Long-Term Survival Impact of High-Grade Complications after Liver Resection for Hepatocellular Carcinoma: A Retrospective Single-Centre Cohort Study

Background and Objectives: Although complications after liver resection for hepatic cancer are common, the long-term impact of these complications on oncological outcomes remains unclear. This study aimed to investigate the potential effect of high-grade postoperative complications on long-term mortality and cancer recurrence after surgical resection of hepatocellular carcinoma. Materials and Methods: In a retrospective cohort study, patients undergoing curative liver resection for primary hepatocellular carcinoma between 2005 and 2016 were evaluated. The Clavien–Dindo (CD) grading system was used to classify patients into two groups of either high-grade complications (grade III or IV) or none or low-grade complications (grade 0 to II) within 30 days after surgery. The primary endpoint was all-cause mortality. Secondary endpoints were cancer-specific mortality and cancer recurrence. Weighted Cox proportional hazards regression models were used to calculate the adjusted hazard ratio (aHR) with a 95% confidence interval (CI) for the outcomes of interest. Results: A total of 1419 patients with a median follow-up time of 46.6 months were analysed. Among them, 93 (6.6%) developed high-grade complications after surgery. The most common complications were bile leakage (n = 30) in CD grade III and respiratory failure (n = 13) in CD grade IV. High-grade complications were significantly associated with all-cause mortality (aHR: 1.78, 95% CI: 1.55–2.06) and cancer-specific mortality (aHR: 1.34, 95% CI: 1.13–1.60), but not cancer recurrence (aHR: 0.92, 95% CI: 0.84–1.02). Independent influential factors for complications were sex, diabetes mellitus, clinically significant portal hypertension, oesophageal varices, multifocal cancer, intraoperative blood loss, and anaesthesia duration. Conclusions: Patients who had high-grade postoperative complications had a greater risk of long-term mortality after liver resection for hepatocellular carcinoma. Prevention of postoperative complications may serve as an effective strategy for improving long-term survival.


Introduction
Hepatocellular carcinoma (HCC) is the most common type of primary liver cancer and is ranked as the third most common cause of cancer-related death worldwide [1]. Surgical resection of primary tumours remains the mainstay treatment modality for patients with resectable HCC. However, the long-term prognosis after surgical resection remains very poor, with rates of 55.9% and 42.3% reported for 1-year and 2-year overall survival, respectively [2,3]. Despite recent advances in surgical techniques and perioperative care, patients undergoing liver resections for HCC are still at high risk of postoperative complications, with a reported rate of 20% to 50% [4][5][6].
Postoperative complication is an established risk factor for short-term adverse events after liver resection, including mortality, readmission, reoperation, prolonged length of hospital stay, and greater medical expenditures [7][8][9][10][11]. Mounting evidence indicates that postoperative complications may also affect long-term survival among patients that survive postoperative complications, irrespectively of the preoperative patient characteristics [12,13]. For patients with cancer, postoperative complications have been demonstrated to be associated with greater long-term mortality and cancer recurrence after surgical resection of lung cancer [14], gastric cancer [15], and pancreatic cancer [16].
This study aimed to evaluate the putative effect of high-grade postoperative complications on long-term oncological outcomes after surgical resection for HCC using the CD classification system [27]. Based on current evidence [28], we hypothesised that high-grade postoperative complications were linked to greater all-cause mortality, cancer-specific mortality, and cancer recurrence in patients with HCC.

Criteria of Patient Selection
We obtained approval from the Institutional Review Board of Taipei Veterans General Hospital in Taiwan (IRB-TPEVGH, No. 2021-07-035BC). Written informed consent was waived by the Institutional Review Board, and all the study methods were conducted in accordance with the institutional ethical standards of the responsible committee on human experimentation. We reviewed the medical records of 2215 patients consecutively undergoing hepatic resections at Taipei Veterans General Hospital between 2005 and 2016. Patients were excluded for the following conditions: repeat surgery, liver transplantation, pathology-proven benign lesion, metastatic cancer, non-HCC cancer, lymph node or distant metastasis, missing data, and follow-up interval or mortality < 30 days after surgery. A total of 1419 patients were included for analyses ( Figure 1).

Postoperative Complications
Complications within a 30-day postoperative period were recorded and ranked using the CD classification, a representative grading system for postoperative complications [27]. This study sought to investigate the association between high-grade complications, defined as a CD grade III or IV (grade III: requiring surgical, endoscopic, or radiological interventions; grade IV: requiring intensive care management), and HCC outcomes, because such patients need invasive interventions [27]. In case of multiple complications, the highest grade was retained for analyses. Patients were divided into two groups of either high-grade complication (CD grade III or IV) or none or low-grade complications (CD grade 0 to II, as controls).

Primary and Secondary Endpoints
The primary endpoint was all-cause mortality. Secondary endpoints were cancerspecific mortality and cancer recurrence. Survival time was defined as the interval between the date of the operation and the date of death or recurrence. For patients without death or recurrence, survival times were the corresponding censored observation. Patients were followed up to 30 September 2018.

Covariates for Adjustment
The institutional electronic medical database was utilised to collect factors potentially associated with oncological outcomes after surgical resection of HCC. The clinical covariates included American Society of Anaesthesiologists (ASA) physical status, viral serology [29], liver cirrhosis and Child-Pugh class [30], clinically significant portal hypertension (hepatic venous pressure gradient ≥ 10 mm Hg), presence of oesophageal varices, and coexisting diseases (alcoholism, diabetes mellitus, and chronic kidney disease) ( Table 1). Preoperative records of radiofrequency ablation, trans-arterial chemoembolization, and percutaneous ethanol injection were also considered for analyses. Preoperative laboratory tests, including serum aspartate aminotransferase (AST), alanine aminotransferase (ALT), total bilirubin, alpha-fetoprotein, and albumin, were also considered [30,31]. Surgical characteristics were extent of hepatectomy (>2 Couinaud liver segments or not), use of laparoscopic or robotic techniques, surgical margin, epidural blockade [32,33], intraoperative blood loss and transfusion (red blood cells, fresh frozen plasma or platelets) [34,35], and anaesthesia duration. Pathology features were tumour size and number, cell differentiation, microvascular invasion, and extracapsular invasion [36,37]. Patients were classified according to the Barcelona Clinic Liver Cancer (BCLC) staging system [38]. The data were collected by a specialist anaesthesiologist not involved in the statistical analysis. The authors verified the quality of data through random sampling.

Surgical Techniques and Cancer Surveillance
At the medical centre, all liver resections were performed by experienced general surgeons. The liver parenchymal transection was performed using a clamp-crushing method, and the Pringle's manoeuvre and argon beam coagulator were used to halt haemorrhage. Laparoscopic or robotic surgery was used in selected patients from July 2011. All surgeries included in this study were performed for curative intent.
For postoperative surveillance, patients received routine ultrasonography, computed tomography, or magnetic resonance imaging every four months for two to three years, then every six months. In addition, the serum concentration of alpha-fetoprotein was checked every four months for two to three years, then every six months. Bone scintigraphy or positron emission tomography were used for suspected locoregional recurrence or distant metastases. If the diagnosis was equivocal, a biopsy was used to confirm the presence of recurrent disease.

Sample Size Estimation
According to previous studies, at least 1175 patients are needed to detect a hazard ratio (HR) of 1.39 for all-cause mortality, accepting a type I error of 5% and a type II error of 20%, with an incidence of high-grade complications of 6.6% in this study [28,39]. We included 1419 patients in the analysis, which met the minimal requirements for sample size.

Statistical Analysis
For baseline patient characteristics, Shapiro-Wilk tests were used as normality tests. Normally distributed variables were presented as the mean ± standard deviation. Nonnormally distributed data were presented as the median with interquartile range. Logarithmic transformation was performed to reduce skewness of non-normal continuous variables, including intraoperative blood loss and anaesthesia duration. The distribution of patient characteristics was compared between the high-grade complication group and the control group using either independent t tests or Mann-Whitney U tests for continuous variables and chi-square tests or Fisher's exact tests for categorical variables, as appropriate. Kaplan-Meier curves and log-rank tests were used to compare the cumulative incidences of all-cause mortality, cancer-specific mortality, and cancer recurrence between groups. Univariate Cox proportional hazard regression analysis was conducted to examine the effects of high-grade complications and other covariates on oncological outcomes.
We used inverse probability treatment weighting (IPTW) as the primary statistical approach because IPTW decreases inherent bias in patient and disease attributes that can affect whether patients do or do not have an exposure factor [40]. IPTW adjusts for the confounding effects that individual covariates could exert on long-term cancer outcomes, whilst retaining the patient sample and statistical power [40]. The IPTW analysis was performed as follows [40]. First, binary logistic regression analysis was used to estimate the probability of developing a high-grade complication based on a list of patient characteristics (Supplementary Table S1). The inverse of the estimated probability was then used for weighted Cox regression analyses, and 1% of subjects at the end of the weighting distribution were truncated to reduce the effect of large weights on study results. The weighted Cox regression model was implemented to assess the independent effects of highgrade postoperative complications on the outcomes of interest. Additionally, multivariable logistic regression analysis was performed using a backward variable elimination process with the entry and removal significance criteria of 0.1 and 0.05, respectively, to identify the influential factors of high-grade complications. We considered p < 0.05 to be statistically significant for a two-sided test. All statistical analyses were conducted using SAS software, version 9.4 (SAS Institute Inc., Cary, NC, USA).

Patient Characteristics
A total of 1419 patients had a median follow-up time of 46.6 months, with an interquartile range of 22.6 to 80.3 months. Among them, 93 (6.6%) patients developed high-grade complications (CD grade III: 72 and grade IV: 21). Table 1 shows the patient, clinical, and pathological characteristics of the included subjects. Compared with patients with none or low-grade postoperative events, those with high-grade complications were more likely to be older and to have diabetes mellitus, ASA class ≥ 3, a higher level of serum AST, and a lower level of serum albumin. CD grade III or IV complications were also more common in patients with multifocal cancer, >2 segments of liver resected, greater intraoperative blood loss, and a higher transfusion rate. The anaesthesia duration was longer in those with a high-grade complication. The most common complications were bile leakage in CD grade III and respiratory failure in CD grade IV (Table 2).

Influential Factors for High-Grade Complications
Preoperative influential factors for high-grade complications included sex (male vs.  (Table 4).

Discussion
In this study, we found that high-grade postoperative complications were significantly associated with greater all-cause mortality and cancer-specific mortality after liver resection for HCC, but cancer recurrence was not affected. Our analyses identified several influential factors for postoperative complications, including sex, diabetes mellitus, Child-Pugh class, multifocal cancer, intraoperative blood loss, and anaesthesia duration. We used a large single-centre cohort and adjusted for a detailed list of patient and disease characteristics to investigate the putative impact of postoperative complications on long-term oncological outcomes after surgical resection of HCC. This evidence supports the importance of preventing postoperative complications as a practicable strategy of improving long-term survival in patients following liver resection for HCC.
Our results suggest that high-grade complications after hepatic resections might increase the risk of all-cause mortality in patients with HCC, in agreement with some previous studies [18][19][20][21][22][23][24] but not others [25,26]. The present study also demonstrated no definite association between postoperative complications and cancer recurrence, in line with some studies [18,20,25,26] but contrasting with others [21][22][23][24]. Differences in the types and severity of complications are potentially responsible for these inconsistent findings. Two Asian studies recently reported a significant association between postoperative infectious complications and cancer recurrence after surgical resection of HCC [23,24]. Host immunity may underlie the increased recurrence among patients who developed infectious complications [23,24]. Compared with these two studies [23,24], most postoperative complications in our study were non-infectious. Regarding the severity of complications according to the CD classification, previous studies demonstrated a significant association between complication severity and increased long-term mortality rather than cancer recurrence [20,22]. Noticeably, few studies have evaluated the severity threshold for when postoperative complications exert a long-term survival impact. More studies are needed to elucidate the prognostic role of postoperative complications with varying severity after cancer surgery.
The biological mechanism underlying greater long-term mortality in patients who develop postoperative complications remains unclear. Prior studies have shown that greater intraoperative blood loss and perioperative transfusion of allogeneic blood were potentially linked to compromised immune function and adversely affected long-term survival and cancer control [34,35,41]. Additionally, it is uncertain whether the association between postoperative complications and cancer outcomes is causative or simply correlative. Our results showed that patients with pre-existing diabetes mellitus and a higher Child-Pugh class were susceptible to severe postoperative complications, and both factors are established prognostic factors for oncological outcomes in HCC [30,42]. Postoperative complications could be a proxy of worse patient health, more aggressive tumours, and more extensive surgical resections, which can increase long-term mortality, instead of being a direct cause of mortality by themselves.
The present findings justify the prevention of postoperative complications as a method of improving long-term survival after liver resection of HCC. Our analyses identified some modifiable factors for postoperative complications, including diabetes mellitus, intraoperative blood loss, and anaesthesia duration. First, Hosokawa et al. have previously indicated that diabetic patients with inadequate glycaemic control were at higher risk of tumour recurrence and early mortality after radiofrequency ablation therapy for HCC [43]. Accordingly, close monitoring and optimal control of perioperative blood glucose levels are pivotal in the prevention of postoperative complications and the improvement of the patient's oncological prognosis. Second, since intraoperative haemorrhage and allogeneic blood transfusions may exert a detrimental effect on long-term survival after cancer surgery, strategies aimed at minimizing surgical bleeding should be further developed, particularly for patients with cirrhotic liver [34,35,41]. Third, implementation of enhanced recovery after surgery protocol has been proven safe and effective in decreasing postoperative complications and improving short-term and long-term survival after cancer surgery [44][45][46]. Reduced operative duration should be regarded as a universal goal for surgeons [47]. Further studies are warranted to develop and assess measures which can shorten operative times.
There were some limitations to the present study. First, this study was retrospective, and therefore unrecorded variables (i.e., intraoperative blood pressure, perioperative myocardial injury, frailty, and sarcopenia) could not be further analysed and controlled. Second, we did not analyse low-grade complications. Consequently, our results were not generalizable to these patient populations. Third, the number of high-grade complications in our study was relatively small, which might produce underpowered statistics in the analyses. Fourth, our medical database did not contain the information about the eligibility and receipt of liver transplantation after liver resection for HCC, which might confound the long-term survival in our cohort. Fifth, the sero-epidemiology of hepatitis virus infection and the coverage of hepatitis B vaccination in Taiwan are different from western countries, which might limit the generalisability of our findings [48]. Last, we only included intraoperative blood transfusions in the analysis. Therefore, the impact of postoperative blood product administration on complications and survival remains unknown in our patient population.

Conclusions
Patients who developed high-grade postoperative complications had greater long-term risks of all-cause mortality and cancer-specific mortality after liver resection for HCC. Female, diabetes mellitus, higher Child-Pugh class, multifocal tumour, greater intraoperative blood loss, and longer anaesthesia duration were risk factors for high-grade complications after surgical resection of HCC. These findings highlight the importance of preventing postoperative complications as a clinical strategy for improving long-term survival after liver resection for HCC. Adequate glycaemic control, optimal surgical haemostasis, and reduced operation duration may play an essential role in decreasing complications and early mortality after liver resection. Future studies are warranted to evaluate the clinical benefits of prophylactic measures for postoperative complications in long-term cancer control and patient survival.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/medicina58040534/s1, Table S1: The result of logistic regression analysis for inverse probability of treatment weighting models. Institutional Review Board Statement: The study was approved by the Institutional Review Board of Taipei Veterans General Hospital, Taipei, Taiwan (IRB-TPEVGH, No. 2021-07-035BC; date of approval on 3 July 2021). All methods of this study were performed in accordance with the Helsinki Declaration and relevant regulations.

Informed Consent Statement: Written informed consent was waived by the Institutional Review
Board, and all the study methods were conducted in accordance with the institutional ethical standards of the responsible committee on human experimentation.

Data Availability Statement:
The data presented in this study are available upon request from the corresponding author. The data are not publicly available due to the regulations of the Institutional Review Board.

Conflicts of Interest:
The authors declare no conflict of interest.