Laparoscopic approach reduces the incidence of postoperative pulmonary complications after hepatectomy for hepatocellular carcinoma: a multicenter overlap propensity score-weighted analysis.

Introduction
Hepatocellular carcinoma (HCC) is a leading cause of cancer-related global mortality. Surgical resection is the primary treatment curative intent option (1-3). Hepatectomy, whether performed through an open or laparoscopic approach, carries the risk of postoperative complications (4,5). In particular, postoperative pulmonary complications (PPCs) are one of the most frequent types of postoperative morbidity. PPCs have been reported to occur in 10–50% of patients and can be responsible for longer postoperative hospitalization, increased perioperative mortality, and worse long-term oncological prognosis (5-17). In turn, the identification and reduction of PPCs are critical to improve short and long-term outcomes among HCC patients treated with hepatectomy.
Risk factors of PCC following hepatectomy include advanced age, diabetes mellitus, overweight/obesity, poorer liver function reserve, larger intraoperative blood loss, and blood transfusion (8,11,13,14,16). Other studies have suggested that a laparoscopic vs. open approach may reduce the incidence of PPCs following hepatectomy (6,15,18,19). Previous data have been limited, however, as these reports were derived from single-center databases (6,19), had small sample sizes (analytic patient number <500) (6,19), enrolled only patients treated with major hepatectomy (15), and had confounding factors that were not adjusted for on multivariable analysis (19). One study did use propensity score matching (PSM) analysis to adjust for confounding factors (15). This statistical method had been criticized, however, due to limitations related to loss of sample size, as well as having overly sensitivity at the expense of specificity (20). In contrast, overlap propensity score-weighted analysis, especially using inverse probability of treatment-weighing (IPTW), offers a more robust method to adjust for confounding factors, potentially providing more accurate estimates of the causal effect (21-23). Additionally, multivariable regression analysis and subgroup analysis can provide further insights into the correlation of two variables and the robustness of the findings.
The objective of the current study was to define the incidence of PPCs among HCC patients undergoing laparoscopic hepatectomy (LH) vs. open hepatectomy (OH) using the IPTW method, multivariable regression analysis, and subgroup analysis. These data may have important implications for surgical decision-making and patient counseling in the surgical management of HCC relative to the risk of PPCs. We present this article in accordance with the STROBE reporting checklist (available at https://hbsn.amegroups.com/article/view/10.21037/hbsn-24-276/rc).

Methods

Study design and participants
This study was a multicenter, retrospective cohort study conducted across 12 hospitals in China. The study period spanned from January 2010 to December 2021. Patients included in the study were individuals who underwent hepatectomy for HCC, which was confirmed by postoperative histopathological examination in the resected specimens. The decision to perform LH or OH was based on the surgeon judgment and the patient preference. Exclusion criteria were patients who had: (I) combined HCC and cholangiocarcinoma; (II) ruptured or recurrent HCC; (III) previous anti-tumor treatment (i.e., portal vein embolization, transcatheter arterial chemoembolization, and associating liver partition and portal vein ligation for staged hepatectomy), or major thoracic or abdominal surgery before hepatectomy; (IV) concomitant biliary reconstruction or gastrointestinal surgical procedures during hepatectomy; and (V) acute bacterial infection within 1 month before surgery. This retrospective study was registered with ResearchRegistry.com (unique identification number: research registry 9610). The study protocol was approved by the institutional review board at each participating center, and the study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments and Good Clinical Practice guidelines (24). Institutions review board approval was obtained from Eastern Hepatobiliary Surgery Hospital of Shanghai (No. EHBHKY2019-01-002). Informed consent for the data to be used for clinical research was obtained from all patients.

Clinical variables
Preoperative variables included age, gender, obesity [body mass index (BMI) ≥28.0 kg/m2], chronic obstructive pulmonary disease (COPD), diabetes mellitus, American Society of Anesthesiologists (ASA) score, infection history of hepatitis B virus (HBV) or hepatitis C virus (HCV), liver cirrhosis, portal hypertension, Child-Pugh grade, gross vascular invasion, maximum tumor size, and tumor number. Intraoperative variables included surgical approach (OH or LH), extent of hepatectomy (major or minor), intraoperative blood loss, and intraoperative blood transfusion. Portal hypertension was defined as presence of splenomegaly with a reduced platelet count (<100×109/L) and/or esophageal varices. Major hepatectomy was defined as partial hepatectomy of three or more Couinaud’s hepatic segments, while minor hepatectomy was defined as fewer than three segments.

Surgical procedures
The perioperative assessment and surgical protocols were generally consistent at each center, with hepatectomy criteria remaining constant throughout the study’s duration. While there were some variations among centers in the application of LH, the following conditions were generally considered relative contraindications: (I) tumor size >10 cm; (II) multiple tumors (>3); (III) tumor invasion of major vessels; (IV) advanced liver cirrhosis; and (V) severe abdominal adhesions. The final decision on surgical approach was made by the operating surgeon based on individual patient factors and preferences. While the technical details of OH have been elucidated previously (25,26), LH involves the creation of a CO2 pneumoperitoneum, maintained at 12–15 mmHg. The procedure typically necessitates the insertion of four to five trocars. Surgeons employ laparoscopic ultrasound or fluorescence-guided navigation to assess tumoral and vascular conditions. All laparoscopic procedures were purely laparoscopic, with no hand-assisted or laparoscopic-assisted cases. Energy devices primarily included ultrasonic scalpels and bipolar electrocautery, with some centers using LigaSure. Liver parenchymal transection was mainly performed using ultrasonic scalpels or CUSA. Intrahepatic vessels and bile ducts exceeding 5 mm in diameter are secured with plastic or titanium clips. Hemostasis is achieved through a combination of bipolar electrocoagulation and suturing techniques. All LH procedures were conducted by surgeons who had undergone specialized training at Chinese provincial-level teaching hospitals and possessed extensive experience in laparoscopic surgery, having performed over 100 laparoscopic cholecystectomies.

Postoperative outcomes
Postoperative mortality was defined as death within 30 days of surgery and the causes of mortality were recorded. Postoperative morbidity (postoperative complications) was standardized as occurring within 30 days of surgery and graded according to the Clavien-Dindo classification (27). Minor and major morbidities were defined as Clavien-Dindo I–II and III–V, respectively. Postoperative complications were further categorized as PPCs and other postoperative complications. PPCs included symptomatic pleural effusion, pulmonary infection, respiratory insufficiency, acute respiratory distress syndrome (ARDS), and pulmonary embolism, which were diagnosed based on clinical, radiological, and laboratory findings (6,8,11,16,28). Symptomatic pleural effusion was defined as pleural effusion confirmed by clinical symptoms (such as dyspnea, chest pain), physical signs (such as dullness to percussion, decreased breath sounds), and imaging examinations [such as X-ray or computed tomography (CT)], requiring medical treatment (such as diuretics) or surgical intervention (such as thoracentesis, chest tube drainage). X-ray examinations were not routinely performed but were decided whether to perform imaging examinations based on the patient’s clinical manifestations (such as hypoxemia, increased respiratory rate, fever). Pleural effusion with simple tachycardia was not considered a positive case unless accompanied by respiratory symptoms, signs, and/or hypoxemia. Other postoperative complications included hepatic dysfunction, abdominal hemorrhage, bile leak, surgical site infection, ascites, and others (26). Postoperative hepatic dysfunction was defined according to the “50-50 criteria” on or after postoperative day (POD) 5 (29). Abdominal hemorrhage was defined by a drop in hemoglobin level of more than 3 g/dL compared with preoperative baseline levels and/or any postoperative transfusion of packed red blood cells units due to decreasing hemoglobin level and/or the need for invasive reintervention (30). Ascites that required diuretics or paracentesis were also included as morbidity (31). Bile leak was diagnosed based on a drain bilirubin concentration more than threefold higher than that in serum (32).
The primary endpoint of this study was the incidence of PPCs, while the secondary endpoints were the incidences of two typical types of PPCs, i.e., symptomatic pleural effusion and pulmonary infection.

Statistical analysis
Statistical analysis was performed using SPSS software (version 26.0) and R software (version 4.3.1). Patients were divided into two groups based on surgical approach of hepatectomy: LH or OH. Categorical variables were expressed as number (n) or proportion (%), while continuous variables were expressed as mean ± standard deviation or median (range). Categorical variables were compared using the χ2 test or the Fisher’s exact test, as appropriate. The Student’s t-test was used to compare continuous variables when applicable, otherwise, the Mann-Whitney U test was applied. The incidences of overall PPCs, symptomatic pleural effusion and pulmonary infection were compared among patients who underwent LH vs. OH.
In order to minimize possible selection bias among patients who underwent LH vs. OH, overlap propensity score-weighted analysis was used (20,22). First, propensity scores were estimated using a logistic regression model that included baseline patient and tumor characteristics as covariates. Then, the inverse probability of treatment weights was calculated, where the weight for each patient was the inverse of the propensity score if the patient received LH, or the inverse of 1 minus the propensity score if the patient received OH (22). After applying the IPTW, a pseudo-population was created, in which the distribution of measured baseline covariates was balanced between the treatment groups (33). The IPTW model preserved the sample size of the study cohort, and no study participants were dropped, which was advantageous compared with the PSM method. A level of P value <0.05 was considered statistically significant. Apart from the P value, standardized mean difference (SMD) was used to measure differences in clinical characteristics among patients who underwent LH vs. OH, with SMD <0.1 to indicate negligible differences, and between 0.1 and 0.3 to indicate small differences.
In addition to the IPTW method, multivariable logistic regression was used to identify independent correlation among variables. Variables with a P value of less than 0.10 in the univariable analysis and the core study variable (surgical approach) were included in the multivariable model. The results were reported as adjusted odds ratios (ORs) with 95% confidence intervals (CIs). Furthermore, subgroup analysis was performed in the IPTW cohort to explore the robustness of the findings across different patient-related and tumor-related characteristics. This allowed for a more nuanced understanding of the impact of LH vs. OH on the incidence of PPCs across different subgroups.

Results
Among 4,694 HCC patients included in the analytic cohort (Figure 1), 766 (16.3%) underwent LH and 3,928 (83.7%) underwent OH. According to the intention-to-treat analysis, 65 (1.4%) patients who underwent conversion from laparoscopic to open approach were included in the LH group; the conversion rate was 8.5% among 766 patients who underwent LH. After applying overlap propensity score-weighted analysis, IPTW created 4,716.2 standardized patients who underwent LH and 4,692.9 who underwent OH.

Comparisons of clinical variables
Patient clinical characteristics among patients who underwent LH vs. OH in the entire and IPTW cohorts are summarized in Table 1. In the entire cohort, patients who underwent LH, vs. patients who underwent OH, had a higher proportion of individuals age >70 years (13.2% vs. 9.2%, P<0.001), COPD (13.7% vs. 7.9%, P<0.001), diabetes mellitus (14.4% vs. 9.8%, P<0.001), ASA score >2 (23.4% vs. 14.2%, P<0.001), and HBV positiveness (82.5% vs. 78.1%, P=0.007), yet a lower proportion of male (82.5% vs. 86.2%, P=0.009), HCV positive serology (4.8% vs. 9.6%, P<0.001), maximum tumor size >5.0 cm (25.3% vs. 52.1%, P<0.001), multiple tumors (11.1% vs. 20.0%, P<0.001), gross vascular invasion (4.7% vs. 11.9%, P<0.001), major hepatectomy (11.7% vs. 24.9%, P<0.001), and intraoperative blood loss >600 mL (15.4% vs. 18.5%, P=0.048). Of note, there were no difference in any covariate among patients who underwent LH vs. OH after matching in the IPTW cohort (all SMDs <0.1 indicated negligible differences) (Figure 2).

Comparisons of postoperative outcomes
Comparisons of postoperative outcomes among patients who underwent LH vs. OH in the overall and IPTW cohorts are noted in Table 2. In each cohort, postoperative morbidity among patients who underwent LH was lower than OH (in the entire cohort: 22.3% vs. 35.2%, and in the IPTW cohort: 22.4% vs. 34.8%, respectively; both P<0.001). Patients who underwent LH also had lower postoperative minor morbidity (in the entire cohort: 14.1% vs. 26.3%, and in the IPTW cohort: 13.4% vs. 26.2%, respectively; both P<0.001), as well as lower incidence of other postoperative complications apart from PPCs (in the entire cohort: 13.1% vs. 23.5%, and in the IPTW cohort: 14.0% vs. 23.3%, respectively; both P<0.001) vs. patients who underwent OH. In addition, the mean postoperative hospital stay after LH was shorter than after OH in the entire cohort (9.3±5.7 vs. 12.6±7.4 days, P<0.001), as well as in the IPTW cohort (9.8±5.2 vs. 12.6±7.3 days, P<0.001).
The incidences of overall PPCs in the entire cohort was 10.9% (511/4,694); the incidence of PPCs among patients who underwent LH was markedly lower vs. individuals who underwent OH (7.3% vs. 11.6%, P=0.001). Compared with OH, LH was also associated with a lower incidence of symptomatic pleural effusion (2.0% vs. 6.2%, P<0.001) but comparable risk of pulmonary infection, respiratory insufficiency/ARDS, and pulmonary embolism in the entire cohort (5.9% vs. 7.0%, P=0.32; 2.0% vs. 2.0%, P>0.99; and 0.3% vs. 0.1%, P=0.41). Similar proportions of overall PPCs, including symptomatic pleural effusion, pulmonary infection, respiratory insufficiency/ARDS, and pulmonary embolism were also noted in the IPTW cohort (7.4% vs. 11.6%, P=0.01; 2.5% vs. 6.2%, P=0.007; 5.3% vs. 7.0%, P=0.16; 1.7% vs. 1.9%, P=0.65; and 0.2% vs. 0.1%, P=0.22). The incidence of overall and each type of PPCs between LH and OH groups in the entire and IPTW cohorts are presented in Figure 3. In addition, there was no significant difference in the incidence of PPCs between right and left hepatectomy (excluding cases of caudate lobe resection) (11.2% vs. 10.5%, P=0.62 in the entire cohort; and 11.8% vs. 10.8%, P=0.59 in the IPTW cohort). Laparoscopic approach showed lower incidence of PPCs in both types of hepatectomy (all P<0.05).

Univariable and multivariable analysis
Univariable and multivariable logistic regression analyses were used to identify factors associated with PPCs; symptomatic pleural effusion and pulmonary infection in the IPTW cohort are noted in Tables 3-5, respectively. After adjustment for other confounding factors, multivariable regression analysis identified laparoscopic approach as an independent protective factor against overall PPCs (OR 0.63, 95% CI: 0.42–0.92, P=0.02) and symptomatic pleural effusion (OR 0.36, 95% CI: 0.18–0.70, P=0.003), but not pulmonary infection (OR 0.82, 95% CI: 0.54–1.26, P=0.88). Similar results were demonstrated in univariable and multivariable logistic regression analyses in the entire cohort (Tables S1-S3).

Subgroup analysis
Subgroup analysis demonstrated similar incidences of overall PPCs in the IPTW cohort across different patient and tumor characteristics. The benefit of LH to reduce the incidence of PPCs was observed in almost all subgroups, with no interaction between the surgical approach and any subgroup variable (Figure 4).

Discussion
The present multicenter, retrospective cohort study aimed to provide a comprehensive examination of the impact of surgical approach (LH vs. OH) on PPCs among patients with HCC. By analyzing data from 12 hospitals over a 12-year span, the findings demonstrated that patients who underwent LH experienced a markedly lower incidence of PPCs, specifically symptomatic pleural effusion compared with individuals who underwent OH. This finding is consistent with several previous studies suggesting laparoscopic approach for other indications may reduce the risk of PPCs (6,15,18,19). The current study adds to the existing literature as the data were derived from a large sample size based on a multicenter database and utilized a more robust statistical approach with IPTW modeling. Of note, the results persisted even after rigorous additional adjustment for potential confounding factors by multivariable analysis, as well as on subgroup analysis.
In the context of PPCs following hepatectomy, our findings further underscore the significance of the surgical approach. PPCs, ranging from pulmonary infections to pleural effusion, can prolong hospital stays and elevate early mortality rates (6). Notably, open approach for abdominal surgery presents a heightened risk due to the inherent surgical trauma (8,9). Our data aligns with existing literature suggesting that minimally invasive surgery (MIS), such as laparoscopic procedures, can mitigate the risk of PPCs (19). The reduced trauma in MIS leads to diminished postoperative pain, facilitating more effective deep breathing and coughing exercises, crucial for preventing PPCs. Additionally, the shorter surgery duration and lesser blood loss in MIS further consolidate its advantages in minimizing PPCs (19). This not only emphasizes better patient outcomes but also hints at potential healthcare cost savings.
LH has emerged as a promising surgical alternative for HCC, which has also been demonstrated to be associated with shorter postoperative hospital stays (mean: 9.8 vs. 12.6 days in the IPTW cohort, P<0.001) and a lower incidence of other postoperative complications (14.0% vs. 23.3%, P<0.001) in the current study. In turn, there is a global trend in surgery with higher utilization of minimally invasive techniques as the preferred approach due to their myriad of benefits including smaller incisions, less pain, faster recovery, and fewer complications (34-39). Our findings add to this body of evidence, suggesting that LH is not only feasible but also advantageous for HCC patients. In fact, despite patients who underwent a less invasive procedure in the LH group tending to have more comorbidities (e.g., older age, COPD, diabetes mellitus, higher ASA score) than those in the OH group, the outcomes were better in the LH group. This may be attributable to surgeons opting for laparoscopic approach for patients deemed at a higher risk for postoperative complications. In spite of potential selection bias, LH demonstrated superior outcomes. After rigorous statistical adjustment via the IPTW method, these discrepancies in baseline characteristics between groups were virtually eliminated, yet the benefit of LH persisted—reinforcing the potential benefits of LH.
In the multivariable analysis of the present study, we included intraoperative blood loss >600 mL and intraoperative blood transfusion as covariates in the model. The results showed that even after adjusting for these factors, laparoscopic approach remained an independent protective factor against PPCs in the IPTW cohort (adjusted OR 0.63, 95% CI: 0.42–0.92, P=0.02). Meanwhile, intraoperative blood transfusion was also identified as an independent risk factor associated with PPCs (adjusted OR 1.81, 95% CI: 1.12–2.93, P=0.02).
A major strength of the present study lies in the application of the IPTW method to rigorously adjust for potential confounding and selection bias. IPTW is a preferred approach for causal inference in observational studies, especially when the treatment groups are unbalanced (23,40-43). Compared with the more commonly-used PSM, IPTW has several advantages. First, IPTW preserves the entire study sample and thus retains more information and statistical power, whereas PSM may exclude a substantial proportion of patients due to the lack of appropriate matches. Second, IPTW creates a pseudo-population that is representative of the original study population, allowing for the estimation of the average treatment effect that is generalizable to the target population. In contrast, PSM estimates the average treatment effect on the treated, which may not be applicable to patients outside the matched sample. Third, IPTW is less sensitive to the misspecification of the propensity score model and the matching algorithm, as it uses the propensity score as a covariate rather than a matching criterion (22,44). Given these strengths, IPTW has been increasingly adopted in high-quality surgical studies published in leading journals (23,40-43,45-51). In the present study, IPTW successfully balanced the baseline characteristics between the LH and OH groups and allowed us to estimate the independent effect of surgical approach on PPCs. We believe that the application of this robust method enhances the validity and generalizability of our findings.
In the present study, we specifically focused on pulmonary complications as the primary outcome because they are one of the most frequent and clinically relevant morbidities following hepatectomy. Therefore, investigating the risk factors and potential interventions for pulmonary complications is of paramount importance to improve the clinical management of HCC patients undergoing hepatectomy. Although our multicenter database contains information on various perioperative outcomes, we believe that a dedicated analysis of pulmonary complications offers several advantages. First, it allows for a more comprehensive and in-depth evaluation of the incidence, risk factors, and impact of pulmonary complications, which may be overlooked in studies focusing on overall morbidity. Second, it enables us to compare our findings with previous studies that have specifically examined pulmonary complications after liver surgery, providing a basis for validation and generalization (11,15). Third, it highlights the importance of pulmonary complications as a key quality indicator and a potential target for perioperative optimization in HCC patients undergoing hepatectomy. We encourage collaborative efforts to address other types of morbidities and outcomes using large-scale, prospective, and multicenter datasets.
Several limitations should be noted. The retrospective nature of the methodology and the institutional variations in surgical practices may have led to potential inherent biases. Additionally, decisions regarding the surgical approach were at the discretion of the surgeon, which may have introduced some unmeasured confounding factors. Furthermore, the study was conducted in a single country, which may limit the generalizability of our findings to other populations. While the results suggest the superiority of LH in terms of PPCs, choice of surgical approach should be individualized, taking into account the tumor’s location, size, and the surgeon’s experience.

Conclusions
In conclusion, for patients with HCC undergoing hepatectomy, the laparoscopic approach appears to be associated with a significantly reduced risk of PPCs, particularly symptomatic pleural effusion. Our findings strongly support the wider adoption of laparoscopic techniques in the management of HCC, wherever feasible. These findings have important implications for surgical decision-making and patient counseling in the surgical management of HCC. Future prospective randomized trials would be invaluable in further substantiating these results and setting forth best practice guidelines.

Supplementary
The article’s supplementary files as