The laparoscopic surgical approach has gained wide acceptance for the treatment of a variety of abdominal pathologies and for several liver resection procedures, mainly for benign diseases. The development of dissection techniques and equipment has allowed us to control anatomical structures such as intrahepatic bile ducts and vessels carefully, permitting safe performance of liver resections [1–4]. Laparoscopic liver resection (LLR) is claimed to be feasible and safe, yielding decreased postoperative pain and disability, reducing hospital stay and shortening patient recovery time. In the case of symptomatic and/or rapidly growing benign hepatic tumors, LLR is an interesting method for the resection of tumors. Major published series are, however, mainly biased by the heterogeneity of the procedures performed (deroofing of biliary cysts, polycystic liver disease, or excision biopsies) and the nature of the resected lesions (benign tumors, hepatocellular carcinoma, and/or metastases). With the evolution of imaging techniques and better understanding of the natural history of hepatocellular tumors, resection of liver cell adenomas (LCA) and focal nodular hyperplasia (FNH) has been progressively restricted. Generally, symptomatic, compressive, or enlarging FNH is considered for resection. Indication to resect LCA is mainly due to the risks of bleeding or the well-documented malignant transformation [5–9]. When surgical treatment is considered for selected patients with benign liver lesions, absence of postoperative morbidity and mortality are of paramount importance [4]. Postoperative morbidity in LLR is principally related to hemorrhaging from cutting edge, ascites (especially in cirrhotic patients or, in case of deroofing, of biliary cysts), and intra-abdominal collections, with rates ranging from 11 to 34%. Mortality is related to major resections and the presence of cirrhosis, with rates ranging between 0 and 5.9% (Table 1). To date there are few reports in the literature focusing on the laparoscopic treatment of benign liver tumors [2, 6, 10]. The aim of this study was to assess overall morbidity and outcome of laparoscopic liver resection of benign liver tumors by performing a case-control matched-pairs analysis with patients undergoing liver resection by open surgery (OS) for the same indication and as a consecutive series in a short time period.

Table 1 Published series of laparoscopic liver resection for solid benign tumors
Full size table

Patients and methods

Between January 2001 and January 2006 a total of 629 liver resections were performed at the Ghent University Hospital. The indication was solid benign liver tumor in 56 (8.9%) patients. Laparoscopic liver resection was performed in 20 (35.7%) cases. Inclusion criteria for LLR were: presumed benign solid liver tumors including LCA and FNH localized to the left and antero-lateral segments (IV to VI) of the liver, and ASA score of 1 or 2. A few cases of enlarging, bulging hepatic hemangiomas were also considered in this analysis. All data concerning consecutive patients undergoing liver resection for benign disease in our institution were collected retrospectively by chart review from January 2001 to January 2006. Data from the LLR group were compared with those from a consecutive control group undergoing open liver surgery for a similar indication in a matched-pair analysis during the same period with the endpoint of investigating overall morbidity and outcome of both procedures. The pairs were matched as closely as possible for age, gender, ASA score, indication for resection, and type and location of the lesions. Results were analyzed by recording blood transfusions, postoperative ileus (expressed by time to oral intake), complication rates, reason for conversion from LLR to OS, hospital stay, and late admission related to surgery. Following the principle of an intention-to-treat analysis, any statistical comparison was done between the laparoscopic group (including the conversions) versus the open group. Twenty patients fulfilled the selection criteria for the open surgery group. All procedures were performed under the direction of the same two attending surgeons (BdH and RT). Indication for surgical treatment was made during multidisciplinary conference. All patients had serum negative tumor markers α-fetoprotein, carcinoembryonic antigen, and carbohydrate antigen. Preoperative workup consisted of liver ultrasonography (US) abdominal evaluation in 37 (92.5%) patients, angio computed tomography (CT) in 34 (85%), magnetic resonance imaging (MRI) in 32 (80%), percutaneous biopsies in nine (22.5%), positron emission tomography (PET) scintigraphy in four (10%), and angiography in three (7.5%). Based on these evaluations, the preoperative indications for liver resection were: LCA, n = 21 (52.5%); FNH, n = 7 (7.5%); undetermined, n = 6 (15%); and hemangioma, n = 6 (15%). Most of the patients undergoing surgery had symptomatic lesions (15, or 75%, in OS group and 14, or 70%, in the LLR group). In the group of patients with LCA, the main indication for surgery included symptoms in 15 (71.4%) (painful palpable mass in 13 with a mean size of 8.5 ± 4 cm, intratumoral hemorrhage in two). In the FNH group the indication for resection was symptomatic mass in six (mean size of 8.7 ± 2.6 cm) and suspicion of metastasis during carcinological follow-up in the other patient. Finally, indication for resection of liver hemangiomas included increasing size of a painful tumor in three (mean size of 9.5 ± 4.2 cm), bulging hemangioma in two (12 and 15 cm) and atypical aspect in one. All patients signed a written informed consent. This study was approved by our hospital ethics committee (project number 2006/035).

Surgical technique

In patients undergoing LLR, a standard French position was required. Open laparoscopy technique was considered in patients with a history of previous abdominal surgery. Four to five trocars (10/12 mm) were usually inserted in the upper abdominal sector and the average carbon dioxide pneumoperitoneum abdominal pressure was around 12 mmHg (optical device 0°). All surgical procedures, either laparoscopic or open techniques, were routinely performed under intra-operative ultrasonography (IOUS) guidance (Aloka SSD 4000, Tokyo, Japan). Parenchymal transection was carried out using the laparoscopic surgical aspirator (CUSA- Excel TM, Ethicon Endosurgery, Cincinnati Ohio, USA). All patients underwent normovolemic hemodilution to a target hematocrit of 22–24% to minimize allogeneic blood transfusion in the intra- and perioperative (first three days) period. Hepatic pedicle clamping (Pringle’s maneuver), consisting of 10 min clamping and 10 min release, was performed with a titanium clamp (OS) or using a tourniquet (LLR). Hemostasis was achieved using titanium clips, an argon beamer coagulator, and an Ultracision dissector (Ethicon). Extraction of the surgical specimen in G1 was performed using an endobag through a Pfannenstiel incision (58%), by enlarging the trocar site (42%). Strasberg’s nomenclature for hepatic resection was used to describe the types of resections [11]. In case of left lateral sectionectomy and left hemihepatectomy (LLR), the vascular endoGIA 35 mm (Ethicon) was used for suturing the vascular pedicle. Cholecystectomy and cholangiography were carried out when indicated. Abdominal silicon drainage was left in place beside the cutting edge. In open surgery, parenchymal transection was exclusively performed by using the surgical aspirator (CUSA) under US guidance. Postoperative hepatic ischemic damage and protein synthesis were assessed based on the postoperative evolution of liver enzymes alanine aminotransferase (ALT) and prothrombin activity.

Statistical analysis

Results were expressed as mean ± standard deviation (SD). Comparisons between groups were analyzed using the chi-square test with Yates correction for continuity or Fisher’s test when appropriate. Differences were evaluated using the Student’s test, the Fisher’s exact test, or the Wilcoxon matched pairs test when appropriate. Statistical analysis was performed using the SPSS 12.0 for Windows program. The statistical significance level was set at α = 0.05.

Results

All patients but one (OS) are alive and well after a mean follow-up of 35 months (range 10–60 months) and all patients but one were female. According to the Strasberg’s nomenclature, liver resections included 13 wedge resections, 10 segmentectomies, 10 bisegmentectomies, four left lateral sectionectomies and three left hemihepatectomies. No differences were noticed between the OS and LLR groups with respect to patient characteristics, preoperative diagnosis, size of the resected lesions (with the larger lesions in this series of 15.8 cm in size) and operative procedures (Table 2). Portal triad clamping was used in 10 (50%) patients with an overall mean duration of 21.6 ± 9.5 min. Conversion to an open approach was required in two out of 20 (10%) cases due to uncontrolled bleeding during laparoscopic resection [two LCA, one 14.5 in size located in SII-III and the second, 7 cm in size, located in SV and close to the middle hepatic vein (MHV)]. According to the intention-to-treat analysis, there were no differences in operation time, heterologous blood transfusion, and total amount of transfused packed cells between the OS and LLR groups. Statistical significance was found in terms of length of postoperative ileus (intended as time to food intake) and length of hospital stay (Table 3). Perioperative bleeding episodes originated from major venous confluents. In one patient, the bleeding occurred at the end of an apparently successful laparoscopic resection of an LCA close to the MHV. Conversion to open laparotomy was made right away. However, hypovolemic shock led to bilateral cortex necrosis of both kidneys. This patient required hemodialysis filtration for four months and is currently dialysis free. Intra- and postoperative blood transfusion was required in six patients (five in OS and two in the LLR group). Overall complications are shown in Table 4. Incidence of incisional hernias (25%) was statistically significant between the two groups (p = 0.047) and a trend for higher cumulative morbidity rate was found in the OS group (45 vs. 20%) although this was not not statistically significant. Late readmission was needed in three out of five patients with incisional hernias (all with previous abdominal surgery) only in the OS group for surgical repair. Mean body mass index (BMI) was significantly higher in patients with incisional hernias compared to those without (28.3 ± 6.8 vs. 21.4 ± 2.9, p = 0.005). One late death was recorded following a metastatic cancer originating from a malignant inflammatory myofibroblastic tumor of the liver diagnosed as FNH. One episode of lung embolism was noticed in a patient in OS, which resolved with medical therapy. Occurrence of a biliary fistula was observed in one patient in the OS group. This complication resolved spontaneously with long-term drain maintenance. Incisional hernia occurred in four patients in the OS group (p = 0.10 versus LLR). Late readmission for surgery was recorded in three patients in OS for repair of a symptomatic incisional hernia. Final histological assessment showed that, despite preoperative evaluation with several diagnostic procedures, there was a large discrepancy in preoperative diagnosis of LCA compared with FNH. From the 21 (52.5%) diagnosed LCA, only 10 (10%) were confirmed as LCA, whereas almost all of the presumed FNH were found to be FNH (Table 5). The postoperative rise of ALT was higher and prothrombin time (%) was lower in the OS compared to the LLR group; however, no statistical correlation was found between the two groups (Fig. 1).

Table 2 Patient’s and procedure’s characteristics
Full size table
Table 3 Perioperative clinical
Full size table
Table 4 Overall complication associated to laparoscopic and open surgery
Full size table
Table 5 Pathological assessment
Full size table
Fig. 1
figure 1

Evolution of biochemical parameters

Full size image

Discussion

Since the introduction of laparoscopic cholecystectomy in 1987, laparoscopic techniques have been applied for solid organ surgery, including partial nephrectomy, prostatectomy, splenectomy, pancreatectomy, and hepatic resections. The first non-anatomical liver resection for FNH was described by Gagner et al. and the first left lateral lobectomy by Azagra et al. [7, 8]. Therefore, despite the widening of indications, and considering that hepatic surgery is basically characterized by resections without the need for reconstruction, the development of laparoscopic liver surgery remained limited during the time. This is probably due to the difficulty to reproduce laparoscopically basic maneuvers in open liver surgery (i.e., mobilization and rotation of the liver, vascular control), the risks of bleeding, pulmonary embolism and, finally, the fear of compromising oncological resection by tumoral seeding when resecting laparoscopically tumoral lesions [9]. Recently, different series of LLR have been reported, showing that laparoscopic resection of liver lesions is feasible, safe, and that LLR has a better postoperative outcome and shorter hospital stay than OS [2, 4, 10]. Biertho et al. reviewed 186 laparoscopic liver resections between 1991 and 2001, confirming the safety of LLR for minor hepatic resections in selected patients, with a morbidity and a mortality rate of 16 and 0.54%, respectively [12]. The safety and feasibility of LLR has been also suggested recently through a meta-analysis conducted on eight non-randomized studies: lower operative blood loss and reduced hospital stay appearing as the most important advantages. However, careful interpretation of the results is mandatory due to a significant selection bias (absence of retrospective randomized studies comparing OS vs. LLR) [13].

The aim of our study is to analyze the value of LLR for benign liver lesions through a case-control study. To our knowledge, this is the largest case-control study ever in the literature of surgical treatment of symptomatic benign liver disease comparing consecutive cases of laparoscopic versus open procedure in a short time period. Hepatic hemangioma and FNH are the most frequently diagnosed benign lesions, usually not representing an indication for resection except when compressing adjacent organs or when rapidly increasing in size [14–17]. LCA is in a different situation. Liver resection is usually indicated for its potential risk of bleeding, especially when the tumor exceeds 5 cm in diameter [18, 19]. Malignant transformation has also been described [17, 20–23]. In some cases, the differential diagnosis between liver adenomas and well-differentiated hepatocellular carcinomas is very difficult, while a conservative approach could facilitate neoplastic growth and intrahepatic dissemination in case of missed diagnosis. In our experience, pathological examination showed FNH in half of the patients with presumed adenoma. Differential diagnosis between LCA and FNH remains difficult to establish, even by combining different imaging modalities. The diagnostic accuracy for FNH ranges from 70 to 90% while it is approximately 90% for hepatic hemangiomas [16, 24–27]. We confirmed LCA histologically in only 40% of cases. In fact, we retained any undefined lesion as LCA when imaging did not unequivocally demonstrate the typical characteristics of an FNH. To improve diagnostic assessment, liver biopsy has claimed to be useful, increasing the accuracy of preoperative diagnosis. The large discrepancy we observed in diagnosing FNH might be lowered by performing routine preoperative biopsies. Its utility should be weighed against the risks of sampling errors, bleeding or misinterpretation (i.e., hyperplastic liver tissue interpreted as adenoma by imaging modalities) [17, 22, 26–29]. As expertise and experience in laparoscopic surgery increase, it is likely that laparoscopic resection will be considered more frequently when diagnostic uncertainty exists [30]. Uncontrolled bleeding is one of the major risks in LLR and this occurred twice in our experience indicating conversion laparotomy. The major risk factor for bleeding is usually related to the size of the lesions but a learning curve factor should also be considered [31–33]. When considering more-recent reports (i.e., after 2003) or studies reporting on 20 or more laparoscopic procedures, a significant reduced operative blood loss, a shorter duration of hospital stay, and a significant decrease in the complications related to liver resection have been shown [13]. Careful dissection by the magnified view of laparoscopy and the hemostatic property of the pneumoperitoneum are also playing a role decreasing blood loss [33]. Lesions in contact with the major hepatic veins are at risk of bleeding and gas embolism [34]. The only two episodes of gas embolism described in the literature have indeed been associated to the large use of the argon beamer coagulator [12]. Parenchymal hemostasis using the Ultracision® and/or bipolar coagulation avoiding argon beamer may reduce the risks of embolism [35]. The use of a surgical aspirator may be proposed when approaching vascular structures in the depth of liver parenchyma and safe hemostasis can be achieved with the use of vascular staplers [34]. In our converted cases, hemostasis achieved initially with the use of clips was probably unsafe, and the bilateral cortex necrosis that led to hemodialysis filtration probably resulted from the combination of low central vein pressure (by pneumoperitoneum) and the need for vasopressor agents. Therefore we do share the opinion of others that careful dissection of deep venous structures by the use of a harmonic scalpel and/or endoscopic linear stapler devices may be preferable to avoid uncontrolled bleeding and pitfalls during liver dissection [32–34]. Our overall complication rate was 20%, which is similar to the average rates reported in the literature and reflects the intention-to-treat basis of the analysis, which included the bleeding episodes with conversion to OS and the acute kidney injury observed in one patient. The incidence of incisional hernia was significantly higher in the OS group. This could be explained by the influence of previous upper abdominal scars and/or by the BMI, which was found significantly to be higher in those patients presenting this complication [35]. However, it is remarkable that, despite laparoscopically resected patients presenting more laparotomy scars, incisional hernias were not recorded in this last group. Laparoscopic liver surgery is an implemented procedure in several hepatobiliary programs worldwide and may be considered as safe as OS for selected patients. The lower operative blood loss and shortened hospital stay has potentially significant implications in terms of overall morbidity and health care costs. Moreover, previous upper abdominal surgeries did not jeopardize overall outcome with a potential advantage of the lower incidence of incisional hernias, fewer adhesions, and minimal scarring [2, 6, 35, 36]. The risk of bleeding should be taken into account and weighed against the benefit of LLR, especially when considering the resection of benign liver tumors. We believe that, in the hands of expert surgeons, LLR may become the gold standard in cases of LCA, symptomatic FNH or undefined lesions located to the anterior and lateral sectors and for minor hepatic resections. The widening of its use is principally limited by its difficultness and the need for specific surgical training. In our opinion, the better outcome with LLR as compared to open surgery should not lead to an extension of the indications and does not justify liver resection as a more-appropriate diagnostic test than history, physical, blood tests, and good-quality imaging.