Survival in Patients with Pyrrolizidine Alkaloids-Induced Hepatic Sinusoidal Obstruction Syndrome Versus Other Causes of Portal Hypertension in the Treatment of Transjugular Intrahepatic Portosystemic Shunt Placement: A Propensity Score Matching Study

Background/purpose: The etiology of portal hypertension may be an important factor affecting patient outcomes after transjugular intrahepatic portosystemic shunt (TIPS) treatment. Pyrrolizidine alkaloid-induced hepatic sinusoidal obstruction syndrome (PA-HSOS) is characterized by acute portal hypertension and liver injury. The aim of this study was to compare the post-TIPS survival between patients with PA-HSOS and those with other etiologies of hepatic portal hypertension. Methods: A retrospective le review of 787 patients with portal hypertension-related complications, including 146 patients with PA-HSOS, who had received TIPS treatment at our institution between August 2008 to September 2020 was performed. We recorded the demographic and clinical characteristics of patients before TIPS. The median follow-up was 20.27 months. Orthotopic liver transplantation (OLT)-free survival was compared between PA-HSOS and non-HSOS patients. Propensity score matching (PSM) was used to minimize bias. Results: Compared to patients with other etiologies of portal hypertension, PA-HSOS patients had a statistically signicantly higher Child-Pugh and model for end-stage liver disease score. No signicant differences were found in the orthotopic liver transplantation (OLT)-free survival between patients with PA-HSOS and patients with other causes of portal hypertension (all P > 0.05). After PSM, OLT-free survival was signicantly better in PA-HSOS patients than in patients with viral hepatitis (P = 0.0011), primary biliary cirrhosis or autoimmune hepatitis (P = 0.021), and schistosomiasis (P = 0.025). Conclusions: TIPS placement can be performed safely in patients with PA-HSOS. PA-HSOS patients had a better or similar prognosis after TIPS treatment than patients with portal hypertension of other etiologies. Trial registration: This study was retrospectively registered chictr.Org in described as numbers and frequencies and compared using Pearson’s Chi-square or Fisher’s exact tests. Survival rates were analyzed using the Kaplan–Meier model. Univariate analysis was used to assess a potential inuence of patient characteristics on the prognosis. Statistically signicant factors were introduced into a multivariate Cox proportional hazards analysis. All tests were two-sided, and a P-value < 0.05 was dened as statistically signicant. All missing data were imputed using the E-M model. Sex, age, the model for end-stage liver disease (MELD) score, CTP score, stent diameter, and pre-TIPS episodes of overt hepatic encephalopathy (OHE) were used as matching factors to adjust for baseline differences, and the selection was performed without replacement. Propensity score matching may become more dicult with more variables, but these variables may be the most critical ones with regard to outcomes. ALT, AST, Tbil, and creatinine serum levels, and the pre-TIPS PVV. In the multivariate analysis, the MELD score, AST, Tbil, and creatinine serum levels, and pre-TIPS PPV remained signicant risk factors (Online Resource 1). In patients with other portal hypertension etiologies, age, the MELD and CTP scores, severity of ascites, PT, INR, Tbil, albumin concentration, PLT, pre-TIPS overt hepatic encephalopathy (OHE), pre-TIPS portal vein thrombosis, pre-TIPS PVV, pre-TIPS PVP (portal vein pressure), and post-operative PVP were signicantly associated with OLT-free survival in the univariate analysis. The Cox proportional hazard model showed that age, CTP score, INR, Tbil, albumin, and pre-TIPS PVV were independent prognostic markers for OLT-free survival in patients with various etiologies of portal hypertension (Online Resource 2).


Introduction
The hepatic sinusoidal obstruction syndrome (HSOS), also known as hepatic veno-occlusive disease (HVOD), is a hepatic vascular disease characterized by swelling and necrosis of sinusoidal endothelial cells in zone 3 of the hepatic acini leading to sinusoidal congestion [1]. It is often associated with hematopoietic stem cell transplantation, the use of certain chemotherapeutic agents in non-transplant conditions (drug-induced liver injury), and immunosuppressive agents after liver transplantation in Western countries [2][3][4]. In China, the leading cause of HSOS is the intake of herbs containing pyrrolizidine alkaloids (PA) [5][6]. There are at least 6,000 plants worldwide that contain PAs, with the most common being Gynura segetum, in the genus Senecio of the family Compositae. Ingestion of Tusanqi contributes to about 50-86.6% of reported PA-HSOS in China [1].
Acute sinusoidal portal hypertension can occur in PA-HSOS because of the progressive narrowing and blocking of the sinusoidal and intrahepatic venous lumen and may further aggravate liver damage [7]. A transjugular intrahepatic portosystemic shunt (TIPS) is an important treatment of portal hypertension-related complications [8]. To date, anticoagulant and supportive treatment is the rst-line treatment for HSOS based on current consensus [1], but the response rate is only approximately 60% [9][10][11]. In patients who do not respond to medication, TIPS is recommended as a second-line salvage therapy, even though this remains controversial [1].
The literature currently provides no data comparing the long-term survival of PA-HSOS patients with that of patients with other etiologies of portal hypertension after TIPS treatment. Given the unique etiology and mechanism of sinusoidal portal hypertension in PA-HSOS patients and its high prevalence in China, We hypothsized that the different etiologies of portal hypertension with their underlying pathogenic mechanisms may result in different long-term survival after TIPS.. Therefore, the aim of this study was to compare the survival of patients with PA-HSOS with that of patients with portal hypertension caused by other diseases after TIPS treatment.

Material And Methods
Patients and data collection A retrospective review of the medical records of 1011 patients diagnosed with portal hypertension and successfully underwent TIPS at our center from August 2008 to September 2020 was performed. The exclusion criteria were as follows: 1) incomplete data; 2) severe, life-threatening disease of another system unrelated to portal hypertension and its complications; 3) previous TIPS or surgical shunting; 4) hepatocellular carcinoma before TIPS; and 5) follow-up time after TIPS therapy of fewer than six months. A total of 224 patients were excluded based on these criteria. Finally, 787 patients were enrolled. The patient ow chart is shown in Fig. 1.
The diagnosis of PA-HSOS was made according to the Nanjing criteria [1], which include a history of ingestion of a PA-containing plant and meeting of the following three criteria or pathological evidence that rules out other known causes of liver injury: abdominal distention and/or pain in the hepatic region, hepatomegaly, and ascites; elevation of total serum bilirubin or abnormal liver function testing; and evidence on enhanced computed tomography or magnetic resonance imaging, such as a map-like or mottle-like nonhomogeneous appearance in the venous phase and equilibrium phase and a halo sign around the portal vein. Typical pathological changes of HSOS include swelling, damage, and shedding of sinusoidal endothelial cells in the acinus zone 3 and signi cant dilation and congestion of the sinusoids.
The diagnostic criteria for other etiologies of portal hypertension were based on the following criteria [12][13][14][15][16][17][18]. A mixed etiology of portal hypertension was de ned as the combination of no less than two etiologies. The diagnostic criteria for portal hypertension were based on the Baveno VI consensus in all patients [19]. All methods were carried out in accordance with relevant guidelines and regulations.
TIPS procedure TIPS procedures were conducted by an experienced team using the standard techniques described previously [20][21]. Brie y, the Rups-100 device (Rösch-Uchida Transjugular Liver Access Set with Check-Flo® Introducer, Cook Medical Inc., Bloomington, IN, USA) was introduced via the right jugular vein into the inferior vena cava and hepatic vein. After successfully passing the portal vein, a composite of a covered stent (Fluency™; Bard Peripheral Vascular Inc., Tempe, AZ, USA) combined with a bare stent (E-Luminexx™; Bard Peripheral Vascular Inc., Tempe, AZ, USA) or a Viatorr® stent (W. L. Gore & Associates, Inc., Flagstaff, AZ, USA) was deployed. The stent diameters used were 6 mm, 7 mm, and 8 mm. In addition, any varicose collateral veins visible on angiography were embolized using tissue glue and/or coils during the TIPS procedure.

Statistical analyses
Quantitative variables with normal distribution were expressed as the mean and standard deviation and compared using the Student's t-test.
Variables that were not normally distributed were presented as the median and range, and the Mann-Whitney-Wilcoxon test was used for comparisons. Categorical variables were described as numbers and frequencies and compared using Pearson's Chi-square or Fisher's exact tests.
Survival rates were analyzed using the Kaplan-Meier model. Univariate analysis was used to assess a potential in uence of patient characteristics on the prognosis. Statistically signi cant factors were introduced into a multivariate Cox proportional hazards analysis. All tests were two-sided, and a P-value < 0.05 was de ned as statistically signi cant.
All missing data were imputed using the E-M model. Sex, age, the model for end-stage liver disease (MELD) score, CTP score, stent diameter, and pre-TIPS episodes of overt hepatic encephalopathy (OHE) were used as matching factors to adjust for baseline differences, and the selection was performed without replacement. Propensity score matching may become more di cult with more variables, but these variables may be the most critical ones with regard to outcomes.
The R software, version 3.5.1 (http://www.r-project.org) was used for all analyses within a range of 0.2 of the standard deviation.

Baseline characteristics of the study population
The baseline characteristics of all patients PA-HSOS and non-PA HSOS are summarized in Table 1. Nearly half of the study population consisted of hepatitis-B virus (HBV) patients (n = 366), followed by PA-HSOS patients (n = 146). Patients with a mixed etiology of liver cirrhosis were rare in our study (n = 14). In general, PA-HSOS patients had more severe liver injury as re ected by a signi cantly higher CTP score and MELD score than all other patients except for HBV patients. The proportion of patients with ascites among PA-HSOS patients was higher than that among patients with all other etiologies except for HBV. PA-HSOS patients had a markedly lower portal vein velocity (PVV) than patients with all other etiologies except for a mixed etiology. The reduction of the portal pressure gradient compared to the intraoperative values was signi cantly higher in PA-HSOS patients than in patients with all other causes of portal hypertension except for cryptogenic portal hypertension.    (Fig. 3e).

Discussion
Our data indicate that TIPS results in better or similar survival rates in patients with PA-HSOS as compared to those of patients with other etiologies of portal hypertension.
As compensated cirrhosis decompensates, a number of complications can occur, among which complications related to portal hypertension are a common cause of death [22,23].
option for lowering portal venous pressure [22]. We hypothsized that the different etiologies of portal hypertension with their underlying pathogenic mechanisms may result in different long-term survival after TIPS. While a few retrospective studies found no signi cant differences in the prognosis of patients with portal hypertension of different etiologies, none of these studies included HSOS patients [24][25][26].
In our study, the OLT-free survival rate for PA-HSOS patients was 86.5% during the follow-up period, which was comparable to the ndings of other studies (64.3-94.6%) [10,11,27]. After accounting for confounding factors using PSM, TIPS achieved a signi cantly better OLT-free survival as compared to that in patients with viral hepatitis, with primary biliary cirrhosis and autoimmune liver hepatitis patients, and with schistosomiasis infection. Careful patient selection remains a critical factor that determines patient outcomes after TIPS, and high CTP and MELD scores have been associated with OHE and high mortality after TIPS treatment [21,28].
In this study, PA-HSOS patients had higher CTP and MELD scores than patients with other etiologies of portal hypertension at baseline. However, remarkably, these patients had a better or similar survival than patients with different other etiologies of portal hypertension after TIPS treatment. This may have different explanations. First, PA-HSOS patients seldom present with gastroesophageal varices and splenomegaly because of the rapid progression of their portal hypertension. Therefore, these patients have a lower risk of dying from massive gastrointestinal bleeding both before and after TIPS treatment [11]. Second, early during PA-HSOS, liver damage is mostly mild to moderate, whereas the abrupt occurrence of severe portal hypertension without effective compensation is seen as the main cause of the occurrence of clinical symptoms and indicate a poor prognosis [11]. TIPS may bene t patients with PA-HSOS through the prompt decompression of the portal vein [22]. Finally, as an acute druginduced liver injury, liver damage in PA-HSOS patients usually does not progress once PA intake is stopped. This is re ected in the rapid decrease in blood pyrrole-protein adducts as biomarkers of PA-induced liver injury, and indicates that the liver damage may be repaired, which favors the longterm survival of patients treated with TIPS [29].
In our study, the MELD score, AST, Tbil, and creatinine serum levels, and pre-TIPS PVV were predictors of survival in PA-HSOS patients. Several studies have also demonstrated that serum bilirubin concentration is independently related to survival in PA-HSOS patients [2,10]. Furthermore, we identi ed the following six factors that were independently associated with survival in patients with other etiologies of portal hypertension: age, CTP score, INR, Tbil and albumin serum levels, and pre-TIPS PVV. Age above 70 years has been reported to be associated with reduced survival within 90 days following TIPS [30]. INR and Tbil and albumin levels are also important factors included in the CTP and MELD scores.
In total, 22 PA-HSOS patients died during the follow-up period. Of note, 11 patients died one month after TIPS because of sepsis (n = 1), heart failure (n = 1), multiple organ dysfunction syndrome (n = 5), or liver failure (n = 4). There were signi cant differences in the MELD score at baseline between patients who died within one month after TIPS and patients who died at least 1 month after TIPS (22.18 ± 7.51 vs. 15.36 ± 6.53, P = 0.034). Therefore, we recommend paying particular attention to the monitoring of PA-HSOS patients during the rst month postoperatively if we are to improve survival rates in these patients.
Our results should be interpreted within the limitations of the study. First, this was a retrospective single-center study. However, this facilitates quality control of the TIPS procedure because all procedures were performed by the same experienced group. Second, the heterogeneity of baseline characteristics among the patients with different etiologies of portal hypertension might have resulted in bias. Although we accounted for confounding factors with PSM, several factors remained incomparable. Third, we did not evaluate the shunt dysfunction rate and post-intervention encephalopathy rate for these parameters had not been recorded in the patients' medical les. Finally, PSM signi cantly reduced the sample size of the analyzed cohort. Therefore, we combined HBV with HCV patients, primary biliary cirrhosis patients with autoimmune hepatitis patients, and patients with mixed etiologies of portal hypertension with cryptogenic portal hypertension patients in three new subgroups.
In summary, TIPS placement can be performed safely in patients with PA-HSOS to treat portal hypertension. The OLT-free survival of PA-HSOS patients after TIPS is better or similar to that of patients with various other causes of portal hypertension. Large, randomized, multicenter, prospective studies are needed to verify our ndings.

Declarations
Ethics approval and consent to participate This study was approved by the ethics committee of the Nanjing Drum Tower Hospital.
The ethics committee of the Nanjing Drum Tower Hospital waived the need for consent because of the retrospective and anonymized nature of the study. All methods were carried out in accordance with relevant guidelines and regulations. All procedures followed were in accordance with the ethical standards of the responsible committee on human experimentation (institutional and national) and with the Helsinki Declaration of 1964, as revised in 2013.