proteomic findings on the AKT-eNOS-NO axis, further implicating disrupted vascular homeostasis in the pathogenesis of SPSS.

Integrated proteomic and metabolomic analysis highlights key molecular pathways in SPSS

By integrating the proteomic and metabolomic datasets using systems biology approaches, we identified 11 significantly enriched KEGG pathways shared between the two datasets (Figure 4A and B). These pathways were grouped into three functional categories: (1) Pathways related to vascular function regulation, including cGMP-PKG signaling and arachidonic acid metabolism; (2) Hepatic metabolic pathways, including primary bile acid biosynthesis, bile secretion, and tyrosine metabolism; and (3) Pathways related to substance transport and energy metabolism, such as ABC transporters and unsaturated fatty acid biosynthesis.

Figure 4 Integrated proteomic and metabolomic analysis highlights key pathways in spontaneous portosystemic shunts. A: Venn diagram of common Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways enriched in both proteomics and metabolomics; B: Bubble plot of common KEGG pathways enriched in both proteomics and metabolomics; C: Enrichment analysis of the cGMP-PKG pathway; D: Immunohistochemical staining of AKT and p-AKT in liver tissues; E: Quantification of AKT staining; F: Quantification of p-AKT staining. Data are presented as mean ± SD (n = 5 per group). ^aP < 0.05. NO: Nitric oxide; SPSS: Spontaneous portosystemic shunts.

Among these pathways, the cGMP-PKG signaling pathway emerged as a particular focus, with AKT and 5'-GMP, two key molecules in this pathway, showing significant dysregulation in SPSS patients (Figure 4C). Given the role of the cGMP-PKG pathway in regulating vascular tone and the well-established involvement of the AKT-eNOS-NO axis in vascular remodeling, we prioritized this cascade for further investigation. Several lines of evidence supported this decision: (1) Immunohistochemical analysis confirmed reduced expression of AKT and p-AKT in samples from patients with SPSS (Figure 4D-F); (2) The AKT pathway is experimentally tractable, with specific tools available such as the activator SC79[18]; and (3) Extensive literature has implicated the AKT-eNOS-NO pathway in portal hypertension and angiogenesis[19-21]. These findings provided a strong rationale and mechanistic basis for subsequent validation in animal models.

Animal models of SPSS and in vivo validation of key pathways

Model establishment and SPSS incidence: The animal experiments consisted of two parts. In the first experiment, we established a model of cirrhosis with spontaneous SPSS using 40 Sprague-Dawley rats (Figure 5A). By the end of week 16, 33 rats had survived (82.5%). Digital subtraction angiography (DSA) via cecal venous puncture (Figure 5B) was performed in all surviving rats, and 20 (60.6%) showed SPSS formation. Representative DSA images from the SPSS and non-SPSS groups are shown in Figure 5C. DSA revealed that rats with SPSS exhibited classic signs of portal hypertension, including dilated and tortuous splenic and mesenteric veins draining directly into the inferior vena cava through spontaneous shunts.

Figure 5 Animal models of spontaneous portosystemic shunts and cecal venography. A: Experimental design of cirrhosis induction and SC79 treatment; B: Schematic diagram of cecal venography and the portal venous system; C: Representative venography images showing splenorenal shunts (orange arrows). SPSS: Spontaneous portosystemic shunts; i.p.: Intraperitoneal.

Liver function and portal hypertension: The rat models with SPSS showed significant liver dysfunction, as evidenced by lower serum albumin levels (28.0 ± 4.7 g/L vs 34.9 ± 1.5 g/L, Figure 6A) and higher alanine aminotransferase (ALT; 110 ± 17.6 U/L vs 75.6 ± 17.7 U/L) and aspartate aminotransferase (AST; 294 ± 70.2 U/L vs 201 ± 33.9 U/L) levels (Supplementary Table 5; all P < 0.05). The rats also exhibited additional signs of portal hypertension, including increased spleen weight [1.65 (1.42-2.12) g vs 1.27 (1.16-1.56) g] and a decreased liver-to-spleen weight ratio [10.81 (10.20-14.91) vs 16.54 (14.19-20.31), Figure 6B and C, P < 0.05].

Figure 6 In vivo validation of AKT-eNOS-nitric oxide signalling. A: Serum albumin in spontaneous portosystemic shunts (SPSS) and non-SPSS rats; B: Spleen weight in SPSS and non-SPSS rats; C: Liver-to-spleen weight ratio in SPSS and non-SPSS rats; D: Representative hematoxylin and eosin, Masson, and Sirius red staining of liver tissues; E: Quantification of Masson staining; F: Quantification of Sirius red staining; G: Expression of AKT, p-AKT, eNOS, and p-eNOS in liver tissues; H: Hepatic nitric oxide levels. Data are presented as mean ± SD (n = 5 per group). ^aP < 0.05; ^bP < 0.01; ^cP < 0.001. H&E: Hematoxylin and eosin; NO: Nitric oxide; SPSS: Spontaneous portosystemic shunts.

Histopathological changes: Masson’s trichrome staining revealed significantly greater collagen deposition in the SPSS group than in the non-SPSS group (22.8% ± 1.8% vs 15.8% ± 1.8%, Figure 6D and E, P < 0.05), consistent with the findings of Sirius red staining (23.0% ± 2.1% vs 15.8% ± 1.1%, Figure 6F, P < 0.05). Other histologic changes observed in rats with SPSS included increased pseudolobules, widened fibrous septa, inflammatory infiltration, and bile duct proliferation.

AKT-eNOS-NO pathway assessment: Western blot analysis revealed significantly reduced expression of phosphorylated AKT (Ser473) and phosphorylated eNOS (Thr495) in SPSS liver tissue (Figure 6G), along with a 36.7% decrease in hepatic NO content (Figure 6H; P < 0.01), suggesting that impaired AKT-eNOS-NO signaling may contribute to SPSS formation. In addition, immunohistochemical staining showed significantly increased expression of CD31, CD34, and VEGFR2 in the SPSS group compared with the non-SPSS group (Supplementary Figure 1, P < 0.05), further indicating enhanced angiogenesis and vascular remodeling associated with SPSS formation.

SC79 treatment ameliorates cirrhosis and reduces SPSS incidence in rats

Experimental design and baseline features: In the second animal experiment, 39 Sprague-Dawley rats were randomized into three groups: Sham control (n = 13), cirrhosis model (n = 13), and SC79-treated cirrhosis model (n = 13). The overall survival rate was 94.9% (37/39), with two rats in the cirrhosis group dying early due to massive ascites. All remaining rats survived to week 16, after which they underwent systematic cecal vein puncture angiography.

Therapeutic outcomes: DSA showed that the incidence of SPSS was 0% in the sham group, 66.7% (8/12) in the cirrhosis control group, and 16.7% (2/12) in the SC79-treated group, with a significantly lower incidence in the SC79-treated group than in the cirrhosis control group (Figure 7A, P < 0.05). Moreover, SC79 treatment improved liver function, as evidenced by increased albumin levels [from 30.0 (27.5-30.0) g/L to 32.0 (32.0-32.0) g/L, Figure 7B, P < 0.05], decreased total bilirubin levels [from 7.2 (2.5-15.7) µmol/L to 1.4 (1.0-2.7) μmol/L, Supplementary Table 6, P < 0.05], and reduced AST levels [from 280.0 (157.0-401.0) U/L to 162.0 (101.0-211.0) U/L, Supplementary Table 6, P < 0.05]. Similarly, the incidence of ascites decreased from 66.7% to 16.7% (Figure 7C, P < 0.05).

Figure 7 SC79 treatment ameliorates cirrhosis and reduces spontaneous portosystemic shunts in vivo. A: Incidence of spontaneous portosystemic shunts in sham, cirrhosis, and SC79-treated groups; B: Serum albumin; C: Ascites incidence; D: Representative liver histology by hematoxylin and eosin, Masson, and Sirius red staining; E: Quantification of Masson staining; F: Quantification of Sirius red staining; G: Expression of AKT, p-AKT, eNOS, and p-eNOS; H: Hepatic nitric oxide levels. Data are showed as mean ± SD (n = 5 per group). ^aP < 0.05; ^bP < 0.01; ^cP < 0.001. NS: Not significant; H&E: Hematoxylin and eosin; NO: Nitric oxide.

Histological improvement: Quantitative Masson’s trichrome staining showed significantly lower collagen deposition in the livers of SC79-treated rats (13.8% ± 0.7%) than in those of cirrhosis controls (22.4% ± 1.9%, Figure 7D and E; P < 0.01). Similar changes were observed on Sirius red staining (22.0% ± 1.9% vs 12.6% ± 0.8%, Figure 7F, P < 0.01). The SC79-treated rats exhibited fewer pseudolobules, markedly reduced inflammatory infiltration, and improved hepatic architecture.

Activation of AKT-eNOS-NO signaling by SC79: Western blot analysis showed that p-AKT and p-eNOS levels were lower in the cirrhosis model group than in the sham group and that SC79 treatment significantly increased the phosphorylation of both proteins (Figure 7G). Correspondingly, hepatic NO content was reduced in the cirrhosis model group but increased following SC79 administration (Figure 7H). In addition, immunohistochemical staining showed that the expression of CD31, CD34, and VEGFR2 was increased in the cirrhosis model group and decreased after SC79 treatment (Supplementary Figure 2, P < 0.05). These results suggest that SC79 may attenuate fibrosis and reduce SPSS formation by activating the AKT-eNOS-NO pathway and suppressing angiogenesis-related vascular remodeling.

DISCUSSION

This multicenter cohort study confirms the close association of SPSS with an increased incidence of decompensation events, such as EGVB, PVT, ascites, and HE, as well as all-cause mortality in patients with cirrhosis. To further explore the molecular mechanisms underlying SPSS formation, we performed integrative multi-omics analyses and identified the AKT-eNOS-NO signaling pathway as a candidate pathway involved in this process. Notably, this finding was further supported by experiments in rat models of cirrhosis, in which the AKT phosphorylation activator SC79 effectively reduced the incidence of SPSS and markedly ameliorated the severity of hepatic fibrosis and portal hypertension. These findings link the clinical significance of SPSS with its potential molecular basis and experimental validation, deepen our understanding of cirrhosis-related complications, and identify novel potential therapeutic targets for clinical intervention.

The clinical significance of SPSS remains controversial; however, the general consensus is that although SPSS serves as a compensatory mechanism in portal hypertension, it provides limited clinical benefit[3,4,22]. In contrast, its pathophysiologic impact is substantial and primarily characterized by progressive hepatic hypoperfusion, which leads to worsening liver function, hepatic atrophy, sparse or absent portal vein branches, and PVT, all of which are salient features of portosystemic shunt syndrome[23-25]. This multicenter cohort analysis systematically evaluated the impact of SPSS on liver function parameters and clinical prognosis in cirrhotic patients. Our results unequivocally demonstrate that patients with SPSS have significantly more severe hepatic impairment and a markedly higher cumulative incidence of major decompensation events during follow-up, along with increased all-cause mortality. These findings corroborate previous reports and strengthen the evidence supporting the detrimental effects of SPSS[7,8], thereby providing compelling justification for enhanced monitoring and management of SPSS in clinical practice.

In terms of mechanistic evidence, the combined proteomic and metabolomic analyses in this study showed a close association between SPSS formation and dysregulation of the AKT-eNOS-NO signaling axis in liver tissue and serum in cirrhosis. Compared with the control group, the SPSS group showed decreased phosphorylation of AKT at Ser473, accompanied by attenuation of eNOS phosphorylation at Thr495 and reduced concentrations of NO metabolites in liver tissue. In addition, immunohistochemical staining showed increased expression of CD31, CD34, and VEGFR2 in the SPSS group, indicating enhanced angiogenesis and vascular remodeling during SPSS formation. These changes may contribute to hepatic sinusoidal endothelial dysfunction, reduced NO bioavailability, and pathological collateral vessel formation.

These findings, particularly those regarding the therapeutic value of targeting the AKT-eNOS-NO pathway, were further validated by our in vivo experiments in rat models of cirrhosis. We found that SC79 administration activated AKT phosphorylation, restored eNOS Thr495 phosphorylation, and improved hepatic NO synthetic capacity. SC79 treatment reduced the incidence of SPSS from 66.7% in control rats to 16.7% in treated rats and was associated with attenuated hepatic fibrosis and reduced collagen deposition. In addition, SC79 treatment reduced the expression of CD31, CD34, and VEGFR2, indicating suppression of angiogenesis-related remodeling. Although body weights were comparable between groups, rats with SPSS exhibited significantly increased spleen weight and decreased liver-to-spleen weight ratios, whereas these changes were partially reversed after SC79 treatment. These results highlight the efficacy of SC79 in mitigating SPSS-associated pathological changes, possibly through improving vascular homeostasis and suppressing angiogenesis-related remodeling.

AKT signaling is known to exert dual biological effects. Previous studies have implicated sustained AKT activation in tumorigenesis[26,27], highlighting the need to carefully evaluate the treatment window, preferably early in compensated cirrhosis, as well as long-term safety, particularly in patients with HCC risk factors such as HBV infection or metabolic dysfunction-associated steatotic liver disease. However, in our study, during the 16-week observation period, SC79 showed no adverse effects on hepatic or renal function, with ALT, AST, and creatinine levels comparable to those in controls (P > 0.05). Therefore, comprehensive risk-benefit assessments are essential before widespread clinical application.

This study has several distinguishing features. First, to our knowledge, it is the first to investigate SPSS pathogenesis using an integrated proteomic and metabolomic approach. Second, by combining clinical analysis, multi-omics profiling, and animal validation, we linked the clinical significance of SPSS with its potential molecular basis and therapeutic modulation. Third, our findings suggest that targeting the AKT-eNOS-NO pathway may represent a potential strategy for reducing SPSS formation and alleviating portal hypertension-related pathological changes. Collectively, these results deepen the current understanding of SPSS in cirrhosis and provide new insight into its pathogenesis and potential treatment.

Despite these merits, this study has several limitations. First, portal venous pressure was not measured during rat angiography, which limited direct hemodynamic assessment of SPSS and the effects of SC79 treatment. Second, the formation of SPSS in rats with chemically induced cirrhosis may not fully reflect its natural progression in human patients. Third, the long-term safety profile of SC79, including its potential impact on hepatocarcinogenesis, remains to be established. Fourth, the proteomic analysis was based on liver tissues from patients undergoing liver transplantation, and the omics cohorts were selected based on sample availability, which may introduce selection bias and affect the representativeness of the omics findings. Finally, the generalizability of our clinical findings is limited by the exclusive inclusion of a cohort with HBV-related cirrhosis. Future studies should incorporate direct portal pressure measurement, optimize experimental models that better mimic human SPSS progression, and further evaluate the long-term safety and therapeutic potential of SC79.

CONCLUSION

This study demonstrates that SPSS is associated with poor prognosis in cirrhosis. Multi-omics analyses identified dysregulation of the AKT-eNOS-NO pathway as a potential mechanism underlying SPSS formation, and this finding was further supported by in vivo validation. These results highlight a potential therapeutic target for reducing SPSS occurrence and delaying disease progression.