Improved Results of Liver Resection for Hepatocellular Carcinoma on Cirrhosis Give the Procedure Added Value

Abstract

Objective

To review a single-center experience to update the performance indexes of liver resection (LR).

Summary Background Data

Several therapies have been proposed in the treatment of hepatocellular carcinoma (HCC) on cirrhosis, although LR was the first to be widely applied.

Methods

Of 408 patients with cirrhosis admitted for HCC in the period 1983 to 1998, 264 had a LR. Patient selection, surgical technique, 30-day deaths, long-term survival, recurrence rate, and recurrence treatment were reviewed after stratifying patients according to the year of surgery. Mean follow-up was 34.5 ± 29.1 months.

Results

The number of Child A patients who underwent surgery after the discovery of the tumor at routine evaluation increased significantly from 64.5% to 87.9% during the study period. Procedures carried out without blood transfusions increased from 31.4% to 76.9%. The overall operative death rate was 4.9%. Actuarial survival rates were 63.1% and 41.1% after 3 and 5 years, respectively; actuarial tumor-free survival rates were 49.3% and 27.9% at the same intervals. After 1992, surgical deaths decreased from 9.3% to 1.3%. Actuarial survival rates increased from 52.9% and 32.3% to 71.7% and 49.4% after 3 and 5 years, respectively. There was no difference in the actuarial recurrence rate between the two periods, but the chance to treat recurrence increased over time from 22.4% to 53.7% with a concomitant, significant improvement in survival.

Conclusions

LR represents a well-established therapy for HCC on cirrhosis. It remains one of the fundamentals in the multidisciplinary approach to this tumor and should be considered as the first option for patients with preserved hepatic function and limited disease. Today, LR should offer a surgical death rate of less than 1.5%, a 5-year survival rate of approximately 50%, and a 5-year tumor-free survival rate of 28% when performed in specialized centers.

During the past two decades, hepatocellular carcinoma (HCC) has been one of the most intriguing diseases to treat. The problem of treatment remains open, especially for patients with underlying cirrhosis of the liver, which substantially reduces the possibility of aggressive treatment. The management of HCC on cirrhosis has progressively involved surgeons, hepatologists, oncologists, and interventional radiologists. Consequently, several different therapies are now available for these patients: liver resection (LR), orthotopic liver transplantation (OLT), percutaneous alcohol injection (PEI), transarterial chemoembolization (TACE), and radiofrequency ablation are applied worldwide by several different specialists.

The routine evaluation of patients with chronic liver disease has led to a significant increase in the number of patients found to have preserved hepatic function and relatively small HCCs. The solitary type is the most common type at presentation. ¹ The disease typically occurs in the seventh decade;² the mean age of a large cohort of patients described by our group for demographic purposes was 61.5 years. ³ In the same group, more than half the patients were Child A. ²

All these considerations explain the increased number of patients with HCC on cirrhosis who require treatment and the number of methods proposed. Unfortunately, the scientific literature lacks prospective comparative evaluations of these treatments, with the exception of those proposed for palliation, ⁴ and there have been no accepted guidelines for managing the disease. Further, the reference standards for results obtained by different methods often refer to historical series that reflect the medical knowledge and expertise of a particular period. ⁵

For these reasons, we decided to perform a retrospective evaluation of the results obtained with a single procedure (LR) over a long period, with the aim of reassessing the standards for this single therapy to be considered when evaluating the wide spectrum of treatments for patients with HCC on cirrhosis.

METHODS

From August 1983 to December 1998, 408 patients received treatment for HCC on cirrhosis at the Department of Surgery and Transplantation of the University of Bologna, Italy. OLT was performed in 54 patients, TACE in 85, and PEI in 5; the remaining 264 patients underwent LR. The first aim of this study was to perform a retrospective review of our single-institution experience with LR in the treatment of patients with HCC on cirrhosis, evaluating changes in patient selection, surgical technique, early results, and long-term survival. The final step was to verify the need to establish new standards when assessing the results of LR for HCC on cirrhosis.

Patients

LR was performed in 216 (81.8%) men and 48 (18.2%) women. Mean age was 61.7 ± 7.8 years (range 22–80). A total of 107 (40.5%) patients received treatment before 1992 and 157 (59.5%) thereafter. Ten patients were treated because of life-threatening symptoms (three hemoperitoneum, one abscess with pain and fever, six severe abdominal pain and discomfort); their data were taken into account only in the analysis of short-term results, and they were excluded from the subsequent analysis of long-term survival and recurrence rate. Table 1 summarizes the clinical characteristics of these patients.

Table 1. CLINICAL CHARACTERISTICS

AFP, alpha-fetoprotein; HBsAg, hepatitis B surface antigen; HCV, hepatitis C virus; TACE, transarterial chemoembolization.

Methods

All patients who underwent LR in the study period were included in the analysis. The follow-up period ended April 30, 1999; thus, every patient had at least 4 months of observation. The year 1992 was arbitrarily chosen as the cutoff value to categorize different periods of our activity. Patients were categorized according to the year of surgery and results were compared accordingly. The main outcome measures were results obtained at 3- and 5-year follow-up after LR.

Indications for Surgery

Laparotomy was indicated in the presence of a single nodule of HCC that was believed to be technically removable by the resection of a limited portion of the liver including an adequate tumor-free margin, according to the size and topographic location of the nodule in the liver. Evaluation of hepatic function was done with the use of the Child-Pugh ⁶ classification of liver dysfunction. Ascites, portosystemic encephalopathy, and the presence of esophageal varices were absolute contraindications to surgery. Thus, indication was basically limited to Child A and B patients who fulfilled these criteria.

Starting in February 1993, liver insufficiency was evaluated with the lidocaine (MEGX) test, which was included in our standard evaluation process. The procedure for this test and the results have been published elsewhere. ⁷ From this experience, a value of MEGX of 25 μg/L or less emerged as a predictor of patients at risk for postoperative liver insufficiency, but this result alone was not considered an absolute contraindication to resection.

Patient Management

No preoperative nutritional supplements were prescribed. Intraoperative blood transfusions were given when the hemoglobin value decreased to less than 8 mg/mL. Postoperative medical treatment included intravenous electrolyte and balanced fluid solutions according to the patient’s weight. Albumin was infused to maintain an albumin blood level of more than 3 mg/L. Branched amino acid infusions were included in the therapy in cases of major hepatic resections or impaired liver function as found before surgery by the Child-Pugh classification or by the MEGX test. No total parenteral nutrition was started in the postoperative period. More recently, we attempted to refeed patients orally early in the postoperative period, with removal of the nasogastric tube on the first postoperative day and oral intake of fluid on the second postoperative day. Patients were usually given a low sodium diet. Diuretics were not given routinely except to patients who were already taking diuretics before surgery. Intravenous furosemide was given at early signs of fluid retention. Ceftriaxone (2 mg/d) was used for antibiotic prophylaxis.

The extent of LR performed followed the segmentary division of the liver as per Couinaud. ⁸ Every adverse event after LR that caused a prolongation of the hospital stay or required adjunctive in-hospital therapy was considered a postoperative complication. Classification of the HCC was made using the fifth edition of the TNM classification of primary liver tumors of the Union Internationale Contre le Cancer (UICC). ⁹

There was a constant evolution of the surgical technique during the study period. The most important was the modification of the technique of parenchymal division, which changed from the original “kellyclasia”¹⁰ to the current slower, painstaking search for every small intrahepatic vessel visualized by the gentle crushing of the hepatic parenchyma with tiny clamps. ¹¹ We always tried to perform anatomical segmentectomies by means of the improved parenchymal mapping achieved with the systematic use of intraoperative ultrasound, with the aim of obtaining an adequate tumor-free margin, according to the size and topographic location of the nodule in the liver. In addition, during recent years we have asked our anesthesiologists to maintain a low central venous pressure (<5 mm/H₂O) to reduce blood loss from the hepatic veins. ¹² Because of the retrospective nature of the present report, we could not identify single variables that could reflect these changes, other than the commonly used intraoperative parameters.

Statistical Analysis

Results are expressed as mean ± standard deviation. The chi-square test was used to compare categorical variables. The Student t test was used to compare continuous variables. Every death occurring within 30 days after surgery was defined as a surgical death. Patients who died before hospital discharge were classified as a hospital death. Survival was computed from the day of surgery to the day of death or to the most recent follow-up visit. The recurrence rate was computed from the day of surgery to the first follow-up visit at which evidence of tumor recurrence was clear, or to the most recent follow-up visit. Tumor-free actuarial survival was defined as the combination of the two variables already mentioned, considering the date of death or the date of tumor recurrence as the time of terminal events. Time-dependent variables were compared by the log-rank test. P < .05 was considered statistically significant.

RESULTS

Patient Selection

There was no difference in the male/female proportion between the two groups. Patients in the more recent period were significantly older (62.2 ± 8.6 years vs. 61.0 ± 6.4 years, P < .05). Table 2 summarizes the differences between the two groups. In the more recent group, an increased proportion of patients had viral hepatitis as the cause of cirrhosis. The percentage of patients who had HCC discovered during routine evaluation of the underlying disease increased from 70.1% to 86.6% (P < .005). The proportion of Child A patients increased from 64.5% to 87.9% (P < .00001). The preoperative workup included a higher percentage of patients who received a Lipiodol injection during angiography in the more recent period (31.6% vs. 82.1%, P < .00001). The proportion of patients undergoing preoperative TACE increased from 14.3% to 30.2% (P < .01). According to the UICC staging of primary liver tumors, ⁹ more patients with stage III and IV tumors underwent surgery in the more recent period.

Table 2. DIFFERENCES IN THE CLINICAL CHARACTERISTICS BY PERIOD

TACE, transarterial chemoembolization.

Surgical Technique

A total of 84 (31.8%) subsegmentectomies, 102 (38.6%) monosegmentectomies, 59 (22.3%) bisegmentectomies, and 19 (7.2%) major hepatectomies (including 7 left hepatectomies and 12 right hepatectomies) were performed. There was no difference in the type of procedures performed between the periods. No vascular control was applied in 109 (41.2%) instances, whereas clamping of the hepatic hilus was done in 139 (52.7%) procedures. Hemihepatic vascular occlusion was applied in only 16 (10.2%) of the 157 resections performed more recently (P < .005). The mean duration of procedures was 187.7 ± 68.8 minutes (range 80–375). During surgery, an average of 350.2 ± 532.9 mL (range 0–3,500) packed red blood cells and 277.4 ± 457.5 mL (range 0–2,500) fresh-frozen plasma were transfused. There was no need for intraoperative transfusion of packed red cells in 153 (58.0%) procedures and no need for intraoperative infusion of fresh-frozen plasma in 164 (62.1%); there was no need for packed red cells nor fresh-frozen plasma in 121 (45.8%). Table 3 shows the differences in these parameters between the two groups. There was a marked reduction in the need for infusion of blood derivatives in the procedures carried out in the more recent period, resulting in 65.0% of the patients undergoing surgery without any of these supports. There was no difference in the mean surgical time between the two periods (191.8 ± 66.4 vs. 184.6 ± 56.3 minutes;P = NS).

Table 3. DIFFERENCES IN SURGICAL PARAMETERS BY GROUP

Early Results

The mean postoperative stay decreased from 14.3 ± 6.5 days in the early period to 11.5 ± 5.1 days in the latter period (P < .0001). The most common postoperative complication was ascites (60 patients, 22.7%). Other adverse events related to the procedure were pleural effusion (n = 20, 7.6%), subphrenic abscess (n = 8, 3.0%), transient impairment of liver function (n = 13, 4.9%), pneumonia (n = 5, 1.9%), hemoperitoneum (n = 6, 2.3%), transient kidney failure (n = 5, 1.9%), gastrointestinal hemorrhage (n = 2, 0.8%), and sepsis (n = 2, 0.8%). There were no statistical differences in the distribution of postoperative complications between the two periods.

There were 13 (4.9%) deaths within 30 days after surgery. Before 1992 there were 10 deaths in 107 patients, giving a surgical death rate of 9.3%. Seven patients died of postoperative liver failure and three of uncontrollable digestive hemorrhage secondary to the ruptured esophageal varices. In the subsequent period there were 3 deaths in 157 patients, giving a surgical death rate of 1.3% (P < .01). Two patients died of postoperative liver failure and myocardial infarction, respectively; the third died at home 25 days after a palliative resection performed for hemoperitoneum resulting from spontaneous rupture of the tumor. All the remaining 251 patients were discharged from the hospital; thus, there were no additional hospital deaths. There was one early death after palliative resection in each group. The surgical death rate in patients who underwent surgery in an elective setting was 11 of 254 (4.3%) patients: 9 of 102 (8.8%) in the first period and 2 of 152 (1.3%) in the second period. After 1992, only one patient died, as a result of liver failure.

Long-Term Results

Of the entire group of 254 patients who underwent curative LR, 125 (49.2%) were alive in April 1999, whereas the remaining 129 (50.8%) had died, including those who died after the operation. Mean follow-up was 34.5 ± 29.1 months (range 1–125.3). The leading cause of death was tumor progression (60 patients, 46.5%). The remainder died of symptoms related to end-stage hepatic disease (30 patients, 23.2%) or other nonliver-related problems (8 patients, 6.2%). The cause of death could not be ascertained in 31 patients. The actuarial survival rate was 63.1% after 3 years and 41.1% after 5 years (Fig. 1). Patients who underwent LR in the early period had survival rates of 52.9% and 32.3% after 3 and 5 years, respectively; patients who underwent surgery in the second period had survival rates of 71.7% and 49.5% after 3 and 5, years respectively (P < .0001). This resulted in an improvement of 18.8% in the 3-year survival rate and of 17.2% in the 5-year one (Fig. 2).

Figure 1. Actuarial survival rate of the group of 254 patients who underwent curative liver resection for hepatocellular carcinoma on cirrhosis between August 1983 and December 1998.

Figure 2. Actuarial survival rates of the group of 254 patients who underwent curative liver resection for hepatocellular carcinoma on cirrhosis according to the year of surgery (square, earlier than 1992; triangle, 1992 or later).

Recurrence

During the follow-up period, recurrence appeared in 109 (42.9%) patients. The actuarial recurrence rate was 36.3% after 3 years and 56.1% after 5 years (Fig. 3). In total, 93 (36.6%) patients remained alive and tumor-free at the end of the study period. The tumor-free actuarial survival rates were 49.3% and 27.9% after 3 and 5 years, respectively. There were no differences for these parameters between the two periods. Recurrence was intrahepatic in 90 (82.6%) patients and extrahepatic in 11 (10.1%) patients; the remaining 8 (7.3%) patients had both intra- and extrahepatic recurrences. The possibility of treating HCC recurrence was assessed on the basis of the site and pattern of the recurrence and the residual liver function. Based on these parameters, further treatment of the tumor could be undertaken in 42 (38.8%) patients. TACE was the therapy applied most frequently (15 patients, 38.5%); 8 (20.5%) patients received PEI, 6 (15.4%) underwent a second liver resection, 3 (7.7%) underwent OLT, 3 (7.7%) received a combination of TACE plus PEI, and the remaining 7 received other treatments.

Figure 3. Actuarial recurrence rate of the entire group of 254 patients who underwent curative liver resection for hepatocellular carcinoma on cirrhosis.

There has been a change in attitude toward the treatment of HCC recurrence over the years. Before 1992, only 22.4% of our patients received treatment for the recurrence, but this proportion increased to 53.7% thereafter (P < .001). The actuarial survival rates of patients who did not receive treatment were 53.2% and 26.3% after 3 and 5 years, respectively. The corresponding figures for patients who received some form of treatment were 83.1% and 52.9% (P < .005) (Fig. 4).

Figure 4. Actuarial survival rates of the group of 109 patients in whom recurrence developed after liver resection, according to the method of treatment of the recurrence (square, treatment of the recurrence [42 patients]; triangle, no treatment [67 patients]).

DISCUSSION

LR was the first treatment applied with success in patients with HCC on cirrhosis. Initially, factors related to the technique itself raised questions as to its true feasibility: the procedure was technically demanding and only a few centers reported acceptable results, with others reporting 30-day death rates of up to 20%. ^13,14 Nevertheless, 5-year survival rates of approximately 30% were considered highly acceptable when compared with the natural history of the disease, which had no survivors 3 years after the initial diagnosis. ¹⁵ A few years later, the report of a 100% recurrence rate after LR carried out for curative purpose further stressed the limited usefulness of conventional surgery for these patients. ¹⁶

Mainly for these reasons, several alternatives to LR were proposed and widely applied for treating HCC on cirrhosis. The growth of OLT began in 1981 with the availability of cyclosporine as a primary immunosuppressant agent, ¹⁷ but the initial results obtained with patients undergoing surgery for HCC on cirrhosis were dismal because of the constant recurrence of the tumor in the grafted liver. A consensus on the indication for OLT in these patients consolidated with the good survivals obtained with tumors of limited dimensions. ¹⁸ Unfortunately, only a few patients can benefit from OLT because of the persistent scarcity of organ donors: access to the waiting lists for OLT is limited worldwide, and the age of the patient is one of the most frequent criteria adopted.

The concept of treating HCC without surgery is based on the use of PEI and TACE. From the surgical point of view, these procedures were considered palliative because there is no evidence of the complete destruction of the tumor in the treated patients. We have found some complete tumoral necrosis in the specimens of patients undergoing resection after TACE, but the observation of the growth of multiple satellite nodules around the main nodule was more common, as we have already described when the PEI technique was applied to patients with hepatic metastases. ¹⁹ Initial reports of PEI were largely positive and even comparable with the results obtained in the first, historical series of LR. ⁵

During the past two decades, several improvements in the clinical management of patients with cirrhosis on HCC have led to better treatment results. The natural history of the disease has changed significantly, and survival rates of up to 30% after 3 years from the initial diagnosis are now reported, with calculated survival rates of approximately 50% for patients with favorable characteristics. ²⁰ Thus, the entire set of clinical parameters used in the evaluation of the effectiveness of therapies applied to treat HCC on cirrhosis must be radically changed. Unfortunately, conducting prospective randomized studies with the aim of assessing the efficacy of every proposed treatment is impractical on ethical grounds. Today, the most rational approach is probably to consider the treatment of this disease as multidisciplinary, with the decision of how to start treating each patient based on several parameters, including the degree of hepatic insufficiency, the number and size of the nodules, and the presence of concomitant disease, with the patient’s age playing a considerable role. Many patients will need more than a single method of treatment.

During the 1990s, LR for HCC on cirrhosis benefited from several improvements in the medical and surgical fields. The widespread practice of clinical monitoring of patients with cirrhosis has led to the surgical evaluation of a larger number of asymptomatic, well-compensated patients with cirrhosis and HCC. ^1,2 The development of new technical tools for the staging and the surgical removal of HCC nodules in cirrhotic livers has significantly benefited patient outcome, with a drastic reduction in the intraoperative blood requirement and in 30-day deaths. The following are noteworthy: the concept of maintaining a low intraoperative central venous pressure to reduce blood loss from hepatic veins ¹²; better parenchymal mapping with the systematic use of intraoperative ultrasound, which simultaneously makes it possible to perform anatomical segmentectomies and to pursue the constant search for a tumor-free margin, thus avoiding the so-called “nodulectomy”; a modified approach to vascular control of the main blood vessels of the liver, ²¹ including the concept of the hemihepatic vascular occlusion; and a more confident approach to tissue division, which we now perform by gentle crushing of the parenchyma using tiny Kelly clamps, thus avoiding intraoperative blood transfusions more and more frequently. ^11,22 Further, the clinical experience gained with such a large number of patients has significantly improved postoperative management, with the more precise use of fluid replacement, albumin, fresh-frozen plasma infusions, and diuretics. ²³

In our hands, these options have increased the 5-year survival rate of our patients by almost 20%. This has been mainly due to the significant reduction of the early surgical risk, subsequent to the above-mentioned more rational surgical approach achieved with the larger number of patients treated. Other factors should not be forgotten, primarily the already mentioned chance of better patient selection. This has led to a better definition of our indication for surgery. We now propose partial resection of the liver to Child A patients in good general health who have a single nodule of HCC. The presence of even mild symptoms of portal hypertension is an absolute contraindication for surgery. In this connection, the MEGX test has been an important refinement in the evaluation process. We have identified a population with low MEGX values (<25 μg/L) in whom the chance of transient postoperative liver decompensation is significantly greater, and this test now represents an added value for the final decision (if not an absolute contraindication if taken alone). ⁷ The possibility of performing surgery on such patients allowed us to avoid performing particular therapeutic measures in the perioperative period and to reduce the hospital stay considerably. Further, most of our patients were already too old for OLT or had contraindications to this procedure. Despite this increased age, they could benefit to the same extent from an increased possibility of long-term survival.

Unfortunately, we could not show a concomitant reduction of the overall recurrence rate over these years. The more confident approach to the disease has led to increased possibilities of treating even postoperative intrahepatic HCC recurrences, with the possibility of performing repeated resections, OLT, or other less invasive percutaneous therapies. The treatment of recurrences represents the ultimate step for an improvement in results. ²⁴

In consideration of the demographic distribution of the disease (typically occurring in the seventh decade) and of its macroscopic presentation (solitary in more than half the cases ^1,2), a large proportion of patients with HCC on cirrhosis should be considered and evaluated for LR. Mainly because of the persistent scarcity of organ donors and also considering the high number of patients requiring transplantation for parenchymal, nonneoplastic diseases, OLT is not at present applicable in a large percentage of patients with HCC on cirrhosis, even if it probably represents a good choice of treatment.

Thus, LR should always be considered among the initial therapies to be proposed to well-compensated patients, even in consideration of the current performance indexes, which include a reduced surgical death rate, a 5-year survival rate approaching 50%, and a 5-year tumor-free survival of 28% when performed in specialized centers and in combination with other therapies for treating possible recurrences.

The review of the performance indexes for resective surgery is of particular importance because of the lack of controlled clinical studies. This situation often leads investigators to undertake complex statistical extrapolations to compare the effectiveness of different treatments and the patient selection criteria for each therapy. ²⁵ Conventional surgery for HCC on cirrhosis must include today’s new parameters, allowing the surgeon to distinguish between the results obtained in tertiary referral centers and those obtained in other centers. ²⁶ Despite the continuous development of new therapeutic tools for HCC on cirrhosis, we strongly believe that the clinical decision of how to start the treatment of a patient with a recently discovered HCC should always include the advice of a surgeon, who can evaluate the possibilities of referring the patient for OLT or LR.

In conclusion, for more than a decade LR has represented a well-established therapy for HCC on cirrhosis. It should be offered as the first option to patients with preserved hepatic function and limited disease who do not fulfill the criteria or are too old for OLT. In consideration of the demographic distribution of the disease, such patients should represent a large proportion of those evaluated for HCC on cirrhosis. In the absence of prospective comparative studies, when considering other therapeutic options, LR offers a surgical death rate of less than 1.5%, a 5-year survival rate of approximately 50%, a 5-year recurrence rate of more than 60%, and a 5-year tumor-free survival rate of 28% when performed in specialized centers.