Introduction

Surgical resection remains the cornerstone curative-intent treatment option for patients with early hepatocellular carcinoma (HCC) and intrahepatic cholangiocarcinoma (IHCC), the two most common primary liver malignancies.1,2 Despite continued improvements in surgical technique and perioperative care, liver resections remain associated with considerable morbidity. In addition to common complications such as bleeding, surgical site infections, and pleural effusions, notable hepatectomy-specific complications include bile leaks and post-hepatectomy liver failure (PHLF).3 PHLF represents the most consequential complication of liver resection and is related to reduced liver remnant function and can be exacerbated by the incidence of other postoperative complications, especially infections.4,5 The incidence of PHLF ranges from 1 to 33% and has been associated mortality of ~ 10%.3,6 In contrast, bile leak occurs in 10–17% of cases and often leads to intra-abdominal sepsis and subsequent risk of mortality.3,7,8

Since the early 2000s, chronic liver disease has been recognized as a risk factor for PHLF, and the morphologic severity of cirrhosis has been correlated with PHLF.9 As such, severe cirrhosis often represents an absolute contraindication to liver resection, with studies suggesting a Model for End-stage Liver Disease (MELD) score of ≥ 11 as a cutoff for an unacceptable high risk of PHLF.10,11 Minimally invasive hepatectomy (MIH) has been gradually introduced as an option for cirrhotic patients despite early criticism of its associated difficulty and technical demand in an already challenging operative cohort.12 Over time, MIH has been demosntrated to be a safe and feasible approach compared with conventional open hepatectomy (OH).13

With a growing acceptance and adoption of MIH, there is a need to properly evaluate its outcomes against the standard open approach with respect to postoperative complications. In the current study, we hypothesize that MIH may provide improved short-term outcomes over OH, especially with respect to PHLF, for patients with early cirrhotic patients who had resectable primary liver malignancy amenable to partial hepatectomy.

Methods

The National Surgical Quality Improvement Program (NSQIP) hepatectomy-targeted database 2014–2021 was utilized for this study. Patients undergoing surgical resection with a partial hepatectomy (CPT code 47120) and American Joint Committee on Cancer (AJCC) 8th edition stage T1 primary liver malignancies of HCC or IHCC were included. Moreover, only patients with confirmed liver cirrhosis who had pre-operative MELD scores ≤ 10 (within 30 days before surgery) were included in the analytic cohort.

The selected population was then subcategorized based on the surgical approach of liver resection into OH and MIH groups. Converted MIH patients were included in the OH group in the final analysis as the aim of the study was to analyze outcomes according to received treatment and not intention-to-treat. PHLF is reported in NSQIP according to the International Study Group of Liver Surgery (ISGLS) consensus definition and is classified as grades A-C according to their published guidelines.5

After comparing baseline characteristics, a 1:1 propensity score matching was performed between OH and MIH patients based on a multivariable logistic regression, which accounted for clinical and demographic variables including age, sex, race, body mass index (BMI), functional status, comorbidities, primary tumor histology, preoperative therapies, and intraoperative factors. Intraoperative details included the performance of additional hepatectomies or ablations in addition to the primary hepatectomy, as well as applying Pringle’s maneuver (i.e., hilar clamping) and drain placement.

Matching was applied following the nearest neighbor method per propensity scores with a caliper width of 0.05. Categorical outcomes were compared using conditional logistic regression, whereas continuous outcomes were compared using mixed-effect modeling. Effect sizes were reported as Hazard Ratios (HR) with 95% confidence intervals for all variables. A backward conditional multivariable logistic regression was performed to identify independent predictors of PHLF. SPSS v29 (Armonk, NY) was used to perform the statistical analysis, and statistical significance was set at α < 0.05 throughout the study.

Results

The NSQIP hepatectomy-targeted database reported 47,275 liver resections between 2014 and 2021. Overall, 922 patients met inclusion criteria and were selected for analysis (Fig. 1). Mean age of the entire cohort was 65.3 ± 8.5 years (median, 65) and 682 (74.0%) were men. Most patients were American Society of Anesthesiology (ASA) class III (73.5%). Regarding primary histology, 836 (90.7%) had HCC and 86 (9.3%) had IHCC. Only 4 (0.4%) patients had preoperative biliary stents, 9 (1.0%) had undergone portal vein embolization (PVE), 13 (1.4%) had arterial infusion therapy, and 4 (0.4%) had prior liver ablation. Eighteen patients (0.2%) were reported to have ascites present at the time of the procedure. 494 (53.6%) patients underwent OH versus 428 (45.4%) who had MIH. In the MIH group, 56 (6.1%) had an open conversion and were crossed over to the OH group for purposes of analysis. Median operative time was 191 ± 93 min, 302 (32.8%) had additional hepatectomies, 78 (8.5%) had additional ablations, 327 (35.5%) had drains, and Pringle’s maneuver was applied in 249 (27.0%) patients during parenchymal transection. Table 1 summarizes the demographic and clinical characteristics of selected patients.

Fig. 1
figure 1

Flow diagram demonstrating the selection steps. AJCC American Joint Commission on Cancer; CPT Current Procedural Terminology; HCC Hepatocellular Carcinoma; IHCC Intrahepatic Cholangiocarcinoma; MELD Model for End-Stage Liver Disease; NSQIP National Surgical Quality Improvement Program

Full size image
Table 1 Demographic and clinical characteristics of the selected patient population
Full size table

Following surgery, a bile leak occurred in 67 patients (7.3%) with grades A, B, and C reported at 36 (3.9%), 30 (3.3%), and 1 (0.1%), respectively. PHLF occurred in 43 (4.7%) patients with grade A, B, and C reported at 18 (2.0%), 14 (1.5%), and 11 (1.2%), respectively. Overall morbidity (Clavien Dindo grade I-V) was documented in 260 (28.2%) patients, major morbidity (grade III-V) in 103 (11.2%), and mortality (grade V) in 12 (1.3%). The median length of stay was 4 days (interquartile range, 3–6), and 70 patients (7.6%) had non-home discharges. In addition, 80 patients (9.0%) had an unplanned readmission within 30 days from surgery. Table 2 demonstrates a summary of 30-day outcomes of the selected patients.

Table 2 Thirty-day outcomes of the selected patients with T1 HCC/IHCC and low MELD who underwent partial hepatectomy
Full size table

Upon comparison of baseline characteristics, OH patients (n = 550) had a higher mean BMI (28.4 ± 5.8 vs. 27.7 ± 5.4; p = 0.037), higher rates of PVE (1.6% vs. 0.0%; p = 0.013), more frequently had Pringle’s clamping (34.0% vs. 16.7%; p < 0.001) and were more likely to have had drains placed (41.3% vs. 26.9%; p < 0.001). After calculating the propensity score and applying matching conditions, 708 patients remained (354 OH vs. 354 MIH) with comparable baseline characteristics and adequate calibration between the groups. Table 3 demonstrates the comparison between OH and MIH patients before and after propensity score matching.

Table 3 Comparison of baseline characteristics between OH and MIH patients (by definitive operative approach) before and after the 1:1 propensity score matching
Full size table

Thirty-day outcomes were then compared between the study groups. Patients undergoing MIH had lower rates of bile leak (HR 0.37 [0.19–0.72)], post-hepatectomy liver failure (HR 0.36 [0.15–0.86]), collections requiring drainage (HR 0.30 [0.15–0.63]), postoperative anemia requiring transfusion (HR 0.36 [0.21–0.61]), major morbidity (HR 0.45 [0.27–0.77]), and overall morbidity (HR 0.44 [0.31–0.63]). MIH patients also had a two-day shorter median length of stay (3 vs. 5 days; HR 0.61 [0.45–0.82]). A trend toward lower rates of non-home discharges was also noted in the MIH group (0.53 [0.28–1.00]). No difference was detected in operative time and other morbidities including wound, respiratory, septic complications, or mortality. Figure 2 shows a summary of 30-day outcomes comparison between the matched OH vs. MIH dataset with the reported HR and 95% confidence intervals.

Fig. 2
figure 2

Comparison of 30-day outcomes between OH and MIH in the matched dataset. MIH Minimally Invasive Hepatectomy; OH Open Hepatectomy

Full size image

Hepatectomy-specific complications were subsequently examined (i.e., PHLF and bile leak) and higher rates of PHLF grades B and C were noted in OH versus MIH (36.8% and 26.3% vs. 14.3% and 14.3%, respectively; p = 0.001). Similarly, higher rates of bile leak grade B were noted in the OH group (55.6% vs. 33.3%; p = 0.005). No grade C bile leaks were documented in either group. Figure 3 is a histogram of PHLF and bile leak grades in both study groups.

Fig. 3
figure 3

Grades of bile leak and post-hepatectomy liver failure in the matched OH vs. MIH dataset. MIH Minimally Invasive Hepatectomy; OH Open Hepatectomy; PHLF Post-Hepatectomy Liver Failure

Full size image

Next, we performed a propensity-score matched analysis comparing outcomes between cases that were performed open (N = 51) versus case that were attempted MIH but converted to open (N = 51). Groups in this comparison were well matched, with matched and un-matched demographics and comorbidities noted in supplementary table 1. After matching, there was no difference in primary outcomes between OH and converted cases, including no difference in rate of PHLF (p = 0.400) or Bile Leak (p = 0.141). There was a difference with respect to operative time (p = 0.002), anemia requiring transfusion (p < 0.001) and overall morbiditiy (p = 0.008), each of which was more frequent in the converted group. There was no difference in mortality (p = 0.315) or major morbidity (p = 0.767) (Table 4). Stepwise backward conditional logistic regression was performed (entry p < 0.05; removal p > 0.1) to identify predictors of PHLF. The presence of ascites, prior PVE, additional hepatectomies, and Pringle’s maneuver were all independent predictors of PHLF occurrence (Table 5). MIH patients, compared with OH, had three-times less likelihood of developing PHLF (HR 0.36 [0.16–0.80]).

Table 4 Comparison of 30-day outcomes between the matched groups of OH and converted MIH
Full size table
Table 5 Final block results of the backward stepwise conditional logistic regression for predictors of post-hepatectomy liver failure in the selected patient population
Full size table

Discussion

To our knowledge, this large database study represents the first study comparing MIH and OH in patients with early-stage cirrhosis, and the first with a specific focus on potentially devastating complications such as PHLF and bile leak. Our comparison demonstrated a higher rate of PHLF and bile leak in OH compared to the MIH approach. The overall rate of both complications was low, less than 5% respectively, yet important given the severity of these complications when they do occur. Interestingly, we noted a slightly higher rate of PHLF in cirrhotic patients with MELD ≤ 10 compared to past works, though the overall rate of PHLF remained below 5%.9,10 Our study, like others, has demonstrated a reduction in overall and major morbidity, and a reduced length-of-stay with MIH, though it is notable that overall LOS in our study was quite high for both arms.14,15,16 MIH cases that were converted to open were not associated with an increase in PHLF, bile leak or major morbidity, though overall morbidity was higher.

The nation- and world-wide rise of minimally invasive (laparoscopic or robotic) liver resection has generated much excitement over the potential of MIS in these generally high-risk patients.16,17,18 Previous work has indicated that MIH is safe and technically feasible in experienced hands, furthering widespread acceptance. For example, Fretland et al.’s randomized controlled trial demonstrated that overall complication rates are lower with MIH, estimated blood loss is reduced, mortality is equivalent or reduced, and there is no added cost to MIH.19 The safety of MIH overall is well established, though similar comparisons in cirrhotic patients, known to be a high-risk cohort, have been limited. Smaller studies by Kaneko (n = 39), Goh (n = 138) and Buell (n = 198) have generally supported the safety of MIH in the cirrhotic patient.11,13,20 The largest study to date by Sahara et al. (n = 481) demonstrated no increased risk of MIH in elderly cirrhotic patients but also did not show any reduction in complications with this approach.12 Conversely, our study demonstrates a reduction in overall and major complications with MIH in the cirrhotic population, with a resulting shorter hospitalization. Additionally, there were no complication categories included for which OH was superior to MIH. This study would support the ongoing use of MIH in patients with early-stage or low-MELD (MELD ≤ 10) cirrhosis, and we would recommend consideration of this approach when technically feasible. MELD > 11 has been shown to be an even worse prognostic factor after hepatectomy.10,21 Future work should consider the outcomes of MIH in higher-risk cirrhotic patients since it may carry a favorable PHLF, and morbidity profile compared to OH.22,23,24 Further, the grade of cirrhosis is correlated with worse outcomes, and Child’s Class C cirrhosis has been associated with exceptionally high rates of PHLF, though there is controversy as to what degree low-MELD or low-grade cirrhosis impacts the risk of PHLF.23 To our knowledge, this is the first study to specifically investigate the risk of PHLF and/or bile leak in the low-MELD cirrhotic population. While the nature of a database study precludes information regarding the choice of the surgical approach, it is reasonable to think that the minimally invasive approach is not only safe in cirrhotic patients with low-MELD but may also be preferred to reduce the risk of PHLF.

This study focused on primary liver cancers (HCC and IHCC), which are more frequently resected along anatomic boundaries versus metastatic lesions that are more commonly approached extra-anatomically.25,26,27 Given the prevalence of colorectal-cancer liver metastasis, many of the previous studies had been conducted in a traditionally extra-anatomic resection population.14,19,25,26 Parenchymal volume included in resection will vary vastly between these approaches, and thus these represent surgically very different populations with respect to PHLF. Therefore, our studies’ specific focus on PHLF after primary liver cancer resection in cirrhotic patients addresses an important topic that has not yet been fully investigated, though the comparative impact of MIH vs OH in patients with metastatic disease has also not been fully elucidated. Furthermore, this study included only partial hepatectomies as opposed to major hepatectomy. Previous work has shown that the number of resected segments is an independent predictor of overall and serious complications.28 It is possible, then, that the impact of MIH is even greater in the setting of major hepatectomy, though confounders such as operative difficulty preclude any definitive conclusions on this topic. Future research should focus on verifying our findings in primary cancers in a prospective manner, expanding this research to include major hepatectomy, and conduct a similar analysis in patients with secondary liver malignanc where the rates of neoadjuvant therapies and associated liver injuries are significantly higher.

There is controversy in the literature regarding the impact of the Pringle Maneuver on post-hepatectomy complications. Fagenson et. al. noted a detrimental impact in partial hepatectomy or in those with metastatic disease.29 Alternatively, Al-Saeedi et al. and Ortiz Galindo et al. both reported no difference in the rate of complications after the maneuver. Of note, the Al-Saeedi study specifically noted no increased risk of PHLF, though this study was not conducted in cirrhotic patients, while the Ortiz study did specifically assess PHLF in the cirrhotic patient.30,31 Contrary to these studies, we did find that Pringle’s Maneuver was associated with an increased risk of PHLF (HR 1.94, 95% CI 1.02 – 3.72). While a surgeon’s choice to utilize this maneuver must be taken on a case-by-case basis, surgeons should be cautioned to take this approach in the low-MELD cirrhotic.

Additional factors associated with PHLF in our study included presence of ascites (HR = 8.0, 95% CI = 27.42), Portal Vein Embolization (HR = 5.35, 1.01–8.73), and additional hepatectomy, with 2 + (HR = 4.02, 1.25–6.14) carrying higher risk than 1 additional hepatectomy (HR 2.12, 1.12–4.85). The pre-operative presence of untreated ascites carried the strongest risk of PHLF, which is consistent with prior studies showing this risk factor.4,22 While PVE was found to be a risk factor, this is a known strategy for prevention of PHLF, and thus this may have been undertaken preferentially in patients known to be at high risk of PHLF based on the nature of the planned resection. Additional hepatectomy is also known to increase the risk of PHLF, as is the extent of the surgical resection.21,32 Our identified risk factors for PHLF do differ slightly from the Dasari et. al. risk factors described in 2019.32 Liver surgeons should take caution when performing hepatectomy in patients with cirrhosis, particularly when those patients have associated ascites, and when multiple hepatectomies may be necessary.

By employing a propensity matched analysis of open cases versus MIH cases that converted to open, we attempted to address potential bias in the OH group. Previous studies have quoted a higher rate of morbidity and longer length of stay with conversion compared with MIH or OH.33,34 Our study did find a technical increase in overall morbidity, nearly all of which was accounted for by an increase in anemia requiring transfusion, which logically may indicate a high rate of conversion for bleeding. The overall similarity between outcomes in OH and Converted cases may indicate that it is the open procedure itself that leads to worse outcomes, versus another confounding factor that led to the initial selection for MIH as opposed to OH, though this cannot be stated definitively. While some rate of conversion is inherent to performing any minimally-invasive surgery, the incidence of morbidity in this study does underscore the need for good technical training and pre-operative selection to minimize the chance of conversion.

This study does have limitations. As mentioned, this is a nation-wide database study, and thus data is not available in granular detail. Thus, patient-specific factors that may contribute to the decision to pursue MIH vs OH cannot be ascertained. We attempted to account for this by analyzing converted cases, as these may represent cases that had pre-operative factors allowing for MIH but have underwent the physiology of an open operation, though we acknowledge this approach also has limits. Additional post-operative or oncologic variables that may play a part in the development of PHLF or bile leak are also unavailable. We attempted to match groups based on available clinically relevant characteristics, though the size of the tumor, number of tumors and amount of parenchymal resection was not available. The relatively large cohort may help balance these across groups, but this cannot be confirmed. The length-of-stay was quite high in both groups, especially MIH, given that these operations are frequently performed with short hospital stays. This may be secondary to the selected patient population as the study was only inclusive of patients with confirmed cirrhosis and may limit broader applicability of these findings.