Pseudomonas pseudomallei Liver Abscesses: A Clinical, Laboratory, and Ultrasonographic Study

412 Pseudomonas pseudomallei Liver Abscesses: A Clinical, Laboratory, and Ultrasonographic Study T. Vatcharapreechasakul, Y. Suputtamongkol, D. A. B. Dance, W. Chaowagul, and N. J. White From Sappasitprasong Hospital. Ubon Ratchatani, and Faculty of Tropical Medicine. Mahidol University. Bangkok. Thailand; and the DepartmentofClinical Sciences, London School ofhygiene and Tropical Medicine. London, and Nuffield Department ofclinical Medicine, John Radcliffi Hospital, Oxford. United Kingdom Ultrasonography revealed evidence of liver abscess in 126 patients who were admitted to one hospital in northeastern Thailand over a 3-year period. There were 50 cases for which a pyogenic bacterial etiology was confirmed; 34 cases (group 1) were caused by Pseudomonas pseudomallei (nine patients died) and 16 cases (group 2) were caused by other bacteria (two patients died). Melioidosis was associated with anemia and underlying diabetes or renal disease; right-upperquadrant pain and jaundice were more common in group 2 (P <.05). Bloodcultures werepositive for bacteria in 68% of group 1 and 50% of group 2. Chest radiographs revealedabnormalities in 17 of 30 group 1 patients and 6 of 12 group 2 patients. The radiographic appearances of a blood-borne pneumonia suggested melioidosis. The serum indirect hemagglutination assay for antibodies to P. pseudomallei was of limited value in differentiating the two types of abscesses. Multiple hypoechoic areas on ultrasonography were significantly associated with melioidosis (P <.01); associated splenic abscess occurred in 19 group 1 patients but only one group 2 patient (2-107,95% confidence interval; odds ratio, 19). In an area where P. pseudomallei is endemic, these characteristic ultrasonographic findings should prompt immediate treatment for melioidosis. Pyogenic liver abscess is a potentially lethal condition. Early diagnosis is essential if mortality is to be reduced. Modern imaging techniques such as radioisotope scanning, ultrasonography, and computed tomography are sensitive and noninvasive diagnostic tools [I, 2]. They increase the accuracy and safety ofneedle aspiration, thereby permitting identification ofthe infecting organisms and allowing therapeutic drainage. Of these methods, ultrasonography is considered by many to be the most practical and is now widely available, even in some parts of the rural tropics. Melioidosis, or infection with Pseudomonas pseudomallei, is an increasingly recognized health problem in Thailand [3, 4]. The clinical manifestations of melioidosis are protean and range from localized infection to fulminant septicemia. Visceral abscesses are common in disseminated melioidosis. Since P. pseudomallei is resistant to the antibiotics usually recommended for empirical treatment of pyogenic liver abscess [5, 6], it is imperative to obtain a precise etiologic diagnosis in suspected cases arising in areas in which the organ- Received 20 February 1991; revised 30 May 1991. Informed consent was obtained from all subjects. and guidelines for human experimentation of the Public Health Ministry of Thailand were followed. Financial support: Wellcome Trust ofgreat Britain (as part of the Wellcome-Mahidol University, Oxford Tropical Medicine Research Fellowship Training Programme). Reprints or correspondence: Dr. Y. Suputtamongkol, Faculty oftropical Medicine, 420/6 Rajvithi Road, Bangkok 10400, Thailand. Clinical Infectious Diseases 1992;14:412-7 1992 by The University of Chicago. All rights reserved. 1058-4838/92/1402-0005$02.00 ism is endemic. Blood culture may be negative for patients with pyogenic liver abscesses, and liver aspiration has been regarded by some as contraindicated in melioidosis because ofthe risk ofprecipitating septic shock [4]. To assess the role ofthe endemic P. pseudo mallei as a cause ofliver abscess in this area, we compared the clinical, ultrasonographic, radiographic, and laboratory findings and the outcome for patients with melioidosis liver abscesses and those with bacteriologically confirmed pyogenic liver abscesses caused by other bacteria. Patients and Methods All patients with liver abscesses who had been admitted to Sappasitprasong Hospital between October 1986 and November 1989 were identified from records of the Radiology Department. Ultrasonographic examination is part of routine diagnostic screening in cases of suspected liver abscess and is usually performed before bacteriologic culture results are available. Liver abscess was diagnosed by the presence of one or more well-circumscribed areas of decreased intrahepatic echogenicity on ultrasonographic examination in patients with a compatible clinical presentation. Only cases in which pathogenic bacteria were isolated in blood cultures or those in which bacteria were isolated or demonstrated by gram stain ofpus aspirated from the liver were documented. Empirically treated patients, those with amebic liver abscess, and those for whom microscopy and culture were negative for bacteria were not included in this comparison because of diagnostic uncertainty. Data from patients with melioidosis were collected pro-

cm 1992; 14 (February) Melioidosis Liver Abscesses 413 spectively during an ongoingstudy ofthe disease in Sappasitprasong Hospital [3, 7]. In every case, a complete history, physical examination, complete blood cell count, and measurements (in serum) ofurea, creatinine, bilirubin, aspartate aminotransferase, alanine aminotransferase, alkaline phosphatase, albumin, globulin, and titers of antibodies to P. pseudomallei (determined by indirect hemagglutination assay [IHA]) had been obtained on admission. In most cases, three blood samples were taken for culture. Abdominal ultrasonography was performed in frontal, right, and left-lateral modes using a real-time machine (Sonolayer V-SSA-90A; Toshiba, Tokyo). Blood and pus were cultured aerobically, and organisms were identified by standard laboratory procedures. Anaerobic cultures were not performed routinely in the laboratory. P. pseudo mallei was identified as described previously [6]. The demographic characteristics, underlying diseases, clinical features, radiographic and ultrasonographic findings, results of laboratory investigations, and outcome for the two groups were compared. The results were compared by eitherthe unpaired t-test for normally distributed quantitative data or Wilcoxon's rank sum test, and proportions were compared by the X 2 test with Yates's correction or by the Fisher exact test. Results There were 126 cases ofliver abscess diagnosed by ultrasonography during the study period. Case notes for 110 of these patients were available for review. A bacterial etiology was confirmed in only 50 cases, so 60 of the 110 patients were not included in the comparative analysis. Three of these 60 had disseminated tuberculosis with liver abscesses; in eight others, amebic abscess was suspected (the diagnosis was supported by good therapeutic response to metronidazole but usually was not confirmed by aspiration or amebic serology). Eight patients were treated for melioidosis, but the diagnosis was not confirmed; all of them had multiple abscesses and two had associated splenic abscesses, but cultures of blood were negative, and liver aspiration was not performed. The remaining 41 patients were suspected ofhaving other pyogenic liver abscesses, but microscopy ofaspirated pus and all cultures were negative. In 25 patients there was a single cavity and in 16 there were multiple abscesses; these patients were treated empirically. Fifty patients had proven pyogenic liver abscesses; these patients comprised two groups. Group I consisted of 34 patients with melioidosis (27 men and seven women) ranging in age from 6 to 64 years (mean, 4\.4 years). Group I patients accounted for 8.7% ofthe melioidosis cases diagnosed in Sappasitprasong Hospital during the same period. Group 2 consisted ofpatients with other confirmed bacterial liver abscesses (II men and five women); their ages ranged from 20 to 78 years (mean, 46.3 years). Underlying diseases. Underlying disease was common in Table 1. Clinical features on admission to the hospital ofpatients with liver abscess due to P. pseudomallei (group I) or other bacteria (group 2). Group I Group 2 Features (n = 34) (n = 16) History Fever 34 (100) 13 (81) Median duration of fever. days (range) 12(1-60) 10 (1-150) Abdominal pain 15 (44) 4 (25) Right-upper-quadrant pain 8 (24) 7 (44)* Nausea. vomiting 9 (26) 4 (25) Signs Fever >37.2 C 23 (68) 10(63) Jaundice 9 (26) 7 (44) Hepatomegaly 26 (76) II (69) Right-upper-quadrant tenderness 16 (47) II (69)* NOTE. Data presented are the no. (%) of patients in each group with a given feature. unless stated otherwise. * p <.05. both groups of patients: 79% of group I (melioidosis patients) and 56% ofgroup 2 (patients with abscesses due to bacteria other than P. pseudomallei) (P <.13). Diabetes mellitus (14 cases, 41%) and renal diseases (12 cases, 35%) were the most frequent underlying conditions associated with melioidosis liver abscess. In contrast, biliary tract disease (19%) was the most prevalent underlying condition in group 2. Other infrequently associated diseases found in both groups were hematologic disorders (thalassemia. glucose-6-phosphate dehydrogenase deficiency, and acute leukemia), chronic steroid use, and pulmonary tuberculosis. Clinical features and laboratory results. The clinical features present on hospital admission for patients of both groups are shown in table 1. There were no symptoms or signs specific to either group. However, a history ofright-upper abdominal pain, right-upper-quadrant tenderness on examination, and jaundice were significantly more common in group 2 patients (P <.05). Laboratory parameters at the time ofadmission are shown in table 2. Anemia (i.e., packed cell volume of <30%) was more common and the mean hematocrit was significantly lower in patients in group I (P <.05); the serum direct bilirubin value was significantly higher in those in group 2 (P <.5). For differentiating melioidosis from other causes ofpyo genic liver abscess, IHAs that measured serum titers of antibodies to P. pseudomallei (figure 1) had the following rates of sensitivity and specificity: 58% and 85%, respectively, at a cutoff titer of ;;.1:640; and 42% and 85%, respectively, at a cutoff titer of o l: 1280. Bacteriologic results. Twenty-three patients (68%) in group 1 and 8 patients (50%) in group 2 were bacteremic (P =.6). On culture, pus from percutaneous aspiration or surgical drainage yielded P. pseudomallei in all II melioidosis

414 Vatcharapreechasakul et al. cro 1992; 14 (February) Table 2. Laboratory findings on admission of patients with liver abscess due to P. pseudomallei (group I) orother bacteria (group 2). Median (range) Laboratory values (normal range) Group I Group 2 PCV,% 25 (15-43) 31 (13-42)* WBe, XlO 9/L 15.4 (2. 7~36.0) 14.1 (7.7-30.4) PMNL, % 85 (27-95) 76 (62-97) BUN, mmol/l 9.5 (2.8-36) 5.9 (2.6-26) Creatinine, JLmol/L 202 (67-1,091) 123 (61-316) Glucose, mrnol/l 7.7 (2.9-36.0) 5.8 (1.7-20.3) Direct bilirubin, JLmol/L 11.0 (2.2-108) 19.8 (4.4-194)* AST, lull (8-40) 49.5 (3-143) 39.5 (7-270) ALT, ru/t. (5-35) 25.5 (4-145) 22 (3-120) Albumin, gil 30 (18-41) 27 (20-52) Globulin, gil 41 (26-56) 40 (19-57) Alkaline phosphatase, lull (8-35) 68 (28-230) 61 (7-138) NOTE. PCV = packed cell volume; WBC = white blood cells; PMNL = polymorphonuclear leukocytes: BUN = blood urea nitrogen: AST = aspartate aminotransferase; ALT = alanine aminotransferase. * P<.05. cases in which pus was obtained from the abscess, while five of the 10 aspirates from patients in group 2 were sterile. Pus from melioidosis liver abscesses varied in color from yellowgreen to brown-red (i.e., blood stained) and did not smell; gram stain revealed small numbersofthe characteristic bipolar gram-negative bacilli in only one of five patients. For seven group I patients and all 16 group 2 patients, a preaspiration dose of an intravenous antimicrobial was given. Melioidosis was diagnosed by positive culture ofblood only for 10 patients, liver pus only for 6, sputum or throat swab specimens for 3, and subphrenic and brain abscess pus for I each. Cultures ofspecimens from multiple sites were positive for P. pseudomallei for the remaining patients in group I. The organisms isolated from patients in group 2 were Klebsiella species (5 patients), Escherichia coli (3), Enterobacter species (4), and coagulase-positive staphylococci (3). For four patients whose bloodcultureswere negative, gram stain ofliver pus demonstrated bacteria (mixed gram-positive cocci and gram-negative rods in three patients and gram-positive cocci alone in one), but cultures ofpus were sterile. It is likely that these abscesses were caused by fastidious organisms (e.g., Streptococcus milleri or anaerobic bacteria) that would not have been isolated by the aerobic culture procedures used routinely in the hospital laboratory. The foul smell ofthe pus in two of these cases supported the diagnosis of anaerobic infection. Radiographic findings. Chest radiographs were normal for approximately one-half the patients examined in both groups (17 of 30 in group I and 6 of 12 in group 2). Of 10 group I patients who had associated pneumonia, 6 had lobar consolidation, 4 had multiple opacities suggesting bloodborne pneumonia, and 3 had pleural effusions with or without elevation of the right dome of the diaphragm. Among group 2 patients, pleural effusions were found in four and basal pneumonia and a cavity suggestive of pulmonary tuberculosis in the left-upper lung were found in one each. Ultrasonographic findings. Liver abscesses characterized by multiple hypoechoic areas (figure 2A) were significantly more common in patients with melioidosis (82%vs. 31%; P <.0 I). In addition, associated splenic abscesses were largely restricted to group I (figure 2B); they occurred in more than one-half of these patients (56%) but in only one patient in group 2 (2-107, 95%confidence interval; odds ratio, 19; P <.00 I). Multiple splenic abscesses were noted in 17 of 19 cases of melioidosis. Details of the type and sites of the abscesses are shown in table 3. Treatment. Patients with melioidosis liver abscess (group I) were treated with a combination of high-dose parenteral chloramphenicol (100 mg/[kg d]), doxycycline (4 mg/ [kg d]), and co-trimoxazole (10/50 mg/[kg d]) or received single-agent therapy with ceftazidime (120 mg/[kg d]). Patients were randomly allocated to either regimen as part ofa prospective clinical trial [7]. The durationofparenteral treatment ranged from 2 days to 4 weeks (median, 10 days). Patients were then treated with oral drugs. A combination of oral chloramphenicol, doxycycline, and co-trimoxazole was used before 1987. Since that time, patients have received therapy with either this conventional four-drug regimen or amoxycillin/clavulanic acid for a median of6 weeks (range, 1-24 weeks). These two regimens are now being evaluated prospectively in our hospital. In group 2, six patients were treated with antimicrobials alone, usually penicillin G sodium or ampicillin plus gentamicin, with or without metronidazole. Parenteral treatment for a median duration of 10 days (range, 9 days to 3 weeks) 14 12 10 8 6 4 2 O-"'----,-----,---,-------,------,----,------,---r-.../ Neg 1:20 1:40 1:80 1:160 1:320 1:640 1:1280 III!Il!ifiI OTHER BACTERIA _ MELIOIDOSIS Figure 1. Serum titers ofantibodies to P. pseudomallei, as determined by indirect hemagglutination assay, in patients with liver abscess due to melioidosis or other bacteria. The number of patients in each group is shown on the vertical axis.

em 1992; 14(February) Melioidosis Liver Abscesses 415 Figure 2. abscesses. Ultrasonograms of melioidosis abscesses. A, a characteristic multiloculated liver abscess. B, liver (left) and splenic (right) was followed by treatment with oral ampicillin, with or without metronidazole, for at least 4 weeks. Outcome. None of the patients developed septicemic shock after aspiration. The median duration of hospitalization was 3 weeks (range, 2 days to 5 weeks) in group I and 14 days (range, 10 days to 6 weeks) in group 2. The duration of follow-up ofthe 20 patients in group I ranged from 8 weeks to 54 months (median, 32 weeks). For five patients either relapse or reinfection occurred during this period (relapse and new infection cannot be distinguished by current techniques), and one of these five patients died from uncon- Table 3. Ultrasonographic findings in patients with pyogenic liver abscesses due to P. pseudomallei (group I) or other bacteria (group 2). No. (%)ofpatients with indicated finding Group I Group 2 Ultrasonographic finding (n = 34) (n = 16) No. ofabscess cavities Multiple 28 (82) 5 (31)' Single 6 (18) II (69) Site ofabscess( es) Right lobe 25 (74) 10 (63) Left lobe 2 (6) 5 (31) Both lobes 9 (26) I (6) Associated splenic abscess 19 (56) I (6)', P <.001. trolled sepsis with recurrence of both liver and splenic abscesses. Details oftreatment and mortality in each group are shown in table 4. Fatal cases. Eleven patients died, nine of whom had melioidosis. Thus the mortality among patients with melioidosis liver abscess (group I) was 9 (26%) of 34 and that of patients with other bacterial abscesses (group 2) was 2 (13%) of 16. The age range in the fatal cases was 38-78 years (median, 48 years). Nine patients (89%) were jaundiced and bacteremic, and all had multiple liver abscesses. All deaths were caused by uncontrolled sepsis despite antimicrobial treatment. No patient died as a direct result of liver aspiration. The median time from hospital admission to death was 9 days (range, 2-39 days). Discussion Previous reports from countries in temperate climates suggest that pyogenic liver abscesses are uncommon, accounting for only 0.007%-0.03% [2, 8, 9] of hospital admissions. In Ubon Rachatani, Thailand, pyogenic liver abscess is more common. accounting for -0.09% ofadmissions. P. pseudomallei is clearly an important cause ofliver abscess, accounting for one-third ofsuspected cases and two-thirds ofall bacteriologically confirmed cases. All ofthe melioidosis cases in this study were documented prospectively, whereas cases in the comparison group were gathered retrospectively with use ofultrasonographic records, which makes selection bias possible. However, as ultrasonography of the liver was usually part of the initial investigation procedures and bacteriologic

416 Vatcharapreechasakul et al. em 1992; 14 (February) Table 4. Treatment and mortality for patients with liver abscesses due to melioidosis (group I) or bacteria other than P. pseudomallei (group 2). Group I Group 2 (n = 34) (n = 16) No. of No. of Treatment method No. treated deaths No. treated deaths Medical treatment only 23 7 6 2 Medical treatment and liver aspiration 9 I 8 0 Medical treatment and surgical drainage 2 I 2 0 Overall mortality (%) 26 13 confirmation followed later, most cases ofliver abscess diagnosed in the hospital are represented here. The primary objective ofthe study was to compare the features ofmelioidosis liver abscesses with those of other proven bacterial liver abscesses. Therefore, we excluded 60 patients for whom a cause ofliverabscess was not identified andwhose treatment was based on clinical suspicion. Diabetes and renal failure are well-known predisposing factors in P. pseudomallei infection [3, 4], and these or other underlying conditions were present in the majority of patients with confirmed melioidosis liver abscess. Patients with melioidosis were also significantly more anemic, a characteristic for which contributory factors included renal impairment (as >90% of patients with a serum creatinine concentration of>250 JImoljL were found to be anemic) and bone marrow suppression due to the chronic infectious process itself. The lesser degree ofjaundice and the lower incidence ofupper-quadrant pain in patients with melioidosis probably reflect both the relative prominence ofsepticemia as a presenting feature of melioidosis (i.e., a more-fulminant disease process) and the lack ofunderlyingbiliary tract or gastrointestinal disease in this group. Apart from these findings, the clinical features of patients with liver abscesses caused by P. pseudomallei were similar to those of patients with pyogenic liver abscesses due to other bacteria. The currently available serological test was not useful in establishing the diagnosis. The IHA for serum antibodies to P. pseudomallei is widely used in Thailand, but it is neither sensitive (58%) nor specific (85%) enough for serological diagnosis ofliverabscess in this area. The lack of routine culture techniques appropriate for the isolation of fastidious organisms that commonly cause pyogenic liver abscess (e.g., S. milleri and obligate anaerobic bacteria) [5, 10-12] and a lower incidence of septicemia in association with these infections must have resulted in their underdiagnosis. The greater antibiotic susceptibility oforganisms other than P. pseudomallei may also have contributed to a reduced isolation rate among patients who had received antimicrobial treatment before admission. In addition, most small liver abscesses were treated empirically with antibiotic regimens including metronidazole, a practice which would not allow distinction between abscesses due to anaerobic bacteria and those caused by Entamoeba histolytica. Thus the range of pathogens recovered in this study differed from that reported in other recent series. The evidence that diagnostic liver aspiration is contraindicated in cases of melioidosis because ofthe risk ofprecipitating septicemic shock is anecdotal [4]. Four patients in this series were not given antimicrobials before aspiration, and the condition ofnone ofthem deteriorated as a result of the procedure. Pretreatment with antibiotics active against P. pseudomallei did not appear to interfere with culture results. This is in accordance with infection at other sites; many days or weeks of treatment are needed to sterilize melioidosis abscesses. Pretreatment with antibiotics may, however, interfere with results of cultures for other pathogens. We usually perform diagnostic aspiration under ultrasonographic guidance and after administration of a parenteral antimicrobial agent, and we have found this a safe and effective method of confirming the diagnosis ofmelioidosis. Despite the advent of new diagnostic techniques, a plain chest radiograph was still diagnostically useful for one-halfof the patients in both groups. Abnormal findings, such as elevation of the dome of the right hemidiaphragm, right-sided pleural effusion, and basal atelectasis, were found in both groups, but multiple opacities suggestive of blood-borne pneumoniawere found only in melioidosis cases. Such opacities have not been described in previous reports of pyogenic liver abscess, even in the presence of bacteremia [5,8, 13], and are therefore highly suggestive of melioidosis in areas in which P. pseudomallei is endemic. Ultrasonography of the abdomen was a valuable procedure; in addition to making possible the diagnosis of liver abscess, it localized the cavity for aspiration and often allowed differentiation of P. pseudomallei liver abscesses from those caused by other bacteria. The ultrasonographic features that suggested melioidosis 'were multiple small cavities in the liver parenchyma (i.e., "Swiss cheese" abscesses) and associated visceral abscesses in other organs. These findings are also seen clearly with computed tomography [14].

em 1992; 14 (February) Melioidosis Liver Abscesses 417 Splenic abscess is an unusual occurrence in most infections, but more than one-halfofthe patients with melioidosis liver abscess also had an abscess in the spleen. Furthermore, in most cases (89%) the splenic abscesses were multiple. Thus, the finding ofa splenic abscess is a strong diagnostic indicator ofmelioidosis and one that, taken togetherwith the other clinical and ultrasonographic findings, allows prescription of an effective antibiotic regimen before isolation ofthe organism from bacteriologic cultures. The mortality due to pyogenic liver abscess has varied between 24% and 40% [5, 8] but has declined in recent years because of improvements in diagnosis and treatment. Factors contributing to a high mortality are cancer, bacteremia, old age, jaundice, and multiple abscesses [8, 15]. In our series, a fatal outcome was associated with P. pseudomallei infection, bacteremia, and multiple liver abscesses. Methods of treatment for melioidosis are still not satisfactory. Most patients with melioidosis were treated medically with antimicrobials alone, for two reasons. First, in cases with multiple small abscesses, diagnostic percutaneous aspiration would have been technically difficult and therapeutic drainage would have been impossible. Initially there was also a reluctance to undertake aspiration because ofthe perceived risk of precipitating septicemia. In such cases, treatment effective against P. pseudomallei was started on the basis ofthe ultrasonographic findings, before isolation of the organism from othersites. Second, neitherpercutaneous needle aspiration nor surgical drainage was performed when the abscesses were small. As a result, use of percutaneous needle aspiration was limited to cases ofsolitary large abscesses and to the few cases of multiple cavities in which the ultrasonographic findings were equivocal. Surgical drainage was reserved for those cases in which the abscess appeared close to the surface of the liver and rupture seemed imminent. Conventional medical treatment ofmelioidosis with a combination ofhighdose chloramphenicol, doxycycline, and co-trimoxazole was given to all patients prior to 1987, but this has now been superceded by treatment with ceftazidime [7, 16]. This change has been associated with a 50%reduction in mortality [7]. Melioidosis is associated with a high mortality [3]. The condition of several patients deteriorated quickly, progressing to irreversible organ failure and death despite appropriate treatment. However, these patients were already seriously ill when they were admitted. Delay in starting appropriate therapy is an important contributor to this high mortality. Drugs used empirically for the treatment ofliverabscess or community-acquired septicemia are usually ineffective against P. pseudomallei. Early diagnosis of pyogenic liver abscess is therefore imperative within areas in which melioidosis is endemic. Where the equipment is available, ultrasonographic examination should be performed in all cases of suspected liver abscess. The findings of multiple "Swiss cheese" abscesses, splenic abscess, and metastatic pneumonia but no obvious visceral source ofinfection are strongly suggestive of melioidosis. Appropriate antimicrobial treatment can then be given while the results of culture are pending. Acknowledgments The authors thank the director of Sappasitprasong Hospital and his staff; the physicians of the Departments of Medicineand Radiology; Ms. Vanaporn Wuthiekanun, Ms. NittayaTeerawattanasook, and the staffof the microbiology laboratory; Ms. Nucharee Cholvilai,who typed the manuscript; and Glaxo Laboratories and Beecham Research Laboratories, which donated the ceftazidime and amoxicillin/clavulanic acid, respectively. References I. Terrier F, Becker CD, Triller JK. Morphologic aspects of hepatic abscesses at computed tomography and ultrasound. Acta Radiol 1983;24: 129-37. 2. Perera MR, Kirk A, Noone P. Presentation, diagnosis and management ofliverabscess. Lancet 1980;2:629-32. 3. Chaowagul W, White NJ, Dance DAB, et al. Melioidosis: a major cause ofcommunity-acquired septicemia in northeastern Thailand. J Infect Dis 1989;159:890-9. 4. Leelarasamee A, Bovornkitti S. Melioidosis: review and update. Rev Infect Dis 1989;11:413-25. 5. McDonald MI, Corey GR, Gallis HA, Durack DT. Single and multiple pyogenic liver abscesses: natural history, diagnosis and treatment, with emphasis on percutaneous drainage. Medicine (Baltimore) 1984;63:291-302. 6. Dance DAB, Wuthiekanun V, Naigowit P, White NJ. Identification of Pseudomonas pseudomallei in clinical practice: use of simple screening tests and API 20NE. J Clin Pathol 1989;42:645-8. 7. White NJ, Dance DAB, Chaowagul W, Wattanagoon Y, Wuthiekanun V, Pitakwatchara N. Halving of mortality of severe melioidosis by ceftazidime. Lancet 1989;2:697-701. 8. Greenstein AJ, Lowenthal D, Hammer GS, Schaffner F, Aufses AH Jr. Continuing changing patterns ofdisease in pyogenic liver abscess: a study of 36 patients. Am J Gastroenterol 1984;79:217-26. 9. Moore-Gillon J'C, Eykyn SJ, Phillips I. Microbiology ofpyogenic liver abscess. BMJ 1981;283:819-21. 10. Chua D, Reinhart HH, Sobel JD. Liver abscess caused by Streptococcus milleri. Rev Infect Dis 1989; 11:197-202. II. Kandel G, Marcon NE. Pyogenic liver abscess: new concepts ofan old disease. Am J Gastroenterol 1984;79:65-71. 12. Verlenden WL III, Frey CF. Management of liver abscess. Am J Surg 1980; 140:53-9. 13. Sabbaj J, Sutter VL, Finegold SM. Anaerobic pyogenic liver abscess. Ann Intern Med 1972;77:629-38. 14. Singcharoen T. CT finding in melioidosis. Australas Radiol 1989;33:376-8. 15. Berger LA, Osborne DR. Treatment ofpyogenic liver abscesses by percutaneous needle aspiration. Lancet 1982; 1:132-4. 16. Camus C, Cartier F, Avril J-L, Journel H. Efficacy ofceftazidime in chronic melioidosis with multiple liver abscesses [letter). Lancet 1990;2:628.