Randomized Comparison of Ultrasonic vs Clamp Transection of the Liver FREE

IN LIVER surgery, transection of the hepatic parenchyma is a technical priority because blood loss during the procedure is a powerful determinant of operative outcome.¹ At high-volume centers, partial hepatectomy has been reported to yield a median blood loss of 700 to 1200 mL,^2- 7 morbidity rates of 23% to 46%,^2- 8 and mortality rates of less than 4%.^2- 9 The hepatic parenchyma can be transected by conventional digitoclasia (finger fracture or clamp crushing)¹⁰ or with instrument-based techniques (ultrasonic or water jet dissector).¹¹ The procedure used is likely to depend on the surgeon's training or preference rather than on objective data.

The ultrasonic dissector, although more costly and time-consuming than digitoclasia, has gained wide acceptance because it may reduce bleeding during hepatectomy.¹¹ Fan et al¹² reported a 30% reduction in blood loss after changing their technique from clamp to ultrasonic transection, with only a 4% prolongation of execution time, consistent with the findings of other specialists.^3,7,9 Reported benefits, however, were based on comparison with historical controls, and conclusive evidence supporting either technique is lacking. Moreover, comparison based solely on classic end points, such as blood loss and time, may disregard other important factors affecting the overall quality of hepatectomy; therefore, a comprehensive evaluation system is required.

We conducted a randomized controlled trial to compare the surgical effectiveness of ultrasonic vs clamp transection of the liver according to a new hepatectomy grading system.

This study was designed to compare ultrasonic vs clamp transection of the liver. Eligibility criteria included partial hepatectomy for tumor resection or graft harvest, hepatic function of Child-Pugh class A or B, acceptable clotting profile (platelet count >5 × 10³/µL and prothrombin activity >60%), and adequate functional reserve of the heart, lungs, and kidneys. In the operating room, eligible patients were randomly assigned to receive hepatectomy with either transection method. Randomization stratified by indocyanine green retention rate at 15 minutes (<20% or ≥20%)⁵ and hepatectomy operation (minor or major resection) was done with a minimization procedure.

The sample size required to detect a difference in blood loss of 300 mL (assuming an SD of 600 mL)^5,12 by t test was 120 patients, with 80% power at the 5% significance level. We anticipated analyzing the data with nonparametric tests and therefore increased the sample to 132 patients. We obtained approval from the local ethics committee and informed consent from the patients.

The hepatectomy procedure was decided according to criteria based on hepatic functional reserve in patients with tumor, or recipients' demands concerning living-related donors, as described elsewhere.^13- 14 Minor hepatectomy was defined as limited resection or resection of less than 2 Couinaud segments, and major hepatectomy consisted of left or right hepatectomy or extended hemihepatectomy (the first was defined as nonanatomical resection, and the others as anatomical resection).⁹ Anatomical resection required access to, landing on, and clear skeletonization of the corresponding hepatic venous trunk.¹⁰ All resections were done with intermittent total (15-minute) or selective (30-minute) occlusion (each with 5-minute release) of inflow vessels and were guided with operative ultrasonography.^13,15

Two consultant surgeons (M.M. and T.T.) performed two thirds of the operations. The first had done 1800 hepatectomies, and the second had performed 550; the experience with clamp crushing was roughly 10-fold that of ultrasonic dissection. Three resident specialists (who had done or assisted with at least 50 hepatectomies) did one third while being assisted by the consultant; there was no specific assignment of surgeon. Neither surgical nor anesthetic technique was modified during the study period, and central venous pressure was not controlled during hepatic transection.

Patients assigned to the ultrasonic dissection group received hepatectomy with an ultrasonic dissector (Cavitron Ultrasonic Surgical Aspirator System 200; Valleylab Inc, Boulder, Colo; amplitude, 40-70 W).^7,12 Those assigned to the clamp-crushing group received hepatectomy by parenchymal crushing with a Péan clamp.^2,10 Technical conversion from the assigned method to the other one was allowed only if the surgeon felt that the safety of the transection had been compromised by major errors. In both groups, tiny vessels were cauterized, vessels larger than 1 mm in diameter were ligated and cut, and bleeding or bile leak point was fine-sutured. Fibrin glue was spread on the cut surface, and a tube measuring 24 F catheter was left at each liver stump. Postoperatively, all patients received uniform care.^5,13

The primary outcome measures were blood loss and hepatectomy grade. Secondary measures included transection time, technical error, surgical margin, landmark appearance, postoperative morbidity and mortality rates, and hospitalization length.

The quality of the operation was assessed by a hepatectomy grading system (Table 1). Our system comprised 6 criteria, each with 3 scoring measures (lower scores indicating higher quality), and gave 4 grades (A = excellent, B = good, C = fair, and D = poor) according to the sum scores. Landmark appearance was not a measure in nonanatomical resection, and surgical margin was not used for donor hepatectomy. All surgical results were recorded by 2 independent surgeons not involved in the operation. Each procedure was scored by a monitoring committee blinded to the assigned method.

The amounts of blood loss were measured both during transection and after all procedures. The speed of transection was calculated as transection area divided by transection time. Whole blood was transfused only when the hematocrit value dropped to less than 25%, and fresh frozen plasma was used for replacement until that point.¹³ Minor technical errors included injury of intrahepatic vessels scheduled to be preserved and congestion or ischemia of the liver remnant; errors requiring conversion of the scheduled procedure were defined as major.

Our principal goal was a surgical clearance of 5 mm or more for any tumor in patients undergoing nonanatomical resection (based on our previous analyses),^16- 17 or a complete 180° appearance of landmark vessels on the cut surface in those undergoing anatomical resection^10,17 (excluding patients receiving extended hemihepatectomy or donor hepatectomy). Landmarks were defined as hepatic veins demarcating the anatomical borders of liver segments (eg, the middle hepatic vein in hemihepatectomy or the right hepatic vein in the resection of the posterior segment; Figure 1). The histologically proven tumor involvement at any surgical margin was defined as tumor exposure. The level of landmark appearance was determined during surgery and from videotapes of the surgical field. All postoperative complications were recorded, and events such as bleeding from the liver stump (requiring reoperation), intra-abdominal abscess (requiring drainage), bile leakage (lasting >2 weeks), and hyperbilirubinemia (>5 mg/dL [>86 µmol/L]) were defined as major morbidity.

All analyses were done on an intention-to-treat basis. The surgical outcomes between the 2 groups were compared using the χ² test for categorical data and the Mann-Whitney U test for continuous or ordinal data. Confidence intervals (95% CIs) were calculated for differences in proportions for categorical data, and approximate 95% CIs for differences in medians for nonparametric data were calculated using the Hodges-Lehman method.¹⁸ The relationship between blood loss and transection time was assessed with the Spearman rank correlation. Logistic regression was used to estimate the influence of transection method on the primary outcome measures (dichotomized with clinically relevant cutoff points¹³: blood loss ≤500 mL vs >500 mL, and hepatectomy grade A or B vs C or D), controlling for 9 potential confounders (Child-Pugh class, indocyanine clearance, noncirrhosis vs cirrhosis, surgeon, inflow occlusion, thoracotomy, minor vs major hepatectomy, nonanatomical vs anatomical resection, and number of resections). The results are presented as adjusted odds ratios (ORs) with 95% CIs and P values from the likelihood ratio test. All P values were 2-sided, and significance was set at P<.05.

From April 1998 through April 1999, 137 patients were scheduled to undergo partial hepatectomy, 5 of whom were found to be inoperable because 4 had extrahepatic metastases and 1 had tumor thrombi in the portal venous trunk. The 132 patients were randomly assigned a hepatic transection method; 66 patients were assigned to the ultrasonic dissection group and 66 to the clamp-crushing group (Figure 2).

The 2 groups were well matched for all baseline characteristics such as age, cause of surgery, hepatic function, tumor status, and surgical details (Table 2).

Between the 2 groups, there was no significant difference in median blood loss, transection time, or transection speed (Table 3). Blood loss correlated with transection time (r = 0.42; 95% CI, 0.27-0.55; P<.001) but not with speed (r = −0.06; 95% CI, −0.23 to 0.11; P = .50). Whole blood was transfused to 1 patient in each group (800 mL and 400 mL). During transection, major technical errors were made in 2 patients (both with the vibrating tip of the ultrasonic dissector): 1 injury to the right hepatic vein and 1 to the hilar bile duct (requiring temporal stenting), resulting in conversion to clamp crushing (Figure 2).

Ultrasonic dissection caused more frequent tumor exposure at the surgical margin (9 vs 3 patients; P = .09) and incomplete landmark appearance on the cut surface (12 vs 4 patients; P = .03) than did clamp crushing (Table 3). These 28 shortfalls in both measures occurred randomly with respect to trial period (15 in first half and 13 in second half) or surgeon (14 by consultant and 14 by resident specialist). Among 105 patients with tumor, there were 12 cases of histologically proven tumor exposure: 6 tumors were deeply located (>2 cm from the surface; 4 in segment VIII and 2 in segment I, all removed by segmentectomy), 3 were smaller multiple tumors (12, 12, and 13 mm; all removed by limited resection), 2 were early hepatocellular carcinoma with ill-defined borders on ultrasound (both removed by limited resection), and 1 was bile duct cancer (removed by extended right hepatectomy). In the other 93 patients, the median length of the surgical margin was not different (2 mm [range, 1-21 mm] for the ultrasonic group vs 4 mm [range, 1-22 mm] for the clamp group; P = .36). Among the 84 patients who underwent anatomical resection, landmark appearance was incomplete in 16 because of tearing of hepatic venous tributaries or the liver parenchyma in 7, misdirection of dissection in 6, and unknown causes in 3.

After surgery, 20 patients (30%) in the ultrasonic group had 25 complications, and 14 (21%) in the clamp group had 16 complications (P = .32) (Table 3). A total of 36 events were minor: 14 were pleural effusion, 5 were hyperamylasemia, 4 were atelectasis, 3 were delayed gastric emptying, 2 were cholangitis, 2 were wound infection, and 6 were other complications (all cured with conservative treatments). The 5 remaining events were major: 1 subphrenic abscess and 1 liver abscess (both requiring drainage) in the ultrasonic group, and 1 liver abscess (requiring drainage) and 2 bile leaks (spontaneously resolved in 21 and 36 days) in the clamp group. Postoperative hepatic function and hospital stay were similar in both groups, and there was no operative mortality in either group.

Between the groups, hepatectomy scores for surgical margin and landmark appearance were different, and the median sum score was significantly lower in the clamp group than in the ultrasonic group (4.0 [range, 0-12] vs 5.0 [range, 0-19]; 95% CI for difference, −2.0 to 0; P = .03) (Table 4). The hepatectomies done with clamp crushing had a significantly higher grade than those done with ultrasonic dissection (P = .05). Clamp crushing resulted in a higher proportion of grade B and a lower proportion of grade D hepatectomies than did ultrasonic dissection.

The transection method was found to independently influence hepatectomy grade; ultrasonic dissection was related to a higher risk of poorer grade (grade C or D) than was clamp crushing (adjusted OR = 3.06; 95% CI, 1.35-6.92; P = .01) but had no significant relevance to blood loss (adjusted OR = 1.73; 95% CI, 0.80-3.76; P = .16). In addition, blood loss was significantly influenced only by indocyanine clearance (≥20% vs <20%; adjusted OR = 5.42; 95% CI, 1.43-20.53; P = .01).

Our randomized trial shows that ultrasonic dissection offers no benefit compared with clamp crushing in partial hepatectomy. Although there was no difference in blood loss or transection time between the 2 groups, the hepatectomies with clamp crushing had significantly higher grades than those with ultrasonic dissection. Clamp crushing was therefore superior to ultrasonic dissection in terms of the overall quality of hepatectomy.

For hepatic parenchymal transection, these 2 methods are the most commonly used worldwide,^1- 11 but randomized comparison remains to be conducted. In this trial, we enrolled both living donors and patients with a diverse range of hepatobiliary tumors to derive broadly applicable results. To assure technical standardization, all operations were performed according to the maneuver by Makuuchi et al.^5,10,13- 15 The study design consisted of randomization in the operating room, uniform perioperative care, independent data collection, and masked outcome assessment.

Surgical experience and skill were important to this trial; the 2 consultants were highly experienced in both techniques, and our resident specialists might have already mastered the learning curve.⁵ An experienced liver surgery group reported a significant reduction in blood loss after changing their technique from clamp to ultrasonic transection, although their experience with ultrasonic dissection was clearly less than that with clamp crushing.¹² Our results do not support such findings of retrospective studies and suggest that surgeons who recommend the use of an ultrasonic dissector^3,6- 7,9,11 overestimate its effectiveness because of expectation bias. A plausible interpretation of our results is that surgeons experienced in hepatic resection will achieve comparable levels of blood loss and transection time whether or not an ultrasonic dissector is used. In this trial, the effect of the surgeon's experience on outcome was uncertain because resident specialists were always assisted by the consultant. Further trials are necessary to determine which method is easier to learn and whether similar outcomes can be reproduced by general surgeons.

The median blood loss with both methods (515 and 452 mL) contrasts sharply with that (800-1200 mL) reported by other specialists^2- 4,6- 7 and was even smaller than in our previous results (700 mL).⁵ The lower blood loss may be due partly to the smaller tumor size and the inclusion of living donors in our trial, but the extent of hepatectomy, a known determinant of bleeding,^19- 20 was comparable with that in other studies.^4,7- 8 Unless major errors occur, most bleeding during transection with inflow occlusion results from tears in the hepatic venous tributaries or liver parenchyma. Similar blood loss in both groups suggests no difference in the risk of such injury between the 2 methods, which conflicts with the concept that clamp crushing is more blunt than ultrasonic dissection.^3,11- 12 Our rule in transection procedures is to stop any bleeding before proceeding¹³; however, this did not prolong the transection time (61 and 54 minutes) as compared with reported results (46-60 minutes).^7,21 Although blood transfusion rates of 25% to 30% are routine even at high-volume centers,^4,6- 8 only 2 of our patients (1.5%) required transfusion even though criteria for this procedure were similar.^4,8 Multivariate analysis revealed that blood loss was unrelated to transection method and was related only to liver consistency, as indicated by indocyanine clearance.²²

We attempted to achieve the protocol requirements for surgical margins and landmark appearance. In this series, 12 histologic tumor exposures may have been caused by deeply located tumors, multiple resections, or early cancer.²³ Because of a similar incidence of baseline characteristics, the higher exposure rate (17%) in the ultrasonic group (although similar to reported rates of approximately 15%^24- 25) suggests inherent drawbacks of ultrasonic transection. Difficulty in handling at the resection bottom, limitations in making a small-range curved plane, or a large transection width (speculated by its narrower median clearance [2 mm vs 4 mm]) may have contributed to the results. We assessed the outcome of anatomical resection by the appearance of intersegmental landmark vasculature, an important technical and prognostic measure.^10,13,17 The 16 cases of incomplete landmark appearance were related to technical problems, such as tearing of hepatic venous tributaries and misdirection of dissection. The significantly higher success rate (90%) with clamp crushing indicates that it is more efficient than ultrasonic dissection for this task. The possibility that our different amounts of experience with the 2 techniques had an influence is unlikely; in terms of both tumor exposure and incomplete landmark appearance, there was no detectable correlation between the 28 shortfalls and their timing or the operating surgeon.

The postoperative morbidity rate in the ultrasonic group (30%) was similar to reported rates (23%-46%),^2- 8 whereas that in the clamp group (21%) was lower. Major complications, occurring in 5 patients (4%), were less frequent than in other studies (8%-15%).^2,6- 7,9 Although the decreased morbidity rate favored clamp crushing, no difference in hospitalization length was found, perhaps because of a difficulty in defining exact criteria for discharge. There was no operative mortality in this series, consistent with our previous results and those of others.^5- 6,9

We first developed the hepatectomy grading system to assess the operation quality. Sum scores based on 6 clinically relevant criteria may provide a benchmark for comprehensive evaluation. A 2-fold weight was assigned to the scores for surgical margin and landmark appearance because both were predefined goals. The significant difference in operation grade, confirmed by multivariate analysis, suggests that clamp crushing affords a higher quality of hepatectomy than does ultrasonic dissection. Whether this finding may be generally acceptable should be clarifed in future trials involving neophyte or inexperienced liver surgeons; we cannot exclude the possibility that the difference in quality is due to our different amounts of experience with the 2 techniques.

Because clamp crushing is simple, inexpensive, and ensures quality of hepatectomy, it is therefore regarded as the option of choice for hepatic parenchymal transection.

This work was supported in part by a grant-in-aid for cancer research from the Ministry of Health and Welfare, Tokyo, Japan.

We thank Seiji Kawasaki, MD (Shinshu University, Matsumoto, Japan), Hidetaka Kawabata, MD, and Takafumi Ueno, MD (University of Tokyo), and Junji Yamamoto, MD (National Cancer Center, Tokyo), for reviewing the manuscript; Nobuyoshi Aoyanagi, MD, Minoru Watanabe, MD, Atsushi Maema, MD, Akira Matsukura, MD, Taku Aoki, MD, Akira Takatsu, MD, and Yasuji Seyama, MD (University of Tokyo), and Wataru Kimura, MD (Yamagata University, Yamagata, Japan), for assessing the surgical outcomes; and Masami Minagawa, MD (University of Tokyo), and Yasuhiko Tsujino, PhD (Taiho Pharmaceutical Co, Tokyo), for statistical help.

Corresponding author: Masatoshi Makuuchi, MD, PhD, Division of Hepato-Biliary-Pancreatic and Transplantation Surgery, Department of Surgery, Faculty of Medicine, University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan (e-mail: makuuchi-tky@umin.u-tokyo.ac.jp).