Leung KL, Lau WY, Lai PBS, Yiu RYC, Meng WCS, Leow CK. Spontaneous Rupture of Hepatocellular CarcinomaConservative Management and Selective Intervention. Arch Surg. 1999;134(10):1103–1107. doi:10.1001/archsurg.134.10.1103
A conservative approach using selective intervention is better than an aggressive approach using nonselective intervention for ruptured hepatocellular carcinoma.
Nonrandomized controlled trial.
A university hospital.
Patients and Interventions
From 1984 to 1990, an aggressive approach was adopted in which 29 and 8 of a total of 40 patients underwent surgical intervention or attempted transarterial embolization (TAE), respectively. From 1991 to 1997, a more conservative approach was used. The initial treatment for 72 patients was conservative with close monitoring. Additional hemostatic procedures consisting of TAE (n=13) or surgical intervention (n=9) were given, depending on the clinical progress, disease status, and liver function of the patients.
Main Outcome Measures
In-hospital mortality, survival.
In-hospital mortality rate was 62% (25 of 40 patients) in the first period and 51% (37 of 72 patients) in the second period. The respective median survival times were 7 and 12 days. If 36 patients with end-stage malignant neoplasms were excluded, the in-hospital mortality rate became 60% (18 of 30 patients) in the first period and 35% (16 of 46 patients) in the second period (P=.03, χ2 test). The respective median survival times became 8 and 72 days (P=.02, log rank test). In the second period, 7 (54%) of 13 patients who underwent TAE and 1 (11%) of 9 patients who underwent surgical intervention died within the same hospital admission (P=.07, Fisher exact test).
Selective intervention was cost-effective and gave better results than an aggressive approach. When intervention was indicated for hemostasis, surgery seemed better than TAE although the difference was not statistically significant.
HEPATOCELLULAR carcinoma (HCC) is common in areas with endemic viral hepatitis B and/or C. Rupture of HCC occurs in about 10% of patients.1- 4 It may occur as a terminal event in patients with advanced disease or it may be the first presentation in an otherwise healthy individual. Some patients may present with progressive abdominal distention and anemia without hemodynamic decompensation while others present with severe hypovolemic shock. The variation in clinical presentations poses a challenge to clinicians in diagnosis and management. Various treatment options have been proposed, which include conservative treatment,5 transarterial embolization (TAE),6- 11 different methods of operative hemostasis,12,13 and emergency liver resection.14 It is still uncertain which procedure is the best treatment option, as most studies were noncomparative and selected groups of patients were treated without predetermined selection criteria. In this study, the outcome of a consecutive series of patients with spontaneous rupture of HCC treated by 2 different approaches was reviewed.
From 1984 to 1997, 112 patients with spontaneous rupture of HCC were admitted to our hospital. The diagnosis of HCC was made by either radiological demonstration of a space-occupying lesion in the liver coupled with an elevation of the α-fetoprotein level above 500 ng/mL or by histological testing confirming HCC. The presence of hemoperitoneum was diagnosed by a sonographic examination showing free fluid in the peritoneum and a dropping hemoglobin level or by abdominal paracentesis. Patients with bleeding after a recent intervention and bleeding incidentally found during elective hepatectomy were not included in this study.
In the first period from 1984 to 1990, an aggressive approach was adopted.2 Patients were actively resuscitated with fluid and blood transfusion. Vital signs, central venous pressure, and urine output were closely monitored. Blood samples were obtained for renal function tests, liver function tests, complete blood cell count, clotting profile, α-fetoprotein level, and hepatitis serology. Any electrolyte disturbances and coagulation defects were corrected. After diagnosis, emergency operation was performed when the patient was satisfactorily prepared. The surgical procedures, which included plication of the bleeders, diathermy coagulation, alcohol injection, packing, hepatic artery ligation, and hepatectomy, were chosen by the surgeon based on the operative findings and the patient's condition. Hepatic angiography (HAG) and embolization were used on candidates who were considered unsuitable for surgical operation. In a few very poor-risk patients, only conservative treatment was offered.
In the second period from 1991 to 1997, a more conservative approach was adopted. All patients were initially treated with conservative treatment. If the patients showed signs of continuous bleeding or rebleeding that resulted in hemodynamic instability or blood transfusion requirement of more than 4 U in 24 hours, HAG and TAE were attempted.10 For patients who were not suitable for embolization or if the facilities were not available within 1 hour because of technical reasons, emergency surgery was performed. The surgical procedure was determined by the surgeon, but the principle was to achieve hemostasis with minimal damage to the liver parenchyma. Ligation of the hepatic artery was not done unless local treatment to the bleeding tumor failed to stop the bleeding. Emergency liver resection was not carried out except under ideal conditions. During this period, patients considered to have end-stage malignant neoplasms were not offered surgical or angiographic intervention.
Patients were discharged home when the bleeding had stopped and liver function had become stable. Patients with potentially resectable tumors were then worked up for resectional surgery.
The following parameters were recorded: sex, age, vital signs, liver function test result, renal function test result, complete blood cell count, clotting profile and α-fetoprotein level on presentation, treatment, in-hospital mortality, and survival. In addition, the clinicians' evaluation of the patient's disease status on presentation and the reasons for such an evaluation were also recorded. The data from the first period were obtained retrospectively by reviewing the hospital records. The data from the second period were collected prospectively.
The results of patient management during the 2 periods are presented in Table 1. Because of the change in policy, more patients were treated conservatively in the second period (69% vs 8% in the first period; P<.001, Fisher exact test). Patients in the second period had significantly lower albumin and plasma sodium levels on presentation. The overall in-hospital mortality was 62% and 51% in the first and second period, respectively. The respective median survival times were 7 and 12 days. The differences were not significant.
Ten patients in the first period and 26 patients in the second period were positively documented as having end-stage malignant neoplasms on presentation (Table 2). The major factors found to affect the clinicians' evaluation were (1) a history of irresectable HCC for more than 1 month before the episode of rupture and (2) the degree of liver function derangement. However, some of these patients were offered surgical or angiographic intervention for psychosocial reasons and some patients survived the rupture episode even without an active interventional procedure. If these 36 patients were excluded, the in-hospital mortality was 60% and 35% in the first and second period, respectively. The difference was significant (P=.03, χ2 test). The respective median survival times were 8 and 72 days, respectively (P=.02, log-rank test).
In the second period, 22 patients were offered an interventional procedure after failed conservative treatment. Seven of 13 patients who underwent HAG and 1 of 9 patients who underwent surgery died within the same hospital admission (P=.07, Fisher exact test).
In the first period, of the 8 patients who underwent HAG, 6 had successful embolization but 4 died within the same hospital admission. In the second period, of the 13 patients who underwent HAG, 11 had successful embolization but 6 died within the same hospital admission.
A combination of procedures was performed for patients who underwent nonresectional hemostatic surgery. In the first period, 6 patients had plication of bleeders, 9 had ligation of hepatic artery, 12 had packing, and 3 had alcohol injection. In the second period, 6 patients had plication of bleeders, 2 patients had packing, and 3 had alcohol injection; no hepatic artery ligation was performed. The types of resections performed in the first period included 6 right hemihepatectomies, 3 left hemihepatectomies, and 2 left lateral segmentectomies. Only 1 left lateral segmentectomy was performed in the second period.
The characteristics of patients who underwent the various treatment options are presented in Table 3. The systolic blood pressure, plasma albumin level, plasma alkaline phosphatase level, plasma alanine transaminase level, plasma sodium level, and plasma urea level of the different groups on presentation were significantly different. There was a tendency for the clinicians to treat patients with higher blood pressure or with poor surgical risks conservatively (which was in accordance with our policy in the second period).
It is difficult to judge how aggressive one should be in treating patients with ruptured HCCs. Treatment for suitable patients, especially those with potentially resectable tumor, will prolong life. On the other hand, aggressive treatment may hasten death in patients with poor liver function and end-stage malignant neoplasms. We were unable to define end-stage malignant neoplasms using absolute objective criteria alone and so the clinicians' subjective evaluations were used. Certainly clinicians' subjective opinions affect how patients are treated, which then affects the final outcome of the patients.
Various treatment options have been described but have only added controversy. Chearanai et al12 reported that conservative treatment had a 100% mortality and advocated hepatic artery ligation, which carries an in-hospital mortality of 54%. In their study, however, they excluded patients with coagulation defects, and some of these were successfully treated with conservative measures. Lai et al13 advocated a policy similar to that used in our first period. They surgically explored 56 patients with an in-hospital mortality of 75%. Owing to the poor result of surgical treatment, TAE was enthusiastically attempted but most of the reported series were small and the selection criteria used were uncertain.6- 11 Okazaki et al9 described a larger series of patients in which the 1-month mortality in patients with a total bilirubin level of less than 51.3 µmol/L (3.0 mg/dL) was 8.3% but was 86% in those with higher bilirubin levels. Chen et al14 suggested immediate hepatectomy with an in-hospital mortality of 4.3% and a 5-year survival rate of 26.5%. Their result was exceptionally good and approached that of elective hepatectomy in nonruptured cases, suggesting that the patients were carefully selected. Xu and Yan5 advocated conservative treatment, with a 1-week and 1-month mortality of 26.5% and 48.5%, respectively, which was comparable to their own control group of surgical intervention. All these reported series suffer from not giving a global picture of the disease and providing only limited information on the patients' background.
In this study, if we compare the result of various treatment options (Table 3) without specifying the selection policy, we would arrive at a conclusion that all the treatment options gave the same result. However, such a conclusion is erroneous. The in-hospital mortality of the various interventional groups cannot be used for comparison with the conservative treatment group, because some patients in the interventional groups had failed the initial conservative treatment while some patients in the conservative treatment group had active intervention withheld as they were considered to have end-stage malignant neoplasms. It is also wrong to combine the results of the individual treatment options in the 2 periods, as the approaches to patient selection were different.
The rate of intervention was significantly lower in the second study period but the in-hospital mortality was also lower. If we exclude the group of patients who were considered to be "terminal" on presentation, the difference in the in-hospital mortality between the first and second period became significant (P=.03, χ2 test). Nonselective aggressive intervention with premature or unnecessary intervention seemed to harm patients. For patients who remained stable after resuscitation and correction of coagulation defects, continuation of conservative treatment seemed to be the best treatment. When bleeding continues, intervention should be offered. It was not an individual treatment option but the logical combination of the options that offered the patients the best chance of surviving the bleeding episode.
When intervention was indicated to stop the bleeding, our experience showed that there was only 1 chance to do it. Although we did not refrain from offering a second intervention, usually when the first intervention failed most patients' conditions would have deteriorated to a nonsalvageable stage. The primary aim of an intervention is to stop bleeding and to preserve as much of the functioning liver parenchyma as possible. Although TAE has the advantage of minimal invasiveness and has been shown to be effective for hemostasis,6- 11 the result was not very impressive in our study. In the second period of this study, 7 of 13 patients eventually died of liver failure after TAE. On the contrary, 8 of 9 patients who underwent operative hemostasis survived. Of significance is the fact that we avoided hepatic artery ligation in the second period. Instead, we used local methods to stop bleeding, such as plication of bleeders, alcohol injection, or microwave coagulation. It was only when local measures failed that hepatic artery ligation was considered. These results suggest that it is important to avoid unnecessary damage to any liver parenchyma during the intervention for hemostasis. In a patient who has marginal liver function due to a cirrhotic liver, any further damage caused by embolizing or ligating the hepatic artery can tip the balance and result in liver failure and death. For the same reason, we do not recommend immediate hepatectomy unless circumstances are ideal. We prefer to stop the bleeding first, work up the patient, and, if the tumor is resectable, perform elective liver resection 6 to 8 weeks later after the patient and the liver have recovered from the rupture episode. We used the in-hospital mortality as the end point because survival of the bleeding episode is the most important consideration in the immediate management of ruptured HCC.
In conclusion, this study showed that, while the intervention rate was significantly less, an approach using conservative and selective intervention gave similar results to an approach using aggressive and nonselective intervention for ruptured HCC. In the subgroup analysis, in which patients who were considered terminal were excluded, the in-hospital mortality and survival in the second approach were significantly better. If an intervention was indicated because of continued bleeding, local operative measures on the tumor for hemostasis gave better results than TAE, probably because of the lesser degree of damage to the remaining liver parenchyma.
We thank J. T. Lau, PhD, and S. Xu, Centre for Clinical Trials and Epidemiological Research, the Chinese University of Hong Kong, for reviewing our manuscript.
Reprints: W. Y. Lau, FRCS(Edin), FRACS, Department of Surgery, The Chinese University of Hong Kong, Prince of Wales Hospital, Shatin, New Territories, Hong Kong (e-mail: email@example.com).