Two Hundred Forty Consecutive Portal Vein Embolizations Before Extended Hepatectomy for Biliary Cancer

Abstract

Objective:

To assess clinical benefit of portal vein embolization (PVE) before extended, complex hepatectomy for biliary cancer.

Summary Background Data:

Many investigators have addressed clinical utility of PVE before simple hepatectomy for metastatic liver cancer or hepatocellular carcinoma, but few have reported PVE before hepatectomy for biliary cancer due to the limited number of surgical cases.

Methods:

This study involved 240 consecutive patients with biliary cancer (150 cholangiocarcinomas and 90 gallbladder cancers) who underwent PVE before an extended hepatectomy (right or left trisectionectomy or right hepatectomy). All PVEs were performed by the “ipsilateral approach” 2 to 3 weeks before surgery. Hepatic volume and function changes after PVE were analyzed, and the outcome also was reviewed.

Results:

There were no procedure-related complications requiring blood transfusion or interventions. Of the 240 patients, 47 (19.6%) did not undergo subsequent hepatectomy. The incidence of unresectability was higher in gallbladder cancer than in cholangiocarcinoma (32.2% versus 12.0%, P < 0.005). The remaining 193 patients (132 cholangiocarcinomas and 61 gallbladder cancers) underwent hepatectomy with resection of the caudate lobe and extrahepatic bile duct (n = 187), pancreatoduodenectomy (n = 42), and/or portal vein resection (n = 63). Seventeen (8.8%) patients died of postoperative complications: mortality was higher in gallbladder cancer than in cholangiocarcinoma (18.0% versus 4.5%, P < 0.05); and it was also higher in patients whose indocyanine green clearance (KICG) of the future liver remnant after PVE was <0.05 than those whose index was ≥0.05 (28.6% versus 5.5%, P < 0.001). The 3- and 5-year survival after hepatectomy was 41.7% and 26.8% in cholangiocarcinoma and 25.3% and 17.1% in gallbladder cancer, respectively (P = 0.011). In 136 other patients with cholangiocarcinoma who underwent a less than 50% resection of the liver without PVE, a mortality of 3.7% and a 5-year survival of 27.6% were observed, which was similar to the 132 patients with cholangiocarcinoma who underwent extended hepatectomy after PVE.

Conclusions:

PVE has the potential benefit for patients with advanced biliary cancer who are to undergo extended, complex hepatectomy. Along with the use of PVE, further improvements in surgical techniques and refinements in perioperative management are necessary to make difficult hepatobiliary resections safer.

Preoperative portal vein embolization was performed on 240 consecutive patients with biliary cancer. Of these patients, 193 underwent extended hepatectomy with resection of the caudate lobe and extrahepatic bile duct (n = 187), pancreatoduodenectomy (n = 42), and/or portal vein resection (n = 63). Overall surgical mortality was 8.8%, and 5-year survival was 26.8% in cholangiocarcinoma and 17.1% in gallbladder cancer, respectively.

Portal vein embolization (PVE), a procedure devised by Makuuchi et al^1,2 and Kinoshita et al,³ is performed widely as a presurgical treatment of patients undergoing extended hepatectomy to minimize postoperative liver dysfunction. We also have used this radiologic intervention and developed the “ipsilateral approach” of percutaneous transhepatic PVE,⁴ a method of “trisegment” PVE,^5,6 and portal plus arterial embolization.⁷ Further, we investigated the changes in volume,⁸ function,⁹ and hemodynamics^10,11 of the liver that occur after PVE.

Many authors have documented clinical utility of PVE before hepatectomy for metastatic liver cancer^12–14 or hepatocellular carcinoma.^15–17 The efficacy, however, of PVE for biliary cancer has been described in few reports with limited numbers of patients.^2,18–20 Hepatectomy for biliary cancer, unlike that for metastatic liver cancer or hepatocellular carcinoma, is a difficult procedure, often requiring caudate lobectomy and extrahepatic bile duct resection with bilioenteric anastomosis, portal vein resection with reconstruction,²¹ and pancreatoduodenectomy.²² This surgery, especially in patients with obstructive jaundice, is still risky and risk reduction remains a primary challenge for hepatobiliary surgeons. In this study, we outlined our experiences of PVE in 240 consecutive patients with biliary cancer and discussed the benefits of PVE before extended hepatectomy.

PATIENTS AND METHODS

Patients

This study involved 240 consecutive patients with biliary cancer who underwent PVE before a planned extended hepatectomy (right or left trisectionectomy or right hepatectomy) at Nagoya University Hospital between January 1991 and May 2005. During this period, we treated 393 patients with perihilar or intrahepatic cholangiocarcinoma (292 resected, 101 unresected) and 231 patients with gallbladder cancer (151 resected, 80 unresected). These 240 patients accounted for 86.2% of a total of 279 patients who received PVE during the same period.

The subjects comprised 125 men and 115 women, with an average age of 63 ± 11 years (range, 35–83 years). A total of 150 patients had cholangiocarcinoma (142 perihilar and 8 intrahepatic carcinomas) and the remaining 90 had advanced gallbladder carcinoma involving the hepatic hilus. Of the 240 patients, 193 (80.4%) had jaundice on admission and then underwent percutaneous transhepatic biliary drainage (PTBD) before PVE. Another 23 patients without jaundice also underwent PTBD to relieve cholangitis and/or to determine the extent of cancer.

PVE

In principle, PVE was indicated when a future liver remnant was estimated to be less than 40% and was performed 2 to 3 weeks before a scheduled liver resection. In jaundiced patients, the intervention was carried out after the serum total bilirubin concentrations had decreased at least less than 5 mg/dL following PTBD. All PVEs were performed by the ipsilateral approach, in which the portal vein was accessed by ultrasound-guided puncture of the portal vein to be embolized.^4–6 The right anterior portal vein was accessed in 236 patients, the left portal vein in 3, and the right posterior portal vein in 1. When right trisectionectomy was planned, the right portal vein plus the left medial portal vein were embolized. Similarly, when left trisectionectomy was scheduled, the left portal vein plus the right anterior portal vein were embolized, as previously reported.^5,6 Four patients whose hepatic functional reserve was markedly poor underwent arterial embolization of the segments to be resected 3 to 6 weeks after PVE.⁷ The actual number of embolized portal veins is summarized in Table 1.

TABLE 1. Type of Planned Hepatectomy and Embolized Portal Vein

From 1991 to 2000, fibrin glue (Bolheal; Fujisawa Pharmaceutical, Tokyo, Japan) mixed with iodized oil (Lipiodol; Kodama Pharmaceutical, Tokyo, Japan) was used as the embolic material (n = 131). After 2001, absolute ethanol with embolization steel coils (Cook, Bloomington, IN) were used (n = 109)¹⁶ because the Prefectural Insurance System prohibited the use of fibrin glue because of its high price. All PVEs were carried out by the authors, not by radiologists.

Liver Volume and Function Measurement

Computed tomography (CT) of the liver was used for volume determination. CT scans at 3- to 10-mm intervals from the dome to the most inferior part of the liver were obtained with enhancement by an intravenous bolus injection of contrast medium. Each slice of the liver was traced with the cursor, and the corresponding area was calculated by computer. Total volume of each lobe was obtained by adding the volumes of individual slices.⁸ The volume change in the nonembolized lobe was expressed as the ratio of the volumes measured after and before PVE and was defined as the hypertrophy ratio, and that of the embolized lobe was defined as the atrophy ratio. In this study, the volume of the caudate lobe was excluded from the volume of the nonembolized lobe because this lobe was not embolized but was resected in almost all cases. All patients underwent CT within 3 weeks after PVE. In 10 (4.2%) patients, laparotomy was postponed because of an insufficient volume of the future remnant liver, and repeat CT was taken 2 to 3 weeks after the first post-PVE CT. Four of the 10 patients underwent arterial embolization after PVE,⁷ as mentioned above. Excluding these 4 patients, the time between PVE and the final CT after PVE was 15 ± 5 days (median, 14 days; range, 6–57 days).

Indocyanine green (ICG) tests were performed 1 to 3 days before and 11 to 13 days after PVE. ICG (0.5 mg/kg of body weight) was administered via a peripheral vein, and venous blood was sampled before and 5, 10, and 15 minutes after injection. Specimens were analyzed for ICG concentrations on a spectrophotometer at 805 nm. The plasma disappearance rate of ICG (KICG) was calculated by linear regression analysis of plasma ICG concentrations.⁹ KICG of the nonembolized (future remnant) lobe was also calculated using the formula, KICG× % volume of the nonembolized lobe/100.

Other liver function parameters, such as the serum total bilirubin, aspartate transaminase, and alanine transaminase concentrations, were measured regularly before and after PVE, using standard laboratory methods.

Statistics

Results are expressed as mean ± SD. Continuous data were evaluated using the paired and unpaired Student t tests. Categorical data were compared using the χ² test and Fisher exact test, where appropriate. Postoperative survival was calculated using the Kaplan-Meier method. Differences in survival curves were compared using the log-rank test. P < 0.05 was considered statistically significant.

RESULTS

PVE-Related Complications

A few patients developed an inflammatory response manifested by mild fever and/or mild abdominal pain or discomfort. Transaminase concentrations were elevated, usually less than triple, in most patients but returned to baseline within a week. Hepatic enzyme variations after PVE had no clinical correlation. Serum total bilirubin remained near pre-embolization concentrations in all patients.

We encountered rare complications in 2 patients. In 1 patient with a hilar cholangiocarcinoma, hypersplenism with splenomegaly developed after PVE because the right portal vein was embolized despite marked stenosis of the left portal vein due to cancer invasion.²³ This patient underwent right hepatectomy with caudate lobectomy and portal vein resection and reconstruction, as scheduled. Postoperative recovery was good, but the patient died of recurrence 37 months after surgery. In another patient with advanced gallbladder cancer, extensive portal and mesenteric vein thrombosis occurred after PVE, due to a protein S deficiency.²⁴ After thrombolytic therapy, surgery was done 51 days after PVE but resulted in only a laparotomy due to locally advanced cancer. The patient died of cancer 12 months after laparotomy.

Overall, there were no complications requiring blood transfusion or radiologic or surgical intervention.

Surgical Outcome After PVE

Of the 240 patients with PVE, 47 (19.6%) did not undergo subsequent hepatectomy (Table 1): 7 patients showed disease progression severe enough to preclude curative resection during the waiting period after PVE and the remaining 40 underwent only laparotomy due to peritoneal dissemination, liver metastasis, and/or locally advanced cancer. The incidence of unresectability after PVE was significantly higher ingallbladder cancer than in cholangiocarcinoma (32.2% = 29/90 versus 12.0% = 18 of 150, P < 0.005). PVE-related inconvenience was not encountered in the outcome of these unresected patients.

A total of 193 patients underwent hepatectomy (Table 2). En bloc resection of the caudate lobe and extrahepatic bile duct was performed in 187 (96.9%) patients, combined pancreatoduodenectomy (PD) in 42 (21.8%), portal vein resection with reconstruction in 63 (32.6%), and hepatic artery resection with reconstruction in 2 (1.0%). Simple hepatectomy without combined resection was performed in only 6 patients. All patients underwent regional lymph node dissection, and some selected patients underwent additional paraaortic lymph node dissection. Operative time was 714 ± 166 minutes (median, 710 minutes; range, 350–1210 minutes), and blood loss was 2860 ± 2529 mL (median, 1940 mL; range, 235–17,114 mL).

TABLE 2. Surgical Procedures in 193 Patients Who Underwent Resection

Of the 193 patients who underwent hepatectomy, 17 (8.8%) patients died of postoperative complications, including intraabdominal bleeding (n = 2), myocardial infarction (n= 1), and liver failure or multiple organ failure (n = 14) (Table 3). Surgical mortality was significantly higher in gallbladder cancer than in cholangiocarcinoma (18.0% versus 4.5%, P < 0.005). Mortality also was higher in hepatectomy with combined PD than in that without PD, although statistically not significant. With time, mortality has improved yearly with the most recent period showing an acceptable mortality of 4.8% (Table 4).

TABLE 3. Details of 17 Hospital Deaths After Extended Hepatectomy

TABLE 4. Mortality After Extended Hepatectomy Following Portal Vein Embolization

During the same period, 136 other patients with cholangiocarcinoma (126 perihilar and 10 intrahepatic carcinomas) underwent a less than 50% resection of the liver without PVE. The type of hepatectomy performed was as follows: resection of segments 1, 2, 3, and 4 in 112 patients, resection of segments 1, 4, 5, and 8 in 8, resection of segments 1, 5, and 8 in 5, resection of segment 1 in 5, and other resections in 6. Extrahepatic bile duct resection with reconstruction was performed in 128 patients, combined portal vein resection in 35, and combined pancreatoduodenectomy in 18. The mortality of these 136 patients was 3.7% (5 of 136), being similar to mortality of the 132 patients with cholangiocarcinoma who underwent extended hepatectomy after PVE.

Changes in Hepatic Lobe Volume After PVE

Changes in hepatic lobe volume after PVE were analyzed in 189 of the 193 hepatectomized patients (4 patients who underwent portal plus arterial embolization were excluded). The calculated volume of the nonembolized lobe increased from 361 ± 119 cm³ (range, 103–700 cm³) before PVE to 460 ± 120 cm³ (range, 239–747 cm³) after PVE (P < 0.0001), whereas the volume of the embolized lobe decreased from 688 ± 167 cm³ (range, 341–1172 cm³) before PVE to 581 ± 149 cm³ (range, 261–1009 cm³) after PVE (P < 0.0001). The volume of the whole liver exhibited no significant changes. In terms of the volumetric ratio to the whole liver, the volume of the nonembolized lobe increased from 33% ± 8% (range, 15%–49%) before PVE to 43% ± 8% (range, 23%–68%) after PVE (P < 0.0001). The hypertrophy ratio of the nonembolized lobe was 133% ± 24% (range, 100%–222%), and the atrophy ratio of the embolized lobe was 85% ± 11% (range, 55%–115%). In the 47 patients with unresectable tumor, hypertrophy and atrophy ratios were 130% ± 27% and 86% ± 13%, respectively, being similar to those in the hepatectomized patients. Gender, embolic material, jaundice on admission, or diabetes mellitus did not affect volume dynamics after PVE (Table 5).

TABLE 5. Potential Factors Affecting Hypertrophy and Atrophy After PVE

Association Between Hepatic Functional Reserve and Surgical Outcome

In the 193 hepatectomized patients, KICG increased from 0.151 ± 0.034 (range, 0.066–0.265) before PVE to 0.157 ± 0.033 (range, 0.080–0.259) after PVE (P < 0.05). KICG of the future remnant (nonembolized) lobe also improved from 0.050 ± 0.016 (range, 0.018–0.094) before PVE to 0.066 ± 0.016 (range, 0.027–0.118) (P < 0.0001). Before PVE 86 (44.6%) patients had a KICG of the future liver remnant of ≥0.05, while after PVE 165 (85.5%) patients had a KICG of ≥0.05 (Fig. 1).

FIGURE 1. Changes in plasma disappearance rate of indocyanine green (KICG) of the nonembolized (future remnant) lobe before and after portal vein embolization (PVE).

The hypertrophy ratio of the nonembolized lobe was not different between the 17 nonsurvivors and the 176 survivors (136% ± 24% versus 132% ± 21%). However, KICG of the future liver remnant after PVE was worse in the 17 nonsurvivors than in the 176 survivors (0.055 ± 0.017 versus 0.066 ± 0.016, P < 0.05). In 28 patients whose KICG of the future liver remnant after PVE was <0.05, 8 (28.6%) patients died of postoperative complications, while in 165 patients whose KICG of the future liver remnant after PVE was ≥0.05, 9 (5.5%) patients died (P < 0.001) (Fig. 2; Table 3). In 3 (cases 10, 11, and 15 in Table 3) of these 9 patients, the cause of death was not directly related to postoperative liver failure. Two other patients (cases 13 and 14) had a markedly fatty liver despite having improved ICG test results, which probably led to organ failure. In another patient (case 17), liver failure developed due to massive intraoperative bleeding. The remaining 3 (cases 9, 12, and 16) had intractable septic complications, which resulted in organ failure.

FIGURE 2. Association between the surgical outcome and plasma disappearance rate of indocyanine green (KICG) of the nonembolized (future remnant) lobe after portal vein embolization (PVE).

Staging and Survival in Resected Patients

Staging of the tumor was described using the TNM Classification of Malignant Tumors by the International Union Against Cancer (6th edition, 2002).²⁵ Most of the patients with cholangiocarcinoma and all of the patients with gallbladder cancer had pT3 or pT4 diseases. More than half of the patients had lymph node metastases. pm1 disease was found in 25 (18.9%) of the patients with cholangiocarcinoma (paraaortic node metastases in 15, liver metastases in 6, and localized dissemination in 4) and 19 (31.1%) of the patients with gallbladder cancer (paraaortic node metastases in 16, liver metastases in 4, and localized dissemination in 4; duplicated in 5 patients) (Table 6).

TABLE 6. TNM Classification and Staging in 193 Resected Patients

The 3- and 5-year survival after extended hepatectomy following PVE (including all deaths) was 41.7% and 26.8% in cholangiocarcinoma and 25.3% and 17.1% in gallbladder cancer, respectively. Survival was significantly improved in cholangiocarcinoma, as compared with gallbladder cancer (P = 0.011). Sixteen patients with cholangiocarcinoma and 5 patients with gallbladder cancer survived more than 5 years after extended hepatectomy. Survival for the 136 patients with cholangiocarcinoma who underwent a less than 50% resection of the liver was almost identical to that of the 132 patients with cholangiocarcinoma who underwent extended hepatectomy after PVE (Fig. 3).

FIGURE 3. Survival after hepatectomy.

DISCUSSION

Today, elective liver resection for hepatocellular carcinoma or metastatic liver cancer can be performed with an acceptable mortality.^26,27 Hepatectomy, however, for biliary hilar malignancies, including perihilar cholangiocarcinoma and gallbladder carcinoma involving the hepatic hilus, remains a greater risk.^28–30 Previously, we studied the risk of postoperative bacteremia in 407 patients who underwent elective liver resection. Multivariate analysis revealed that presence of obstructive jaundice on admission increased postoperative bacteremic risk by 4.7 times and performance of extended (>50%) hepatectomy increased the risk by 2.7 times.³¹ This indicates that, when a biliary cancer patient with obstructive jaundice undergoes extended hepatectomy, bacteremic risk increases by 12.7 times. Belghiti et al³² also analyzed operative risks in 747 hepatectomized patients and showed that the mortality of patients with jaundice was extremely high (21%), as compared with that of patients with a normal liver. In addition, they demonstrated that a combined extrahepatic procedure was the only independent predictor of in-hospital death in patients with malignancy (relative risk of 7.5). In our series, 80% of the subjects had obstructive jaundice; all of the resected patients underwent extended hepatectomy; and 97% of the hepatectomized patients received extrahepatic procedures. Considering difficulties arising from such a study population, the overall mortality of 8.8% in this study is acceptable.

Many studies have shown that PVE is effective in inducing hypertrophy of nonembolized hepatic segments.1–8,12–20 This study also demonstrated a sizable atrophy/hypertrophy complex after PVE. The potential benefits, however, of this intervention have not been validated by a randomized controlled trial. In addition, indications for this intervention are still unclear because of the few data regarding the minimal hepatic volume required to tolerate surgery without serious complications. For normal livers, some authors have stressed that more than 40% of the total liver volume must be preserved to make the resection safe,³³ but the general consensus is that a 25% to 30% remnant is the minimum essential volume.19,20,34,35 Vauthey et al³⁵ reported that major postoperative complications increased when the estimated hepatic remnant was less than 25%. A larger remnant may be necessary even in normal livers when complex hepatectomy is planned. Farges et al³⁶ conducted a prospective, but not randomized, study to assess the benefits of PVE before right hepatectomy. They demonstrated that PVE has no beneficial effect on the postoperative course in patients with normal livers but significantly reduced postoperative complications in patients with chronic liver diseases. Based on these previous studies, PVE before extended hepatectomy may be considered in patients with damaged livers, and also even in patients with normal livers when planned procedure is complex or estimated hepatic remnant is less than 25% to 30%.^19,34

We demonstrated that the KICG of the future liver remnant increased after PVE. Before PVE only 86 (44.6%) of the patients had KICG of the future liver remnant of ≥0.05, while after PVE 165 (85.5%) of the patients had KICG ≥0.05. We also showed that the postoperative mortality was significantly greater when the KICG of the future liver remnant was <0.05 (28.6% versus 5.5%). Recently, Cherqui et al³⁷ recommended “rapid hepatectomy” without preoperative biliary decompression and portal vein embolization in jaundiced patients with biliary hilar malignancies. If our patients had undergone “rapid hepatectomy” without preoperative intervention, the operative mortality may have increased up to at least 18.3% [= (86 × 0.055 + 107 × 0.286)/193]. In addition, this study demonstrated that in cholangiocarcinoma the mortality was similar in patients who underwent extended hepatectomy following PVE and those who underwent a less than 50% resection of the liver without PVE (4.5% versus 3.7%). Furthermore, the survival for these 2 patients groups was similar. Although this study was retrospective, our observations clearly indicate that PVE has potential benefits clinically. There is a possibility, however, that PVE was overused. For example, PVE might have been unnecessary for patients whose KICG of the future liver remnant before PVE was good, ie, >0.08. We agree with Abdalla et al³⁸ who stated that “randomized trial cannot be recommended to test the efficacy of PVE, for it would be unethical to deny the benefit of the technique and safer resection to patients who are otherwise poor candidates for resection based on inadequate liver size or function.” Continued prospective analyses of the association between hepatic volume and function and morbidity and mortality should clarify the appropriate indication of PVE.

In this study, the hypertrophy ratio of the nonembolized lobe was not different between the 17 nonsurvivors and the 176 survivors, while KICG of the future liver remnant after PVE was significantly lower in the former than in the latter. This observation indicates that hepatic function as well as liver volume is important as a guide to decision-making in liver resection.³³ In this context, KICG of the future liver remnant is useful, as this index includes elements of both function and volume. Based on our results, 0.05 of KICG of the future liver remnant may be a “cutoff” value for deciding liver resection. Actually, in the most recent 3 years, we have used this value for this sake; however, further study is required to validate whether this cutoff value is appropriate.

Results of PVE in cholangiocarcinoma, including both the surgical outcome and survival, were acceptable, whereas the results of advanced gallbladder cancer were poor. First, the incidence of unresectability after PVE was 32.2% in gallbladder cancer, which indicated that approximately one third of PVEs in gallbladder cancer were unnecessary. Therefore, staging laparoscopy before PVE is essential, and recently we have used this diagnostic modality selectively for patients with gallbladder cancer. Weber et al³⁹ demonstrated that staging laparoscopy before surgery in patients with biliary cancer was more useful in cases of gallbladder cancer, than in cholangiocarcinoma. Second, mortality after extended hepatectomy was significantly higher (approximately 4 times) in gallbladder cancer than in cholangiocarcinoma. The main reason for this poor outcome is that combined resections, including pancreatoduodenectomy and/or portal vein resection, were performed more frequently in patients with gallbladder cancer. Some of the patients with gallbladder cancer had extremely advanced disease, which impaired the host immune function⁴⁰ and may have contributed to the poor outcome. Survival also was worse in gallbladder cancer compared with cholangiocarcinoma. All of our patients with gallbladder cancer had advanced disease with biliary hilar invasion. Most of these patients were considered to be beyond the scope of surgical resection in Western centers with few 5-year survivors reported.^41,42 Thus, the fact that 5 patients survived more than 5 years suggests a potential benefit of the aggressive surgery after PVE.

CONCLUSION

Although possible overutilization cannot be denied, PVE may benefit patients with advanced biliary cancer undergoing extended, complex hepatectomy. PVE, however, is not an omnipotent tool; therefore, further improvements in surgical techniques and refinements in perioperative management are necessary to make difficult hepatobiliary resections safer.