Is Extended Hepatectomy for Hepatobiliary Malignancy Justified?

Abstract

Background:

Extended hepatectomy may be required to provide the best chance for cure of hepatobiliary malignancies. However, the procedure may be associated with significant morbidity and mortality.

Methods:

We analyzed the outcome of 127 consecutive patients who underwent extended hepatectomy (resection of ≥ 5 liver segments) for hepatobiliary malignancies.

Results:

The patients underwent extended hepatectomy for colorectal metastases (n = 86; 67.7%), hepatocellular carcinoma (n =12; 9.4%), cholangiocarcinoma (n =14; 11.0%), and other malignant diseases (n =15; 11.5%). Thirty-two left and ninety-five right extended hepatectomies were performed. Eight patients also underwent caudate lobe resection, and 40 patients underwent a synchronous intraabdominal procedure. Twenty patients underwent radiofrequency ablation, and 31 underwent preoperative portal vein embolization. The median blood loss was 300 mL for right hepatectomy and 600 mL for left hepatectomy (P = 0.02). Thirty-six patients (28.3%) received a blood transfusion. The overall complication rate was 30.7% (n = 39), and the operative mortality rate was 0.8% (n = 1). Significant liver insufficiency (total bilirubin level > 10 mg/dL or international normalized ratio > 2) occurred in 6 patients (4.7%). Multivariate analysis showed that a synchronous intraabdominal procedure was the only factor associated with an increased risk of morbidity (hazard ratio [HR], 4.9; P = 0.02). The median survival was 41.9 months. The overall 5-year survival rate was 25.5%.

Conclusions:

Extended hepatectomy can be performed with a near-zero operative mortality rate and is associated with long-term survival in a subset of patients with malignant hepatobiliary disease. Combining extended hepatectomy with another intraabdominal procedure increases the risk of postoperative morbidity.

The perioperative and long-term survival of 127 patients who underwent extended hepatectomy for hepatobiliary malignancy were analyzed. The authors report that extended hepatectomy can be performed with a near-zero operative mortality rate and is associated with long-term survival in a subset of patients with malignant hepatobiliary disease.

Without treatment, the prognosis of primary and secondary hepatic malignancies is dismal. Because primary treatment with medical therapy does not yield durable results, hepatic resection is considered to be the optimal treatment modality with a cure rate between 20% to 35%.^1,2 Unfortunately, many patients with malignant hepatic tumors are not candidates for resection due to the extent of their underlying liver disease or the burden of tumor present in their liver.

Extended hepatectomy (resection of ≥ 5 liver segments) may be the only means available to achieve complete resection and provide a chance for cure in some patients with multiple or large tumors. Although recent advances in patient selection and surgical technique have resulted in low mortality after liver resection,^3,4 the mortality rate for extended hepatectomy is ≥ 5% and the morbidity rate is ≥ 50%.^5–9 A recent study reported an association between perioperative complications and the number of segments resected.⁵

The objective of this study was to review our experience with extended hepatic resection to identify the variables associated with morbidity, mortality, and survival. A more detailed understanding of the factors associated with perioperative risks could optimize the selection of patients most likely to benefit from extended hepatic resection versus alternative surgical and/or medical treatment.

PATIENTS AND METHODS

Between October 1993 and April 2003 a series of 127 consecutive patients underwent extended hepatic resection (resection of ≥ 5 liver segments) performed by 2 surgeons (JNV and SAC). All patients with histologically confirmed primary or metastatic hepatic malignancies with no clinical, radiographic, or intraoperative evidence of unresectable extrahepatic disease were eligible. In all cases, extended hepatic resection was performed with curative intent.

Preoperative Assessment

All patients were evaluated preoperatively with a baseline history and physical examination; serum laboratory tests including liver function tests and hepatitis B and C virus serology were performed as appropriate. In the first 5 years of the study, a computed tomography (CT) with arterial portography and more recently a helical CT with rapid injection of intravenous contrast (3 to 5 mL/s) and 5-mm cuts through the liver were performed; if needed, further imaging was obtained as previously described.¹⁰ Patients presenting with biliary obstruction underwent preoperative endoscopic and/or percutaneous biliary drainage to provide effective clearance of jaundice and/or cholangitis.

In the absence of definite compensatory hypertrophy as a result of tumor growth, patients who were candidates for an extended right hepatectomy underwent 3-dimensional reconstruction of the future liver remnant (FLR) (Couinaud segments II + III ± I) as previously described.^11,12 The ratio of the measured FLR volume:total estimated liver volume was determined using a formula that is derived from the association between total liver volume and body surface area (BSA): Total Liver Volume (cm³) = −794.41 + 1267.28 × BSA (m²).^13,14 On the basis of previous volumetric studies, we used > 20% of the total estimated liver volume as the minimal safe FLR in most patients.¹² Those patients with an anticipated small FLR underwent embolization of the right and segment IV portal vein branches using an ipsilateral percutaneous approach to initiate hypertrophy of the FLR before resection.^13,15 After portal vein embolization (PVE), repeat CT scans were obtained at 3 to 4 weeks to assess for the extent of compensatory hepatic hypertrophy. Surgical decision-making was then appropriately based on post-PVE CT volumetric analysis.

Surgical Techniques

In most cases, an initial incision was made 2 cm below the right costal margin extending from the midline to the right flank. If there was no evidence of extrahepatic disease, the incision was extended to a bilateral subcostal incision or up along the midline (hockey stick incision). Upon entering the abdomen, a more thorough exploration was performed to rule out contraindications to resection (eg, extensive or unresectable extrahepatic disease and/or gross lymphadenopathy). An intraoperative ultrasound (IOUS) was systematically performed to confirm the preoperative imaging and to review the intrahepatic portal and hepatic vein anatomy.

We have previously described the technique of extended right and left hepatectomy in detail.¹⁶ In most patients extrahepatic division of the vascular inflow vessels was performed (portal vein and hepatic artery). Control of the right or left hepatic veins and of the middle hepatic vein, as appropriate, was also performed in most cases. The hepatic parenchyma was divided using the ultrasonic dissector and cautery (ValleyLab, Boulder, CO), as well as more recently the Tissuelink (TissueLink Medical Inc., Dover, NH). Major portal triad and hepatic veins were ligated or suture ligated with nonabsorbable sutures. During parenchymal dissection a Pringle maneuver was performed to reduce blood loss.¹⁷ The Pringle maneuver was applied by intermittently occluding the hepatic artery and portal vein for periods up to 15 minutes separated by 5 minutes of restored flow.

For those patients undergoing simultaneous radiofrequency ablation (RFA), the RF 2000 or 3000 generator system (Boston Scientific Corp., Natick, MA) and a LeVeen monopolar needle electrode (4.0-cm maximum array diameter) was used. The RF 2000 system consists of a generator that supplies up to 100 W of power, while the RF 3000 provides up to 200 W of power. The LeVeen needle electrode is a 15-gauge insulated cannula that contains 10 to 12 individual hook-shaped electrode arms that are deployed in situ. For tumors < 2.5 cm in diameter, the multiple array was deployed into the center of the tumor. For larger lesions, the array was first deployed at the most posterior interface (ultrasonographically) between tumor and normal liver parenchyma; it was subsequently withdrawn and redeployed at 1.5-cm intervals in the tumor. The electrode was optimally positioned to achieve complete destruction of tumor and at least a 1-cm zone of normal liver parenchyma when possible.

Two anesthesiologists (JFA and DLK) dedicated to the liver resection team assisted in most cases. In all patients, the primary goal was to minimize infusions and/or transfusions until the parenchymal transection was completed. Transient hypotension associated with hepatic mobilization (eg, in case of large hepatic tumors) was treated with vasopressor support (usually phenylephrine) rather than infusion or transfusion. In most instances, infusions or transfusions were delayed until the end of the parenchymal transection to maintain a low central venous pressure with a minimal acceptable urine output of 0.5 mL/kg per hour. After completion of parenchymal transection and hemostasis, the patients received infusion of crystalloids and/or albumin to render them euvolemic.

Postoperative Care

After hepatic resection, a predetermined clinical care pathway that included the timing of patient mobilization, diet progression, and the ordering of laboratory studies guided postoperative care. Patients were seen in follow-up within 1 month of discharge, every 3 to 4 months up to 3 years after treatment, and every 6 months thereafter. CT scan of the abdomen, a chest radiograph, and serum laboratory tests were obtained at the time of each follow-up visit.

For purposes of this study, the following were collected for all patients: patient age and sex; tumor histology, number, location, and size; type of surgical resection; operative details; estimated blood loss (EBL); the total number of units of blood or blood products transfused within the same hospital stay; disease status; follow-up date; death date; complication data. Distribution of complication rates and survival time were analyzed in relation to the different variables collected. Univariate tests (log-rank) were used to test for differences in these distributions by any single factor. Those factors that appeared to have significant impact on survival or time to progression were entered into a Cox proportional hazards model to test for significant effects while adjusting for multiple factors simultaneously. Perioperative mortality was defined as mortality within 30 days of surgery or within the same hospital stay. P values were considered as significant for P < 0.05. Continuous variables were reported as median and range unless stated otherwise.

RESULTS

A total of 127 extended hepatic resections were performed. There were 71 men (55.9%) and 56 women (44.1%) with a median age of 58 years (range, 28 to 81 years). Extended hepatic resection was used to treat metastatic colorectal cancer in the majority of cases: 86 patients (67.7%). Cholangiocarcinoma was the second most common diagnosis (n = 15; 11.8%), and hepatocellular carcinoma (HCC) was the third (n = 13; 10.2%), whereas other lesions were less common (Table 1). The median tumor size was 5.5 cm (range, 0.7 to 26.0 cm), and the median number of tumors was 2 (range, 1 to 12). Thirty-one patients (all with right hepatectomies) underwent preoperative PVE, with the majority of the PVEs occurring after 1999 (n = 24; 77.4%) (Fig. 1A). There was an 8.8% increase in the median percent FLR after PVE (25.8%) compared with before PVE (17%). Seventeen patients (13.4%) presenting with obstructive jaundice underwent preoperative endoscopic and/or percutaneous biliary drainage before resection.

TABLE 1. Demographics and Perioperative Results

FIGURE 1. (A) RFA, PVE, and an associated intraabdominal procedure were all significantly more common within the last 5 years of the study period. (B) Thirty-two extended left hepatic resections (25.2%) and 95 extended right hepatic resections (74.8%) were performed. The number of extended resections significantly increased over time (P = 0.04).

Thirty-two extended left hepatic resections (25.2%) and 95 extended right hepatic resections (74.8%) were performed. The number of extended resections significantly increased over time (P = 0.04) (Fig. 1B). In addition to an extended hepatic resection, 8 patients (6.3%) underwent simultaneous caudate lobe resection and 40 (31.5%) underwent a synchronous intraabdominal procedure (common bile duct resection, n = 16; resection/thrombectomy of the vena cava or portal vein with or without reconstruction, n = 6; small and large bowel resection, n = 6; major [> 50%] diaphragmatic resection, n = 5; right adrenalectomy, n = 4; pancreatectomy, n = 2; and Nissen fundoplication, n = 1). There was a significant increase in the number of synchronous procedures performed during the last 5 years of the study (1999 to 2003) (n = 29; 72.5%) compared with the first (1993 to 1998) (n = 11; 27.5%) (P = 0.001) (Fig. 1A). RFA was also carried out in 20 patients (15.7%) at the same time as the extended hepatic resection. Twelve (60%) of these patients underwent an extended left hepatic resection, and 8 (40%) had an extended right hepatectomy. In general, there was no difference in the median number of tumors (n = 2; range, 1 to 12) in patients who underwent RFA compared with those who did not. However, subset analysis of only patients with colorectal metastases revealed that patients who underwent synchronous extended resection and RFA were more likely to have ≥ 5 tumors (P = 0.004). Of the 20 RFA procedures, 19 (95%) were performed in the year 2000 or later (P = 0.008) (Fig. 1A).

The median operative time was 385 minutes (range, 180 to 977 minutes). The median central venous pressure before parenchymal transection was 6 (range, 1 to 14). A Pringle maneuver was used in 116 (91.3%) patients, with a median clamp time of 27 minutes (range, 4 to 94 minutes). Only 5 patients had a clamp time longer than 60 minutes; 1 of these 5 patients experienced postoperative hepatic dysfunction (bilirubin > 3 mg/dL). During the course of the operation, the hepatic vein was isolated in 23 (63.9%) of 36 extended left resections and 77 (84.6%) of 91 extended right resections. The median operative blood loss was 600 mL (range, 100 to 3500 mL) for an extended left hepatectomy and 300 mL (range, 100 to 1500 mL) for an extended right hepatectomy (P = 0.02). Of note, the median blood loss was the same in those patients who underwent a synchronous intraabdominal procedure compared with those who did not (median, 550 mL in both groups; P = 0.54). There was no statistical trend in blood loss with regard to the year in which surgery was performed (P = 0.82). Of the 127 patients, 36 (28.3%) received a blood transfusion. Patients who underwent an extended left hepatectomy were more likely to require a transfusion (P = 0.03).

The median hospital stay after extended hepatic resection was 8 days (range, 4 to 39 days) with no difference between extended left and right hepatectomy patients. The overall postoperative complication rate was 30.7% (n = 39). Twenty-four complications (18.9%) were liver-related, whereas 15 (11.8%) were not (Table 2). Of the 12 patients with hepatic insufficiency, only 6 (4.7%) had severe hepatic insufficiency (total bilirubin > 10 mg/dL and/or international normalized ratio (INR) > 2 more than 2 days after resection),^18,19 whereas the other 6 experienced clinically significant ascites and/or encephalopathy. Univariate analysis revealed that male gender (P = 0.02), age > 65 years (P = 0.04), and performance of a synchronous intraabdominal procedure (P = 0.04) were all associated with a higher complication rate. Other factors, such as tumor size, number of tumors resected, date of surgery, preoperative PVE, extended right versus left hepatectomy, EBL, RFA, and type of hepatic malignancy, did not significantly affect the postoperative complication rate (all P > 0.05). On multivariate analysis, the only factor associated with an increased risk of morbidity was the history of a synchronous intraabdominal procedure performed at the time of the extended hepatic resection (hazard ratio (HR) = 4.9; P = 0.02). Male gender and age failed to maintain significance on multivariate analysis.

TABLE 2. Postoperative Complications

Only 1 postoperative death occurred (n = 1; 0.8%). The patient was a 49-year-old, otherwise healthy woman with HCC. Due to extensive disease, she had received prolonged preoperative chemotherapy consisting of 5-fluorouracil, α-interferon, and capecitabine followed by hepatic artery chemoembolization. After responding to preoperative therapy, she underwent an extended right hepatectomy combined with RFA of segment III. Her operative course was uncomplicated; blood loss was 900 mL with an in-flow occlusion time of only 16 minutes. Her initial postoperative course was notable for transient hepatic dysfunction (total bilirubin level, 5.8 mg/dL). However, the patient shortly thereafter developed pneumonia with subsequent worsening hepatic insufficiency. Over the ensuing weeks, her total bilirubin rose to 25.1 mg/dL and she developed hepatorenal syndrome, which led to her demise on postoperative day 39.

At a median follow-up of 18.4 months, the median and 5-year overall survival were 41.9 months and 25.5%, respectively (Fig. 2). Univariate analysis revealed that age, margin status, and tumor size were significant factors affecting overall survival. Younger patients (< 65-year-old) had a median survival of 71.9 months compared with 31.0 months for patients ≥ 65 years (P = 0.01). Likewise, patients with a positive resection margin (n = 21; 16.5%) had a median survival of only 17.6 months compared with 55.5 months for those with a negative margin (P = 0.02). Whereas patients with smaller tumors (≤5 cm) had a median survival of 59.9 months, patients with larger tumors (>5 cm) had a median survival of 36.9 months (P = 0.01) (Fig. 3A). Although not statistically significant, patients who had RFA at the time of extended hepatic resection showed a trend towards worse survival. Patients undergoing extended hepatic resection plus RFA had a median survival of 14.8 months compared with 44.5 months for all other patients (P = 0.15) (Fig. 3B). Other factors, such as gender, history of PVE, extended right versus left hepatic resection, a simultaneous associated procedure, EBL, and tumor number, did not significantly affect survival on univariate analysis (all P > 0.05) (Fig. 4). On univariate analysis, the overall survival of patients with colorectal metastasis (55.6 months) did not differ significantly from patients with noncolorectal disease (42.0 months) (P = 0.35).

FIGURE 2. Overall and median survival for the entire cohort.

FIGURE 3. (A) Patients with a tumor size > 5.0 cm had a significantly worse median survival (36.9 months) compared with those patients with tumors ≤ 5 cm (59.9 months) (P = 0.01). (B) Similarly, although not statistically significant on univariate analysis, those patients who had RFA at the time of extended hepatic resection had a worse median survival (14.8 months) compared with patients who underwent only extended hepatic resection (44.5 months) (P = 0.15). On multivariate analysis, a history of concomitant RFA and extended hepatic resection was a significant risk for worse overall survival (HR = 3.34; P = 0.03).

FIGURE 4. Neither a history of preoperative PVE (A) nor the type of extended resection performed (right versus left) (B) had an effect on overall survival.

On multivariate analysis, both the size of the tumor and a history of RFA performed at the time of the extended hepatic resection significantly affected survival. Patients who had a tumor size > 5 cm had a higher likelihood of death than those who underwent resection for a lesion ≤ 5 cm (HR = 3.54; confidence interval [CI] = 1.44 to 8.72; P = 0.01). Similarly, the need to perform RFA for additional disease in the remaining unresected segment of liver was associated with a poor outcome (HR = 3.34; CI = 1.15 to 9.72; P = 0.03).

DISCUSSION

The first elective liver resection is credited to Langenbuch, who in 1888 performed a bisegmentectomy II + III.²⁰ It was not until 1952, however, that Lortat-Jacob reported the first anatomic extended right hepatectomy. The procedure included a thoracoabdominal incision, extrahepatic control of both the inflow and outflow vessels of the right liver, as well as resection in the plane between segment IV and bisegment II + III.²¹ Since then, the techniques of extended right hepatectomy (Couinaud segments IV - VIII) and extended left hepatectomy (Couinaud segments II–V and VIII) have been described in detail.^16,22 In fewer than 2 decades, hepatic resection has evolved from a procedure with an associated mortality rate of up to 20% to less than 5%.^23,24 Extended hepatectomy, however, remains a complex procedure with recent reports of mortality ≥ 5% and morbidity ≥ 50%, even at specialized centers.^5,6 In the current study, we report a near zero (0.8%) mortality rate and a 30.7% morbidity rate for 127 consecutive extended hepatectomies. This improvement is multifactorial and undoubtedly related to better patient selection, improved anesthetic monitoring, greater understanding of hepatic anatomy, advances in surgical technique, and improved perioperative critical care.

One improvement has come in the area of patient selection for extended hepatectomy. In the past, patients with cholangitis and obstructive jaundice underwent extended hepatic resection. Several studies, however, have shown that cholangitis and hyperbilirubinemia are both strongly associated with increased in-hospital mortality.^6,25 In one study, the presence of preoperative cholangitis or an elevated bilirubin (>6 mg/dL) was associated with a mortality rate of ≥ 40% after extended hepatectomy.⁶ In another study of 218 patients undergoing major hepatic resection for biliary tract carcinoma, Kanai et al found that preoperative cholangitis resulted in a greater than 2-fold increase in hospital mortality.²⁵ Nimura et al^26,27 and Makuuchi et al²⁸ have recommended routine preoperative biliary drainage before resection to achieve a bilirubin level less than 3 mg/dL. In the series reported here, all patients presenting with biliary tract obstruction underwent preoperative biliary drainage to achieve clearance of cholangitis and/or jaundice. In general, it appears that hyperbilirubinemia adversely affects the ability of the liver to regenerate, causing an inhibition in such critical transcription factors as hepatocyte nuclear factor-6 (HNF-6) and HNF-4, both of which are important for hepatocyte function, repair, and regeneration.^29,30

Traditionally, measures such as indocyanine green (ICG) retention rate, galactose elimination, and aminopyrine clearance have been used to evaluate hepatic metabolic function and hepatic functional reserve. Most experience with ICG comes from Japan, with this test not being widely used in the West, though retention rates of 15 minutes after intravenous injection of ICG (0.5 mg/kg) correlate with outcome in some series.^31,32 We believe, however, that these tests are impractical for surgical planning before extended hepatectomy, as they provide an overall measurement of function and do not differentiate between the liver to be resected and the anticipated liver remnant. Our group has therefore focused on the evaluation of the FLR volume as the most important factor indicative of an increased risk for complications.^12,13

In general, > 20% of the total liver volume appears to be the minimum safe volume that can be left after extended resection in patients with normal underlying liver. CT can now provide an accurate and reproducible method for preoperative liver volume calculation using three-dimensional CT volumetry.^12,14 In 48 patients without chronic liver disease undergoing extended hepatectomy with and without preoperative portal vein embolization, the postoperative complication rate was significantly increased in patients with FLR volume ≤ 20% of the total estimated liver volume.¹² Shirabe et al has also demonstrated correlation between liver volume and outcome in patients with chronic liver disease using a standardized method of calculation based on body surface area.³³ In this study, all deaths from liver failure occurred in patients with FLR < 300 mL/m². Small liver remnant size has been associated with increases in portal pressure and flow, endothelial and Kupffer cell injury, and the release of pro-inflammatory cytokines.³⁴ Surrogate measures of the overall postoperative course, such as hospital stay and intensive care unit stay, also appear to be increased as the FLR decreases.¹³ Thus, unlike other investigators who have previously suggested that the number of segments resected dictates perioperative morbidity and mortality,⁵ it is now becoming clear that the true critical factor is the size of the FLR.³⁵ Our data support this contention, as one of the reasons for the low morbidity and mortality rates seen in the current series may be our systematic use of preoperative volumetry in the selection of patients for extended hepatectomy.

To avoid operating on patients with small FLR, patients with FLR ≤ 20% of the total liver volume underwent preoperative embolization to induce hypertrophy of the contralateral liver. PVE is safe with less than a 5% complication rate, causes little periportal reaction, and generates durable portal vein occlusion, especially when used in combination with coils.¹⁵ PVE has been shown to increase both the size and function of the FLR.^36,37 In the current series, 31 patients underwent preoperative PVE, all of whom had an increase in FLR before extended resection. PVE was exclusively performed in patients undergoing an extended right hepatic resection, as PVE is rarely necessary before extended left hepatectomy because the right posterior sector constitutes ∼30% of the total liver volume.^38,39

Although clearly important, an adequate liver remnant is not an absolute safeguard against postoperative complications. Regardless of the FLR volume, massive intraoperative bleeding can result in postoperative hepatic compromise. Multiple studies have shown a relationship between intraoperative blood loss and poor outcomes after major hepatic resection.^6,40,41 Massive blood transfusions can add to the risk of coagulopathy as well as exert immunosuppressive effects. Melendez et al⁶ reported that blood loss > 3 L was a key operative variable associated with increased in-hospital mortality. Similarly, Matsumata el al⁴¹ found an increase in abdominal septic complications in patients experiencing blood loss > 3 L. Didolkar et al⁴⁰ also reported that blood loss > 5 L resulted in increased postoperative mortality. In our series, only 1 patient experienced a blood loss > 3 L, whereas the median blood loss was considerably lower: 600 mL (range, 100 to 3500 mL) for a left extended hepatectomy and 300 mL (range, 100 to 1500 mL) for an extended right hepatectomy (P = 0.01). Of the 127 resections reported here, only 28.3% of patients received a transfusion.

A number of factors contributed to the low blood loss seen in the current series. One important factor is the relation between intraoperative blood loss and the pressure within the inferior vena cava (IVC). In the current series, we minimized infusions and transfusions until the end of the parenchymal dissection. In a prospective study examining blood loss and IVC pressure, Johnson et al⁴² showed a direct linear correlation between mean caval pressure and blood loss. Surgical technique, however, is likely the most important factor in minimizing blood loss. In extended right hepatectomy, a complete dissection of the anterolateral vena cava and control of the right hepatic vein may reduce blood loss. This maneuver not only provides vascular control, but it also allows for an anterior mobilization of the liver to be resected, which facilitates visualization and intraparenchymal dissection. Hepatic vein control was performed in the majority of cases (78.7%) in the current series, and this may have also contributed to the low intraoperative blood loss. Other studies have corroborated this, with one study showing that use of hepatic vein control considerably reduced blood loss.⁴³ Another technique proven to reduce blood loss and consequently improve surgical outcome is the Pringle maneuver.⁴⁴ In the current study, a Pringle maneuver was used in 116 patients (91.3%), with a median clamp time of 27 minutes (range, 4 to 94 minutes).

The complication rate for extended hepatic resection in this study was 30.7%, which is comparable to the 21% to 45% morbidity rates reported overall for hepatic resections.^5,7 Most of the complications were minor or intermediate, including perihepatic collections that were amenable to percutaneous drainage. Serious complications were less common, including 6 patients (4.7%) who had significant hepatic insufficiency as defined as a total bilirubin > 10 mg/dL or an INR > 2.^19,32 On univariate analysis, male gender (P = 0.02), age > 65 years (P = 0.04), and performance of a synchronous intraabdominal procedure (P = 0.04) were all associated with a higher complication rate.

Previous studies have reported conflicting results about the importance of gender with regards to perioperative morbidity.^45,46 In the current analysis, gender did not withstand competing risk in the multivariate analysis. Similarly, there have been divergent reports on the association between advanced age and morbidity after hepatic resection.^6,47,48 In the current study, on multivariate analysis, age also failed to remain a significant factor in determining morbidity. One factor that did maintain significance on both univariate and multivariate analysis was the history of a synchronous intraabdominal procedure performed at the time of the extended hepatic resection (HR = 4.9; P = 0.02). Others have also shown that concomitant procedures, such as biliary or vascular resections, are associated with higher morbidity.^3,5 Of note, there was a significant increase in the number of synchronous procedures performed during the last 5 years of the study (P = 0.001), which corresponded to an increase in associated morbidity. Caution, therefore, needs to be taken when combining extended hepatectomy with another intraabdominal procedure, and staged procedures deserve consideration in selected patients when possible.

At a median follow-up of 18.4 months, the median overall survival was 41.9 months. Two factors seemed to affect overall survival. On multivariate survival analysis, both the size of the tumor and a history of RFA performed at the time of the extended hepatic resection significantly affected survival. Patients who had a tumor size > 5 cm had a higher likelihood of death than those who underwent resection for a lesion ≤ 5 cm (HR = 3.5; P = 0.01). Previous studies have also shown a correlation between tumor size and survival. Similarly, the need to perform RFA for additional disease in the remaining unresected segment of liver was associated with a poor long-term outcome (HR = 3.3; P = 0.03). In the current study, tumor size and the need to RFA residual disease in the contralateral lobe after extended hepatectomy may have acted as surrogate markers for overall tumor burden.

Some have argued that extended hepatic resection for colorectal metastases is unwarranted because the combination of high perioperative mortality and inferior survival did not translate into a long-term benefit.⁴⁹ In the current study, we showed that extended hepatectomy can be accomplished safely and is associated with long-term survival. At times, preoperative PVE will be needed to ensure an adequate FLR. When combining resection with an additional intraabdominal procedure, the peri-operative risk is increased and the approach should be individualized.

Discussions

Dr. Alan W. Hemming (Gainesville, Florida): When first reading the manuscript I thought the title “Is Extended Hepatectomy for Hepatobiliary Malignancy Justified?” was clearly rhetorical, since any liver surgeon has no doubt that the answer is yes.

When asked to discuss this paper, I informally asked my non-liver surgery medical and surgical colleagues what they thought about taking out three-quarters of the liver for cancer, and it was surprising to find out that the general feeling was that it was probably all we had to offer but that the results were not very good.

This paper from Dr. Vauthey is important in that it demonstrates the excellent results that can be currently obtained with extended hepatic resections by specialized hepatobiliary surgeons. Dr. Vauthey has presented a series of 127 consecutive patients undergoing extended hepatectomy over a 10-year period with a commendable operative mortality of less than 1%. These results were achieved by applying techniques that are becoming standards for liver surgery, including the ability to control inflow and venous outflow to the liver, low CVP during parenchymal transection, dedicated liver anesthesiologists, and preoperative portal vein embolization to optimize the projected liver remnant post-resection. Dr. Vauthey, I have several questions regarding the paper.

In the manuscript you made a point of relating complications from liver resection not to the amount of liver resected but to the amount of liver left behind. You put the emphasis on calculating FLR and using this information to select patients for extended resection and seemed to be against using functional tests such as indocyanine green clearance. Indocyanine green clearance is generally used to assess the amount of hepatocellular reserve in cirrhotic patients rather than in patients with normal liver function. Your series was largely if not completely performed in noncirrhotic patients with colorectal liver metastases and presumably normal liver function. Do you think that FLR assessment alone is adequate information to assess the ability of a cirrhotic liver to tolerate an extended resection? In our cirrhotics we would currently perform calculations of both FLR and ICG retention at 15 minutes, using ICGR-15 cut-off of 15% with extended resection.

Second question. The results of extended hepatectomy with RFA to lesions in the remnant liver appear to be relatively worse than without RFA. Have your recurrences been primarily hepatic or extrahepatic? Our experience has been there is a fair amount of recurrence within the liver and we have begun placing hepatic artery infusion pumps for adjuvant therapy in these patients. What is your current approach to HAI pumps in these patients, or what other therapy are you giving to these patients at high risk for recurrence?

Third question. Your operative mortality of less than 1% is commendable. And as surgeons I suppose that we all should be aiming at a zero percent operative mortality. This is perhaps more of a philosophical question. But is zero percent operative mortality truly what we are aiming at, or would that suggest that we are overly selecting our patients and at the same time not performing resections in patients that may benefit? I have no real answer to that. I would be interested in your opinion.

Dr. John S. Bolton (New Orleans, Louisiana): The presentation by Dr. Vauthey is notable for an admirably low operative mortality rate, less than 1%, and low rate of postoperative hepatic insufficiency after extended hepatic resection. Given the median overall survival of almost 5 years for patients with metastatic colorectal cancer and 42 months overall for all patients, the authors confirm that extended hepatic resection is an endeavor worth pursuing.

The data suggest that preoperative measurement of future liver remnant and the application of selective portal vein embolization to patients with a predicted low FLR enhances safety. But the data presented doesn't prove this, since many other measures, including fastidious biliary drainage and preoperative treatment of cholangitis, reduction in the numbers of patients with massive blood loss, a dedicated anesthesiology team, and perhaps unidentified patient selection factors might be responsible for some or all of the mortality reduction.

Also, the paper doesn't provide good confidence intervals given the small sample size for selected subpopulations.

One such subpopulation is patients with hilar cholangiocarcinoma, especially if hepatic artery or portal vein reconstruction is required. They had 15 patients in their series with hilar cholangiocarcinoma; a sample size this small creates wide confidence intervals for the operative mortality endpoint. A second subgroup, patients who required concomitant bowel resection or major vascular reconstruction of any type, consists of only 12 patients.

For a third subgroup, patients over the age of 70 or 75, my teaching and training always was beware of liver regeneration in this elderly population. But what is their mortality from an extended hepatic resection? We probably don't have the sample size to give us good confidence intervals in this paper.

I have several questions for Dr. Vauthey. Among the 13 patients with hepatocellular carcinoma, how many had underlying chronic liver disease and how many of these patients underwent portal vein embolization?

Second, for hilar cholangiocarcinoma were multiple biliary drainage catheters placed in most patients? That is, did you not only drain the future liver remnant but also drain the site to be resected? And how about segment 1 drainage? That always seems to be sort of a well of sepsis that is difficult to drain preoperatively.

And my third question, did you encounter NASH or non-alcoholic steatohepatitis in your patient groups? How do you recognize it? And does it preclude extended hepatic resection?

Dr. Layton F. Rikkers (Madison, Wisconsin): In calculating the residual volume of the liver, is the liver occupied by tumor included in the calculation?

Dr. Martin J. Heslin (Birmingham, Alabama): I rise to congratulate Dr. Vauthey and his colleagues. This is a large and impressive series.

Many patients undergoing extensive resections have large tumors or tumors that are placed in poor locations. We find that a lot of these patients undergo preoperative chemotherapy, and we have noticed anecdotally that some patients with more extended resections and preoperative chemotherapy have either a higher risk of hepatic failure or an increased risk of complications subsequently.

I may have missed it in your talk, but did you mention how many people received preoperative chemotherapy and have you noted if there is a relationship between liver failure and any one particular chemotherapeutic agent?

Dr. Jean-Nicolas Vauthey (Houston, Texas): First, in answer to Dr. Hemming, we would agree that ICG is a tool for estimating hepatic reserve. It has been demonstrated to be useful in Japan especially for resection of small tumors in patients with advanced cirrhosis. However, in the West we are talking very often of patients with large tumors, and perhaps compensatory hypertrophy, and measuring the hepatic function globally with ICG may not be meaningful in these patients. You really need to focus on the part that is going to remain after resection, and you cannot determine accurately how much functional parenchyma you will be removing by doing ICG. So it is a general estimate that is good in cirrhosis and in the resection of small tumors. The Japanese have used, in fact, the future remnant in addition to ICG in patients with cirrhosis who need major hepatectomy.

Regarding radiofrequency ablation, Dr. Abdalla will talk about the intrahepatic recurrence rate and our current results. We have used intra-arterial chemotherapy as part of a phase II adjuvant chemotherapy protocol after resection and ablation but have been disappointed by the results (Ann Surg Oncol 2003; 10:348–354). We have essentially discontinued the use of intra-arterial chemotherapy as adjuvant therapy in patients who undergo complete resection of colorectal metastasis given the recent advances in effective systemic chemotherapy such as oxaliplatin or irinotecan based regimens.

Are we selecting the patients? Maybe. I would like to note that the last patient presented here, who is a 10-year survivor, was a patient for whom I sent the x-ray to Dr. Blumgart in New York and Dr. Blumgart asked me to send him the patient for resection. Do we have patients with chronic liver disease in this series? Yes, we have patients with steatosis. But we don't have patients with fibrosis or cirrhosis.

How do we drain our patients before surgery? We initially focus on draining the future liver remnant in patients with hilar cholangiocarcinoma. In some patients, we have had difficulty in doing portal vein embolization when there was dilation on the side to be embolized. So for this reason, and because we use portal vein embolization and an ipsilateral approach for portal vein embolization we have been draining both sides recently. But we would certainly accept a drain, a single endoscopic drain, well placed going to the left side in a patient, who is going to undergo an extended right hepatectomy in the absence of jaundice if the future liver remnant is adequate in size.

Dr. Rikkers is asking about our method of measurement of the future liver remnant. We have developed a method of measurement that is similar to that used in transplantation and is essentially based on the future liver remnant. And for this liver remnant measurement, the numerator is the absolute volume obtained by CT volumetric reconstruction, which for an extended right hepatectomy is segment II and III volume directly measured from CT, and as a denominator we use the total liver volume based on a formula that is based on body surface area which we have developed (Liver Transpl 2002; 8:233–240). So our measurements are all standardized for the size of the patients. Small patients need small liver remnants. Large patients need large liver remnants.

Dr. Heslin asked about preoperative chemotherapy and the risk of steatosis and damage from chemotherapy. We recently published a paper specifically looking at steatosis in patients who received irinotecan based systemic chemotherapy (J Gastrointest Surg 2003; 7:1082–1088). We did not find a significant increase in steatosis in these patients. We have graded the steatosis in this study and I think if you use 3 or 4 cycles of systemic chemotherapy before resection this is a safe approach.