Analysis of clinical features and prognostic factors in patients with hepatic hydrothorax: a single-center study from China

We conducted a retrospective cohort analysis of 5698 patients diagnosed with decompensated cirrhosis from January 2013 to June 2021 at the Department of Hepatology, First Hospital of Jilin University, China. Follow-up data were collected until December 30, 2021. All relevant clinical and laboratory data at the time of first diagnostic admission, including a complete medical history, were collected. Decompensated cirrhosis and decompensating events were defined according to the latest EASL Clinical Practice Guidelines on Decompensated Cirrhosis [7]. The inclusion criteria were as follows: (1) known diagnosis of cirrhosis, either biopsy confirmed or based on clinical complications present in the clinic consistent with cirrhosis; (2) known diagnosis of pleural effusion on chest radiography or lung computed tomography (CT); (3) pleural effusion consistent with known features of hepatic pleural fluid but not considered consistent with the presence of infection, malignancy or other known chronic disease; (4) no history of primary cardiopulmonary dysfunction, including but not limited to congestive heart failure; and (5) portal hypertension as determined by esophageal varices, portal hypertensive gastropathy, ascites, portal vein thrombosis or elevated hepatic venous pressure gradient (HVPG). The criteria for the exclusion of subjects were as follows: (1) combined malignancy or major organ failure; (2) patients on hormonal or immunosuppressive agents; and (3) incomplete clinical data or information on additional tests. All patients were screened for ascites and pleural effusion using ultrasound on admission. In addition, all patients underwent standard chest radiographs or CT of the lungs to detect underlying lung disease (pneumonia, tumors, and other lesions). Patients with clinical, laboratory or electrocardiographic suspicion of heart failure underwent cardiac ultrasound to rule out decompensated heart disease. Finally, 131 adult HH patients were included in this analysis (Fig. 1). The study protocol was approved by the Ethics Committee of the First Hospital of Jilin University and was conducted according to the principles of the Declaration of Helsinki. Written general informed consent was obtained from all participants.

All demographic and clinical information was confirmed directly from the electronic medical record, including demographic data (age and sex), serum biochemical information (alanine aminotransferase (ALT), aspartate aminotransferase (AST), alkaline phosphatase (ALP), gamma-glutamyl transpeptidase (GGT), total bilirubin (TBIL), albumin (ALB), creatinine (Cr), serum sodium, international normalized ratio (INR), lymphocyte absolute count (LY), neutrophil absolute count (NE), neutrophil-to-lymphocyte ratio (NLR), platelet count (PLT), prothrombin time (PT), and imaging features. The diagnosis of cirrhosis was confirmed by liver biopsy or ultrasound without regard to the presence of portal hypertension. The Child–Pugh score was calculated from the original article published by Pugh et al. in 1973 [8]. The MELD-Na score was calculated from Biggins et al. [9]. The ALBI scoring and grading system was implemented as described by Johnson et al. [10]. The Child–Pugh score is based on five indicators: serum ALB, TBIL, PT, ascites and hepatic encephalopathy, with a score of 5–6 for grade A, 7–9 for grade B and 10–15 for grade C. The criteria for judging the amount of pleural effusion are as follows: If the amount of pleural fluid is greater than 6 cm or greater than the 7th rib space by ultrasound or chest radiograph, it is a large amount; If the fluid level is 3–6 cm or greater than the 8th rib space, it is a medium amount; If the fluid level is less than 3 cm or the angle of the rib diaphragm is blunt, it is a small amount. Overt encephalopathy was defined as grade 2 to 4 liver encephalopathy according to the West Haven criteria [11]. Ascites was classified according to the most recent position paper published by the International Ascites Club [12].

All blood samples were tested in the Clinical Laboratory of Jilin University’s First Hospital. An automated biochemical analyzer (7600–210, Hitachi, Japan) was used to detect blood biochemical indices. According to the manufacturer’s instructions, the total blood count was determined using a SYSMEX XN-9000 hematological analyzer (Sysmex Corporation, Kobe, Japan). Clotting tests were carried out with the automated coagulometer “SYSMEX CS-5100” utilizing the clotting technique (Sysmex Corporation, Kobe, Japan).

We collected details of HH treatment, date of death, LT, and final follow-up. The primary outcome was liver-related death [death from liver failure, or hepatocellular carcinoma (HCC)] or LT. Our primary time point analysis identified the relationship between factors that may have affected prognosis and death, which was calculated from the date of first diagnosis of HH to the date of death.

Categorical data were expressed as numbers (n) and proportions (%). Continuous variables were expressed as the means (± standard deviation) when data were normally distributed and medians (quartile 25-quartile 75) when data were not normally distributed. The Shapiro–Wilk (W test) prevailed for the normality test. The proportional hazard assumption was assessed using the Schoenfeld residuals for continuous variables and using the graphical method for categorical variables. The variables that seemed to satisfy the proportional hazard assumption were incorporated into the univariate Cox proportional hazard model to assess the effect of different variables on survival time. The contribution of each variable was estimated by the risk ratio and its 95% confidence interval. All variables with a significant effect on survival (P < 0.1) were included in the multivariate Cox proportional hazard model. In the multivariate Cox regression model, the variables that might be considered were further removed step by step, and the change of hazard ratio (HR) value of the target variable (the first variable) was observed after removing a variable. If the change in HR value after removal was more than 10%, the variable was ultimately retained. If not, the variable was removed. For categorical variables, the survival outcomes were compared with both methods (the Cox regression models and the Kaplan–Meier method of log-rank tests), and they used the Kaplan–Meier estimates to display survival curves. The results of the Cox proportional hazards regression analysis are presented as hazard ratios (HRs) and their corresponding 95% confidence intervals (CIs). Our analysis for identifying factors associated with mortality/survival is an exploratory analysis. All statistical analyses were performed using SPSS Statistics 26 (IBM, New York, NY, USA) and R version 4.1.3. A two-sided P value < 0.05 was considered statistically significant.

During the study period, 5698 patients with decompensated cirrhosis were screened, and according to the inclusion and exclusion criteria, follow-up data were obtained for 131 patients who met the diagnostic criteria for hepatic hydrothorax. There were more males than females, with a male to female ratio of 1.67:1 in HH patients, with 80 men (61.1%) and 51 women (38.9%). The minimum age was 25 years and the maximum age was 81 years, with a mean age of 52.76 ± 11.88 years and a median age of 58 years. At the time of primary treatment, post-hepatitis B cirrhosis was the most common cause in 55 cases (42%), followed by alcoholic cirrhosis in 29 cases (22.1%), post-hepatitis C cirrhosis in 15 cases (11.5%), primary biliary cirrhosis in 10 cases (7.6%), and other etiologies in 22 cases (16.8%) (see Table 1).

At the time of first diagnosis, most patients had multiple complaints, with the most common symptoms being abdominal distention (82.4%), difficulty breathing (67.9%), poor appetite (41.2%), and cough (22.9%). Splenomegaly (84%) was the most common sign. Only pleural effusion on the right side was found in more than three quarters of the patients (107 out of 131; 81.7%). Left-sided effusion alone was found in 16 cases (12.2%), and bilateral effusion was found in 8 cases (6.1%). According to radiological criteria, massive pleural fluid accumulation predominated in 108 cases (82.4%), followed by moderate amounts in 19 cases (14.5%) and small amounts in 4 cases (3.1%). All patients had ascites, including 87 cases (66.4%) with massive pleural effusion combined with massive ascites. Approximately 96 cases (73.3%) presented with manifestations of compressive pulmonary atelectasis. Ascites was present in varying amounts in all patients and was the most common complication of the decompensated phase. This was followed by hyponatremia in 52 cases (39.7%), pleural fluid infection identified by pleural fluid bacterial culture in 30 cases (22.9%), renal insufficiency in 24 cases (18.3%), liver failure in 21 cases (16%), peritonitis in 18 cases (13.7%), hepatic encephalopathy and gastrointestinal bleeding in 16 cases each (12.2%) and hepatorenal syndrome in 9 cases (6.9%) (see Table 1).

Routine blood collection and prognostic score data are shown in Table 2. The laboratory data were noteworthy in that most patients did not show neutrophilia, and thrombocytopenia was typical. Serum cholinesterase and serum albumin levels were generally low. Almost all patients had significant liver dysfunction, including elevated bilirubin levels, prothrombin time (PT), thromboplastin time (TT), activated partial thromboplastin time (APTT), and international normalized ratio (INR), and significantly lower than normal prothrombin activity (PTA) (58.86 [55.06–62.65] %).

There were 131 patients at first diagnosis, of whom 20 (15%) were treated with drug management alone, with a mean age of 53.35 ± 11.91 years, a median MELD score of 10.14 (0.47–24.76) and an estimated median time of 19 months. A total of 111 (75%) patients were treated with medication and thoracic intubation, with a mean age of 57.32 ± 11.83 years, a median MELD score of 10.22 (0.0-30.50), and an estimated median time of 22 months. During the later stages of treatment, one patient was treated with TIPS, with a MELD score of 16 and a survival time of 32 months, and three patients were treated with liver transplantation, with a mean MELD score of 18, all of whom are currently alive (see Table 3). For overall survival without transplantation, the median follow-up time was 45 (interquartile range [IQR]: 25–96) months, the estimated median survival time was 21 (95% CI:18–25) months, and the survival rates at 6 months, 1 year, and 2 years were 77.2%, 62.4%, and 29.7%, respectively (Fig. 2).

We used both univariate and multivariate Cox regression analyses to identify the indicators related to liver-related death. In the univariate Cox regression analysis, the factors associated with liver-related death were male sex (HR: 1.772, 95% CI: 1.171–2.682, P = 0.007), combined hepatic encephalopathy (HR: 2.361, 95% CI: 1.354–4.116, P = 0.002), combined liver failure (HR: 1.820, 95% CI: 1.040–3.185, P = 0.036), combined hyponatremia (HR: 1.721, 95% CI: 1.156–2.562, P = 0.007), and higher TBIL level (HR: 1.004, 95% CI: 1.000- 1.007, P = 0.045) at baseline. In the multivariate Cox regression model, the possible variables considered were liver failure, hyponatremia, TBIL, and PT, and these four variables were further removed to see the change in HR value of the target variable (male) after the removal of a variable. After controlling for covariates that were associated with liver-related mortality in the univariate analysis, male sex (HR: 1.721, 95% CI: 1.114–2.658, P = 0.005), and combined hepatic encephalopathy (HR: 2.016, 95% CI: 1.101–3.693, P = 0.001) were found to be associated with an increase in liver-related mortality (see Table 4).

As depicted in Fig. 3, we compared the transplant-free survival of patients by sex using the Kaplan–Meier method. We found that the probability of transplant-free survival was significantly lower in female patients than in male patients (log-rank P = 0.0048). The estimated median survival time was 18 (95% CI:13–22) months for the male group and 26 (95% CI:20–29) months for the female group. As depicted in Fig. 4, we compared transplant-free survival in patients with or without combined hepatic encephalopathy using the Kaplan–Meier method. We found that patients with hepatic encephalopathy had a significantly lower probability of transplant-free survival than patients without hepatic encephalopathy (log-rank P = 0.0013). The estimated median survival times for patients with and without hepatic encephalopathy were 10 (95% CI:3–15) months and 23 (95% CI:19–26) months, respectively.