Introduction
Hepatic sinusoidal obstruction syndrome (HSOS) is a specific form of drug-induced liver injury (DILI) that mainly targets endothelial cells in the hepatic sinusoids, hepatic venules, and interlobular veins [1]. In most situations, liver injury and acute portal hypertension (PH) are typical clinical features of HSOS. In Western countries, HSOS is often caused by certain chemotherapeutic agents during myeloablative pretreatment prior to hematopoietic stem cell transplantation (HSCT). However, in China, HSOS induced by HSCT is rarely reported; furthermore, HSOS is frequently associated with the ingestion of plants containing pyrrolizidine alkaloids (PA) [2], especially ‘Tusanqi’, which accounts for 50.0-88.6% of all cases. This form of HSOS is also referred to as pyrrolizidine alkaloid-induced HSOS (PA-HSOS) [3-5].
Hepatic sinusoidal obstruction syndrome induced by pyrrolizidine alkaloids can present as mild, moderate or severe cases, although there is a clear lack of reliable prognostic indicators to predict the severity of these patients. Several previous studies have explored risk factors related to the prognosis of PA-HSOS. The hepatic venous pressure gradient (HVPG) remains the gold standard for the diagnosis of portal hypertension [6, 7]. Previous studies have demonstrated that an HVPG > 10 mmHg has a sensitivity of 52% and a specificity of 91% for the diagnosis of HSOS [8]. In a previous study, Xiao et al. reported that a prolonged prothrombin time at baseline and increased serum total bilirubin levels five days after a transjugular intrahepatic portosystemic shunt (TIPS) are independent risk factors for predicting death after TIPS treatment in PA-HSOS patients [9]. A severity system, referred to as the Drum Tower Severity Scoring (DTSS) system, was developed to evaluate the severity of PA-HSOS and select appropriate treatment. The DTSS system considers aspartate aminotransferase, total bilirubin, fibrinogen, and peak portal vein velocity and can predict the outcomes of patients with PA-HSOS, at least to some extent [4]. However, there is a clear need for more simplistic prediction factors to predict severity.
The serum-ascites albumin gradient (SAAG) is defined as the difference between serum albumin and ascites albumin when measured on the same day. Previous research has shown that SAAG is an appropriate discriminator of PH-related ascites [10]. The value of SAAG is ≥ 11 g/l in the presence of PH while a value < 11 g/l can rule out PH [11]. In addition, SAAG has been proposed to represent a factor that can determine the prognosis of patients with cirrhosis. Furthermore, some previous studies have investigated the relationship between SAAG and HVPG, although it has never been investigated in patients with PA-HSOS.
In the present study, we aimed to investigate the association between SAAG and HVPG, portal pressure gradient (PPG), and portal pressure (PP) in patients with PA-HSOS and to explore the clinical value of SAAG to predict the severity of PA-HSOS.
Material and methods
Patients
This was a retrospective study that was performed in a single center and collated clinical data from patients with PA-HSOS undergoing PP measurements and abdominocentesis at Nanjing Drum Tower Hospital between 2015 and December 2021. The severity of these cases was defined by assessment of the DTSS system: mild (4-6 points), moderate (7-10 points), and severe (11-16 points) [4].
For each patient, we recorded the intraoperative wedge hepatic vein pressure (WHVP), free hepatic venous pressure (FHVP), inferior vena cava pressure (IVCP), and other relevant data. HVPG was calculated and referred to IVCP. The baseline portal vein pressure (PP) was measured in all patients undergoing TIPS, and the PPG was calculated. Physicians performing HVPG and TIPS were blinded to the other technique. The inclusion criteria were as follows: 1) 18-75 years of age; 2) diagnosed with PA-HSOS based on the Nanjing criteria for PA-HSOS. The specific contents of the Nanjing criteria include a confirmed history of PA-containing plant ingestion together with the following 3 requirements: i) abdominal distention and/or pain in the hepatic region, hepatomegaly and ascites; ii) increased serum total bilirubin or abnormal liver function testing; iii) enhanced computed tomography (CT) or magnetic resonance imaging (MRI) features, as shown in Figure 1, or pathological evidence that rules out other known causes of liver injury (the typical pathology features of PA-HSOS include notable dilation and congestion of the hepatic sinusoids, in addition to inflammation, injury, and discharge of hepatic sinusoid endothelial cells in hepatic acinus zone III) [1]; 3) HVPG measurements available; 4) availability of ascites and biochemical data for ascites; and 5) provision of informed consent. The exclusion criteria were as follows: 1) previous history of liver transplantation; 2) severe portal vein thrombosis and portal vein cavernous lesions; and the 3) existence of malignant tumors or other serious diseases. The study complied with the Declaration of Helsinki.
Fig. 1
Computed tomography (CT) images of a patient diagnosed with pyrrolizidine alkaloid-induced hepatic sinusoidal obstruction syndrome (PA-HSOS). A) Upper abdominal CT plain scanning showed hepatomegaly, uniform decrease in liver parenchymal density, and ascites. B) Upper abdominal CT plain scanning showed full liver morphology, uneven decrease in liver parenchymal density, and ascites. C) Upper abdominal CT scanning showed uneven distribution of patchy low-density shadows in the liver, splenomegaly, and fluid accumulation around the liver, spleen, and ascites. D) Upper abdominal CT scanning showed uneven enhancement of the liver in a map-like and spotted manner, enhancement of the gastric mucosa, and a large amount of ascites
Intervention procedure
Hepatic venous pressure gradient measurement and the TIPS procedure were performed under local anesthesia. The external zero reference point was set at the patient’s midaxillary line. A RupS-100 access tool (COOK, Bloomington, IN, USA) was inserted into the inferior vena cava through the right internal jugular vein and a 5.5-7 F balloon catheter (Edwards Lifesciences, Irvine, CA, USA) was inserted into the middle or right hepatic vein. WHVP, FHVP, and IVCP measurements were repeated three times. Then, the floating catheter was removed and IVCP was measured close to the entrance of the hepatic vein and inferior vena cava. Once pressure measurement had been completed, we injected a contrast agent to confirm whether the obstruction was complete and whether there was communication between the hepatic vein-to-vein. Once HVPG measurement was complete, we performed portal vein puncture. A pigtail catheter was placed in the main trunk of the portal vein and the PP was measured [12, 13]. HVPG and PPG were calculated as shown by the formulae below:
The surgical procedure was carried out in strict accordance with the manufacturer’s protocol.
Statistical analysis
All statistical analyses were conducted using IBM SPSS Statistics version 22.0 (IBM, Armonk, NY, USA). The Shapiro-Wilk test and normal distribution plots were used to determine whether the data conformed to a normal distribution. Quantitative variables conforming to a normal distribution are given as the mean (SD) and median (IQR) and compared by the independent sample t-test or Mann-Whitney U test. Qualitative variables are presented as numbers (%) and were compared by the χ2 test or Fisher’s exact test. Correlations between SAAG and HVPG, PPG, or PP were evaluated by scatter plots, and by calculating Pearson’s correlation coefficient (R), intraclass correlation coefficient (ICC), and confidence interval (CI). In order to stratify patients by ‘high’ and ‘low’ levels of SAAG and to observe the differences between patients with ‘high’ and ‘low’ levels of SAAG, we calculated a Youden’s optimized cut-off for ROC curves, which was based on SAAG-guided treatment response stratification. ROC curves showed that the optimal cut-off value for predicting treatment non-response using SAAG at the maximum Youden index was 16 g/l. A comparison of SAAG levels between patients of different severity was conducted. Double-tailed p-values < 0.05 were considered statistically significant.
Results
Patient characteristics
According to the inclusion and exclusion criteria, 60 patients with PA-HSOS who underwent HVPG were initially included. However, we excluded 27 patients due to abdominocentesis after treatment, and five patients due to incomplete data. Consequently, our final analysis included 28 patients. Technical success was achieved in all patients, and no complications related to the procedure were recorded. The characteristics of the patients are summarized in Table 1. Mean patient age was 66 years, with a range of 47 to 79 years. Sixteen (57.1%) patients were male and 12 (42.9%) were female. All of the patients had ascites, and 27 (96.4%) patients had moderate to severe ascites. The median CTP score was 9 points (IQR: 8-10) and 10 patients (35.7%) had CTP C. Using a Youden’s optimized index, the patients were divided into two groups based on an SAAG cut-off of 16 g/l. Ten patients (35.7%) had an SAAG < 16 g/l, while 18 (64.3%) had an SAAG > 16 g/l.
Table 1
Characteristics of patients with hepatic sinusoidal obstruction syndrome induced by pyrrolizidine alkaloids (PA-HSOS)
[i] SAAG – serum albumin ascites gradient, HVPG – hepatic venous pressure gradient, WHVP – wedge hepatic vein pressure, FHVP – free hepatic vein pressure, IVCP – inferior vena cava pressure, WBC – white blood cells, HB – hemoglobin, PLT – platelets, PT – prothrombin time, INR – international normalized ratio, ALT – alanine aminotransferase, AST – aspartate aminotransferase, TBil – total bilirubin, ALB – albumin, Scr – serum creatinine, FIB – fibrinogen, DTSS – Drum Tower Severity S coring, CTP – Child-Turcotte-Pugh
Correlation between SAAG and HVPG
The mean SAAG for the two patient groups was 14.12 ±1.41 g/l and 18.78 ±1.62 g/l, respectively; these values were significantly different (p < 0.001). The mean HVPG in the SAAG > 16 g/l group (23.54 ±3.09 mmHg) was significantly higher than that in the SAAG < 16 g/l group (20.87 ±2.89 mmHg) (p = 0.034). Figure 2 depicts the correlations between SAAG, HVPG, and WHVP. The correlation between SAAG and HVPG was moderate in patients with PA-HSOS (R = 0.41, ICC = 0.40, 95% CI: 0.02-0.74, p = 0.031). However, the correlation between SAAG and WHVP was poor (R = 0.36, ICC = 0.34, 95% CI: 0.05-0.63, p = 0.035).
Fig. 2
Correlation between serum albumin ascites gradient (SAAG) and hepatic venous pressure gradient (HVPG), as demonstrated by a scatter plot. A) In patients with PA-HSOS (n = 28), the strength of the linear correlation between SAAG and HVPG was moderate (R = 0.41, ICC = 0.40). B) In patients with PA-HSOS (n = 28), the strength of the linear correlation between SAAG and wedge hepatic vein pressure (WHVP) was poor (R = 0.36, ICC = 0.34)
Correlation between SAAG and PPG
Of the entire cohort, 20 (71.4%) patients underwent the TIPS procedure and PP was measured prior to shunt insertion, with a technical success rate of 100%. PPG was calculated for these patients. The mean PP in these patients was 17.65 ±4.03 mmHg, and the mean PPG was 11.15 ±4.42 mmHg. The correlation between SAAG and PPG was moderate (R = 0.49, ICC = 0.42, 95% CI: 0.12-0.78, p = 0.030). The correlation between SAAG and PP was good (R = 0.57, ICC = 0.52, 95% CI: 0.26-0.78, p = 0.008) (Fig. 3).
Fig. 3
Correlation between serum albumin ascites gradient (SAAG) and portal pressure gradient (PPG), as demonstrated by a scatter plot. A) In patients with PA-HSOS (n = 28), the strength of the linear correlation between SAAG and PPG was moderate (R = 0.49, ICC = 0.42). B) The strength of the linear correlation between SAAG and portal pressure (PP) was good (R = 0.57, ICC = 0.52)
SAAG level and DTSS severity
The DTSS system was used to evaluate the severity of disease in each patient; the DTSS score ranges from 4 to 16 points. The DTSS score of our patients ranged from 5 to 14 points. Seven (25%) cases were mild (DTSS score: 4-6), while twenty-one (75%) were moderate (DTSS score: 7-10) or severe (DTSS score: 11-16). Seventeen cases were moderate and the remaining four were severe. Next, we compared the mean SAAG level between the two groups. The mean SAAG in the mild patients was 15.31 (IQR: 12.70-19.10) g/l, which was significantly lower than the mean SAAG of 17.72 (IQR: 15.95-19.00) g/l in the moderate and severe patient groups (p = 0.042). The distribution of SAAG level across differing severities of PA-HSOS is shown in Figure 4.
Discussion
In this study, we aimed to identify the correlations between SAAG and HVPG, PPG, or PP in patients with PA-HSOS; these associations have not been reported previously. There is a clear lack of a simple and straightforward factor with which to predict the severity of PA-HSOS. In the present study, we investigated the utility of SAAG for predicting the severity of PA-HSOS and predicting prognosis. Our findings demonstrated that SAAG correlated well with PP in PA-HSOS patients, while the correlation between SAAG and HVPG was only moderate. Therefore, SAAG could serve as a marker for portal hypertension in patients with PA-HSOS. This is the first study to investigate the association between SAAG and HVPG or PPG in patients with PA-HSOS.
The direct measurement of PP is an invasive method and cannot be easily tolerated by patients [14-17]. Therefore, researchers have attempted to develop a simple, non-invasive, and easy-to-implement method that reflect PP in an accurate manner. In this study, the agreement between SAAG and PPG or PP in PA-HSOS patients was good, although SAAG was not correlated with HVPG or WHVP. According to some previous studies, HVPG cannot accurately reflect the degree of portal hypertension and needs to be evaluated in combination with other examinations [18]. Cheng et al. reported that HVPG and PPG were not significantly correlated in patients with PA-HSOS [8], while another study demonstrated that the HVPG of most patients with portal hypertension did not accurately represent PPG, and that the former was lower than the latter, implying that HVPG was underestimated [19, 20]. Formation of the hepatic vein collateral branches is one of the most important factors responsible for the underestimation of HVPG [18]. Therefore, in the presence of collateral branches, HVPG is not representative of PP. Furthermore, the measurement of HVPG is an invasive procedure that requires local anesthesia and cannot be easily repeated [21, 22]. A number of factors may affect the accuracy of HVPG measurements [22]. Our study identified that the correlations between SAAG and PPG or PP were better, thus providing a theoretical basis for SAAG reflecting PP in patients with PA-HSOS. Due to the common presence of ascites in patients with PA-HSOS, the acquisition of SAAG data is very convenient.
The severity criteria for HSOS proposed by the European Blood and Bone Marrow Transplantation (EBMT) Collaboration Group were previously used to evaluate the severity of HSOS [23]. However, the criteria proposed by the EBMT are associated with certain drawbacks and were inaccurate for patients with PA-HSOS. Acute portal hypertension is a significant feature of HSOS, although the criteria failed to consider hemodynamic parameters for the portal vein. In 2022, our team developed the DTSS system to determine the severity of PA-HSOS [4]. Using the DTSS system, the severity of cases of HSOS can be measured accurately, and can allow the development of personalized treatment plans. However, in order to better estimate the severity of PA-HSOS, further tests of the specific correlations between DTSS and the clinical manifestations of PA-HSOS or its biomarkers are necessary. The DTSS system was specifically developed to determine the severity of PA-HSOS [4]. Although the DTSS system performed impressively in internal validation, more evidence is needed to demonstrate its reliability and correlation with portal hypertension in patients with PA-HSOS. In a previous study, Li et al. found that HVPG has low sensitivity when determining the clinical severity of PA-HSOS, and it is not recommended to use HVPG alone to evaluate the specific conditions of PA-HSOS patients [18]. In this study, the SAAG of patients with moderate and severe PA-HSOS was significantly higher than that of patients with mild PA-HSOS. The SAAG of patients with moderate and severe PA-HSOS was generally > 16 g/l, while the SAAG of patients with severe PA-HSOS was generally < 16 g/l, thus indicating the relationship between SAAG in PA-HSOS patients and the severity of their condition. Our data suggest that SAAG level increases significantly with the severity of disease. Patients with an SAAG > 16 g/l may be more likely to have moderate or severe conditions and require timely treatment. Previous studies demonstrated that SAAG is correlated with PP [24-26], and that acute portal hypertension is one of the typical clinical symptoms of PA-HSOS [1]. Therefore, our results provide further confirmation of the reliability of the DTSS to determine the severity of PA-HSOS. Our results support the notion that the DTSS system can precisely reflect the severity of PA-HSOS. Furthermore, our data confirm that the DTSS score can reliably evaluate hemodynamic condition in patients with PA-HSOS [4]. Moreover, SAAG can be obtained by simply puncturing a patient’s ascites, which is a minimally invasive method. This method is simple to perform and can be easily tolerated by patients. When necessary, SAAG measurements are also more convenient to repeat.
There are some limitations to this study that need to be considered. First, this was a retrospective study performed in a single center; furthermore, the sample size was small. Our findings need to be validated in a larger prospective cohort. Moreover, the impact of the time interval between HVPG measurements and SAAG measurements cannot be ignored. In addition, some patients may have already taken medicine to reduce portal vein pressure before hospitalization; the specific impact of these medications on pressure measurement remains unclear, and further research is needed to investigate these effects. Finally, the accuracy and stability of SAAG need to be further validated [27, 28]. Further prospective validation in a large cohort is necessary.
Conclusions
In this study, we detected a satisfactory correlation between SAAG and PPG in patients with PA-HSOS, although the correlation between SAAG and HVPG was poorer. Furthermore, we observed significant differences in the SAAG level between patients with different disease severities. Our findings demonstrate that SAAG is clinically valuable, as it can predict the severity of PA-HSOS, which may be helpful for predicting patient prognosis and selecting individualized treatment. Compared with HVPG, SAAG exhibits a stronger correlation with PPG and is simple, repeatable, and minimally invasive. Thus, SAAG represents a valid alternative to HVPG for certain patients with PA-HSOS. The potential capability of SAAG to predict disease severity and prognosis in patients with PA-HSOS should now be validated in prospective studies.
