Yeh T, Jan Y, Jeng L, Hwang T, Chao T, Chien R, Chen M. Pyogenic Liver Abscesses in Patients With Malignant DiseaseA Report of 52 Cases Treated at a Single Institution. Arch Surg. 1998;133(3):242-245. doi:10.1001/archsurg.133.3.242
Prognosis of pyogenic liver abscesses in patients with malignant disease is generally considered poor. The discrepancy between the outcomes of liver abscesses caused by hepatopancreatobiliary malignant disease and those caused by other malignant diseases, however, to our knowledge has never been investigated.
To clarify the clinical course of pyogenic liver abscess in patients with different types of cancer, and to compare outcomes in abscesses caused by hepatopancreatobiliary malignant disease and other malignant disease.
Retrospective review of case series in our experience from 1980 through 1993.
Tertiary care university teaching hospital.
Fifty-two patients with pyogenic liver abscess related to the underlying cancer were divided into 2 groups. Group 1 (n=32) was composed of patients with cancer originating from the hepatic parenchyma, bile duct, and pancreas; group 2 (n=20) was composed of patients with cancer originating from other sites.
Parenteral antibiotics, percutaneous drainage, surgical drainage, or hepatectomy, in combinations, were employed.
Main Outcome Measures
Patient characteristics, symptoms, laboratory data, abscess characteristics, microbiological study, management, and outcome of the 2 groups were analyzed.
Thirteen patients (41%) in group 1 and 16 patients (80%) in group 2 had undergone prior anticancer treatment. Jaundice was encountered more often in group 1 than in group 2 (29 patients  vs 6 patients , respectively, P=.001), whereas nausea and vomiting were more frequently seen in group 2 than in group 1 (17 patients  vs 6 patients, respectively, P=.04). Leukocytosis, hypoalbuminemia, hyperbilirubinemia, and reversed albumin-globulin ratio were more pronounced in group 1 than in group 2 (P=.001, .02, .003, and .03, respectively). Abscesses communicating with the intrahepatic biliary tree were more frequently encountered in group 1 than in group 2 (11 patients  vs 2 patients , respectively, P=.03). Escherichia coli and Klebsiella pneumoniae predominated in group 1, while the bacteria species in group 2 were more diverse. The hospital mortality rates of group 1 and group 2 were 28% (9 of 32 patients) vs 10% (2 of 20 patients) (P=.04), respectively. Twenty-three patients (72%) of group 1 died of uncontrolled biliary sepsis or progressive cancer or both within 6 months after the diagnosis, while 17 patients (85%) of group 2 survived longer than 1 year without relapse of the abscess and continued with anticancer treatment.
Pyogenic liver abscess could be a presentation of hepatopancreatobiliary malignant disease at the preterminal stage, and carries a grave prognosis. Pyogenic liver abscess in patients with nonhepatopancreatobiliary malignant disease has a better chance of favorable outcome.
PYOGENIC LIVER abscess was recognized as early as the era of Hippocrates (460-370 BC). In 1938, Ochsner et al1 noted portal venous suppuration secondary to acute appendicitis in 34% of a collected series of 575 cases with pyogenic liver abscess. By 1960, Sherman and Robbins2 had noted that biliary tract disease and malignant disease supplanted portal vein seeding as the most common mode of bacterial entry into the liver. During the last decade, despite advances in antibiotics, microbiologic study, diagnostic modalities, and nonoperative percutaneous drainage, the mortality rate has remained from 12% to 26% and from 2% to 15% in surgical patients and in those with a percutaneous drain, respectively.3- 7 These unfavorable results are mainly attributed to the increasing number of patients with neoplastic disease, biliary tract disease, multiple liver abscess, old age, and immunocompromise following various anticancer treatments. Reports from institutions worldwide have indicated that malignant disease is the major risk factor for liver abscess,3,4,7,8 although Marcus et al5 claim that there is still a good prognosis even in patients who have cancer and a liver abscess. The aim of this study was to clarify the clinical course of pyogenic liver abscess in patients with different types of cancer. We also compared outcomes in abscesses caused by hepatopancreatobiliary malignant disease and other malignant disease.
From January 1980 through December 1993, a total of 542 patients with pyogenic liver abscess were admitted to Chang Gung Memorial Hospital, Taipei, Taiwan. Of these patients, 52 (9.6%) who had pyogenic liver abscess related to preexisting or coexisting malignant disease were included in this study. All of their types of cancers were confirmed histopathologically. Pyogenic liver abscess was defined based on clinical manifestations; evidence from ultrasonography or computed tomography, or both, of solitary or multiple space-occupying lesions; pus from liver aspirates with microorganisms isolated; positive blood culture results; disappearance or shrinkage of the corresponding lesions after antibiotic treatment and drainage procedures; or laparotomy findings. These 52 patients were further divided into 2 groups based on their primary neoplastic sites. Group 1 was composed of patients whose cancers originated from the hepatic parenchyma, biliary system, and pancreas. Group 2 was composed of patients with cancer originating from other sites. Patients' characteristics, symptoms, laboratory data, abscess characteristics, microbiological study, management, and outcomes of the 2 groups were compared. To determine whether malignant disease affected the prognosis of the liver abscess, the age, causes, management, and outcomes of the other 490 patients with pyogenic liver abscess without malignant disease were compared with those of the 52 patients in the study. All continuous variables are expressed as mean±SD. Hospital mortality is defined as death during the same hospital admission for management of the disease. Statistical analysis was done by χ2test for qualitative data and Student t test for quantitative data. A P<.05 was considered statistically significant.
The types of cancers of patients in group 1 and group 2 are given in Table 1. There were 17 men and 15 women in group 1, with a mean (±SD) age of 58.2 (±2.5) years (range,40-73 years), and 12 men and 8 women in group 2, with a mean (±SD) age of 53.9 (±3.1) years (range, 9-77 years). Thirteen (41%) of the 32 patients in group 1 and 16 (80%) of the 20 patients in group 2 had undergone anticancer treatment (Table 2).
Clinical manifestations of both groups are listed in Table 3. Group 1 had more cases of jaundice compared with group 2 (29 patients  vs 6 patients , respectively, P<.001), while nausea and vomiting were more frequently seen in group 2 than in group 1 (17 patients  vs 7 patients , respectively, P=.02).
Leukocytosis was more prominent in group 1 than in group 2, in which leukopenia was frequently noted (P<.001). Hypoalbuminemia and hyperbilirubinemia were also more pronounced in group 1 than in group 2 (P=.02, P=.003, respectively). Group 1 had a more strongly inverted albumin-globulin ratio compared with group 2 (P=.03).
Multiplicity, loculation, size, gas formation, and rupture of the liver abscess were similar in the 2 groups. Abscesses communicating with the intrahepatic biliary tree, which were identified by presence of bilious fluid within the collection bag or by a subsequent sinogram, were more often encountered in group 1 than in group 2 (11 patients  vs 2 patients , respectively, P=.03).
Abscess and blood cultures were acquired for all patients. Group 1 patients had more polymicrobial pus cultures than did group 2 patients (16 patients  vs 5 patients , respectively, P=.02). Escherichia coli and Klebsiella pneumoniae were the predominant bacteria in both pus and blood cultures in group 1 patients, whereas bacterial species in group 2 patients were more diverse.
The detailed procedures of management are given in Table 4. The hospital mortality rates of group 1 and group 2 were 28% (9/32) and 10% (2/20), respectively (P=.04). The overall hospital mortality rate for patients with liver abscesses and malignant disease was, thus, 21% (11/52), which was significantly worse than that for patients with abscesses but no malignant disease, 11% (54/490, P=.02) (Table 5).
Twenty-three (72%) of 32 patients in group 1 died of uncontrolled sepsis or progressive cancer or both within 6 months after the diagnosis. In contrast, 17 (85%) of 20 patients in group 2 survived for longer than 1 year following the liver abscess, with continuing anticancer treatment.
In agreement with other reports,3,4,7 the mortality rate of pyogenic liver abscess caused by underlying malignant disease was almost twice as high as that without malignant disease (Table 5). We attempted to discover whether pyogenic liver abscesses with various kinds of cancer had all of the same poor results as those without cancer or just some of those results. The pathogeneses of liver abscesses in group 1 patients were mainly malignant biliary obstruction and spontaneous or iatrogenic necrosis of primary hepatic neoplasms with superimposed bacterial infection. In contrast, the pathogeneses of the abscesses in group 2 patients included portal venous suppuration secondary to colorectal cancer, postgastrectomy for gastric cancer, and hepatic artery seeding with uncertain infectious nidi following systemic anticancer treatment.
Clinical manifestations of liver abscess caused by underlying malignant disease were fever, chills, right-sided hypochondrial pain, hepatomegaly, nausea and vomiting, malaise, body weight loss, and jaundice, all of which were indistinguishable from those with benign causes. Nevertheless, the presence of unexplained anemia; severely reversed albumin-globulin ratio; abnormal tumor markers; imaging studies including ultrasonography, computed tomography, cholangiography, and sinogram; and aggressive tissue proof provide the foundation for diagnosis.
Although gas bubbles usually suggest the presence of abscesses, they can also occur in necrotic tissue without infection in certain conditions.9 In our experience, gas bubble formation with dispersed distribution is frequently seen within hepatic tumor following effective transhepatic arterial embolization.9 Therefore, sterile necrotic hepatic tumors after transhepatic arterial embolization should be cautiously interpreted and differentiated from superimposed bacterial abscess, in which the gas formation appears with focal air/fluid level.10 Attention to clinical signs of toxic reaction, aspiration, and bacteriological study aids in precisely differentiating between the two. Abscesses communicating with the intrahepatic biliary tree were encountered more often in group 1 than in group 2 (11 patients  vs 2 patients , respectively). Most of these cases were related to malignant biliary obstruction. Do et al11 reported that the duration of drainage was longer in patients with intrahepatic biliary communication (mean, 22 days) than in patients without communication (mean, 14 days), but the cure rates for percutaneous drainage in both groups were similar. Our experience concurred with these findings.
Argument over the superiority of computed tomography–guided percutaneous drainage and surgical drainage has not been settled.7,11- 15 We concur with the opinion of Bertel et al15 that percutaneous drainage should be the initial choice of treatment whenever no concomitant surgical procedure is planned. Because most patients in group 1 were so ill and in the preterminal stage of their underlying malignant disease when admitted to the hospital, successful treatment of the abscess and the cancer simultaneously was unrealistic. In this series, the only 3 cases treated by hepatectomy had survived for only 3, 3, and 8 months after the operation. Our data indicated that hepatectomy was not justified in these cases and should not be encouraged in other such cases. The optimal management of concurrent liver abscess and cancer is to resolve life-threatening infection as simply as possible. Percutaneous drainage of the abscess, in addition to percutaneous transhepatic biliary drainage if necessary, works well in most circumstances.
Twenty-three patients (72%) in group 1 died of uncontrolled biliary sepsis or progressive cancer or both within 6 months after diagnosis of the abscess. In comparison, 17 patients (85%) of group 2 survived longer than 1 year without relapse of the infection, and continued to receive anticancer treatment. Branum and et al,7 in their series of 73 patients with pyogenic liver abscess, noted that, in the largest group of patients with hepatic abscess, malignant disease was present and was also the most important determinant of outcome. Also, in a report from Hong Kong,16 malignant disease, jaundice, and elevated activated partial thromboplastin time were found to be important risk factors. Our study indicates that hepatopancreatobiliary malignant disease, but not all cancers, results in a dismal prognosis in these patients. This reasonably explains why patients with both cancer and a pyogenic liver abscess were still able to attain satisfactory results in the series reported by Marcus et al.5 In that series, the malignancies of only 5 (29%) of 17 patients who had cancer and a bacterial abscess were of hepatopancreatobiliary origin.
In conclusion, pyogenic liver abscess can be a presentation of underlying hepatopancreatobiliary malignant disease at a preterminal stage and carries a grave prognosis. In contrast, patients who have pyogenic liver abscess and nonhepatopancreatobiliary malignant disease have a greater chance of a favorable outcome.
Reprints: Ta-Sen Yeh, MD, Department of Surgery, Chang Gung Memorial Hospital, 5, Fu-Hsing St, Kwei-Shan, Taoyuan, Taiwan.