Predictive model for postoperative pleural effusion after hepatectomy

Abstract Aim Severe postoperative pleural effusion (sPOPE) after hepatectomy can lead to respiratory distress and may require thoracic drainage, leading to prolonged hospitalization. Preventive chest tube insertion may be useful for patients at high risk for sPOPE. We aimed to develop a predictive model for sPOPE after hepatectomy and evaluate indications for preventive chest tube insertion using our model. Methods We evaluated all patients who underwent hepatectomy from 2013 to 2020. Risk factors for sPOPE were used to develop a predictive model for sPOPE, which was validated in a cohort that received preventative chest tube placement postoperatively. Results A total of 325 patients were analyzed. Thirty‐one (9.5%) patients had a preventive chest tube placed at the end of their operation. Twenty‐one patients out of the remaining 294 patients developed sPOPE. Multivariate analysis identified resection containing segment 8 [relative risk (RR) 3.24, P = .022], intraoperative bleeding ≥ 500 g (RR 4.02, P = .008), intraoperative diaphragmatic incision (RR 6.96, P = .042) and open hepatectomy (RR 7.51, P = .016) as independently associated with sPOPE. The estimated probability of sPOPE ranged from 0.4% in patients with none of these factors to 73.4% in the presence of all factors. Among the 31 patients who received a preventive chest tube, more patients in the high‐risk group defined by the model had postoperative pleural effusions compared to the low‐risk group (P = .012). Conclusion Our predictive model for sPOPE using four risk factors allows for reliable prediction and may be useful for selection of preventive chest tube in patients undergoing hepatectomy.


| INTRODUC TI ON
Although hepatectomy has become a relatively safe operative procedure, the emergence of pleural effusion afterwards is one of the commonly observed complications, occurring in 18%-71% of cases. 1-4. The clinical presentation of postoperative pleural effusion (POPE) is wide and ranges from asymptomatic to respiratory distress with dyspnea or respiratory failure requiring reintubation and mechanical ventilation. [5][6][7] There are several studies pertaining to the risk factors of posthepatectomy pleural effusion. 2,[8][9][10] However, most of these studies analyzed not only severe pleural effusion but also mild pleural effusion managed with noninvasive therapy, such as fluid or salt restriction, and diuretic agent administration. Controllable, or mild, effusion is not so problematic in clinical practice. However, symptomatic pleural effusion requiring thoracic drainage accounts for 9.4% of cases. 1 Such severe POPE (sPOPE) influences patient outcomes by delaying postoperative recovery and leading to increased hospital stay and higher associated costs. 2,10 Therefore, analysis of risk factors for sPOPE is of strategic significance for postoperative management. However, there are currently little data available regarding risk factors of sPOPE and its management, making management and prevention a challenge for clinicians.
Intraoperative chest tube insertion and continuous drainage of the right thoracic cavity may be effective to prevent sPOPE for high-risk patients as nearly all sPOPE post-hepatectomy develops within the right thoracic cavity. 8 Prevention of pleural effusions post-hepatectomy may allow for improved pulmonary function and less subjective patient dyspnea, thus promoting postoperative rehabilitation and better postoperative course. However, no reports or guidelines currently exist on the role of preventative chest tube placement for hepatectomy patients or which patients may benefit from this intervention. Our study, therefore, aims to identify independent predictors of sPOPE and establish a model to predict sPOPE based on pre-or intraoperative valuables. We then aim to evaluate candidates for preventive chest tube placement using our model.

| Surgical procedure
Of the patients, 204 underwent laparotomies and 121 underwent laparoscopic hepatic resections. Tumor location, size, and its relationship to the portal and hepatic veins were examined using intraoperative ultrasound, regardless of whether laparotomy or laparoscopic methods were used. Hepatic parenchymal transection was performed with a Cavitron Ultrasonic Surgical Aspirator (Olympus, Tokyo, Japan) or clamp crush technique. Twenty-two patients required diaphragmatic incision intraoperatively due to large right-sided tumors, tumor invasion to the diaphragm, or adhesiolysis between the diaphragm and liver. 31 patients (9.5%) had a chest tube

| Pleural effusion
Postoperative pleural effusion was evaluated by chest X-ray on postoperative days 1 and 3, or computed tomography (CT) at postoperative day 5, for all patients. Once pleural effusion was detected by imaging, clinical evaluation including pulmonary auscultation and oxygen saturation monitoring were carried out. Patients showing signs of dyspnea or respiratory distress received diuretics. sPOPE was defined as accumulation of moderate to large pleural effusion with symptoms regardless of diuretic agent administration and was managed with ultrasound-guided chest tube insertion with continuous drainage using a MERA continuous suction unit (Senko Medical Instrument). Daily chest tube output was monitored and the chest tube was removed when output was less than 100 mL/day. The same criteria for chest tube removal applied to patients who underwent preventive chest tube insertion at the end of their operation.

| Statistical analysis
Categorical variables were compared using χ 2 or Fisher's exact tests.
Continuous variables are depicted as mean ± standard deviation (SD).
Continuous variables were compared using Mann-Whitney U tests.
Variables that showed a P-value of <0.05 on univariate analysis were included in our multivariate analysis using a logistic regression model.
All variables associated with sPOPE were candidates using a stepwise backward elimination procedure with a threshold of P < .05. The level of significance for all tests was set at P < .05. The predictive performance of the model was measured by the area under the curve of the receiver operating characteristic (ROC) curve analysis. The percentages of sPOPE probability in the predictive model were calculated based on the coefficients that were obtained by the multivariate logistic regression model. All statistical analyses were carried out using JMP version 12 (SAS Institute, Cary, NC, USA), and R version 3.1.1.

| Patient characteristics
Baseline characteristics of all included patients (n = 325) are shown in Table 1 Mean postoperative hospital stay of patients without sPOPE (n = 273), with sPOPE (n = 21), and with preventive chest tube were 10 ± 7, 29 ± 31 and 13 ± 9 days, respectively. Postoperative hospital stay of patients with sPOPE was significantly longer than that of patients without sPOPE (P < .0001) and patients with preventive chest tube (P = .002). Baseline characteristics of the validation cohort (n = 290) are shown in Table S1.

| Predictive model for sPOPE
Per the predictive model based on the results of our logistic regression analysis, the probability of developing sPOPE was calculated using the following formula: Abbreviation: sPOPE, severe postoperative pleural effusion.

| Risk classification of POPE using the predictive model
The optimal cut-off value of the sPOPE probability was 26.9% by ROC analysis (sensitivity, 0.62; specificity 0.91). We therefore compared the short-term outcomes of the 31 patients who underwent preventive chest tube insertion with sPOPE probabilities < 26.9% (low-risk group, n = 11) and ≥ 26.9% (high-risk group, n = 20). Total amount of drained pleural effusion until tube removal was higher in the high-risk group (1200 ± 602 mL) than in the low-risk group (643 ± 565mL; P = .012) (Figure 2A). Duration of chest tube insertion was longer in the high-risk group (7.0 ± 1.8 days) compared to the low-risk group (4.9 ± 2.3 days, P = .003) ( Figure 2B). Moreover, 6 patients (55%) out of the 11 patients classified as low risk drained less than 500 mL of pleural fluid ( Figure 3A). Five patients (45%) out of the 11 low-risk patients had their chest tube removed within 4 days of placement ( Figure 3B). In contrast, 18 patients (90%) out of 20 patients classified as high risk drained over 500ml of pleural fluid.

F I G U R E 2
Comparison of total amount of drained pleural effusion (mL) and duration of preventive chest tube placement (days) between low-and high-risk groups. A, Total amount of drained pleural effusion (mL) in the high-risk group was significantly greater than that of the low-risk group (P = .012). B, Duration (days) of preventive chest tube insertion in the high-risk group was longer than that of the low-risk group (P = .003) F I G U R E 3 A, Total amount of drained (mL) pleural effusion in patients who underwent preventive chest tube insertion (n = 31) according to risk classification. B, Duration (days) of drainage for pleural effusion for patients who underwent preventive chest tube insertion (n = 31) according to risk classification Nineteen patients (95%) out of the 20 high-risk patients required drainage for more than 5 days.

| D ISCUSS I ON
Postoperative pleural effusion after hepatectomy is a common postoperative complication and occurs in 18% to 71% of patients. [1][2][3][4] Symptomatic pleural effusion requiring thoracic drainage accounts for 9.4% of cases. which may have some effect on the amount of drained pleural effusion. Third, the data which were used to establish the predictive model did not involve patients with preventive chest tube insertion, and it could be argued this allowed for some statistical bias.

| CON CLUS ION
Severe postoperative pleural effusion causes prolongation of hospital stay. Our predictive model consisting of four risk factors allows for reliable prediction of sPOPE. This model may be a valuable tool in deciding when to place a preventive chest tube in patients undergoing hepatectomy.

CO N FLI C T S O F I NTE R E S T
The authors declare no conflicts of interest.