Original article
Learning curve for laparoscopic major hepatectomy T. Nomi, D. Fuks, Y. Kawaguchi, F. Mal, Y. Nakajima and B. Gayet Department of Digestive Disease, Institut Mutualiste Montsouris, Université Paris-Descartes, 42 Boulevard Jourdan, 75014 Paris, France Correspondence to: Professor B. Gayet (e-mail: [email protected]) Background: Laparoscopic major hepatectomy (LMH) is evolving as an important surgical approach in
hepatopancreatobiliary surgery. The present study aimed to evaluate the learning curve for LMH at a single centre. Methods: Data for all patients undergoing LMH between January 1998 and September 2013 were recorded in a prospective database and analysed. The learning curve for operating time (OT) was evaluated using the cumulative sum (CUSUM) method. Results: Of 173 patients undergoing major hepatectomy, left hepatectomy was performed in 28 (16⋅2 per cent), left trisectionectomy in nine (5⋅2 per cent), right hepatectomy in 115 (66⋅5 per cent), right trisectionectomy in 13 (7⋅5 per cent) and central hepatectomy in eight (4⋅6 per cent). Median duration of surgery was 270 (range 100–540) min and median blood loss was 300 (10–4500) ml. There were 20 conversions to an open procedure (11⋅6 per cent). Vascular clamping was independently associated with conversion on multivariable analysis (hazard ratio 5⋅95, 95 per cent c.i. 1⋅24 to 28⋅56; P = 0⋅026). The CUSUMOT learning curve was modelled as a parabola (CUSUMOT = 0⋅2149 × patient number2 − 30⋅586 × patient number − 1118⋅3; R2 = 0⋅7356). The learning curve comprised three phases: phase 1 (45 initial patients), phase 2 (30 intermediate patients) and phase 3 (the subsequent 98 patients). Although right hepatectomy was most common in phase 1, a significant decrease was observed from phase 1 to 3 (P = 0⋅007) in favour of more complex procedures. Conclusion: The learning curve for LMH consisted of three characteristic phases identified by CUSUM analysis. The data suggest that the learning phase of LMH included 45 to 75 patients. Paper accepted 4 February 2015 Published online 15 April 2015 in Wiley Online Library (www.bjs.co.uk). DOI: 10.1002/bjs.9798
Introduction
Since the first successful report of a laparoscopic liver wedge resection for a benign lesion in 19911 , diffusion of laparoscopic liver resection (LLR), particularly laparoscopic major hepatectomy (LMH), has remained limited because of technical difficulties, fear of haemorrhage and embolism, and concern for oncological inadequacy. Although many studies have suggested the technical and oncological safety of LLR2 – 6 , this technique is still confined to a few expert centres and its reproducibility remains doubtful. Buell and colleagues7 reported at the Louisville Consensus Conference in 2008 that major hepatectomy should be performed by expert surgeons experienced in both open and laparoscopic liver resection. These data were essentially provided from small series of patients or multicentre studies that may have included some biases8,9 . Very few studies reporting large experiences from single institutions have been undertaken. More evidence is certainly required, including the assessment of both short- and long-term outcomes. Laparoscopic atypical © 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
liver resection has been developed at the Institut Mutualiste Montsouris since 1992, with the first laparoscopic major liver resections in 19982 . In the 15 years since the first LMH, more than 170 procedures have been performed, and the technique for LMH has gradually been standardized. The cumulative sum (CUSUM) technique is a method originally devised for monitoring performance and detecting areas for improvement in the industrial sector. This method was adopted by the medical profession in the 1970s to analyse the learning curve for surgical procedures10,11 , particularly laparoscopic procedures. CUSUM analysis transforms raw data into the running total of data deviations from the group mean, enabling investigators to visualize the data for trends not discernible with other approaches. The same effect should be expected in LMH; however, at present, no specific data are available. The present study was designed to evaluate the learning curve effect for LMH using CUSUM methodology to define its evolution in terms of feasibility and reproducibility. BJS 2015; 102: 796–804
Learning curve in laparoscopic major hepatectomy
Methods
All consecutive patients who underwent LMH between January 1998 and September 2013 at the Institut Mutualiste Montsouris in Paris were selected for the study. Major hepatectomy was defined as the resection of three or more contiguous Couinaud segments12 . Suitability for the laparoscopic approach was based on tumour size and location, type of planned resection, and patient co-morbidities. Bilobar procedures, including wedge resection or radiofrequency ablation in the contralateral liver, were also included. In all patients considered for LMH, future remnant liver volume was determined by liver volumetry using three-dimensional CT volumetry. Hepatic function was assessed before surgery with liver enzyme tests (serum bilirubin level, prothrombin time, platelet count) and imaging (absence of portal hypertension, spleen size). The indocyanine green retention test was not used routinely. Patients under consideration for major resection underwent portal vein embolization (PVE) when the future remnant liver volume was 25 per cent or less in normal liver (fewer than 6 cycles of chemotherapy) and 40 per cent or less in advanced fibrosis or cirrhosis (Child grade A with no portal hypertension), or in patients who had more than six cycles of chemotherapy. Resection was not considered if a sufficient liver volume could not be attained even after PVE. Selection of the laparoscopic approach did not depend on underlying liver disease but was considered each time the indication for resection was decided in the multidisciplinary staff meeting. This study was approved by the institutional review board.
Surgical procedures The surgical technique of LMH, including positioning of the trocars, has been described previously13 – 17 . All resections were performed with curative intent. The types of hepatectomy were classified according to the Brisbane 2000 terminology18 . Over time and with increasing surgical experience, the approach to LMH changed from posterior to anterior. Laparoscopic ultrasonography was performed routinely to confirm the number and size of lesions, and their relationship with the intrahepatic vascular structures. An intermittent Pringle manoeuvre was used only in cases of failure of bleeding control. For all procedures, tissue dissection and haemostasis were performed by the ultrasonic dissector, mainly the SonoSurg™ or Thunderbeat® (Olympus, Tokyo, Japan) or Harmonic® Scalpel® (Ethicon Endo-Surgery, Cincinnati, Ohio, USA); the Gayet bipolar forceps (MicroFrance CEVBG134; Medtronic, Minneapolis, Minnesota, USA) provided retraction and rescue haemostasis. All operations © 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
797
were performed by the senior surgeon. Conversion was defined as the requirement for laparotomy at any time during the procedure, with the exception of extraction of the resected specimen.
Postoperative outcomes Postoperative complications were stratified according to the Dindo–Clavien classification19 , which defines major complications by a score of 3 or more. If the patient had two or more postoperative complications, the most severe one was recorded. The specific liver complications that were encountered more often after major liver procedures were detailed as follows: liver failure was defined according to the ‘50–50 criteria’ on postoperative day 520 ; ascites was defined as an abdominal drainage output of more than 10 ml per kg per day after the third postoperative day21 ; and biliary leakage was defined by a bilirubin concentration in the drainage fluid more than threefold greater than that in the serum22 . Both complications and operative mortality were recorded as events occurring within 90 days of surgery or at any time during the postoperative hospital stay. Surgical margins were classified as either negative (R0) or positive (R1).
Learning curve analysis Two analyses were performed. In the first analysis, the CUSUM technique was used for quantitative assessment of the learning curve. The CUSUM technique was used when data on duration of surgery (defined as the time from first incision to final closure), blood loss (estimated by anaesthetists), vascular clamping, complications and length of stay were available. First, the patients were ordered chronologically, from the earliest to the latest date of surgery. Data for each patient in the series were plotted on a chart from left to right on the horizontal axis. The line ascended for every patient who was operated on within a certain operating time (OT), and descended for every patient in whom the operation took longer than the defined cut-off. Each patient included in the CUSUMOT analysis shows the real time required for the operation minus the expected time to complete the procedure, in chronological order. CUSUMOT of the first patient was the difference between OT for the first patient and the mean OT for all patients. CUSUMOT of the second patient was CUSUMOT of the previous patient plus the difference between OT for the second patient and the mean OT. This recursive process was continued until CUSUMOT of the last patient was calculated as zero. In the second analysis, learning curve analysis was based on the conversion rate. Univariable analysis of the factors predictive of conversion was performed; the comparison of conversion rates into three consecutive www.bjs.co.uk
BJS 2015; 102: 796–804
798
T. Nomi, D. Fuks, Y. Kawaguchi, F. Mal, Y. Nakajima and B. Gayet
25
No. of LMH procedures
20
15
10
5
0
1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year
Fig. 1
Number of laparoscopic major hepatectomy (LMH) procedures performed per year
groups of patients was then adjusted using multivariable logistic analysis.
Statistical analysis Patient baseline characteristics are expressed as median (range) for continuous data and as numbers with percentages for categorical data. Fisher’s exact test was used to compare differences in categorical variables, and the Wilcoxon rank sum test for continuous variables. Variables achieving statistical significance at the 0⋅1 level in univariable analysis were considered for multivariable analysis. A backward variable procedure was used to identify independent predictive factors. All statistical analyses were performed using SPSS® for Windows® version 20⋅0 (IBM, Armonk, New York, USA), and statistical significance was set at the 5 per cent level. Results
Among the 173 selected patients, there were 104 men (60⋅1 per cent) and 69 women (39⋅9 per cent) with a median age of 63 (range 24–86) years. The median number of LMH resections was 11 per year, with a maximum of 20 in 2008 (Fig. 1). The indication for liver resection was malignancy in 156 patients (90⋅2 per cent) and benign disease in 17 (9⋅8 per cent). In total, 106 patients (61⋅3 per cent) had undergone abdominal surgery previously, including 31 hepatectomies (17⋅9 per cent).
Surgical procedures and pathological specimens The types of major hepatectomy included left-sided hepatectomy in 37 patients (21⋅4 per cent), right-sided © 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
hepatectomy in 128 (74⋅0 per cent) and central hepatectomy in eight (4⋅6 per cent). Vascular clamping was required in 18 patients (10⋅4 per cent). Median duration of surgery was 270 (range 100–540) min. Median blood loss was 300 (10–4500) ml, with a loss of more than 500 ml in 45 patients (26⋅0 per cent) and more than 1000 ml in 17 (9⋅8 per cent). Eighteen patients (10⋅4 per cent) needed an intraoperative transfusion.
Postoperative outcomes Five patients died during the early postoperative period, giving an overall rate of 2⋅9 per cent. Major complications occurred in 43 patients (24⋅9 per cent), including biliary leakage in 23 (13⋅3 per cent), liver failure in four (2⋅3 per cent), intra-abdominal abscess in four (2⋅3 per cent), ascites in three (1⋅7 per cent) and intra-abdominal bleeding in two (1⋅2 per cent). Median hospital stay was 9 (range 3–57) days, and median follow-up 25 (6–83) months.
Learning curve and comparison of the three consecutive phases The raw operating times were plotted in chronological order (Fig. 2a). The CUSUMOT learning curve was best modelled as a second-order polynomial (parabola) with the following equation: CUSUMOT (in minutes) = 0⋅2149 × patient number2 − 30⋅586 × patient number − 1118⋅3. This had a high R2 value of 0⋅7356 (Fig. 2b). The CUSUMOT learning curve comprised three unique phases: phase 1 (the initial 45 patients), phase 2 (the middle 30 patients) and phase 3 (the final 98 patients) (Fig. 2b). Comparisons between the three phases of various www.bjs.co.uk
BJS 2015; 102: 796–804
Learning curve in laparoscopic major hepatectomy
799
Operating time (min)
600 500 400 300 200 100 0
1 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 105 110 115 120 125 130 135 140 145 150 155 160 165 170 175
Phase 1
Phase 2
January 1998
a
December 2005
Phase 3 September 2013
March 2008
Operating time versus no. of patients No. of patients 0
1 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95100 105 110 115120125130135140145 150 155 160 165 170175
CUSUMOT (min)
–500 –1000 –1500 –2000 –2500 –3000 Phase 1 January 1998
b
Phase 3
Phase 2 December 2005
March 2008
September 2013
CUSUMOT versus no. of patients
Operating time (min)
600 500 400 300 200 100 0
1
5
10
15
20
25
Phase 1 January 1998
c
30
35
40
45
50
55
60
65
70
75
Phase 2 December 2005
80
85
90
95
100
105
110
115
Phase 3 March 2008
September 2013
Operating time versus no. of patients having laparoscopic right hepatectomy
a Evolution of operating time (OT) plotted against number of patients for all laparoscopic major hepatectomy (LMH) procedures. b Cumulative sum (CUSUM)OT plotted against number of patients. The black line represents the curve of best fit for the plot (a second-order polynomial using the equation CUSUMOT = 0⋅2149 × patient number2 − 30⋅586 × patient number − 1118⋅3; R2 = 0⋅7356). c Evolution of operating time plotted against number of patients who had laparoscopic right hepatectomy
Fig. 2
© 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
www.bjs.co.uk
BJS 2015; 102: 796–804
800
Table 1
T. Nomi, D. Fuks, Y. Kawaguchi, F. Mal, Y. Nakajima and B. Gayet
Interphase comparisons of characteristics of the 173 patients undergoing laparoscopic major liver resection P‡
Preop. characteristics Age (years)* Male sex Body mass index (kg/m2 )* Child grade B–C Diagnosis Malignancy CRLM HCC Cholangiocarcinoma Benign lesion Tumour characteristics Tumour number* Tumour size (mm)* Previous abdominal operation Hepatectomy Surgical procedures Type of hepatectomy Left hepatectomy Right hepatectomy Left trisectionectomy Right trisectionectomy Central hepatectomy Additional hepatic procedures Radiofrequency ablation Wedge resection Combined resection of adjacent organs Abdominal drainage Postop. outcomes Mortality Morbidity Major morbidity† Background liver disease Steatosis (>60%) Cirrhosis Reoperation
Phase 1 (n = 45)
Phase 2 (n = 30)
Phase 3 (n = 98)
Phase 1 versus 2
Phase 2 versus 3
Phase 1 versus 3
59 (27–85) 31 (69) 23⋅3 (18⋅4–29⋅0) 1 (2)
60 (24–84) 19 (63) 24⋅2 (22⋅5–26⋅4) 1 (3)
67 (32–86) 54 (55) 25⋅1 (15⋅9–35⋅5) 2 (2)
0⋅245§ 0⋅627 0⋅500§ 1⋅000
0⋅861§ 0⋅528 0⋅500§ 0⋅554
0⋅048§ 0⋅143 0⋅437§ 1⋅000
38 (84) 26 (58) 7 (16) 4 (9) 7 (16)
23 (77) 14 (47) 7 (23) 0 (0) 7 (23)
95 (97) 75 (77) 6 (6) 12 (12) 3 (3)
0⋅546 0⋅357 0⋅546 0⋅145 0⋅546
0⋅001 0⋅003 0⋅012 0⋅067 0⋅001
0⋅011 0⋅029 0⋅112 0⋅776 0⋅011
1 (1–4) 40 (5–170) 28 (62) 6 (13)
2 (1–5) 50 (15–160) 12 (40) 4 (13)
2 (1–20) 35 (5–150) 66 (67) 21 (21)
0⋅197§ 0⋅416§ 0⋅097 1⋅000
0⋅393§ 0⋅391§ 0⋅010 0⋅434
0⋅034§ 0⋅676§ 0⋅573 0⋅357
6 (13) 37 (82) 0 (0) 1 (2) 1 (2)
8 (27) 20 (67) 1 (3) 0 (0) 1 (3)
14 (14) 58 (59) 8 (8) 12 (12) 6 (6)
0⋅225 0⋅168 0⋅400 1⋅000 1⋅000
0⋅164 0⋅525 0⋅684 0⋅067 1⋅000
1⋅000 0⋅007 0⋅056 0⋅063 0⋅433
7 (16) 6 (13) 4 (9) 12 (27)
5 (17) 10 (33) 4 (13) 5 (17)
7 (7) 23 (23) 12 (12) 23 (23)
1⋅000 0⋅047 0⋅706 0⋅403
0⋅150 0⋅340 1⋅000 0⋅614
0⋅135 0⋅185 0⋅776 0⋅680
1 (2) 30 (67) 10 (22)
1 (3) 18 (60) 8 (27)
3 (3) 51 (52) 25 (26)
1⋅000 0⋅627 0⋅783
1⋅000 0⋅531 1⋅000
1⋅000 0⋅107 0⋅834
11 (24) 1 (2) 1 (2)
7 (23) 1 (3) 1 (3)
27 (28) 3 (3) 4 (4)
1⋅000 1⋅000 1⋅000
0⋅814 1⋅000 1⋅000
0⋅838 1⋅000 1⋅000
Values in parentheses are percentages unless indicated otherwise; *values are median (range). †Postoperative complications were stratified according to the Dindo–Clavien classification14 . CRLM, colorectal liver metastases; HCC, hepatocellular carcinoma. ‡Fisher’s exact test, except §Wilcoxon rank sum test.
parameters that were identified by CUSUMOT analysis are presented in Table 1. Patient age was higher in phase 3 than in phase 1 (P = 0⋅048). No difference in terms of male sex, body mass index or American Society of Anesthesiologists fitness grade (median 2 for each phase; P = 0⋅874) was found among the three groups. The proportion of patients who had undergone previous abdominal surgery increased between phases 2 and 3 (P = 0⋅010). Considering indications for surgery, the rate of colorectal liver metastases had increased significantly by later years, from 58 to 77 per cent (P = 0⋅029) and hepatocellular carcinoma decreased significantly between phases 2 and 3 (P = 0⋅012). Twenty patients underwent associated laparoscopic extrahepatic resection: seven diaphragmatic
resections, six common bile duct resections, two inferior vena cava resections, one duodenal resection, one pancreatic resection, one colonic resection, one adrenal gland resection and one kidney resection, with no difference over time. The need for pedicle clamping decreased from eight (18 per cent) in phase 1 to five (5 per cent) in phase 3 (P = 0⋅024); whenever clamping was required, the clamping duration became progressively shorter (from 45 to 10 min). Similarly, the median duration of surgery and blood loss decreased significantly between phases 1 and 3 (from 360 to 240 min, and from 500 to 200 ml, respectively). Moreover, a reduction in the conversion rate was observed in phase 3 compared with phase 1 (P = 0⋅037) and phase 2 (P = 0⋅033). A trend toward
© 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
www.bjs.co.uk
BJS 2015; 102: 796–804
Learning curve in laparoscopic major hepatectomy
801
decreased morbidity was observed (P = 0⋅107). Hospital stay was also slightly shorter during phase 3 (9 days) than in phase 1 (11 days) (P = 0⋅046).
Comparison of standard (right or left hepatectomy) and more complex (extended right or left hepatectomy and central hepatectomy) laparoscopic major hepatectomy
Table 2
More complex Standard hepatectomy hepatectomy (n = 30) (n = 143) Intraoperative data Combined resection of adjacent organs Duration of operation (min)* Blood loss (ml)* Transfusion Use of Pringle manoeuvre Conversion Abdominal drainage Postoperative outcomes Mortality Morbidity Major morbidity† Bile leakage Reoperation Length of hospital stay (days)*
Specific learning curve and outcomes of right hepatectomy
P‡
15 (10⋅5)
5 (17)
0⋅349
270 250 15 (10⋅5) 12 (8⋅4) 14 (9⋅8) 30 (21⋅0)
280 400 3 (10) 6 (20) 6 (20) 10 (33)
0⋅367§ 0⋅567§ 1⋅000 0⋅092 0⋅122 0⋅157
3 (2⋅1) 82 (57⋅3) 30 (21⋅0) 16 (11⋅2) 5 (3⋅5) 9 (6–48)
2 (7) 17 (57) 13 (43) 8 (27) 1 (3) 11 (5–57)
0⋅207 1⋅000 0⋅018 0⋅039 1⋅000 0⋅914§
Values in parentheses are percentages unless indicated otherwise; *values are median (range). †Postoperative complications were stratified according to the Dindo–Clavien classification14 . ‡Fisher’s exact test, except §Wilcoxon rank sum test.
Table 3
Right hepatectomy was the most common laparoscopic procedure (115 patients; 66⋅5 per cent) (Table 1). However, a decrease in the number of right hepatectomies was observed from phase 1 to 3 (P = 0⋅007). On the other hand, the number of extended procedures such as trisectionectomies tended to increase in phase 3. When comparing standard (right or left hepatectomy) with more complex procedures (extended right or left hepatectomy and central hepatectomy), higher rates of major complications (21⋅0 versus 43 per cent respectively; P = 0⋅018) including bile leakage (11⋅2 versus 27 per cent; P = 0⋅039) were observed in patients undergoing more complex hepatectomy (Table 2). In the subgroup of 115 patients having a right hepatectomy, there were nine conversions, with a decrease in later years (11, 10 and 5 per cent in phases 1, 2 and 3) (Table 3). The morbidity rate (65, 59 and 38 per cent) decreased significantly whereas length of stay (10, 9 and 9
Interphase comparisons of characteristics of the 115 patients undergoing laparoscopic right hepatectomy P†
Preop. characteristics Age (years)* Male sex Body mass index (kg/m2 )* ASA fitness grade ≥ II Tumour characteristics Tumour number* Tumour size (mm)* Previous abdominal operation Previous hepatectomy Preop. PVE Surgical procedures Combined resection of adjacent organs Duration of surgery (min)* Blood loss (ml)* Transfusion Use of Pringle manoeuvre Conversion Abdominal drainage Postop. outcomes Length of hospital stay (days)*
Phase 1 (n = 37)
Phase 2 (n = 20)
Phase 3 (n = 58)
Phase 1 versus 2
Phase 2 versus 3
Phase 1 versus 3
59 (27–85) 25 (68) 23⋅3 (20⋅1–29⋅0) 30 (81)
59 (35–84) 13 (65) 23⋅6 (22⋅7–26⋅4) 18 (90)
66 (47–86) 33 (57) 25⋅4 (16⋅4–33⋅1) 45 (78)
0⋅432‡ 0⋅579 0⋅892‡ 1⋅000
0⋅083‡ 0⋅877 1⋅000‡ 0⋅922
0⋅020‡ 0⋅512 1⋅000‡ 1⋅000
1 (1–4) 42 (5–130) 23 (62) 6 (16) 8 (22)
3 (1–5) 40 (10–140) 7 (35) 4 (20) 5 (25)
2 (1–8) 35 (7–120) 35 (60) 16 (28) 9 (16)
0⋅051‡ 0⋅600‡ 0⋅032 1⋅000 1⋅000
0⋅919‡ 0⋅859‡ 0⋅022 0⋅402 0⋅521
0⋅027‡ 0⋅510‡ 1⋅000 0⋅216 0⋅586
4 (11) 345 (180–540) 600 (10–4500) 4 (11) 7 (19) 4 (11) 10 (27)
3 (15) 210 (180–480) 300 (50–2000) 3 (15) 2 (10) 2 (10) 4 (20)
7 (12) 240 (100–480) 185 (10–1500) 5 (9) 1 (2) 3 (5) 9 (16)
1⋅000 0⋅004‡ 0⋅174‡ 1⋅000 0⋅461 1⋅000 0⋅537
1⋅000 0⋅407‡ 1⋅000‡ 0⋅680 0⋅190 0⋅617 1⋅000
1⋅000 < 0⋅001‡ 0⋅002‡ 1⋅000 0⋅006 0⋅430 0⋅293
10 (5–35)
9 (5–38)
9 (5–50)
0⋅856‡
0⋅231‡
0⋅098‡
Values in parentheses are percentages unless indicated otherwise; *values are median (range). ASA, American Society of Anesthesiologists; PVE, portal vein embolization. †Fisher’s exact test, except ‡Wilcoxon rank sum test.
© 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
www.bjs.co.uk
BJS 2015; 102: 796–804
802
Table 4
T. Nomi, D. Fuks, Y. Kawaguchi, F. Mal, Y. Nakajima and B. Gayet
Risk factors for conversion to laparotomy Univariable analysis No conversion to laparotomy (n = 153)*
Age > 70 years Male sex Body mass index (kg/m2 ) ASA fitness grade > II Malignancy Tumour diameter > 5 cm Previous abdominal operation Prehepatectomy chemotherapy Type of hepatectomy Left hepatectomy Right hepatectomy Left trisectionectomy Right trisectionectomy Central hepatectomy Additional hepatic procedures Radiofrequency ablation Wedge resection Combined resection of adjacent organs Blood loss > 500 ml Intraoperative transfusion Use of Pringle manoeuvre Steatosis > 60% Cirrhosis
Conversion to laparotomy (n = 20)*
Multivariable analysis
P‡
47 (30⋅7) 94 (61⋅4) 17 (11⋅1) 116 (75⋅8) 140 (91⋅5) 62 (40⋅5) 99 (64⋅7) 65 (42⋅5)
5 (25) 10 (50) 0 (0) 17 (85) 16 (80) 5 (25) 7 (35) 9 (45)
0⋅796 0⋅341 0⋅224 0⋅572 0⋅114 0⋅226 0⋅014 1⋅000
23 (15⋅0) 105 (68⋅6) 7 (4⋅6) 12 (7⋅8) 6 (3⋅9)
5 (25) 10 (50) 2 (10) 1 (5) 2 (10)
0⋅328 0⋅129 0⋅288 1⋅000 0⋅232
16 (10⋅5) 35 (22⋅9) 14 (9⋅2) 35 (22⋅9) 13 (8⋅5) 12 (7⋅8) 39 (25⋅5) 4 (2⋅6)
3 (15) 4 (20) 6 (30) 10 (50) 5 (25) 6 (30) 6 (30) 1 (5)
0⋅464 1⋅000 0⋅015 0⋅014 0⋅039 0⋅008 0⋅786 0⋅463
Odds ratio†
P§
0⋅31 (0⋅08, 1⋅18)
0⋅085
1⋅65 (0⋅27, 9⋅99) 1⋅88 (0⋅34, 10⋅36) 2⋅61 (0⋅43, 15⋅85) 5⋅95 (1⋅24, 28⋅56)
0⋅588 0⋅470 0⋅296 0⋅026
Values in parentheses are *percentages and †95 per cent c.i. ASA, American Society of Anesthesiologists. ‡Fisher’s exact test; §backward variable analysis.
days respectively) was stable over time. The percentage of R0 resection was also stable over the different phases (97 per cent for all phases).
Conversion to open surgery: risk factors and learning curve Twenty patients (11⋅6 per cent) required conversion to open surgery, because of haemorrhage (11 patients), oncological reasons (6) and lack of progress (3). The learning curve adjusted for the risk factors of conversion demonstrated that the rate of conversion to open surgery decreased in later years (18, 20 and 6 per cent in phases 1, 2 and 3 respectively). Univariable analysis showed that previous abdominal surgery, resection of adjacent organs, blood loss greater than 500 ml, intraoperative transfusion and vascular clamping were associated with a significantly higher risk of conversion (Table 4). Only vascular clamping was independently associated with conversion on multivariable analysis (odds ratio 5⋅95, 95 per cent c.i. 1⋅24 to 28⋅56; P = 0⋅026). Discussion
The learning curve is defined as the improvement in performance over time or the change in the ability to complete © 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
a task until failure is reduced to a constant minimum acceptable rate23 – 25 . The learning curve of a pioneer surgeon who is technically proficient or has advanced training in complex minimally invasive surgery techniques will be different from that of a novice surgeon25 . The level of both laparoscopic and liver surgical experience at the beginning of LMH is essential when evaluating the effect of the learning curve. Given that LMH is a technically demanding procedure, the surgeon should have advanced previous experiences in both open liver resection and laparoscopic procedures, such as upper gastrointestinal and colorectal surgery, before starting LMH. Although some articles26,27 have reported improvement of results during the second half of surgeons’ experience in laparoscopic minor hepatectomy, the learning curve has never been studied for LMH. In the present study, according to the CUSUM plot, data were categorized into three unique groups to confirm the learning curve effect on LMH: phase 1 (45 initial patients), representing the initial learning curve; phase 2 plateau (30 middle patients), representing increased competence with laparoscopy; and phase 3 (the subsequent 98 patients), representing the mastery phase in which more complex procedures were undertaken. Despite the expansion of indications for LMH and the increasing complexity www.bjs.co.uk
BJS 2015; 102: 796–804
Learning curve in laparoscopic major hepatectomy
803
of the resections performed, the surgical outcomes, including blood loss and duration of surgery, improved over time. A better understanding of the technical difficulties, and concerns regarding life-threatening operative risks of LMH, is imperative to carry out these procedures successfully. Given that nearly 95 per cent of all hepatectomies are performed laparoscopically in the authors’ institution, it should be taken into account that the present results might not be reproducible in every institution, or for all surgeons. Indeed, it is important to keep in mind that non-expert surgeons should start LLR with simple procedures, with a gradual increase in technical demand. Laparoscopy was considered for almost all major hepatectomies in the authors’ institution. The present study clearly suggests a learning curve effect due to technique improvement and standardization, especially for parenchymal transection, which represents the most difficult part of LLR. During the 15-year interval starting in 1998 and ending in 2013, many new devices have been developed including ultrasonic scalpels, sealing devices, coagulation systems and staplers, and these have facilitated expansion of the indications. In the authors’ institution, CUSA® (ValleyLab, Boulder, Colorado, USA) has never been used, and tissue dissection and haemostasis were performed with an ultrasonic dissector, first SonoSurg™, then Harmonic® Scalpel® and finally Thunderbeat®, with frequent use of the Gayet bipolar forceps. Even though these devices do not seem to improve the quality of transection in open surgery, difficulty in controlling potential bleeding during parenchymal transection warrants their use in laparoscopy. Nevertheless, the present study shows that the speed and ease of transection improves with experience, with a significant decrease in the duration of surgery over time. Given that conversion to open hepatectomy is associated with longer procedure times and a potentially increased overall morbidity, the ability accurately to identify preoperative risk factors for conversion may improve the care of patients undergoing LMH. In the present study, the conversion rate decreased progressively to 6 per cent in phase 3, which is relatively low compared with that in previous reports of LMH8,9,28 . In the present series, haemorrhage, unclear limits of tumour extension and failure to progress represented the three major reasons for conversion to open hepatectomy. Even though vascular clamping was performed mainly when blood loss was excessive during parenchymal transection, multivariable analysis identified vascular clamping to be associated with conversion. This result indicates the need to use pedicle clamping more widely, and earlier, during parenchymal transection. The effect of conversion on postoperative outcomes during LMH remains unclear. Unlike most
studies, no significant decrease in complication rates with increasing experience was observed. The complication rate may depend more on the selection of patients than on technical skills. In the present study, the main reason for conversion was bleeding in the early era; however, oncological reasons (tumour extent, the need for biliary reconstruction) increased during later years. A conversion rate of zero should not be considered a realistic goal, although an institution’s policy on conversion could be an important key to this problem. The present study indicated that 45 LMH procedures are required in order to reduce operating time, and to move from phase 1 to phase 2. As right and left hepatectomies represented more than 95 per cent of hepatectomies, it would make sense to start LMH with standard right or left hepatectomy. Indeed, the learning curve in LMH should start with anatomical resection completely mastered by an open approach. The surgeon should probably select patients with limited tumour load (for example in phase 1 a single tumour with a 4-cm diameter), low co-morbidities and no overweight. Additionally, the surgeon might be helped by a new technique of laparoscopic right hepatectomy29 , allowing low blood loss and morbidity. Finally, phase 3 of the learning curve included a higher rate of extended or atypical major hepatectomy, which may explain the pronounced lack of reproducibility for the procedure. The present series has several limitations. First, this report was composed of a single-surgeon’s experience. This senior surgeon is a pioneer in minimally invasive surgery from the early 1990s, and had performed many other gastrointestinal procedures by laparoscopy before starting LLR. Ideally, the present study should have included different operators, but the number of hepatopancreatobiliary surgeons is limited and the surgical techniques vary greatly from one centre to another.
© 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
www.bjs.co.uk
Acknowledgements
The authors thank M. Govindasamy, K. Araki and S. Ogiso for coordinating patients’ follow-up and maintaining the prospective database that formed the basis of this study. B.G. has received royalties for Gayet bipolar forceps (MicroFrance BG-CEV134; Medtronic, Minneapolis, Minnesota, USA). Disclosure: The authors declare no other conflict of interest. References 1 Reich H, McGlynn F, DeCaprio J, Budin R. Laparoscopic excision of benign liver lesions. Obstet Gynecol 1991; 78: 956–958.
BJS 2015; 102: 796–804
804
T. Nomi, D. Fuks, Y. Kawaguchi, F. Mal, Y. Nakajima and B. Gayet
2 Castaing D, Vibert E, Ricca L, Azoulay D, Adam R, Gayet B. Oncologic results of laparoscopic versus open hepatectomy for colorectal liver metastases in two specialized centers. Ann Surg 2009; 250: 849–855. 3 Ito K, Ito H, Are C, Allen PJ, Fong Y, DeMatteo RP et al. Laparoscopic versus open liver resection: a matched-pair case control study. J Gastrointest Surg 2009; 13: 2276–2283. 4 Belli G, Limongelli P, Fantini C, D’Agostino A, Cioffi L, Belli A et al. Laparoscopic and open treatment of hepatocellular carcinoma in patients with cirrhosis. Br J Surg 2009; 96: 1041–1048. 5 Topal B, Fieuws S, Aerts R, Vandeweyer H, Penninckx F. Laparoscopic versus open liver resection of hepatic neoplasms: comparative analysis of short-term results. Surg Endosc 2008; 22: 2208–2213. 6 Sarpel U, Hefti MM, Wisnievsky JP, Roayaie S, Schwartz ME, Labow DM. Outcome for patients treated with laparoscopic versus open resection of hepatocellular carcinoma: case-matched analysis. Ann Surg Oncol 2009; 16: 1572–1577. 7 Buell JF, Cherqui D, Geller DA, O’Rourke N, Iannitti D, Dagher I et al. The international position on laparoscopic liver surgery: the Louisville Statement, 2008. Ann Surg 2009; 250: 825–830. 8 Dagher I, O’Rourke N, Geller DA, Cherqui D, Belli G, Gamblin TC et al. Laparoscopic major hepatectomy: an evolution in standard of care. Ann Surg 2009; 250: 856–860. 9 Abu Hilal M, Di Fabio F, Teng MJ, Lykoudis P, Primrose JN, Pearce NW. Single-centre comparative study of laparoscopic versus open right hepatectomy. J Gastrointest Surg 2011; 15: 818–823. 10 Chaput de Saintonge DM, Vere DW. Why don’t doctors use cusums? Lancet 1974; 1: 120–121. 11 Wohl H. The cusum plot: its utility in the analysis of clinical data. N Engl J Med 1977; 296: 1044–1045. 12 Reddy SK, Barbas AS, Turley RS, Steel JL, Tsung A, Marsh JW et al. A standard definition of major hepatectomy: resection of four or more liver segments. HPB (Oxford) 2011; 13: 494–502. 13 Gayet B, Cavaliere D, Vibert E, Perniceni T, Levard H, Denet C et al. Totally laparoscopic right hepatectomy. Am J Surg 2007; 194: 685–689. 14 Gumbs AA, Gayet B. Totally laparoscopic left hepatectomy. Surg Endosc 2007; 21: 1221. 15 Gumbs AA, Gayet B. Totally laparoscopic extended right hepatectomy. Surg Endosc 2008; 22: 2076–2077. 16 Gumbs AA, Bar-Zakai B, Gayet B. Totally laparoscopic extended left hepatectomy. J Gastrointest Surg 2008; 12: 1152.
17 Araki K, Conrad C, Ogiso S, Kuwano H, Gayet B. Intraoperative ultrasonography of laparoscopic hepatectomy: key technique for safe liver transection. J Am Coll Surg 2014; 218: e37–e41. 18 Pang YY. The Brisbane 2000 terminology of liver anatomy and resections. HPB 2000; 2: 333–39. HPB (Oxford) 2002; 4: 99; author reply 99–100. 19 Dindo D, Demartines N, Clavien PA. Classification of surgical complications: a new proposal with evaluation in a cohort of 6336 patients and results of a survey. Ann Surg 2004; 240: 205–213. 20 Balzan S, Belghiti J, Farges O, Ogata S, Sauvanet A, Delefosse D et al. The ‘50–50 criteria’ on postoperative day 5: an accurate predictor of liver failure and death after hepatectomy. Ann Surg 2005; 242: 824–828. 21 Ishizawa T, Hasegawa K, Kokudo N, Sano K, Imamura H, Beck Y et al. Risk factors and management of ascites after liver resection to treat hepatocellular carcinoma. Arch Surg 2009; 144: 46–51. 22 Koch M, Garden OJ, Padbury R, Rahbari NN, Adam R, Capussotti L et al. Bile leakage after hepatobiliary and pancreatic surgery: a definition and grading of severity by the International Study Group of Liver Surgery. Surgery 2011; 149: 680–688. 23 Tekkis PP, Senagore AJ, Delaney CP, Fazio VW. Evaluation of the learning curve in laparoscopic colorectal surgery. Ann Surg 2005; 242: 83–91. 24 Cook JA, Ramsay CR, Fayers P. Statistical evaluation of learning curve effects in surgical trials. Clin Trials 2004; 1: 421–427. 25 Sudan R, Bennett KM, Jacobs DO, Sudan DL. Multifactorial analysis of the learning curve for robotassisted laparoscopic biliopancreatic diversion with duodenal switch. Ann Surg 2012; 255: 940–945. 26 Otsuka Y, Tsuchiya M, Maeda T, Katagiri T, Isii J, Tamura A et al. Laparoscopic hepatectomy for liver tumors: proposals for standardization. J Hepatobiliary Pancreat Surg 2009; 16: 720–725. 27 Bryant R, Laurent A, Tayar C, Cherqui D. Laparoscopic liver resection – understanding its role in current practice: the Henri Mondor Hospital experience. Ann Surg 2009; 250: 103–111. 28 Cai X-J, Wang Y-F, Liang Y-L, Yu H, Liang X. Laparoscopic left hemihepatectomy: a safety and feasibility study of 19 cases. Surg Endosc 2009; 23: 2556–2562. 29 Soubrane O, Schwarz L, Cauchy F, Perotto LO, Brustia R, Bernard D et al. A conceptual technique for laparoscopic right hepatectomy based on facts and oncologic principles: the caudal approach. Ann Surg 2014; [Epub ahead of print].
Supporting information
Additional supporting information may be found in the online version of this article: Fig. S1 Cumulative number of LMH procedures, from 1998 to 2013 (Pdf document)
© 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
www.bjs.co.uk
BJS 2015; 102: 796–804
Learning curve for laparoscopic major hepatectomy T. Nomi, D. Fuks, Y. Kawaguchi, F. Mal, Y. Nakajima and B. Gayet Department of Digestive Disease, Institut Mutualiste Montsouris, Université Paris-Descartes, 42 Boulevard Jourdan, 75014 Paris, France Correspondence to: Professor B. Gayet (e-mail: [email protected]) Background: Laparoscopic major hepatectomy (LMH) is evolving as an important surgical approach in
hepatopancreatobiliary surgery. The present study aimed to evaluate the learning curve for LMH at a single centre. Methods: Data for all patients undergoing LMH between January 1998 and September 2013 were recorded in a prospective database and analysed. The learning curve for operating time (OT) was evaluated using the cumulative sum (CUSUM) method. Results: Of 173 patients undergoing major hepatectomy, left hepatectomy was performed in 28 (16⋅2 per cent), left trisectionectomy in nine (5⋅2 per cent), right hepatectomy in 115 (66⋅5 per cent), right trisectionectomy in 13 (7⋅5 per cent) and central hepatectomy in eight (4⋅6 per cent). Median duration of surgery was 270 (range 100–540) min and median blood loss was 300 (10–4500) ml. There were 20 conversions to an open procedure (11⋅6 per cent). Vascular clamping was independently associated with conversion on multivariable analysis (hazard ratio 5⋅95, 95 per cent c.i. 1⋅24 to 28⋅56; P = 0⋅026). The CUSUMOT learning curve was modelled as a parabola (CUSUMOT = 0⋅2149 × patient number2 − 30⋅586 × patient number − 1118⋅3; R2 = 0⋅7356). The learning curve comprised three phases: phase 1 (45 initial patients), phase 2 (30 intermediate patients) and phase 3 (the subsequent 98 patients). Although right hepatectomy was most common in phase 1, a significant decrease was observed from phase 1 to 3 (P = 0⋅007) in favour of more complex procedures. Conclusion: The learning curve for LMH consisted of three characteristic phases identified by CUSUM analysis. The data suggest that the learning phase of LMH included 45 to 75 patients. Paper accepted 4 February 2015 Published online 15 April 2015 in Wiley Online Library (www.bjs.co.uk). DOI: 10.1002/bjs.9798
Introduction
Since the first successful report of a laparoscopic liver wedge resection for a benign lesion in 19911 , diffusion of laparoscopic liver resection (LLR), particularly laparoscopic major hepatectomy (LMH), has remained limited because of technical difficulties, fear of haemorrhage and embolism, and concern for oncological inadequacy. Although many studies have suggested the technical and oncological safety of LLR2 – 6 , this technique is still confined to a few expert centres and its reproducibility remains doubtful. Buell and colleagues7 reported at the Louisville Consensus Conference in 2008 that major hepatectomy should be performed by expert surgeons experienced in both open and laparoscopic liver resection. These data were essentially provided from small series of patients or multicentre studies that may have included some biases8,9 . Very few studies reporting large experiences from single institutions have been undertaken. More evidence is certainly required, including the assessment of both short- and long-term outcomes. Laparoscopic atypical © 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
liver resection has been developed at the Institut Mutualiste Montsouris since 1992, with the first laparoscopic major liver resections in 19982 . In the 15 years since the first LMH, more than 170 procedures have been performed, and the technique for LMH has gradually been standardized. The cumulative sum (CUSUM) technique is a method originally devised for monitoring performance and detecting areas for improvement in the industrial sector. This method was adopted by the medical profession in the 1970s to analyse the learning curve for surgical procedures10,11 , particularly laparoscopic procedures. CUSUM analysis transforms raw data into the running total of data deviations from the group mean, enabling investigators to visualize the data for trends not discernible with other approaches. The same effect should be expected in LMH; however, at present, no specific data are available. The present study was designed to evaluate the learning curve effect for LMH using CUSUM methodology to define its evolution in terms of feasibility and reproducibility. BJS 2015; 102: 796–804
Learning curve in laparoscopic major hepatectomy
Methods
All consecutive patients who underwent LMH between January 1998 and September 2013 at the Institut Mutualiste Montsouris in Paris were selected for the study. Major hepatectomy was defined as the resection of three or more contiguous Couinaud segments12 . Suitability for the laparoscopic approach was based on tumour size and location, type of planned resection, and patient co-morbidities. Bilobar procedures, including wedge resection or radiofrequency ablation in the contralateral liver, were also included. In all patients considered for LMH, future remnant liver volume was determined by liver volumetry using three-dimensional CT volumetry. Hepatic function was assessed before surgery with liver enzyme tests (serum bilirubin level, prothrombin time, platelet count) and imaging (absence of portal hypertension, spleen size). The indocyanine green retention test was not used routinely. Patients under consideration for major resection underwent portal vein embolization (PVE) when the future remnant liver volume was 25 per cent or less in normal liver (fewer than 6 cycles of chemotherapy) and 40 per cent or less in advanced fibrosis or cirrhosis (Child grade A with no portal hypertension), or in patients who had more than six cycles of chemotherapy. Resection was not considered if a sufficient liver volume could not be attained even after PVE. Selection of the laparoscopic approach did not depend on underlying liver disease but was considered each time the indication for resection was decided in the multidisciplinary staff meeting. This study was approved by the institutional review board.
Surgical procedures The surgical technique of LMH, including positioning of the trocars, has been described previously13 – 17 . All resections were performed with curative intent. The types of hepatectomy were classified according to the Brisbane 2000 terminology18 . Over time and with increasing surgical experience, the approach to LMH changed from posterior to anterior. Laparoscopic ultrasonography was performed routinely to confirm the number and size of lesions, and their relationship with the intrahepatic vascular structures. An intermittent Pringle manoeuvre was used only in cases of failure of bleeding control. For all procedures, tissue dissection and haemostasis were performed by the ultrasonic dissector, mainly the SonoSurg™ or Thunderbeat® (Olympus, Tokyo, Japan) or Harmonic® Scalpel® (Ethicon Endo-Surgery, Cincinnati, Ohio, USA); the Gayet bipolar forceps (MicroFrance CEVBG134; Medtronic, Minneapolis, Minnesota, USA) provided retraction and rescue haemostasis. All operations © 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
797
were performed by the senior surgeon. Conversion was defined as the requirement for laparotomy at any time during the procedure, with the exception of extraction of the resected specimen.
Postoperative outcomes Postoperative complications were stratified according to the Dindo–Clavien classification19 , which defines major complications by a score of 3 or more. If the patient had two or more postoperative complications, the most severe one was recorded. The specific liver complications that were encountered more often after major liver procedures were detailed as follows: liver failure was defined according to the ‘50–50 criteria’ on postoperative day 520 ; ascites was defined as an abdominal drainage output of more than 10 ml per kg per day after the third postoperative day21 ; and biliary leakage was defined by a bilirubin concentration in the drainage fluid more than threefold greater than that in the serum22 . Both complications and operative mortality were recorded as events occurring within 90 days of surgery or at any time during the postoperative hospital stay. Surgical margins were classified as either negative (R0) or positive (R1).
Learning curve analysis Two analyses were performed. In the first analysis, the CUSUM technique was used for quantitative assessment of the learning curve. The CUSUM technique was used when data on duration of surgery (defined as the time from first incision to final closure), blood loss (estimated by anaesthetists), vascular clamping, complications and length of stay were available. First, the patients were ordered chronologically, from the earliest to the latest date of surgery. Data for each patient in the series were plotted on a chart from left to right on the horizontal axis. The line ascended for every patient who was operated on within a certain operating time (OT), and descended for every patient in whom the operation took longer than the defined cut-off. Each patient included in the CUSUMOT analysis shows the real time required for the operation minus the expected time to complete the procedure, in chronological order. CUSUMOT of the first patient was the difference between OT for the first patient and the mean OT for all patients. CUSUMOT of the second patient was CUSUMOT of the previous patient plus the difference between OT for the second patient and the mean OT. This recursive process was continued until CUSUMOT of the last patient was calculated as zero. In the second analysis, learning curve analysis was based on the conversion rate. Univariable analysis of the factors predictive of conversion was performed; the comparison of conversion rates into three consecutive www.bjs.co.uk
BJS 2015; 102: 796–804
798
T. Nomi, D. Fuks, Y. Kawaguchi, F. Mal, Y. Nakajima and B. Gayet
25
No. of LMH procedures
20
15
10
5
0
1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year
Fig. 1
Number of laparoscopic major hepatectomy (LMH) procedures performed per year
groups of patients was then adjusted using multivariable logistic analysis.
Statistical analysis Patient baseline characteristics are expressed as median (range) for continuous data and as numbers with percentages for categorical data. Fisher’s exact test was used to compare differences in categorical variables, and the Wilcoxon rank sum test for continuous variables. Variables achieving statistical significance at the 0⋅1 level in univariable analysis were considered for multivariable analysis. A backward variable procedure was used to identify independent predictive factors. All statistical analyses were performed using SPSS® for Windows® version 20⋅0 (IBM, Armonk, New York, USA), and statistical significance was set at the 5 per cent level. Results
Among the 173 selected patients, there were 104 men (60⋅1 per cent) and 69 women (39⋅9 per cent) with a median age of 63 (range 24–86) years. The median number of LMH resections was 11 per year, with a maximum of 20 in 2008 (Fig. 1). The indication for liver resection was malignancy in 156 patients (90⋅2 per cent) and benign disease in 17 (9⋅8 per cent). In total, 106 patients (61⋅3 per cent) had undergone abdominal surgery previously, including 31 hepatectomies (17⋅9 per cent).
Surgical procedures and pathological specimens The types of major hepatectomy included left-sided hepatectomy in 37 patients (21⋅4 per cent), right-sided © 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
hepatectomy in 128 (74⋅0 per cent) and central hepatectomy in eight (4⋅6 per cent). Vascular clamping was required in 18 patients (10⋅4 per cent). Median duration of surgery was 270 (range 100–540) min. Median blood loss was 300 (10–4500) ml, with a loss of more than 500 ml in 45 patients (26⋅0 per cent) and more than 1000 ml in 17 (9⋅8 per cent). Eighteen patients (10⋅4 per cent) needed an intraoperative transfusion.
Postoperative outcomes Five patients died during the early postoperative period, giving an overall rate of 2⋅9 per cent. Major complications occurred in 43 patients (24⋅9 per cent), including biliary leakage in 23 (13⋅3 per cent), liver failure in four (2⋅3 per cent), intra-abdominal abscess in four (2⋅3 per cent), ascites in three (1⋅7 per cent) and intra-abdominal bleeding in two (1⋅2 per cent). Median hospital stay was 9 (range 3–57) days, and median follow-up 25 (6–83) months.
Learning curve and comparison of the three consecutive phases The raw operating times were plotted in chronological order (Fig. 2a). The CUSUMOT learning curve was best modelled as a second-order polynomial (parabola) with the following equation: CUSUMOT (in minutes) = 0⋅2149 × patient number2 − 30⋅586 × patient number − 1118⋅3. This had a high R2 value of 0⋅7356 (Fig. 2b). The CUSUMOT learning curve comprised three unique phases: phase 1 (the initial 45 patients), phase 2 (the middle 30 patients) and phase 3 (the final 98 patients) (Fig. 2b). Comparisons between the three phases of various www.bjs.co.uk
BJS 2015; 102: 796–804
Learning curve in laparoscopic major hepatectomy
799
Operating time (min)
600 500 400 300 200 100 0
1 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 105 110 115 120 125 130 135 140 145 150 155 160 165 170 175
Phase 1
Phase 2
January 1998
a
December 2005
Phase 3 September 2013
March 2008
Operating time versus no. of patients No. of patients 0
1 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95100 105 110 115120125130135140145 150 155 160 165 170175
CUSUMOT (min)
–500 –1000 –1500 –2000 –2500 –3000 Phase 1 January 1998
b
Phase 3
Phase 2 December 2005
March 2008
September 2013
CUSUMOT versus no. of patients
Operating time (min)
600 500 400 300 200 100 0
1
5
10
15
20
25
Phase 1 January 1998
c
30
35
40
45
50
55
60
65
70
75
Phase 2 December 2005
80
85
90
95
100
105
110
115
Phase 3 March 2008
September 2013
Operating time versus no. of patients having laparoscopic right hepatectomy
a Evolution of operating time (OT) plotted against number of patients for all laparoscopic major hepatectomy (LMH) procedures. b Cumulative sum (CUSUM)OT plotted against number of patients. The black line represents the curve of best fit for the plot (a second-order polynomial using the equation CUSUMOT = 0⋅2149 × patient number2 − 30⋅586 × patient number − 1118⋅3; R2 = 0⋅7356). c Evolution of operating time plotted against number of patients who had laparoscopic right hepatectomy
Fig. 2
© 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
www.bjs.co.uk
BJS 2015; 102: 796–804
800
Table 1
T. Nomi, D. Fuks, Y. Kawaguchi, F. Mal, Y. Nakajima and B. Gayet
Interphase comparisons of characteristics of the 173 patients undergoing laparoscopic major liver resection P‡
Preop. characteristics Age (years)* Male sex Body mass index (kg/m2 )* Child grade B–C Diagnosis Malignancy CRLM HCC Cholangiocarcinoma Benign lesion Tumour characteristics Tumour number* Tumour size (mm)* Previous abdominal operation Hepatectomy Surgical procedures Type of hepatectomy Left hepatectomy Right hepatectomy Left trisectionectomy Right trisectionectomy Central hepatectomy Additional hepatic procedures Radiofrequency ablation Wedge resection Combined resection of adjacent organs Abdominal drainage Postop. outcomes Mortality Morbidity Major morbidity† Background liver disease Steatosis (>60%) Cirrhosis Reoperation
Phase 1 (n = 45)
Phase 2 (n = 30)
Phase 3 (n = 98)
Phase 1 versus 2
Phase 2 versus 3
Phase 1 versus 3
59 (27–85) 31 (69) 23⋅3 (18⋅4–29⋅0) 1 (2)
60 (24–84) 19 (63) 24⋅2 (22⋅5–26⋅4) 1 (3)
67 (32–86) 54 (55) 25⋅1 (15⋅9–35⋅5) 2 (2)
0⋅245§ 0⋅627 0⋅500§ 1⋅000
0⋅861§ 0⋅528 0⋅500§ 0⋅554
0⋅048§ 0⋅143 0⋅437§ 1⋅000
38 (84) 26 (58) 7 (16) 4 (9) 7 (16)
23 (77) 14 (47) 7 (23) 0 (0) 7 (23)
95 (97) 75 (77) 6 (6) 12 (12) 3 (3)
0⋅546 0⋅357 0⋅546 0⋅145 0⋅546
0⋅001 0⋅003 0⋅012 0⋅067 0⋅001
0⋅011 0⋅029 0⋅112 0⋅776 0⋅011
1 (1–4) 40 (5–170) 28 (62) 6 (13)
2 (1–5) 50 (15–160) 12 (40) 4 (13)
2 (1–20) 35 (5–150) 66 (67) 21 (21)
0⋅197§ 0⋅416§ 0⋅097 1⋅000
0⋅393§ 0⋅391§ 0⋅010 0⋅434
0⋅034§ 0⋅676§ 0⋅573 0⋅357
6 (13) 37 (82) 0 (0) 1 (2) 1 (2)
8 (27) 20 (67) 1 (3) 0 (0) 1 (3)
14 (14) 58 (59) 8 (8) 12 (12) 6 (6)
0⋅225 0⋅168 0⋅400 1⋅000 1⋅000
0⋅164 0⋅525 0⋅684 0⋅067 1⋅000
1⋅000 0⋅007 0⋅056 0⋅063 0⋅433
7 (16) 6 (13) 4 (9) 12 (27)
5 (17) 10 (33) 4 (13) 5 (17)
7 (7) 23 (23) 12 (12) 23 (23)
1⋅000 0⋅047 0⋅706 0⋅403
0⋅150 0⋅340 1⋅000 0⋅614
0⋅135 0⋅185 0⋅776 0⋅680
1 (2) 30 (67) 10 (22)
1 (3) 18 (60) 8 (27)
3 (3) 51 (52) 25 (26)
1⋅000 0⋅627 0⋅783
1⋅000 0⋅531 1⋅000
1⋅000 0⋅107 0⋅834
11 (24) 1 (2) 1 (2)
7 (23) 1 (3) 1 (3)
27 (28) 3 (3) 4 (4)
1⋅000 1⋅000 1⋅000
0⋅814 1⋅000 1⋅000
0⋅838 1⋅000 1⋅000
Values in parentheses are percentages unless indicated otherwise; *values are median (range). †Postoperative complications were stratified according to the Dindo–Clavien classification14 . CRLM, colorectal liver metastases; HCC, hepatocellular carcinoma. ‡Fisher’s exact test, except §Wilcoxon rank sum test.
parameters that were identified by CUSUMOT analysis are presented in Table 1. Patient age was higher in phase 3 than in phase 1 (P = 0⋅048). No difference in terms of male sex, body mass index or American Society of Anesthesiologists fitness grade (median 2 for each phase; P = 0⋅874) was found among the three groups. The proportion of patients who had undergone previous abdominal surgery increased between phases 2 and 3 (P = 0⋅010). Considering indications for surgery, the rate of colorectal liver metastases had increased significantly by later years, from 58 to 77 per cent (P = 0⋅029) and hepatocellular carcinoma decreased significantly between phases 2 and 3 (P = 0⋅012). Twenty patients underwent associated laparoscopic extrahepatic resection: seven diaphragmatic
resections, six common bile duct resections, two inferior vena cava resections, one duodenal resection, one pancreatic resection, one colonic resection, one adrenal gland resection and one kidney resection, with no difference over time. The need for pedicle clamping decreased from eight (18 per cent) in phase 1 to five (5 per cent) in phase 3 (P = 0⋅024); whenever clamping was required, the clamping duration became progressively shorter (from 45 to 10 min). Similarly, the median duration of surgery and blood loss decreased significantly between phases 1 and 3 (from 360 to 240 min, and from 500 to 200 ml, respectively). Moreover, a reduction in the conversion rate was observed in phase 3 compared with phase 1 (P = 0⋅037) and phase 2 (P = 0⋅033). A trend toward
© 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
www.bjs.co.uk
BJS 2015; 102: 796–804
Learning curve in laparoscopic major hepatectomy
801
decreased morbidity was observed (P = 0⋅107). Hospital stay was also slightly shorter during phase 3 (9 days) than in phase 1 (11 days) (P = 0⋅046).
Comparison of standard (right or left hepatectomy) and more complex (extended right or left hepatectomy and central hepatectomy) laparoscopic major hepatectomy
Table 2
More complex Standard hepatectomy hepatectomy (n = 30) (n = 143) Intraoperative data Combined resection of adjacent organs Duration of operation (min)* Blood loss (ml)* Transfusion Use of Pringle manoeuvre Conversion Abdominal drainage Postoperative outcomes Mortality Morbidity Major morbidity† Bile leakage Reoperation Length of hospital stay (days)*
Specific learning curve and outcomes of right hepatectomy
P‡
15 (10⋅5)
5 (17)
0⋅349
270 250 15 (10⋅5) 12 (8⋅4) 14 (9⋅8) 30 (21⋅0)
280 400 3 (10) 6 (20) 6 (20) 10 (33)
0⋅367§ 0⋅567§ 1⋅000 0⋅092 0⋅122 0⋅157
3 (2⋅1) 82 (57⋅3) 30 (21⋅0) 16 (11⋅2) 5 (3⋅5) 9 (6–48)
2 (7) 17 (57) 13 (43) 8 (27) 1 (3) 11 (5–57)
0⋅207 1⋅000 0⋅018 0⋅039 1⋅000 0⋅914§
Values in parentheses are percentages unless indicated otherwise; *values are median (range). †Postoperative complications were stratified according to the Dindo–Clavien classification14 . ‡Fisher’s exact test, except §Wilcoxon rank sum test.
Table 3
Right hepatectomy was the most common laparoscopic procedure (115 patients; 66⋅5 per cent) (Table 1). However, a decrease in the number of right hepatectomies was observed from phase 1 to 3 (P = 0⋅007). On the other hand, the number of extended procedures such as trisectionectomies tended to increase in phase 3. When comparing standard (right or left hepatectomy) with more complex procedures (extended right or left hepatectomy and central hepatectomy), higher rates of major complications (21⋅0 versus 43 per cent respectively; P = 0⋅018) including bile leakage (11⋅2 versus 27 per cent; P = 0⋅039) were observed in patients undergoing more complex hepatectomy (Table 2). In the subgroup of 115 patients having a right hepatectomy, there were nine conversions, with a decrease in later years (11, 10 and 5 per cent in phases 1, 2 and 3) (Table 3). The morbidity rate (65, 59 and 38 per cent) decreased significantly whereas length of stay (10, 9 and 9
Interphase comparisons of characteristics of the 115 patients undergoing laparoscopic right hepatectomy P†
Preop. characteristics Age (years)* Male sex Body mass index (kg/m2 )* ASA fitness grade ≥ II Tumour characteristics Tumour number* Tumour size (mm)* Previous abdominal operation Previous hepatectomy Preop. PVE Surgical procedures Combined resection of adjacent organs Duration of surgery (min)* Blood loss (ml)* Transfusion Use of Pringle manoeuvre Conversion Abdominal drainage Postop. outcomes Length of hospital stay (days)*
Phase 1 (n = 37)
Phase 2 (n = 20)
Phase 3 (n = 58)
Phase 1 versus 2
Phase 2 versus 3
Phase 1 versus 3
59 (27–85) 25 (68) 23⋅3 (20⋅1–29⋅0) 30 (81)
59 (35–84) 13 (65) 23⋅6 (22⋅7–26⋅4) 18 (90)
66 (47–86) 33 (57) 25⋅4 (16⋅4–33⋅1) 45 (78)
0⋅432‡ 0⋅579 0⋅892‡ 1⋅000
0⋅083‡ 0⋅877 1⋅000‡ 0⋅922
0⋅020‡ 0⋅512 1⋅000‡ 1⋅000
1 (1–4) 42 (5–130) 23 (62) 6 (16) 8 (22)
3 (1–5) 40 (10–140) 7 (35) 4 (20) 5 (25)
2 (1–8) 35 (7–120) 35 (60) 16 (28) 9 (16)
0⋅051‡ 0⋅600‡ 0⋅032 1⋅000 1⋅000
0⋅919‡ 0⋅859‡ 0⋅022 0⋅402 0⋅521
0⋅027‡ 0⋅510‡ 1⋅000 0⋅216 0⋅586
4 (11) 345 (180–540) 600 (10–4500) 4 (11) 7 (19) 4 (11) 10 (27)
3 (15) 210 (180–480) 300 (50–2000) 3 (15) 2 (10) 2 (10) 4 (20)
7 (12) 240 (100–480) 185 (10–1500) 5 (9) 1 (2) 3 (5) 9 (16)
1⋅000 0⋅004‡ 0⋅174‡ 1⋅000 0⋅461 1⋅000 0⋅537
1⋅000 0⋅407‡ 1⋅000‡ 0⋅680 0⋅190 0⋅617 1⋅000
1⋅000 < 0⋅001‡ 0⋅002‡ 1⋅000 0⋅006 0⋅430 0⋅293
10 (5–35)
9 (5–38)
9 (5–50)
0⋅856‡
0⋅231‡
0⋅098‡
Values in parentheses are percentages unless indicated otherwise; *values are median (range). ASA, American Society of Anesthesiologists; PVE, portal vein embolization. †Fisher’s exact test, except ‡Wilcoxon rank sum test.
© 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
www.bjs.co.uk
BJS 2015; 102: 796–804
802
Table 4
T. Nomi, D. Fuks, Y. Kawaguchi, F. Mal, Y. Nakajima and B. Gayet
Risk factors for conversion to laparotomy Univariable analysis No conversion to laparotomy (n = 153)*
Age > 70 years Male sex Body mass index (kg/m2 ) ASA fitness grade > II Malignancy Tumour diameter > 5 cm Previous abdominal operation Prehepatectomy chemotherapy Type of hepatectomy Left hepatectomy Right hepatectomy Left trisectionectomy Right trisectionectomy Central hepatectomy Additional hepatic procedures Radiofrequency ablation Wedge resection Combined resection of adjacent organs Blood loss > 500 ml Intraoperative transfusion Use of Pringle manoeuvre Steatosis > 60% Cirrhosis
Conversion to laparotomy (n = 20)*
Multivariable analysis
P‡
47 (30⋅7) 94 (61⋅4) 17 (11⋅1) 116 (75⋅8) 140 (91⋅5) 62 (40⋅5) 99 (64⋅7) 65 (42⋅5)
5 (25) 10 (50) 0 (0) 17 (85) 16 (80) 5 (25) 7 (35) 9 (45)
0⋅796 0⋅341 0⋅224 0⋅572 0⋅114 0⋅226 0⋅014 1⋅000
23 (15⋅0) 105 (68⋅6) 7 (4⋅6) 12 (7⋅8) 6 (3⋅9)
5 (25) 10 (50) 2 (10) 1 (5) 2 (10)
0⋅328 0⋅129 0⋅288 1⋅000 0⋅232
16 (10⋅5) 35 (22⋅9) 14 (9⋅2) 35 (22⋅9) 13 (8⋅5) 12 (7⋅8) 39 (25⋅5) 4 (2⋅6)
3 (15) 4 (20) 6 (30) 10 (50) 5 (25) 6 (30) 6 (30) 1 (5)
0⋅464 1⋅000 0⋅015 0⋅014 0⋅039 0⋅008 0⋅786 0⋅463
Odds ratio†
P§
0⋅31 (0⋅08, 1⋅18)
0⋅085
1⋅65 (0⋅27, 9⋅99) 1⋅88 (0⋅34, 10⋅36) 2⋅61 (0⋅43, 15⋅85) 5⋅95 (1⋅24, 28⋅56)
0⋅588 0⋅470 0⋅296 0⋅026
Values in parentheses are *percentages and †95 per cent c.i. ASA, American Society of Anesthesiologists. ‡Fisher’s exact test; §backward variable analysis.
days respectively) was stable over time. The percentage of R0 resection was also stable over the different phases (97 per cent for all phases).
Conversion to open surgery: risk factors and learning curve Twenty patients (11⋅6 per cent) required conversion to open surgery, because of haemorrhage (11 patients), oncological reasons (6) and lack of progress (3). The learning curve adjusted for the risk factors of conversion demonstrated that the rate of conversion to open surgery decreased in later years (18, 20 and 6 per cent in phases 1, 2 and 3 respectively). Univariable analysis showed that previous abdominal surgery, resection of adjacent organs, blood loss greater than 500 ml, intraoperative transfusion and vascular clamping were associated with a significantly higher risk of conversion (Table 4). Only vascular clamping was independently associated with conversion on multivariable analysis (odds ratio 5⋅95, 95 per cent c.i. 1⋅24 to 28⋅56; P = 0⋅026). Discussion
The learning curve is defined as the improvement in performance over time or the change in the ability to complete © 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
a task until failure is reduced to a constant minimum acceptable rate23 – 25 . The learning curve of a pioneer surgeon who is technically proficient or has advanced training in complex minimally invasive surgery techniques will be different from that of a novice surgeon25 . The level of both laparoscopic and liver surgical experience at the beginning of LMH is essential when evaluating the effect of the learning curve. Given that LMH is a technically demanding procedure, the surgeon should have advanced previous experiences in both open liver resection and laparoscopic procedures, such as upper gastrointestinal and colorectal surgery, before starting LMH. Although some articles26,27 have reported improvement of results during the second half of surgeons’ experience in laparoscopic minor hepatectomy, the learning curve has never been studied for LMH. In the present study, according to the CUSUM plot, data were categorized into three unique groups to confirm the learning curve effect on LMH: phase 1 (45 initial patients), representing the initial learning curve; phase 2 plateau (30 middle patients), representing increased competence with laparoscopy; and phase 3 (the subsequent 98 patients), representing the mastery phase in which more complex procedures were undertaken. Despite the expansion of indications for LMH and the increasing complexity www.bjs.co.uk
BJS 2015; 102: 796–804
Learning curve in laparoscopic major hepatectomy
803
of the resections performed, the surgical outcomes, including blood loss and duration of surgery, improved over time. A better understanding of the technical difficulties, and concerns regarding life-threatening operative risks of LMH, is imperative to carry out these procedures successfully. Given that nearly 95 per cent of all hepatectomies are performed laparoscopically in the authors’ institution, it should be taken into account that the present results might not be reproducible in every institution, or for all surgeons. Indeed, it is important to keep in mind that non-expert surgeons should start LLR with simple procedures, with a gradual increase in technical demand. Laparoscopy was considered for almost all major hepatectomies in the authors’ institution. The present study clearly suggests a learning curve effect due to technique improvement and standardization, especially for parenchymal transection, which represents the most difficult part of LLR. During the 15-year interval starting in 1998 and ending in 2013, many new devices have been developed including ultrasonic scalpels, sealing devices, coagulation systems and staplers, and these have facilitated expansion of the indications. In the authors’ institution, CUSA® (ValleyLab, Boulder, Colorado, USA) has never been used, and tissue dissection and haemostasis were performed with an ultrasonic dissector, first SonoSurg™, then Harmonic® Scalpel® and finally Thunderbeat®, with frequent use of the Gayet bipolar forceps. Even though these devices do not seem to improve the quality of transection in open surgery, difficulty in controlling potential bleeding during parenchymal transection warrants their use in laparoscopy. Nevertheless, the present study shows that the speed and ease of transection improves with experience, with a significant decrease in the duration of surgery over time. Given that conversion to open hepatectomy is associated with longer procedure times and a potentially increased overall morbidity, the ability accurately to identify preoperative risk factors for conversion may improve the care of patients undergoing LMH. In the present study, the conversion rate decreased progressively to 6 per cent in phase 3, which is relatively low compared with that in previous reports of LMH8,9,28 . In the present series, haemorrhage, unclear limits of tumour extension and failure to progress represented the three major reasons for conversion to open hepatectomy. Even though vascular clamping was performed mainly when blood loss was excessive during parenchymal transection, multivariable analysis identified vascular clamping to be associated with conversion. This result indicates the need to use pedicle clamping more widely, and earlier, during parenchymal transection. The effect of conversion on postoperative outcomes during LMH remains unclear. Unlike most
studies, no significant decrease in complication rates with increasing experience was observed. The complication rate may depend more on the selection of patients than on technical skills. In the present study, the main reason for conversion was bleeding in the early era; however, oncological reasons (tumour extent, the need for biliary reconstruction) increased during later years. A conversion rate of zero should not be considered a realistic goal, although an institution’s policy on conversion could be an important key to this problem. The present study indicated that 45 LMH procedures are required in order to reduce operating time, and to move from phase 1 to phase 2. As right and left hepatectomies represented more than 95 per cent of hepatectomies, it would make sense to start LMH with standard right or left hepatectomy. Indeed, the learning curve in LMH should start with anatomical resection completely mastered by an open approach. The surgeon should probably select patients with limited tumour load (for example in phase 1 a single tumour with a 4-cm diameter), low co-morbidities and no overweight. Additionally, the surgeon might be helped by a new technique of laparoscopic right hepatectomy29 , allowing low blood loss and morbidity. Finally, phase 3 of the learning curve included a higher rate of extended or atypical major hepatectomy, which may explain the pronounced lack of reproducibility for the procedure. The present series has several limitations. First, this report was composed of a single-surgeon’s experience. This senior surgeon is a pioneer in minimally invasive surgery from the early 1990s, and had performed many other gastrointestinal procedures by laparoscopy before starting LLR. Ideally, the present study should have included different operators, but the number of hepatopancreatobiliary surgeons is limited and the surgical techniques vary greatly from one centre to another.
© 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
www.bjs.co.uk
Acknowledgements
The authors thank M. Govindasamy, K. Araki and S. Ogiso for coordinating patients’ follow-up and maintaining the prospective database that formed the basis of this study. B.G. has received royalties for Gayet bipolar forceps (MicroFrance BG-CEV134; Medtronic, Minneapolis, Minnesota, USA). Disclosure: The authors declare no other conflict of interest. References 1 Reich H, McGlynn F, DeCaprio J, Budin R. Laparoscopic excision of benign liver lesions. Obstet Gynecol 1991; 78: 956–958.
BJS 2015; 102: 796–804
804
T. Nomi, D. Fuks, Y. Kawaguchi, F. Mal, Y. Nakajima and B. Gayet
2 Castaing D, Vibert E, Ricca L, Azoulay D, Adam R, Gayet B. Oncologic results of laparoscopic versus open hepatectomy for colorectal liver metastases in two specialized centers. Ann Surg 2009; 250: 849–855. 3 Ito K, Ito H, Are C, Allen PJ, Fong Y, DeMatteo RP et al. Laparoscopic versus open liver resection: a matched-pair case control study. J Gastrointest Surg 2009; 13: 2276–2283. 4 Belli G, Limongelli P, Fantini C, D’Agostino A, Cioffi L, Belli A et al. Laparoscopic and open treatment of hepatocellular carcinoma in patients with cirrhosis. Br J Surg 2009; 96: 1041–1048. 5 Topal B, Fieuws S, Aerts R, Vandeweyer H, Penninckx F. Laparoscopic versus open liver resection of hepatic neoplasms: comparative analysis of short-term results. Surg Endosc 2008; 22: 2208–2213. 6 Sarpel U, Hefti MM, Wisnievsky JP, Roayaie S, Schwartz ME, Labow DM. Outcome for patients treated with laparoscopic versus open resection of hepatocellular carcinoma: case-matched analysis. Ann Surg Oncol 2009; 16: 1572–1577. 7 Buell JF, Cherqui D, Geller DA, O’Rourke N, Iannitti D, Dagher I et al. The international position on laparoscopic liver surgery: the Louisville Statement, 2008. Ann Surg 2009; 250: 825–830. 8 Dagher I, O’Rourke N, Geller DA, Cherqui D, Belli G, Gamblin TC et al. Laparoscopic major hepatectomy: an evolution in standard of care. Ann Surg 2009; 250: 856–860. 9 Abu Hilal M, Di Fabio F, Teng MJ, Lykoudis P, Primrose JN, Pearce NW. Single-centre comparative study of laparoscopic versus open right hepatectomy. J Gastrointest Surg 2011; 15: 818–823. 10 Chaput de Saintonge DM, Vere DW. Why don’t doctors use cusums? Lancet 1974; 1: 120–121. 11 Wohl H. The cusum plot: its utility in the analysis of clinical data. N Engl J Med 1977; 296: 1044–1045. 12 Reddy SK, Barbas AS, Turley RS, Steel JL, Tsung A, Marsh JW et al. A standard definition of major hepatectomy: resection of four or more liver segments. HPB (Oxford) 2011; 13: 494–502. 13 Gayet B, Cavaliere D, Vibert E, Perniceni T, Levard H, Denet C et al. Totally laparoscopic right hepatectomy. Am J Surg 2007; 194: 685–689. 14 Gumbs AA, Gayet B. Totally laparoscopic left hepatectomy. Surg Endosc 2007; 21: 1221. 15 Gumbs AA, Gayet B. Totally laparoscopic extended right hepatectomy. Surg Endosc 2008; 22: 2076–2077. 16 Gumbs AA, Bar-Zakai B, Gayet B. Totally laparoscopic extended left hepatectomy. J Gastrointest Surg 2008; 12: 1152.
17 Araki K, Conrad C, Ogiso S, Kuwano H, Gayet B. Intraoperative ultrasonography of laparoscopic hepatectomy: key technique for safe liver transection. J Am Coll Surg 2014; 218: e37–e41. 18 Pang YY. The Brisbane 2000 terminology of liver anatomy and resections. HPB 2000; 2: 333–39. HPB (Oxford) 2002; 4: 99; author reply 99–100. 19 Dindo D, Demartines N, Clavien PA. Classification of surgical complications: a new proposal with evaluation in a cohort of 6336 patients and results of a survey. Ann Surg 2004; 240: 205–213. 20 Balzan S, Belghiti J, Farges O, Ogata S, Sauvanet A, Delefosse D et al. The ‘50–50 criteria’ on postoperative day 5: an accurate predictor of liver failure and death after hepatectomy. Ann Surg 2005; 242: 824–828. 21 Ishizawa T, Hasegawa K, Kokudo N, Sano K, Imamura H, Beck Y et al. Risk factors and management of ascites after liver resection to treat hepatocellular carcinoma. Arch Surg 2009; 144: 46–51. 22 Koch M, Garden OJ, Padbury R, Rahbari NN, Adam R, Capussotti L et al. Bile leakage after hepatobiliary and pancreatic surgery: a definition and grading of severity by the International Study Group of Liver Surgery. Surgery 2011; 149: 680–688. 23 Tekkis PP, Senagore AJ, Delaney CP, Fazio VW. Evaluation of the learning curve in laparoscopic colorectal surgery. Ann Surg 2005; 242: 83–91. 24 Cook JA, Ramsay CR, Fayers P. Statistical evaluation of learning curve effects in surgical trials. Clin Trials 2004; 1: 421–427. 25 Sudan R, Bennett KM, Jacobs DO, Sudan DL. Multifactorial analysis of the learning curve for robotassisted laparoscopic biliopancreatic diversion with duodenal switch. Ann Surg 2012; 255: 940–945. 26 Otsuka Y, Tsuchiya M, Maeda T, Katagiri T, Isii J, Tamura A et al. Laparoscopic hepatectomy for liver tumors: proposals for standardization. J Hepatobiliary Pancreat Surg 2009; 16: 720–725. 27 Bryant R, Laurent A, Tayar C, Cherqui D. Laparoscopic liver resection – understanding its role in current practice: the Henri Mondor Hospital experience. Ann Surg 2009; 250: 103–111. 28 Cai X-J, Wang Y-F, Liang Y-L, Yu H, Liang X. Laparoscopic left hemihepatectomy: a safety and feasibility study of 19 cases. Surg Endosc 2009; 23: 2556–2562. 29 Soubrane O, Schwarz L, Cauchy F, Perotto LO, Brustia R, Bernard D et al. A conceptual technique for laparoscopic right hepatectomy based on facts and oncologic principles: the caudal approach. Ann Surg 2014; [Epub ahead of print].
Supporting information
Additional supporting information may be found in the online version of this article: Fig. S1 Cumulative number of LMH procedures, from 1998 to 2013 (Pdf document)
© 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
www.bjs.co.uk
BJS 2015; 102: 796–804
