SEARCH

SEARCH BY CITATION

Keywords:

  • liver abscess;
  • amoebic abscess;
  • pyogenic

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Patients and methods
  5. Statistical analysis
  6. Results
  7. Discussion
  8. Conclusion
  9. References

Distinguishing amoebic from pyogenic liver abscesses is crucial because their treatments and prognoses differ. We retrospectively reviewed the medical records of 577 adults with liver abscess in order to identify clinical, laboratory, and radiographic factors useful in differentiating these microbial aetiologies. Presumptive diagnoses of amoebic (n = 471; 82%) vs. pyogenic (n = 106; 18%) abscess were based upon amoebic serology, microbiological culture results, and response to therapy. Patients with amoebic abscess were more likely to be young males with a tender, solitary, right lobe abscess (P = 0.012). Univariate analysis found patients with pyogenic abscess more likely to be over 50 years old, with a history of diabetes and jaundice, with pulmonary findings, multiple abscesses, amoebic serology titres <1:256 IU, and lower levels of serum albumin (P < 0.04). Multivariate logistic regression analysis confirmed that age >50 years, pulmonary findings on examination, multiple abscesses, and amebic serology titres <1:256 IU were predictive of pyogenic infection. Several clinical and laboratory parameters can aid in the differentiation of amebic and pyogenic liver abscess. In our setting, amebic abscess is more prevalent and, in most circumstances, can be identified and managed without percutaneous aspiration.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Patients and methods
  5. Statistical analysis
  6. Results
  7. Discussion
  8. Conclusion
  9. References

Despite refinement in diagnostic and therapeutic modalities, liver abscess remains a serious condition with high morbidity and mortality. Pyogenic abscesses of the liver occur secondary to biliary or intestinal tract infections, haematogenous seeding or extension of contiguous infection, and carry a mortality rate of 20–60% even with appropriate medical–surgical management. In contrast, amoebic abscesses respond well to chemotherapy and rarely require drainage (Blessmann et al. 2003; Stanley 2003). As the management and prognoses of these entities are markedly different, their differentiation is crucial.

Several studies have compared the clinical presentation and course of patients with amoebic and pyogenic liver abscess, but only a few have included more than 30 patients in each of these microbiological groups (Ogden et al. 1961; Barbour & Juniper 1972; Balasegaram 1981; Conter et al. 1986; Barnes et al. 1987). We reviewed 577 cases of liver abscess admitted to our tertiary care academic teaching hospital in an attempt to determine clinical, laboratory and radiographic features that might be useful in distinguishing amoebic from pyogenic abscess, thereby obviating the need for diagnostic aspiration.

Patients and methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Patients and methods
  5. Statistical analysis
  6. Results
  7. Discussion
  8. Conclusion
  9. References

The medical records of all patients more than 15 years of age with liver abscess, seen at the Aga Khan University Hospital in Karachi, Pakistan, between 1988 and 1998, were retrospectively reviewed. The diagnosis was established in all cases by hepatic sonography. Inpatient medical records were reviewed for all study subjects in order to determine clinical features, results of laboratory and radiographic studies and hospital course. Where clinical or laboratory findings were not specified, data entries were classified as unknown rather than absent. Antibodies to Entamoeba histolytica were measured by an indirect haemagglutination (IHA) assay and values >1:32 IU were considered positive.

A total of 577 subjects with hepatic abscess were identified. One hundred and six patients (18%) had lesions classified as pyogenic based upon: (i) blood and/or abscess cultures, positive for bacterial pathogens, irrespective of amoebic serology, or; (ii) negative or absent blood and abscess cultures, negative amoebic serology, and a clinical response to antibiotics that did or did not include metronidazole or chloroquine. Four hundred and seventy one patients (82%) were classified as amoebic based on negative blood and abscess cultures (where available) and a good clinical response to metronidazole or chloroquine alone, with either positive (77%), negative (14%), or absent (9%) amoebic serology. Twelve patients (6%) with blood and abscess cultures positive for bacteria but also with very high IHA titres (>1:512 IU) were included in the amoebic group, and were considered to have amoebic abscesses with superimposed bacterial infection.

Statistical analysis

  1. Top of page
  2. Summary
  3. Introduction
  4. Patients and methods
  5. Statistical analysis
  6. Results
  7. Discussion
  8. Conclusion
  9. References

Data were entered into EpiInfo version 6.04 and analysed by Statistical Package for Social Sciences (SPSS) standard version 8.0. Descriptive analysis was performed for demographic, clinical, and radiographic features. Results were expressed as number (percentage), mean ± SD, odds ratio, and 95% confidence interval (CI) for odds ratio. Univariate analysis was performed by using the independent sample t-test, Pearson's chi-square, and Fisher's exact test wherever appropriate. Multivariate analysis was used to identify factors that significantly predicted the presence of pyogenic liver abscess. Initially all the variables were included but the best model was derived using stepwise logistic regression analysis. After developing the main effect model, plausible interaction terms between the variables were assessed and Wald statistics were used to assess the importance of each variable in the model. The final model was tested for goodness-of-fit using Pearson's chi-square statistics. Results with P-values <0.05 were considered as statistically significant; all P-values were two-sided.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Patients and methods
  5. Statistical analysis
  6. Results
  7. Discussion
  8. Conclusion
  9. References

Demographic and clinical features of all 577 study subjects with liver abscess are shown in Table 1. Using univariate analysis, patients with amoebic abscess were significantly younger than those with pyogenic abscess, with mean ages of 40 years and 51 years, respectively (P = 0.004). Males were affected more often than females in both groups with a significantly higher male:female ratio in the amoebic vs. the pyogenic group. The seasonal distribution of both groups was similar with cases occurring throughout the year.

Table 1.  Demographic and clinical features of 577 adult patients admitted with liver abscess
CharacteristicNumber (%) of cases, by aetiology
Amoebic (n = 471)Pyogenic (n = 106)Odds ratioP-value*
  1. * All P-values not shown were non-significant and >0.1.

Demographic
Sex
 Male405 (86)79 (75)1
 Female 66 (14)27 (25)2.10.004
Age (years)
 ≤50352 (75)47 (45)1
 >50119 (25)58 (55)3.65<0.001
Historic
Alcohol use 57/163 (35)16/41 (39)1.12
Diabetes mellitus 52/235 (22)26/66 (39)2.290.005
Prior diarrhoea 91/300 (30)14/64 (22)0.64
Signs and symptoms
Chills275/362 (76)50/75 (67)0.6330.093
Abdominal pain/tenderness407/467 (87)79/102 (77)0.510.012
Cough 78/208 (38)26/60 (43)1.27
Temperature    
 ≤102 °F148/447 (33)41/98 (42)1
 >102 °F299/447 (67)57/98 (48)0.69
Rales/rhonchi 64/464 (14)25/104 (24)1.980.009
Jaundice148/464 (32)43/101 (43)1.580.04
Hepatomegaly341/460 (74)68/102 (67)0.70

The most common presenting complaints among patients with liver abscess of either aetiology were fever, chills, and abdominal pain, with the latter significantly more common in patients with amoebic abscess (P = 0.012). Patients with pyogenic abscess were much more likely to have a history of diabetes mellitus (P = 0.005), to present with jaundice (P = 0.04), and to have abnormal pulmonary findings on physical examination (P = 0.009).

Laboratory and radiographic findings for study subjects are shown in Table 2. The two groups were similar in their mean peripheral white blood cell (WBC) count, erythrocyte sedimentation rate, total serum bilirubin, serum alanine transaminase, and alkaline phosphatase levels. However, the mean albumin level in the pyogenic group was significantly lower than that in the amoebic group (2.1 vs. 2.4 mg/dl, P = 0.02). Patients (90%) with amoebic abscess had positive amoebic serology, while 54% of patients with pyogenic abscess had IHA titres >1:32 IU (not significantly different). However, serological titres below 1:256 IU were predictive of pyogenic infection (P < 0.001).

Table 2.  Laboratory and radiographic features of 577 adult patients with liver abscess
 Number (%) of cases, by aetiology
Amoebic (n = 471)Pyogenic (n = 106)Odds ratioP-value*
  1. * All P-values not shown are >0.1.

  2. † Mean values.

  3. WBC, white blood cell; ESR, Erythrocyte sedimentation rate; ALT, alanine aminotransferase; ALP, alkaline phosphatase; IHA; Indirect haemagglutination assay, in IU.

WBC count (×103/ml)† 19.1 18.9
ESR (per h)† 79 90
Albumin (mg/dl)†  2.4  2.10.02
Total bilirubin†  1.9  2.4
ALT (IU/ml)† 66 63
ALP (IU/ml)†211236
IHA titre
 >1:256290/403 (72) 25/76 (33)1
 ≤1:256113/403 (28) 51/76 (67)5.24<0.001
Abnormal chest X-ray164/343 (48) 40/80 (50)1.09
Abscess
 Solitary348/451 (77) 53/102 (52)1
 Multiple103/451 (23) 49/102 (48)3.12<0.001
Liver lobe affected
 Right330/451 (73) 56/99 (57)1
 Left 75/451 (17) 16/99 (16)1.26
 Both 46/451 (10) 27/99 (27)3.46<0.001
Largest abscess diameter (cm)†  7.0  7.2
Percutaneous aspiration183/471 (39) 45/106 (41)1.07

Chest X-ray abnormalities including atelectasis, pulmonary infiltrates, and pleural effusion or elevated right haemidiaphragm were found with comparable frequency in both groups. Patients (48%) with pyogenic abscess had multiple lesions compared with only 23% of those with amoebic infection (P < 0.001). Involvement of the right lobe of the liver was seen more frequently with amebic vs. pyogenic abscess, while involvement of both lobes was more common in patients with pyogenic abscess (P < 0.001). Sonographic features (e.g. round or oval shape, hypoechoic appearance) were found to be of no value in suggesting the microbiological aetiology of the liver abscess (data not shown). Similarly, abscess size was not significantly different between the two study cohorts.

The above univariate analysis was followed by multivariate logistic regression analysis which confirmed that age >50 years, pulmonary findings on examination, multiple abscesses, and amoebic serology titres <1:256 IU were predictive of pyogenic infection (Table 3).

Table 3.  Features associated with pyogenic liver abscess based on multivariate analysis (n = 452)
FactorAdjusted odds ratio95% CIP-value
  1. CI, confidence interval; IHA, indirect haemagglutination assay, in IU.

IHA titre
 >1:2561  
 ≤1:2568.55(4.4–16.3)<0.001
Abscess
 Solitary1  
 Multiple5.34(2.86–9.96)<0.001
Age (years)
 ≤501  
 >502.86(1.59–5.17)<0.001
Rales/rhonchi
 No1  
 Yes2.43(1.2–4.9)<0.013

Percutaneous aspiration was performed in a minority of instances in each group (183 of 471 and 45 of 106 amoebic and pyogenic cases, respectively); diagnostic aspiration for suspected amoebic cases was generally undertaken because of slow response to antiamoebic therapy. Forty-four percent (20 of 45) of abscess cultures in the pyogenic group showed growth of bacteria compared with 6.6% (12 of 183) in the amoebic group (P < 0.001). Abscess cultures most frequently yielded Streptococcus species (e.g. viridans species and non-enterococcal group D streptococci) and Staphylococcus species (e.g. S. aureus, S. saprophyticus and S. epidermidis). Other recovered microorganisms included Candida albicans, Enterococcus species, Escherichia coli, Pseudomonas aeruginosa, Proteus mirabilis, Klebsiella pneumoniae, Bacillus species, and anaerobes. Blood cultures were obtained from more than 60% of patients from each group and were positive for Streptococcus and Enterobacteriaceae in 23% of patients with pyogenic abscess.

Thirty percent of all study subjects were treated with antiprotozoal agents alone, while the rest received a combination of antiprotozoals and antibacterials. Patients (85%) with amoebic abscess who received metronidazole and/or chloroquine alone and who completed follow-up (245 of 289 cases) had complete clinical cure with their first course of antimicrobials. Patients in the pyogenic abscess group were treated with antibacterials (usually cefotaxime and clindamycin) alone (9%) or in combination with antiprotozoal drugs (89%). In our hospital, metronidazole is generally chosen because of its significantly lower cost than clindamycin, with its antiamoebic activity being only a secondary consideration. Forty-five patients (42%) in the pyogenic group underwent diagnostic and/or therapeutic drainage of their abscesses. Antimicrobial treatment of patients with pyogenic abscesses was generally continued for 2–6 weeks, with a mean of about 4 weeks.

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Patients and methods
  5. Statistical analysis
  6. Results
  7. Discussion
  8. Conclusion
  9. References

Only a few literature reports have comparatively studied a sizeable number of patients with amoebic and pyogenic liver abscess (Ogden et al. 1961; Barbour & Juniper 1972; Balasegaram 1981; Conter et al. 1986; Barnes et al. 1987). Our report of 577 patients with hepatic abscess represents the single largest series comparing clinical, laboratory, and radiographic features of both amoebic and pyogenic abscesses, and we have observed or confirmed a number of criteria that seem to be of value in non-invasively differentiating the microbial aetiologies of these abscesses (Table 4).

Table 4.  Useful features in distinguishing amebic from pyogenic liver abscess
Characteristic*Study [reference]
Present series (n = 577)Barnes et al. (1987) [5] (n = 144)Conter et al. (1986) [3] (n = 82)Barbour and Juniper (1972) [2] (n = 66)
  1. * All features listed display statistically significant differences between the amebic and pyogenic groups (P < 0.05).

  2. † Not significant.

  3. ALP, alkaline phosphatase.

Age (>50 years)PyogenicPyogenicPyogenicPyogenic
Diabetes mellitusPyogenicPyogenic
Chest pain/ coughNS†PyogenicNS
Rales/rhonchiPyogenicNSNS
JaundicePyogenicPyogenicPyogenicNS
Low albuminPyogenicPyogenicNS
Elevated bilirubinNSPyogenicPyogenic
Elevated ALPNSNSPyogenicAmoebic
Male predominanceAmoebicAmoebicNSNS
Abdominal painNSAmoebicAmoebicAmoebic
DiarrhoeaNSAmoebicAmoebicAmoebic
HepatomegalyNSNSAmoebicNS
Abdominal tendernessAmoebicAmoebicAmoebicNS
High IHA titresAmoebicAmoebic
Solitary abscessAmoebicNSNS
Right lobe abscessAmoebicNSNSNS

Males predominated over females in both groups; however, a marked male predominance was seen in patients with amoebic abscess, a finding previously well-documented (Barbour & Juniper 1972; Balasegaram 1981; Abuabara et al. 1982; Katzenstein et al. 1982; Conter et al. 1986; Barnes et al. 1987). Differences in gender incidence could reflect varying environmental exposures related to occupation, although the precise underlying explanations remain unclear. Like us, many investigators have noted patients with pyogenic abscess to be older; with amoebic abscess being a disease of young adults aged 28–44 years (Ogden et al. 1961; Levitt et al. 1979; Abuabara et al. 1982; Conter et al. 1986; Blessmann et al. 2002). Literature also reports high-grade fever (>102 °F) with chills, abdominal pain, and prior gastroenteritis more commonly in persons with amoebic abscess (May et al. 1967; Barbour & Juniper 1972; Conter et al. 1986; Barnes et al. 1987). Our finding of an association between pyogenic liver abscess and diabetes mellitus confirmed that of previous investigators (Verlenden & Frey 1980; Greenstein et al. 1984; McDonald 1984; Miedema & Dineen 1984; Conter et al. 1986). Jaundice was more common among our patients with pyogenic abscess; previous studies have found jaundice to be relatively uncommon in patients with amoebic liver abscess (Ogden et al. 1962; McDonald 1984; Conter et al. 1986; Barnes et al. 1987). The greater likelihood of abnormal chest findings in patients with pyogenic liver abscess noted in the present study has not been reported before.

While our study confirmed that there was no difference between the peripheral white blood cell counts in patients with amoebic or pyogenic abscesses (May et al. 1967; Barbour & Juniper 1972; Ribaudo & Ochsner 1973; Conter et al. 1986), hypoalbuminaemia was more severe in those with bacterial abscess. Our results found serum liver enzymes to be of no value in distinguishing between amoebic and pyogenic abscess, although previous observers have suggested that bacterial abscesses may be more frequently associated with greater elevations of serum alanine transferase and alkaline phosphatase (Barbour & Juniper 1972; Barnes et al. 1987).

In non-endemic areas, a positive amoebic serology almost always reflects acute infection, and in the setting of hepatic abscess is essentially diagnostic of an amoebic aetiology. However, in regions where amoebiasis is very prevalent, positive serology does not carry the same diagnostic usefulness as seropositivity may reflect prior invasive intestinal infection. Serological IHA titres usually become negative within 1 year of acute infection, but may remain elevated for 5–6 years after cure in some patients. In the present study, a substantial percentage of patients (54%) categorized as having pyogenic liver abscess had elevated IHA titres (>1:32 IU) suggesting prior invasive amoebic disease. However, serological titres below 1:256 IU were predictive of pyogenic abscess in both univariate and multivariate analyses.

Although a few previous studies have noted certain ultrasonographic criteria to be more characteristic of amoebic abscess (Ralls et al. 1987), we could find no such features diagnostically useful in suggesting the microbial aetiology of hepatic abscess. Solitary right lobe lesions have been repeatedly found to be more common in amoebic disease (Ogden et al. 1961; May et al. 1967; Barbour & Juniper 1972; Ribaudo & Ochsner 1973; Levitt et al. 1979; Verlenden & Frey 1980; Balasegaram 1981; Conter et al. 1986) and we verified this finding. Ultasonography can, however, be useful in identifying aetiologies commonly associated with pyogenic liver abscess such as cholangitis, bile duct occlusion and underlying liver cirrhosis.

Blood and/or liver aspirate cultures that grow bacteria, especially enteric aerobic or anaerobic organisms or Sta. aureus, are generally indicative of a pyogenic aetiology. As suggested by others (Levitt et al. 1979; McDonald 1984; Donovan et al. 1991), we believe that aspiration should be performed in all patients with suspected pyogenic abscess to define the microbial flora; diagnostic aspiration of a presumed amoebic abscess is generally not indicated (Blessmann et al. 2003). Bacterial superinfection of an amoebic abscess may occur but is rare enough (12 of 471, or 2.5% of our subjects) to counter indicate routine sampling of suspected amoebic liver abscess. Interestingly, needle aspiration of a non-pyogenic liver lesion may actually introduce secondary bacterial infection (McDonald 1984).

Amoebic liver abscesses can be effectively treated with a 10-day course of antiprotozoal agents alone (Abuabara et al. 1982; Katzenstein et al. 1982; Conter et al. 1986). Therapeutic aspiration should be reserved for the occasional patient responding poorly to antibiotic therapy (Donovan et al. 1991), and for those with large left lobe abscesses that may rupture into the pericardium (Pitt 1990). In contrast, a combined medical–surgical approach, typically employing percutaneous drainage (Herbert et al. 1982; Pitt 1990) is indicated for many or most bacterial hepatic abscesses. In our series, open surgical drainage was required in only two patients with amoebic abscess and in four patients with pyogenic abscess; in all cases surgery was performed because of fever persisting for more than 2 weeks despite percutaneous drainage and appropriate medical therapy. Empiric antibiotic therapy for pyogenic abscess should provide coverage against Enterobacteriaceae, anaerobic cocci including microaerophilic streptococci, and anaerobic bacilli including Bacteroides fragilis (e.g. regimens such as metronidazole or clindamycin plus a third generation cephalosporin, fluoroquinolone, or aminoglycoside). Therapy is frequently continued for 4–6 weeks.

Conclusion

  1. Top of page
  2. Summary
  3. Introduction
  4. Patients and methods
  5. Statistical analysis
  6. Results
  7. Discussion
  8. Conclusion
  9. References

We found several clinical and laboratory features that can help to differentiate pyogenic from amoebic liver abscess. In our setting patients older than 50 years, with pulmonary findings on physical examination, multiple abscesses, and amoebic serology titres <1:256 IU are more likely to have a pyogenic liver abscess. Amoebic abscesses tend to occur in young males, are frequently accompanied by a history of diarrhoea and abdominal pain/tenderness, and have amebic serology titres above 1:256 IU. The consistency of our results with previous literature studies suggests that a presumptive diagnosis based on the above features is reliable, and that diagnostic aspiration of all patients presenting with hepatic abscess is not required.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Patients and methods
  5. Statistical analysis
  6. Results
  7. Discussion
  8. Conclusion
  9. References
  • Abuabara SF, Barrett JA, Hau T et al. (1982) Amebic liver abscess. Archives of Surgery 117, 239244.
  • Balasegaram M (1981) Management of hepatic abscess. Current Problems in Surgery 18, 282340.
  • Barbour GL & Juniper K Jr (1972) A clinical comparison of amebic and pyogenic abscess of the liver in sixty-six patients. American Journal of Medicine 53, 323334.
  • Barnes PF, DeCock KM, Reynolds TN et al. (1987) A comparison of amoebic and pyogenic abscess of the liver. Medicine 66, 472484.
  • Blessmann J, Van Linh P, Nu PA et al. (2002) Epidemiology of amebiasis in a region of high incidence of amoebic liver abscess in central Vietnam. American Journal of Tropical Medicine and Hygiene 66, 578583.
  • Blessmann J, Binh HD, Hung DM et al. (2003) Treatment of amoebic liver abscess with metronidazole alone or in combination with ultrasound-guided needle aspiration: a comparative, prospective and randomised study. Tropical Medicine and International Health 8, 10301034.
  • Conter RL, Pitt HA, Tompkins RK et al. (1986) Differentiation of pyogenic from amoebic hepatic abscesses. Surgical Gynecology and Obstetrics 162, 114120.
  • Donovan AJ, Yellin AE & Ralls PW (1991) Hepatic abscess. World Journal of Surgery 15, 162169.
  • Greenstein AJ, Lowenthal D, Hammer GS et al. (1984) Continuing changing patterns of disease in pyogenic liver abscess: a study of 38 patients. American Journal of Gastroentrology 79, 217226
  • Herbert DA, Rotham J, Simmons F et al. (1982) Pyogenic liver abscesses: successful nonsurgical therapy. Lancet 1, 134136.
  • Katzenstein D, Rickerson V & Braude A (1982) New concepts of amebic liver abscess derived from hepatic imaging, serodiagnosis and hepatic enzymes in 67 consecutive cases in San Diego. Medicine 61, 237246.
  • Levitt MD, Quintan MF & Sheiner HJ (1979) Liver abscess in Western Australia (1974–1983). Australian and New Zealand Journal of Surgery 56, 341346.
  • May RP, Lehmann JD & Sanford JP (1967) Difficulties in differentiating amebic from pyogenic liver abscess. Archives of Internal Medicine 119, 6974.
  • McDonald MI (1984) Pyogenic liver abscess: diagnosis, bacteriology and treatment. European Journal of Clinical Microbiology 3, 506509.
  • Miedema BW & Dineen P (1984) The diagnosis and treatment of pyogenic liver abscesses. Annals of Surgery 200, 328335.
  • Ogden WW, Hunter PR & River JD (1961) Liver abscess. Postgraduate Medicine 30, 1119.
  • Ogden WW, Hunter P & River JD (1962) Pyogenic liver abscess. Journal of the Louisiana State Medical Society 114, 139.
  • Pitt HA (1990) Surgical management of hepatic abscesses. World Journal of Surgery 14, 498504.
  • Ralls PW, Barnes PF, Radin DR et al. (1987) Sonographic features of amoebic and pyogenic liver abscesses: a blinded comparison. American Journal of Roentgenology 149, 499501.
  • Ribaudo JM & Ochsner A (1973) Intrahepatic abscesses: amebic and pyogenic. American Journal of Surgery 125, 570574.
  • Stanley SL Jr (2003) Amebiasis. Lancet 361, 10251034.
  • Verlenden WI III & Frey CF (1980) Management of liver abscess. American Journal of Surgery 140, 5259.