Introduction
Sinusoidal obstruction syndrome (SOS), also known as veno-occlusive disease, is a hepatic vascular disease characterized by liver dysfunction, intrahepatic congestion, and sinusoidal portal hypertension.1 SOS is typically observed in Western countries after hematopoietic stem cell transplantation, whereas the ingestion of products containing pyrrolizidine alkaloids (PAs), such as Gynura segetum, is a leading cause of SOS in China. Untreated PA-SOS is associated with high mortality.2 However, since the implementation of anticoagulation-transjugular intrahepatic portosystemic shunt (TIPS) stepwise therapy, mortality has markedly decreased to approximately 10%, resulting in the survival of most patients.3 Although several studies have extensively evaluated the pathophysiology, diagnosis, liver histology, and natural history of patients with PA-SOS,4–6 the long-term outcomes of surviving patients remain poorly understood.
In recent years, we have identified a substantial number of patients who survived the acute phase of PA-SOS. In some of these patients, a second liver biopsy was performed while they were completely asymptomatic. This cohort of patients who survived PA-SOS offers a unique opportunity to assess the long-term consequences of the disease, which is the aim of our study. In addition, we aimed to evaluate findings observed in liver biopsies performed during complete or partial recovery from PA-SOS.
Methods
Study population
The Department of Gastroenterology, Drum Tower Hospital affiliated with Nanjing University Medical School, has maintained a prospective registry of patients diagnosed with PA-SOS since January 2017. These patients were followed up regularly until the end of follow-up for this study, which was April 20, 2025. Long-term follow-up was based on outpatient clinic visits and telephone interviews.
PA-SOS is diagnosed according to the Nanjing criteria (a combination of a definitive history of exposure to plants containing PAs before the appearance of symptoms, typical radiological findings, abnormal laboratory indicators, and clinical manifestations),3 which have been confirmed to have a diagnostic accuracy close to 100% compared with pathological diagnosis.7 Patients must be older than 18 years and must not have any of the following exclusion criteria: (1) the presence of two or more additional etiologies of liver injury, such as HBV positivity, HCV positivity, alcohol abuse, or autoimmune liver disease; (2) other serious diseases (cardiopulmonary failure, serious cardiovascular or cerebrovascular diseases, or malignant tumors); or (3) a large amount of incomplete clinical data. To ensure a comprehensive study and fully informed consent, our protocol stipulated that only patients under regular outpatient follow-up would be considered for convalescent liver biopsy. As such, most patients were not asked about liver biopsy because their follow-up was conducted via telephone or through data uploaded from local hospitals. Finally, 14 patients who were willing to undergo liver biopsy during the clinical recovery period and had good-quality biopsy specimens (greater than 1.5 cm in length or otherwise considered adequate for interpretation by an expert pathologist)8 were identified. The clinical recovery period was defined as beginning at least 3 months after the initiation of anticoagulation or TIPS treatment if the following additional conditions were met: (1) significant improvement in abdominal distention and upper abdominal pain; (2) a decrease in serum alanine aminotransferase levels to 50% of the baseline value or within the normal range (9–40 U/L); (3) disappearance of ascites confirmed by abdominal ultrasound; and (4) a Drum Tower Severity Scoring (DTSS) of 4–6 points, reflecting at most mild disease severity.9
As previously mentioned, a liver biopsy is not mandatory for the diagnosis of PA-SOS. Consequently, only 3 of the 14 patients also had liver biopsy results available at the time of diagnosis of PA-SOS. To compare diagnostic and recovery biopsy findings, we selected patients from our registered PA-SOS cohort who had undergone a diagnostic liver biopsy. For each of the 14 patients with a recovery liver biopsy, two PA-SOS patients with a diagnostic liver biopsy were matched on the basis of age, sex, treatment method, and baseline DTSS score.
This study protocol was reviewed and approved by the Ethics Committee of Nanjing Drum Tower Hospital (2022-486-02). The flow diagram is shown in Figure 1.
Data collection
Clinical data of the 14 included patients were retrospectively extracted from the clinical records of the patients using a case report form specifically predesigned for the study. Demographic, clinical, biochemical, and imaging (ultrasound and/or computed tomography [CT] scan) data were collected at baseline and at 1, 3, 6, 12, 24, and 48 months or longer whenever possible for a comprehensive assessment of baseline clinical characteristics and follow-up changes. Liver histology data were obtained from 28 diagnostic liver biopsies (3/14 in our study and another 25 PA-SOS cases) and 14 recovery liver biopsies.
Histopathologic assessment
Liver biopsy specimens that had been fixed in 10% buffered formaldehyde were processed for paraffin embedding according to standard procedures. Subsequently, 5 µm sections were cut from the paraffin-embedded samples and stained with hematoxylin and eosin (H&E), Sirius red, and immunohistochemistry (IHC) in accordance with a previously described protocol. Antibodies against CD34 (Proteintech, 14486-1-AP) and CD34 (Servicebio, GB15013) were used for human and rat IHC, respectively. Normal CD34 staining was defined as strong cytoplasmic expression confined to the endothelial cells of the portal tract and central vein. Cases were scored from 0 (normal) to 3+ (markedly aberrant, defined as strong centrilobular sinusoidal expression).10
Histopathological evaluation was jointly performed by two experienced hepatic pathologists (JC and YL) who were blinded to the clinical data and biopsy time points. They reviewed the slides together, resolved discrepancies through on-site discussion, and finalized unified semiquantitative and quantitative results. The primary observational features of specific and nonspecific histologic findings in porto-sinusoidal vascular disease (PSVD) were as follows: specific findings included obliterative portal venopathy (OPV), incomplete septal fibrosis, and nodular regenerative hyperplasia; nonspecific findings included portal vein herniation, an increased number of portal vessels, nonzonal sinusoidal dilation, perisinusoidal fibrosis, collagen fiber hyperplasia, and portal tract remnants. Moreover, other histological signs, such as bile ductular proliferation, sinusoidal hemorrhage, steatosis, portal inflammation, and lobular inflammation, were also recorded. In addition, the Scheuer scoring system was used for grading inflammation (G0–G4) and staging fibrosis (S0–S4) for overall evaluation. The expression was evaluated throughout the entire area of each sample, both quantitatively and qualitatively. Detailed definitions of the indices mentioned above are presented in Supplementary Table 1 according to a literature review and group consensus.10,11
Animal models of PA-induced liver injury
Male Sprague-Dawley rats were purchased from Beijing Vital River Laboratory Animal Technology Co. Ltd. (Beijing, China). They were kept in a 12-h dark/light cycle at a temperature of 23 °C and a humidity of 65%. Rats were gavaged with monocrotaline (MCT) (Sigma-Aldrich, St. Louis, MO, USA) to establish a model of PA-induced liver injury, which is a well-established and reproducible model of SOS. The MCT solution was prepared according to previously reported methods. We established a rat model of acute resolving liver injury as described in a previous study.12 Briefly, the rats were intragastrically administered a single dose of MCT (130 mg/kg) on Day 0. Equal volumes of vehicle were used as controls. After 2 days, 1 week, 2 weeks, or 4 weeks, the rats were euthanized for blood collection. These time points, covering the recognized acute and early chronic stages of MCT-induced SOS as well as recovery phases, were chosen based on our preliminary experiments and a published study.12 Afterwards, the livers were collected for H&E staining, Sirius red staining, IHC, Western blotting, and quantitative real-time polymerase chain reaction (PCR).13
Western blotting
Liver tissues were lysed with RIPA lysis buffer (Beyotime, Shanghai, China) containing a protease inhibitor cocktail and phosphatase inhibitor. Total protein was quantified using the BCA Protein Assay Kit (Thermo Fisher Scientific, MA, USA). Equal amounts of protein were added to SDS-PAGE gels for electrophoresis, blocked with 5% nonfat milk, and incubated with a suitable concentration of primary antibody overnight at 4 °C. The antibodies used were CD34 (1:800, Servicebio, GB15013) and β-actin (1:3,000, Proteintech, 81115-1-RR). After incubation with the corresponding secondary antibodies, the protein was visualized with UltraSignal hypersensitive ECL chemiluminescence substrate (4A Biotech, Beijing, China).14
RNA extraction and quantitative reverse-transcriptase (qRT)-PCR
Total RNA from tissues and cells was extracted using the Takara MiniBEST Universal RNA Extraction Kit, and cDNA was synthesized using the Takara PrimeScript RT Reagent Kit (Takara, Kyoto, Japan) as described previously.14,15 Then, qRT-PCR assays were performed using SYBR Green qPCR MasterMix (abm, Vancouver, Canada) according to the manufacturer’s instructions. The primer sequences used were as follows: CD34 (Rat), forward, 5′-GCCTGCCGTCTGTCAATG-3′, reverse, 5′-TCCTCACAACTAGATGCTTCACTT-3′; β-actin (Rat), forward, 5′-CGTTGACATCCGTAAAGACCTC-3′, reverse, 5′-TAGGAGCCAGGGCAGTAATCT-3′.
Statistical analysis
Normally distributed continuous variables are expressed as means and standard deviations (means ± standard deviations), whereas medians and interquartile ranges were calculated for variables that did not meet the assumption of normality. Continuous variables were compared using paired-sample t tests or Wilcoxon signed-rank tests for paired data, and Student’s t tests or Mann–Whitney U tests for unpaired data, as appropriate. Categorical data were presented as frequencies and percentages and compared using the chi-square or Fisher’s exact test, as appropriate. Propensity score matching was conducted using the MatchIt package for R software. All statistical analyses were conducted using SPSS 23.0 or GraphPad Prism 8 (GraphPad Software). P-values < 0.05 were considered statistically significant.
Results
Histological findings at recovery liver biopsy
Fourteen PA-SOS patients, with a mean age of 67 ± 8 years and a male-to-female ratio of 1.8:1, were included in the study. As summarized in Supplementary Table 2, among the 14 patients included, 11 patients (78.6%) received TIPS after 2 weeks of ineffective anticoagulation treatment, while 3 patients (21.4%) were in the control group and received anticoagulation alone. The indications for TIPS were defined as (1) either total bilirubin ≥ 5 mg/dL or peak portal vein velocity < 10 cm/s or (2) any two of the following four criteria being met during 2 weeks of anticoagulation therapy: no improvement in total bilirubin (<5 mg/dL), an increase of less than 10% in the peak portal vein velocity from baseline, no significant improvement in ascites, or concurrent renal or coagulation deterioration.16 The median time from symptom onset to TIPS intervention was 38 days (range, 11–95 days). The changes in serological and imaging indicators at different stages can be found in Supplementary Tables 3 and 4. All the above clinical, biochemical, and imaging alterations gradually improved until they disappeared after a median of 3 months (range, 1.5–6 months) following TIPS or after a median of 2.5 months of anticoagulation therapy (range, 2–3 months) (Supplementary Figs. 1–3). Given that most patients declined liver biopsy, we also compared patients who underwent biopsy with those who did not, using the available data, to rule out potential selection bias, and no significant differences were detected (Supplementary Table 5).
In the 14 patients who underwent liver biopsy at recovery, the mean time from the first manifestation to liver biopsy was 20.5 months (range, 3–61 months), and the time from recovery to biopsy was 17.5 months (range, 0–58 months). As previously mentioned, only 3 patients underwent liver biopsy at the time of PA-SOS diagnosis, allowing direct comparison of pathological changes before and after recovery. For this reason, and to better characterize the acute lesions of PA-SOS, we selected a group of 25 additional PA-SOS patients from our prospective registry who, although they did not undergo liver biopsy at recovery, had a liver biopsy during the acute phase. The characteristics of the two groups are shown in Supplementary Table 6, indicating that both groups were well matched at the time of acute PA-SOS diagnosis.
Overall, compared with the findings of PA-SOS in diagnostic liver biopsies, there was a high prevalence of specific and nonspecific histological signs described in PSVD in liver biopsies performed at recovery. Eight patients (57%) showed at least 1 of the 3 histological features designated by the Vascular Liver Disease Group as “specific for the diagnosis of PSVD,” i.e., enabling a diagnosis of PSVD irrespective of clinical findings (Table 1). Fifty percent of the patients exhibited OPV, and 1 patient had incomplete septal fibrosis; representative images are shown in Figure 2. However, the duration of follow-up might have been too short to detect any changes in nodular regenerative hyperplasia. Among the 6 remaining cases, 3 met the alternative criteria for a PSVD diagnosis, i.e., the joint presence of 1 less-specific clinical sign of portal hypertension. Specifically, one patient had a platelet count of 85 × 109/L accompanied by portal vein herniation and perisinusoidal fibrosis; the second had a platelet count of 65 × 109/L with portal vein herniation alone; and the third had a platelet count of 106 × 109/L, with both portal vein herniation and an increased number of portal vessels. With the inclusion of these 3 cases, a total of 11 cases (79%) exhibited histological PSVD-like changes on liver biopsy. In particular, all patients had collagen fiber hyperplasia, while 43% of them had increased portal vessel counts, 50% had portal vein herniation, 86% had portal tract remnants, and 79% had perisinusoidal fibrosis. Furthermore, nonzonal sinusoidal dilation was present in 64% of the patients. Representative histopathological images with matched pathology from the same patient at diagnostic and recovery liver biopsy are also shown in Figure 3. Overall, patients manifested hepatic sinusoidal dilation and congestion in the acute phase. CD34 expression was absent or minimal, with scores ranging from 0 to 1, and no prominent portal vein abnormalities were observed. In contrast, most cases presented with portal vein abnormalities, such as herniation, stenosis, or occlusion, during the recovery phase, and CD34 scores rose to 2–3.
Table 1Evaluation of pathological characteristics in patients with PA-induced liver injury between liver biopsies at diagnosis and recovery.
| Variable | Levels | Liver biopsy at diagnosis (n = 28) | Liver biopsy at recovery (n = 14) | P-value |
|---|
| Specific histological signs | | | | |
| Obliterative portal venopathy | No | 26 (93%) | 7 (50%) | 0.003 |
| Yes | 2 (7%) | 7 (50%) | |
| Incomplete septal fibrosis | No | 28 (100%) | 13 (93%) | 0.406 |
| Yes | 0 (0%) | 1 (7%) | |
| Nodular regenerative hyperplasia | No | 28 (100%) | 14 (100%) | 0.999 |
| Yes | 0 (0%) | 0 (0%) | |
| Nonspecific histological signs | | | | |
| Portal vein herniation | No | 26 (93%) | 7 (50%) | 0.003 |
| Yes | 2 (7%) | 7 (50%) | |
| Increased number of portal vessels | No | 26 (93%) | 8 (57%) | 0.011 |
| Yes | 2 (7%) | 6 (43%) | |
| Non-zonal sinusoidal dilation | 0 point | 1 (4%) | 5 (36%) | <0.001 |
| 1 point | 13 (46%) | 8 (57%) | |
| 2 points | 14 (50%) | 1 (7%) | |
| Perisinusoidal fibrosis* | 0 point | 4 (40%) | 3 (21%) | 0.146 |
| 1 point | 5 (50%) | 5 (36%) | |
| 2 points | 1 (10%) | 5 (36%) | |
| 3 points | 0 (0%) | 1 (7%) | |
| Collagen fiber hyperplasia | 0 point | 14 (50%) | 0 (0%) | <0.001 |
| 1 point | 11 (39%) | 2 (14%) | |
| 2 points | 3 (11%) | 8 (57%) | |
| 3 points | 0 (0%) | 4 (29%) | |
| Portal tract remnant | No | 18 (64%) | 2 (14%) | 0.003 |
| Yes | 10 (36%) | 12 (86%) | |
| Other histological signs | | | | |
| CD34-aberrant centrilobular sinusoidal staining* | 0 point | 12 (60%) | 0 (0%) | <0.001 |
| 1 point | 7 (35%) | 6 (46%) | |
| 2 points | 1 (5%) | 6 (46%) | |
| 3 points | 0 (0%) | 1 (8%) | |
| Bile ductular proliferation | No | 24 (86%) | 4 (29%) | <0.001 |
| Yes | 4 (14%) | 10 (71%) | |
| Sinusoidal hemorrhage | % | 37.14 ± 24.09 | 0.71 ± 2.67 | <0.001 |
| Steatosis | No | 26 (93%) | 8 (57%) | 0.011 |
| Yes | 2 (7%) | 6 (43%) | |
| Portal inflammation | 0 point | 25 (89%) | 0 (0%) | <0.001 |
| 1 point | 3 (11%) | 14 (100%) | |
| Lobular inflammation | 0 point | 24 (86%) | 9 (64%) | 0.092 |
| 1 point | 4 (14%) | 4 (29%) | |
| 2 points | 0 (0%) | 0 (0%) | |
| 3 points | 0 (0%) | 1 (7%) | |
| Grading of inflammation | 0 point | 21 (75%) | 0 (0%) | <0.001 |
| 1 point | 7 (25%) | 12 (86%) | |
| 2 points | 0 (0%) | 1 (7%) | |
| 3 points | 0 (0%) | 1 (7%) | |
| Staging of liver fibrosis | 0 point | 15 (54%) | 1 (7%) | <0.001 |
| 1 point | 13 (46%) | 9 (64%) | |
| 2 points | 0 (0%) | 3 (21%) | |
| 3 points | 0 (0%) | 1 (7%) | |
Interestingly, hepatic steatosis was detected in 6 of 14 patients (43%), comprising 4 cases of microvesicular steatosis and 2 cases of mixed-type steatosis, involving approximately 5%–8% of hepatocytes. Among the 13 evaluable recovery biopsies, 46% of patients had aberrant expression of CD34 in centrilobular sinusoids (score of 1+), 46% had a score of 2+, and 8% had a score of 3+ (Supplementary Figs. 4 and 5).
In addition, we conducted a further analysis of the characteristics of pathological alterations at different time points of liver biopsy when the patients achieved recovery (from a clinical and biochemical point of view). Detailed pathological information for the 14 patients is provided in Supplementary Table 7. The interval between initial clinical recovery and liver biopsy at recovery ranged from 0 to 58 months. After stratification by time period, we found 3 cases within 3 months, 3 cases between 3 and 12 months, 4 cases between 12 and 24 months, and 4 cases beyond 24 months. As shown in Supplementary Figure 6, within 3 months, patients experienced portal vein stenosis, accompanied by incomplete repair of sinusoidal hemorrhage. After the initial 3-month period, patients developed obvious OPV, increased portal vessel counts, and proliferation of collagen fibers in the portal tract, with the disappearance of sinusoidal hemorrhage; however, in some cases, sinusoidal dilation could occur. On the basis of clinical recovery and liver biopsy time, we also conducted a correlation analysis and found that, with prolongation of the time from first manifestation to clinical recovery, collagen fiber proliferation tended to increase, but no significant correlation was found between this proliferation and the time from recovery to liver biopsy (Fig. 4). In addition, the data also suggested a consistent trend toward longer recovery times among patients with concomitant OPV or portal vein herniation, although the difference was not significant.
Follow-up
The mean duration from disease initiation to the last follow-up was 50.9 months (range, 3–81 months). All 14 patients recovered well and achieved symptom improvement. Only one patient in our study group who underwent TIPS developed splenomegaly and persistent thrombocytopenia during the 23-month follow-up period after the initial visit. Notably, none of the three patients in the anticoagulation therapy group developed clinically evident portal hypertension-related symptoms during the follow-up period.
Animal model analysis
To further confirm these findings, an animal model of PA-induced SOS was created through the administration of MCT following the flowchart shown in Figure 5A. The serum levels of alanine aminotransferase and aspartate aminotransferase gradually decreased as the recovery period progressed and returned to baseline at 2 weeks (Fig. 5B). Surprisingly, in contrast to the serological results, both the H&E and Sirius Red staining results revealed significant histopathological abnormalities at the second week (Fig. 5C and D). Unlike the pathological features on the second day of modeling, rats showed no significant hepatic sinusoidal dilation or congestion after 2 weeks of treatment. Instead, they exhibited obvious fibrous septa formation and portal vein stenosis (Supplementary Table 8), which were similar to the pathological evolution observed in PA-induced SOS patients. In addition, the percentage of CD34-positive cells and the RNA and protein expression levels of CD34 significantly increased at 2 or 4 weeks (Fig. 5E and Supplementary Figs. 7 and 8). The main histological alterations observed in the MCT-induced SOS animal models are summarized in Figure 6.
Discussion
Studies of the pathological features of both hematopoietic stem cell transplantation- and PA-related SOS have focused on the initial stages of the disease. Understanding of the natural history of SOS remains limited, especially regarding the histopathological features of the liver during the clinical recovery period or even after clinical cure. In this study, we first reported that PSVD-specific changes, predominantly OPV, were observed in more than half of the patients at recovery. Additional nonspecific alterations, including portal vein herniation, an increased number of portal vessels, and hyperplasia of collagen fibers, were also identified among the 14 consecutive patients. Our animal experiments also revealed this phenomenon.
We found that the vast majority of serological markers returned to normal, with no map-like enhancement on CT. Furthermore, during the month after the initiation of treatment, the results of most liver function tests returned to within normal limits, accompanied by regression of ascites, which is consistent with the findings of other studies.17 Only a small number of patients had mild biochemical abnormalities at the time of liver biopsy, which may potentially confound the pathological findings. With respect to the imaging findings during the follow-up of patients with SOS, anticoagulation therapy or TIPS not only improved blood flow velocity in the portal vein system but also reduced the degree of patchy and map-like enhancement. Similar results were previously reported by Huang et al., who reported that liver parenchymal enhancement tended to become homogeneous again, with disappearance of ascites 1 month after TIPS.18 In particular, although liver biopsy provides clear evidence for the diagnosis of SOS, only a small number of patients underwent percutaneous liver biopsy because of coagulopathy and the high risk associated with excessive ascites. In addition, considering its invasive nature, high cost, and technical requirements, the application of transjugular liver biopsy was also limited, and patients in the recovery stage were often reluctant to undergo liver biopsy.19
We delineated the pathological features of different phases of PA-induced liver injury for the first time. The evolution from hepatocellular and hepatic sinusoidal injury in the acute stage to hepatic fibrosis, portal vein stenosis, and even atresia and herniation in the recovery stage raises a series of questions: 1. The pathological changes in PA-related SOS during the convalescent period are obviously similar to the specific pathological features of PSVD. Does this mean that SOS and PSVD are essentially two stages of the same disease? 2. This alteration in PA-related liver pathology during the convalescent phase is rarely observed in the convalescent phase in patients treated with acetaminophen or in most other cases of drug-related acute liver injury20,21; therefore, what are the mechanisms responsible for this difference?
With the advent of specific histopathological and clinical diagnostic criteria, PSVD, a spectrum of diseases that encompasses a multiplicity of subtle abnormalities of the hepatic portal tract, sinusoids, and parenchymal architecture,22,23 is no longer solely a diagnosis of exclusion. This realization provides a new perspective in which PSVD and other diseases can coexist.24,25 To date, the evolution of the pathological features of PA-SOS has not been reported. The most intuitive and common alteration was that all patients had portal vein stenosis to varying degrees, with more than half of them showing strong OPV, which was a typical histologic finding of PSVD. Additionally, CD34 staining, proposed in the literature to indicate abnormal vascular marker expression in livers with PSVD-type alterations,26,27 was at least mildly aberrant (1+) in patients during the recovery period, revealing capillarization of the sinusoids, which was associated with the development of PSVD.28–30 Similarly, a study by Jafari et al. suggested that abnormal CD34 expression may be a valuable adjunct in the workup of PSVD.10 Although CD34 is not specific for PSVD, its increased expression in our cohort may also be associated with PSVD-like changes. Moreover, hepatic steatosis was observed in a subset of patients and may represent a potential confounding factor, as mild steatosis may contribute to perisinusoidal fibrosis. However, it is unlikely to fully explain the portal tract abnormalities observed in our cohort.
Furthermore, three cases were diagnosed using alternative nonspecific clinical and histological criteria in our cohort, which may overestimate the true prevalence of PSVD. Larger cohorts with longer follow-up are required for further validation. In addition, no specific signs of portal hypertension that could be clearly attributed to these pathological changes were observed during the follow-up period in our cohort. The reasons might include the following two aspects. On the one hand, 79% of patients underwent TIPS at disease initiation, during which stent implantation in the portal vein can establish a shunt pathway and significantly reduce portal vein pressure. On the other hand, as indicated by Zhang et al.,31 PSVD patients without portal hypertension experienced slow progression and did not exhibit any new signs of portal hypertension during follow-up; moreover, only a small number of patients were diagnosed after a median follow-up of 8.6 years in a study in France.32 To address this issue, a multicenter study with a larger sample size and longer duration of follow-up is needed.
Nonetheless, an unresolved question remains: why do certain patients experience this transition in PA-induced liver injury? Previous studies have shown that PA metabolites primarily have direct toxic effects on sinusoidal endothelial cells, which can persist in the blood for 2–4 months.33 This indicates that even during the clinical recovery phase, toxic metabolites of PAs may persist in the circulation and possibly remain in liver tissues for an extended period, thereby continuing to cause injury. Nevertheless, PSVD-like pathological changes have been observed in patients even several years after the recovery phase, suggesting that other factors may play important roles in this pathological transition process. Our recent study revealed that the gut microbiota and intestinal barrier play vital roles in maintaining SOS homeostasis.34 Similarly, PSVD patients exhibit altered intestinal permeability, and in particular, gut-derived endotoxins are of paramount importance in this process.35–37 Recent studies have proposed that gut microbiota dysbiosis may contribute to hepatic microvascular abnormalities and PSVD development, while experimental and clinical evidence also supports a regulatory role of the gut microbiota in endothelial and microvascular homeostasis.38–40 With this background, it is possible to speculate that gut microbiota dysbiosis in patients with SOS might not be fully restored during the clinical recovery period, during which bacterial translocation and their metabolites through the enterohepatic axis cause repeated insults to the liver blood vessels, culminating in portal fibrosis and obliteration of the portal vein. Of course, this is just speculation, and specifically designed studies are needed to shed light on these aspects. Given the available data, we first present evidence for the possible existence of PSVD in PA-induced liver injury, reflecting the natural course of the disease.
Considering that alterations in liver pathology in PA-SOS patients may be influenced by confounding factors such as treatment methods and specific diseases, we constructed an animal model to simulate the natural evolution of MCT-induced SOS in rats without any intervention. A similar phenomenon was reported in a previous study.12 They reported typical pathological features of SOS, such as sinusoidal dilation and endothelial cell damage on Day 2 and obvious liver fibrosis at 1–2 weeks if there was no further PA exposure. Moreover, PSVD-like changes in MCT-induced rats at 2–4 weeks were also identified, showing histopathological parallels with those in established PSVD models.41 Although assessments of portal vein pressure and splenomegaly were not conducted in our study, experiments are currently underway in our laboratory to develop a more convenient and stable animal model that could comprehensively mimic key clinicopathological dimensions of PSVD.
However, several limitations of the present study need to be acknowledged, primarily attributable to its insufficient sample size and retrospective design. Given the relative rarity of PA-induced liver injury and the lack of access to pathological specimens due to the noninvasive criteria for SOS, the availability of samples was limited, especially matched pathology specimens, which may consequently decrease the statistical power. Notably, the observation of the above phenomenon in 14 consecutively collected cases highlights certain issues. In addition, compared with that of PSVD, the follow-up period was not long enough. Hence, a high-quality prospective study with a larger sample size and longer follow-up duration is needed to confirm our findings. Another noteworthy limitation is that most included patients had relatively severe PA-SOS and thus received TIPS intervention.
Given that TIPS stents markedly alter portal hemodynamics, this may interfere with the observation of natural portal hypertension progression. Accordingly, our findings are less generalizable to patients without TIPS intervention. Studies enrolling a larger sample of patients receiving anticoagulation therapy are needed, although we also performed further experimental validation.
Supporting information
Supplementary Table 1
Histological features of liver biopsies and definition during PAs-induced liver injury.
(DOCX)
Supplementary Table 2
Baseline characteristics of the 14 patients with PA-SOS.
(DOCX)
Supplementary Table 3
Comparison of serological indexes of the 14 patients with PA-SOS.
(DOCX)
Supplementary Table 4
Evaluation of imaging findings in the 14 patients with PAs-induced liver injury between the acute and recovery phase.
(DOCX)
Supplementary Table 5
Baseline characteristics of PA-SOS patients with and without liver biopsy in the recovery phase.
(DOCX)
Supplementary Table 6
Baseline characteristics of PA-SOS patients at disease initiation between the acute cases (cohort 2) and the matched controls (cohort 1).
(DOCX)
Supplementary Table 7
Detailed pathological characteristics of the 14 PA-HSOS patients in recovery phase.
(DOCX)
Supplementary Table 8
Incidence of typical histopathological features in individual rats at different points after MCT exposure.
(DOCX)
Supplementary Fig. 1
Comparison of clinical indicators at different points of PA-SOS patients. (A) PLT; (B) ALT; (C) AST; (D) TBIL; (E) ALB; (F) FIB; (G) Peak PVV; (H) the diameter of portal vein; (I) DTSS score; (J) Child-Pugh score.
ns, no significance, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. PLT, platelets; ALT, alanine aminotransferase; AST, aspartate aminotransferase; TB, total bilirubin; ALB, albumin; FIB, fibrinogen; PVV, portal vein velocity; DTSS, Drum Tower Severity Scoring.
(DOCX)
Supplementary Fig. 2
Ultrasound images of PA-SOS patients.
(A-B) A 67-year-old female patient who received anticoagulation at disease initiation (A) and 11 months (B); (C-D) A 73-year-old male patient who received TIPS at disease initiation (C) and 20 months (D). TIPS: transjugular intrahepatic portosystemic shunt.
(DOCX)
Supplementary Fig. 3
Representative CT images for PA-SOS patients.
(A-B) A 66-year-old male patient who received anticoagulation at disease initiation (A) and 18 months (B); (C-D) A 65-year-old female patient who received TIPS at disease initiation (C) and 40 months (D). The asterisks represent a map-like or mottle-like nonhomogeneous appearance in the equilibrium phase.
(DOCX)
Supplementary Fig. 4
Results of immunohistochemical staining for CD34.
Representative images depicting scores of (A) 0 (normal staining pattern with sparing of sinusoids), (B) 1+ (mildly aberrant staining involving minority of centrilobular sinusoids), (C) 2+ (moderately aberrant staining involving part of centrilobular sinusoids), and (D) 3+ (markedly aberrant staining involving majority of centrilobular sinusoids). Scale bars represent 25 µm.
(DOCX)
Supplementary Fig. 5
Quantitative analysis of CD34 immunohistochemical staining during the acute and recovery phases for PA-SOS patients.
(DOCX)
Supplementary Fig. 6
Representative histopathological images of patients at different points. Control means relative healthy patients with hepatic hemangioma.
<3 months in recovery phase means that the interval between the initial clinical recovery and the liver biopsy at recovery is less than 3 months. Scale bars are 50µm. CV, central vein; PV, portal vein. BD: bile duct. HA: hepatic interlobular artery.
(DOCX)
Supplementary Fig. 7
(A) Violin plot illustrating the distribution of rats with or without OPV at each time point (0=absent, 1=present); (B) Violin plot illustrating the distribution of mice with or without ISC at each time point (0=absent, 1=present); (C) Quantitative analysis of CD34 immunohistochemical staining in liver tissues at different points after MCT exposure; (D) Relative CD34 mRNA levels in liver tissues at different points after MCT exposure.
OPV: obliterative portal venopathy, ISC: incomplete septal cirrhosis.
(DOCX)
Supplementary Fig. 8
Protein expression of CD34 in liver tissues at different points after MCT exposure.
(DOCX)