Liver abscesses in Southern Europe: clinical and microbiological characterization of a retrospective cohort

Abstract

Background

Liver abscess is a potentially life-threatening condition and differentiating between pyogenic (PLA) and amoebic liver abscess (ALA) can be challenging. Besides, increasing antimicrobial resistance and the lack of standardized guidelines complicate its management.

Methods

We conducted an observational study of adult individuals diagnosed with liver abscess at Vall d’Hebron University Hospital (Barcelona) from 2016 to 2020. Patients were classified as cryptogenic (CLA) or non-cryptogenic (NCLA) based on presumed pathophysiology. Data on demographics, microbiology, treatment and outcomes were collected. Multivariate logistic regression identified predictors of ALA, recurrence, and mortality.

Results

145 patients were included. Median age was 71 years; 74% were male. NCLA represented 72% of cases, mainly due to biliary tract disease (65%). Microbiological diagnosis was obtained in 117 (80%) cases, with Enterobacterales present in 75/103 (73%) of PLA. Enterobacterales showed high resistance rates, with amoxicillin-clavulanate resistance of 56% in NCLA vs. 29% in CLA, and to piperacillin-tazobactam 49% vs. 14%, respectively. ALA accounted for 14/40 (35%) of CLA. Migration or recent international travel was the only predictor (OR 11.2, p = 0.01), but 4/14 (29%) had no risk factors. Recurrence occurred in 7/92 (8%) NCLA cases. Thirty-day mortality for pyogenic abscesses was 14/106 (12%). Solid (OR 6.5, p = 0.003) and hematologic malignancies (OR 6.1, p = 0.043) were independently associated with early mortality.

Conclusion

CLA and NCLA differ significantly in presentation and microbiological isolations with high antimicrobial resistance among NCLA. In CLA cases, amoebic aetiology should be considered, even in the absence of classical risk factors.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12879-026-12698-9.

Introduction

Pyogenic liver abscess (PLA) is a potentially life-threatening infection of hepatic parenchyma, particularly in individuals with underlying comorbidities. In Europe, the incidence of PLA is estimated at 3.7 to 7 cases per 100.000 population annually [1, 2]. Diagnosis of PLA can be challenging due to its nonspecific clinical presentation. Despite advances in diagnostic methods, drainage techniques, and antimicrobial therapy, the 30-day mortality rate remains notable, ranging from 5.3% to 14% [1–4]. To date, no standardized clinical practice guidelines exist for the management of PLA, likely owing to its heterogeneous aetiology and the scarcity of randomized controlled trials.

The most commonly reported causes of PLA are related to biliary tract pathologies, such as acute cholecystitis, cholangitis, choledocholitiasis, malignancies, or prior biliary instrumentation [5]. Haematogenous dissemination via the portal circulation following intra-abdominal infections is also a recognized mechanism. However, in approximately 29–35% of cases, no clear aetiology is identified, and these are classified as cryptogenic abscesses [1, 5].

In Europe, the predominant pathogens isolated from PLA are Enterobacterales, mainly Escherichia coli and Klebsiella spp. Between 30 and 60% of infections are polymicrobial, with anaerobic bacteria present in approximately 10% of cases [2, 4, 6, 7]. In a subset of patients, cultures may fail to yield any organisms. Given the global rise in antimicrobial resistance and the limited available data, further studies are needed to characterize local microbiological profiles and guide empirical therapy.

Amoebic liver abscess (ALA), caused by the protozoan Entamoeba histolytica, represents another well-recognized cause of hepatic abscess. ALA is the most common extraintestinal manifestation of E. histolytica infection. Its incidence varies geographically; while it is rare in Europe, it accounts for the majority of liver abscess cases in endemic regions such as India [8, 9]. Risk factors for ALA include younger age, male sex, alcohol consumption and travel to high endemic regions. The standard therapy for ALA is benzimidazoles (metronidazole or tinidazole). Few studies have characterized amoebic abscess in non-endemic areas. Compared to PLA, ALA is generally associated with shorter hospital stays and lower mortality rates as suggested by comparative studies [5, 6].

The aim of this study is to characterize the clinical, microbiological and antimicrobial resistance profiles of a large cohort of patients diagnosed with liver abscess at a European tertiary hospital. Additionally, we aim to identify risk factors of ALA to aid in early diagnosis and management.

Methods

Study population and setting

We conducted a retrospective, single-centre study of all adult patients (≥ 18 years) diagnosed with a liver abscess at Vall d’Hebron University Hospital in Barcelona, Spain, between January 2016 and December 2020. Cases were identified through a systematic search of hospital records using clinical coding data from the International Classification of Diseases, 9th and 10th revisions (ICD-9 and ICD-10), with the search terms “liver abscess” (ICD K75.0) and “amoebic liver abscess” (ICD A06.4). Additional amoebic cases were identified through positive test results reported by the microbiology department. Duplicate cases were excluded.

Case definitions

Liver abscess was defined based on imaging findings consistent with abscess formation, detected by ultrasound, computed tomography (CT) or magnetic resonance imaging, in combination with microbiological confirmation from either peripheral blood or aspirated pus. In cases lacking microbiological confirmation, the diagnosis was accepted based on compatible clinical findings (fever and right hypochondrium pain), together with imaging evidence of a liver abscess, and after exclusion of alternative diagnoses [3, 5]. Patients with superinfected hepatic neoplasms, extrahepatic abscesses, or incomplete clinical records were excluded. For patients with multiple episodes, baseline characteristics were recorded only once. All episodes were considered for clinical, microbiological, and outcome analyses.

From a clinical point of view, and based on presumed pathophysiology at diagnosis we classified patients as having either cryptogenic liver abscess (CLA) or non-cryptogenic liver abscess (NCLA). Cryptogenic liver abscesses were defined as those with no identifiable source of infection following comprehensive evaluation. Non-cryptogenic abscesses were attributed to identifiable causes such as biliary tract disease, intra-abdominal infections, or endovascular infections. We defined recurrence as a new liver abscess occurring within the first year with the same microbiological isolate as the initial episode.

Study variables and data collection

Data were extracted from the hospital’s electronic health record system. Collected variables included demographic characteristics, comorbidities and Charlson comorbidity index [10], clinical presentation, and initial laboratory findings. Information regarding abscess morphology, isolated microorganisms, and their antimicrobial resistance pattern was documented. Treatment-related variables included empirical and targeted antibiotic regimens, treatment duration, use of percutaneous aspiration or drainage, and adequacy of treatment. Percutaneous drainage was indicated in abscesses ≥ 5 cm, multiloculated lesions or clinical non-response to antibiotics after 48–72 h. Outcomes of interest were recurrence within the first year, 30-day mortality, and overall mortality. Follow-up data included endoscopic evaluations performed for colon cancer screening, as well as the identification of new digestive or abdominal malignancies diagnosed within 12 months after the liver abscess episode.

Microbiological procedures

Blood cultures were processed using the BacT/ALERT 3D automated system (bioMérieux, Marcy l’Étoile, France). When available, aerobic and anaerobic cultures and Gram stains of aspirated pus were performed according to standard hospital protocols. Identification of the pathogen was performed by mass spectrometry. In cases with negative cultures, 16 S rRNA gene sequencing was used whenever possible. Antimicrobial susceptibility testing followed the guidelines of the European Committee on Antimicrobial Susceptibility Testing (EUCAST). In the analysis of antimicrobial resistance, changes introduced in 2019 to the definition of the ‘intermediate’ (I) category have been taken into account. Thus, category ‘I’, which was previously assessed together with category ‘Resistant’ due to its association with an uncertain therapeutic effect, is now considered ‘Susceptible’ when exposure to the antimicrobial agent is increased. When applying the current interpretation criteria to isolates obtained since 2019, most antimicrobial susceptibility percentages remained unchanged.

In cases of suspected E. histolytica infection, amoebic serology was determined by enzyme-linked immunosorbent assay (ELISA): NovaLisa Entamoeba histolytica IgG (NovaTec Immunodiagnostica GmbH. Dietzenbach, Germany) with titers > 1:128 IU deemed positive. Additional diagnostic methods included antigen detection in pus or faeces by Entamoeba CELISA Path1 test (Cellabs, Brookvale, New South Wales, Australia), PCR-based identification of E. histolytica DNA, and microscopic observation of cysts or trophozoites in pus. ALA was diagnosed when any of these tests were positive.

Statistical analysis

Categorical variables were summarized as frequencies and percentages, while continuous variables were expressed as means and standard deviations or medians with interquartile ranges, depending on distribution. Categorical comparisons were made using Fisher’s exact test or Chi-square test, as appropriate. Variables showing potential association in univariate analysis were entered into multivariable logistic regression models using backward elimination to identify independent predictors of recurrence, mortality, and ALA. A two-tailed p-value of < 0.05 was considered statistically significant, and 95% confidence intervals were reported. Data were analysed using SPSS software (version 20.0; SPSS Inc., Chicago, IL).

Results

Overall, 337 episodes of liver abscesses were initially reviewed, from which 145 patients (146 episodes) met inclusion criteria and included in the final analysis (Fig. 1). The median age was 71 years (interquartile range [IQR] 57–82), with a predominance of male individuals (108/145; 74%). In 105 cases (72%), a predisposing condition was identified classifying the case as NCLA. The most common condition was biliary tract abnormalities, present in 95/145 cases (65%), followed by non-biliary intra-abdominal infections (8/145, 2%), and endovascular infections (2/145; 1.4%) (Table 1).

Fig. 1.

Flowchart of patients included and excluded in the study

Table 1.

Predisposing conditions that led to allocate the abscess to the NCLA group

Risk factor	Ratio (%)
Biliary tract abnormalities	95/145 (65)
Choledocholithiasis/cholecystitis	42/145 (29)
Biliodigestive derivation	14/145 (10)
Biliary endoprosthesis	11/145 (8)
Malignant biliary stricture	7/145 (5)
Benign biliary stricture	6/145 (4)
Bile duct ectasia without identifiable cause	6/145 (4)
Endoscopic retrograde cholangiopancreatography	4/145 (3)
Percutaneous biliary drainage	4/145 (3)
Biliary hamartomatosis	1/145 (1)
Intra-abdominal infection	8/145 (5)
Postoperative infection	6/145 (4)
Neutropenic colitis	2/145 (1)
Endovascular infection	2/145 (1)
Infective endocarditis	1/145 (1)
Septic thrombophlebitis	1/145 (1)

A comparison between CLA and NCLA cases revealed that NCLA patients were older (median age: 74 vs. 60 years, p < 0.01), had a higher comorbidity burden (Charlson Comorbidity Index: 5 vs. 2, p < 0.01) and more frequently presented with a solid organ malignancy (26% vs. 0%, p < 0.01). Fever and diarrhoea were more commonly reported among CLA patients (95% vs. 73%, p = 0.03; and 35% vs. 7%, p < 0.01, respectively). No significant differences were observed in the number or location of abscesses; however, CLA lesions tended to be larger (median size 60 mm vs. 30 mm, p < 0.01) (Table 2).

Table 2.

Basal and clinical characteristics

Age, years; (median; IQR)	NCLA (n = 105)	CLA (n = 40)	p value
	74 (61–84)	60 (45–71)	< 0.01
Male sex	76 (72)	32 (80)	0.35
Comorbidities
Charlson Comorbity Index; (median; IQR)	5 (4-6.5)	2 (0–3)	< 0.01
Obesity	6 (6)	4 (10)	0.46
Diabetes mellitus	27 (26)	7 (17)	0.3
Liver cirrhosis	9 (9)	1 (2)	0.29
Proton pump inhibitors use	56 (53)	12 (30)	0.12
Immunosuppression	42 (40)	3 (7)	< 0.01
Solid organ neoplasm	26 (25)	0	< 0.01
Pancreatic neoplasm	11 (10)	0
Cholangiocarcinoma	6 (6)	0
Hepatocarcinoma	1 (1)	0
Colorectal neoplasm	3 (3)	0
Gastric cancer	2 (2)	0
Breast cancer	2 (2)	0
Renal neoplasm	1 (1)6 (6)	0
Hematological neoplasia	5 (5)	2 (5)	1
Solid organ transplantation	2 (2)	0	0.32
Monoclonal antibodies	1 (1)	0	1
Bone marrow aplasia	1 (1)	0	1
HIV infection	0	0	1
Primary immunodeficiency		1 (2)	0.28
Clinical presentation a
Fever	77 (73)	36 (90)	0.03
Abdominal pain	67 (63)	24 (60)	0.85
Nausea/vomiting	29 (27)	14 (35)	0.37
Diarrhea	8 (8)	13 (32)	< 0.01
Pleural effusion	31 (29)	11 (27)	0.84
Septic shock	14 (13)	6 (15)	0.78
Abscess characteristics a
Number of abscesses			0.48
1	49 (46)	23 (57)
2	13 (12)	4 (9)
≥3	43 (41)	13 (33)
Location			0.11
Right lobe	47 (44)	25 (62)
Left lobe	17 (16)	6 (15)
Both lobes	42 (40)	9 (22)
Size, mm (median; IQR)	30 (15–64)	60 (40–85)	0.01

Data are shown as No. (%) unless otherwise indicated

Abbreviations: CLA, cryptogenic liver abscess; HIV, human immunodeficiency virus; IQR, interquartile range, NCLA, non-cryptogenic liver abscess

^aCalculated on 106 episodes of liver abscess

Among the 40 CLA cases, 26 (65%) were pyogenic and 14 (35%) amoebic. Patients with amoebic liver abscess (ALA) were younger (median age: 44 years [IQR 32–59] vs. 67 years [IQR 55–78], p < 0.01), more frequently had a history of migration or recent international travel (71% vs. 15%, p < 0.01), and more commonly presented with diarrhoea (57% vs. 19%, p = 0.03). ALA also presented as significantly larger lesions (median size 79 mm [IQR 64–101] vs. 48 mm [IQR 35–67], p = 0.003). In a multivariate logistic regression, only migration or recent international travel was independently associated with ALA (OR 11.2 95%CI [1.8–68.2], p = 0.01) (Table S1 of Supplementary material). In 4/14 (29%) of ALA cases, no risk factors were found.

Microbiological confirmation was achieved in 38/40 (95%) CLA cases compared to 79/106 (74%) in NCLA (p = 0.01). Amoebic diagnostic tests were performed in 21/40 (52%) CLA cases, yielding positive results in 14 (67%). Among ALA cases, all tested samples were positive by serology (14/14), PCR (10/10), and antigen detection in drained pus (6/6). Antigen detection in faeces was positive in 2/4 cases, while amoebic cysts were visualized in pus in 1/7 individuals. Amoebic typhlitis could not be ruled out as none of them underwent endoscopic evaluation during the acute phase.

Among pyogenic CLA cases, monomicrobial infections were more common than in NCLA (83% vs. 52%, p = 0.01). Enterobacterales were the most frequently isolated pathogens in both CLA and NCLA (58% and 77% respectively), followed by anaerobes (29% and 25%) and the Streptococcus anginosus group (25% and 24%). Hypervirulent strains of Klebsiella pneumoniae were only isolated in two patients, one case imported from Philippines with a solitary liver abscess; and another locally acquired with disseminated infection in prostate, lungs and eyes. NCLA cases showed greater microbial diversity, with additional isolates of Enterococci (24%), Candida spp. (5%), Staphylococcus aureus (4%) and Pseudomonas aeruginosa (2%) (Table 3). Anaerobic isolates were more frequent among patients with cancer (8/21 [38%] vs. 19/96 [20%]), although this difference did not reach statistical significance (p = 0.08). Antimicrobial resistance among Enterobacterales was more prevalent in NCLA, with significant differences for piperacillin-tazobactam (49% vs. 14%, p = 0.02) and third-generation cephalosporins (28% vs. 0%, p = 0.02) (Table 4).

Table 3.

Microbiological characteristics

	NCLA (n = 106)	CLA (n = 40)	p value
Etiology
Pyogenic	106/106 (100)	26/40 (65)
Amoebic	0	14/40 (35)
Microbiological diagnosis	79/106 (74)	38/40 (95)	0.01
Amoebic diagnostic tests performed	0	21/40 (52)	< 0.01
Positive amoebic diagnostic test
Serology	-	14/14 (100)
PCR in drained abscess	-	10/10 (100)
Antigen in drained abscess	-	7/7 (100)
Microscopy in drained abscess	-	1/7 (14)
Antigen in faeces	-	2/4 (50)
Microscopy in faeces	-	4/10 (40)
Positive microbiological sample (pyogenic)			0.35
Blood cultures	39/79 (49)	10/24 (42)
Abscess culture	18/79 (23)	9/24 (37)
Blood and abscess culture	22/79 (28)	5/24 (21)
Number of isolated microorganisms (pyogenic)			0.01
Monomicrobial	41/79 (52)	20/24 (83)
Polymicrobial	38/79 (48)	4/24 (17)
Etiology (pyogenic)
Enterobacterales	61/79 (77)	14/24 (58)	0.07
Escherichia coli	40/79 (51)	6/24 (25)	0.03
Klebsiella spp.	27/79 (34)	5/24 (21)	0.22
Hypervirulent strain	0/27	2/5 (40)
Enterobacter spp.	8/79 (10)	0	0.19
Other Enterobacteralesa	6/79 (8)	3/24 (12)	0.43
Anaerobesb	20/79 (25)	7/24 (29)	0.71
Streptococcus anginosus group	19/79 (24)	6/24 (25)	0.92
Enterococcus spp.c	19/79 (24)	0	0.01
Staphylococcus spp d	3/79 (4)	0	1
Pseudomonas aeruginosa	2/79 (2)	0	1
Candida spp.	4/79 (5)	0	0.57

Data are shown as No. (%) unless otherwise indicated

Abbreviations: CLA, cryptogenic liver abscess; NCLA, non-cryptogenic liver abscess

^a NCLA: Morganella morganii (3 cases), Citrobacter freundii (1 case), Proteus mirabilis (1 case), Serratia marcescens (1 case). CLA: Serratia marcescens (2 cases), Proteus mirabilis (1 case)

^b NCLA: Bacteroides spp. (6 cases), Clostridium perfringens (3 cases), Fusobacterium spp. (3 cases), polymicrobial anaerobic flora (3 cases), Parvimonas spp. (2 cases), Prevotella spp. (2 cases), Actinomyces meyeri (1 case). CLA: Fusobacterium spp. (3 cases), Bacteroides spp. (1 case), Parvimonas spp. (1 case), polymicrobial anaerobic flora (1 case), Prevotella spp. (1 case)

^c Enterococcus spp: E. faecalis (5 cases), E. faecium (14 cases)

^DStaphylococcus spp: S. aureus (2 cases), S. epidermidis (1 case)

Table 4.

Antibiotic resistance pattern in enterobacterales

Antibiotic resistance	NCLA (N = 61)	CLA (N = 14)	p value
Amoxicillin-clavulanic acid	34 (56)	4 (29)	0.07
Piperacillin-tazobactam	30 (49)	2 (14)	0.02
3rd generation Cephalosporin	17 (28)	0	0.02
E. coli ESBL	2
Klebsiella spp. AmpC	2
Klebsiella spp. ESBL	6
Klebsiella pneumoniae OXA	1
Klebsiella pneumoniae NDM	1
Enterobacter cloacae AmpC	3
Morganella morganii AmpC	1
Serratia marcescens AmpC	1
Carbapenem	3 (5)	0	0.39
Klebsiella pneumoniae OXA	1
Klebsiella pneumoniae NDM	1
Enterobacter cloacae VIM	1
Quinolones	18 (29)	1 (7)	0.08

Data are shown as No. (%) unless otherwise indicated

Abbreviations: CLA, cryptogenic liver abscess; ESBL: extended-spectrum β-lactamase; AmpC: class C β-lactamase; OXA: Oxacillinase class D β-lactamase; NCLA, Non-cryptogenic liver-abscess; NDM: New Delhi class B metallo-β-lactamase; VIM: Verona integro-encoded class B metallo-β-lactamase

All patients received antimicrobial treatment. Percutaneous image-guided drainage was performed in 68/146 (46.6%) cases, predominantly in those with larger abscesses (mean size 71 mm [SD 32] vs. 24 mm [SD 20], p < 0.01). Drainage was more commonly performed in CLA compared to NCLA (75% vs. 35%, p < 0.01). Among the drained cases, only one pyogenic liver abscess had a negative pus culture; 16 S rRNA gene sequencing was subsequently performed, yielding a negative result. Empirical therapy with amoebicidal coverage was more frequently initiated in CLA compared to NCLA (20% vs. 4%, p = 0.01), although only 6/14 (43%) ALA cases received appropriate empirical treatment. Among microbiologically confirmed pyogenic abscesses, empirical therapy was adequate in 87% of CLA cases and 71% of NCLA cases, with no differences between groups. Empirical antibiotic regimens and treatment outcomes are detailed in Table 5.

Table 5.

Treatment and outcomes

Drainage	NCLA (n = 106)	CLA (n = 40)	p value
	38 (35.8)	30 (75)	< 0.01
Empiric antibiotic
Amoxicillin-clavulanate	27 (25)	21 (52)a	< 0.01
Piperacillin-tazobactam	52 (49)	2 (5)b	< 0.01
Third generation cephalosporin	2 (1.9)	8 (20)c	< 0.01
Carbapenem	17 (16)	8 (20)	0.57
Quinolone	7 (7)d	1 (2)	0.45
Other	1 (1)e	0	1
Empiric amoebicidal treatment (overall)	4 (4)	8 (20)	0.01
Empiric amoebicidal treatment (amoebic abscesses)	-	6/14 (43)
Adequate empiric treatment (overall)	56/79 (71)	26/38 (68)	0.79
Adequate empiric treatment (pyogenic abscesses)	56/79 (71)	21/24 (87)	0.1
Duration of antibiotic treatment, days (median; IQR) (pyogenic abscesses)	32 (22–44)	45 (28–52)	0.04
30-day all-cause mortality (overall)	14 (13)	2 (5)	0.24
30-day all-cause mortality (pyogenic abscesses)	14/106 (13)	2/26 (8)	0.74
1-year all-cause cumulative mortality (overall)	35 (33)	4 (10)	0.04
1-year all-cause cumulative mortality (pyogenic abscesses)	35/106 (33)	4/26 (15)	0.08
1- year recurrence	7/92 (8)f	0	0.1

Data are shown as No. (%) unless otherwise indicated

Abbreviations: CLA, cryptogenic liver abscess; IQR, interquartile range; NCLA, Non-cryptogenic liver abscess

^a In combination with metronidazole in 1 case

^b In combination with metronidazole in 1 case

^c In combination with metronidazole in 6 cases

^d In combination with metronidazole in 4 cases, with clindamycin in 1 case and with teicoplanin in 1 case

^e Ceftazidime/avibactam plus aztreonam

^f All cases had an alteration of the biliary tract: 4 biliary lithiasis, 1 biliodigestive derivation surgery, 1 biliary endoprosthesis and 1 percutaneous biliary drainage

Univariate and multivariate analysis of 30-days mortality, cumulative mortality and recurrence are presented in Supplementary Tables S2-S4. The 30-day all-cause mortality rate was 12% (16/132) among pyogenic abscesses, while no deaths occurred among ALA cases, despite 4/14 (28%) requiring ICU admission. When comparing 30-day mortality between patients with positive and negative microbiological results, no significant differences were observed [11/106 (9%) vs. 5/24 (17%), p = 0.31]. Among those who died due to the infection (7/16), 71% (5/7) had an active neoplasm. In the multivariate model, both solid organ and hematologic malignancies were independently associated with early mortality [OR 6.5 95%CI (2–23), p < 0.01 and OR 6.1 95%CI (1–35), p = 0.04, respectively]. Cumulative one-year all-cause mortality was 39/145 (27%), with solid organ malignancy being the only variable independently associated in multivariate analysis [OR 8.7 95%CI (3–26), p < 0.01]. Seven cases of recurrence were observed within the first year, all in patients with NCLA. Among those with positive cultures, recurrence occurred only in patients who received inadequate tailored treatment (4/31; 13%), compared to none among those who received adequate treatment (p = 0.01). However, multivariate analysis was not performed due to the limited number of events. During follow-up, 17 patients underwent endoscopic evaluation, 12 as part of colon cancer screening and 5 for routine surveillance of pre-existing disease. Nine patients were found to have colonic polyposis, but no new malignancy was diagnosed in any case.

Discussion

In our cohort, 72% of patients presented identifiable risk factors for liver abscess development, with biliary tract disorders being the most frequent aetiology. Among cryptogenic cases, it is worth noting that one-third were of amoebic origin; however, only 43% of these received empirical amoebicidal treatment. Among pyogenic abscesses, we found a considerable rate of antimicrobial resistance among Enterobacterales, particularly to amoxicillin-clavulanate and to piperacillin-tazobactam, leading to inappropriate empirical therapy in 25% of cases.

Differentiating between pyogenic and amoebic liver abscesses remains clinically challenging. Pyogenic origin is often suspected in patients with identifiable risk factors such as biliary abnormality, intra-abdominal infection, or endovascular seeding. In our study, all patients with one of these conditions were ultimately diagnosed with pyogenic abscesses. However, in cryptogenic cases, the distinction between pyogenic and amoebic aetiologies is less evident.

Previous literature has associated ALA with younger male patients and recent travel to endemic areas. Radiologically, ALA tends to involve the right hepatic lobe and presents as solitary lesions [6, 11–13]. Our findings partially align with these reports: patients with ALA were younger, had more frequently had a history of migration or recent international travel, and more often presented with solitary and larger lesions. However, these characteristics are not sufficiently specific for clinical application. Importantly, 29% of ALA cases in our series occurred in patients without a travel history or migrant background, consistent with previous local reports [14]. These findings support the recommendation to systematically rule out amoebic aetiology in all cases of CLA, even in non-endemic settings. E. histolytica serology, which showed excellent diagnostic performance in our study, remains a non-invasive and widely accessible diagnostic tool [15].

Regarding pyogenic abscesses, our data reaffirm Enterobacterales as the predominant pathogens, followed by anaerobes and Streptococcus anginosus, in agreement with prior European and North American studies [2, 4, 7, 16–18]. Notably, Escherichia coli was more common than K. pneumoniae, and only 2 of them were hypervirulent strains, one in a patient travelling from Philippines and another locally acquired. This contrasts with the epidemiology in certain Asian countries, where hypervirulent K. pneumoniae strains are the most prevalent and showed distinct clinical behaviours, including increased metastatic complications and severe presentations, highlighting important regional differences that should be considered when interpreting our findings [13, 19–21]. Pathogens such as Enterococcus spp., Staphylococcus spp., Pseudomonas aeruginosa, and Candida spp. were isolated exclusively in NCLA cases, likely reflecting the higher prevalence of malignancy, priori biliary interventions, and antibiotic exposure in this group. These factors may also contribute to the higher resistance rates observed in NCLA isolates. Nonetheless, even among CLA, 29% of Enterobacterales were resistant to amoxicillin-clavulanate, representing a 16% of all isolates. These findings discourage the use of this agent as empirical therapy. In contrast, ceftriaxone combined with metronidazole appears to be an appropriate first-line option, providing broad coverage against E. histolytica and the local resistance pattern.

The 30-day all-cause mortality in patients with pyogenic liver abscesses in our cohort was 12.1%, whereas no deaths occurred among ALA cases. This aligns with previously reported mortality rates ranging from 5% to 14% for pyogenic abscesses and negligible mortality for amoebic abscesses [2, 6, 8]. Notably, NCLA patients had a significantly higher Charlson Comorbidity Index, which may contribute to their clinical outcome. When we analysed the included items, malignancy was the strongest predictor of early mortality, with 69% of deceased patients having an underlying cancer diagnosis, underscoring the poor prognosis of oncologic patients presenting with liver abscess. However, no new malignancies were identified during follow-up, and none of the cryptogenic cases developed cancer. Although previous reports linked liver abscess and underlying colonic malignancies, our results do not support this hypothesis [22]. Further prospective studies are needed to determine whether specific subgroups could benefit from targeted oncologic screening. Finally, although other studies have identified additional risk factors such as advanced age, diabetes mellitus, chronic liver disease, polymicrobial bacteraemia, absence of drainage, and hyperbilirubinemia, our analysis highlights neoplastic disease as the predominant determinant of poor outcomes [1, 23, 24].

This study has several limitations. The retrospective design and reliance on diagnostic coding may have led to selection bias, potentially missing unrecorded cases. The diagnosis of ALA depended on microbiological confirmation, which might have reduced the likelihood of misclassification compared to pyogenic cases, possibly resulting in a relative overrepresentation of amoebic abscesses. On the other hand, amoebic testing was not systematically performed. We acknowledge that some ALA cases may therefore have been missed. However, most CLA patients showed a favourable clinical response to empirical antibiotic regimens that do not include amoebic coverage, suggesting that the real extent of misclassification is likely small. Furthermore, in some cases microbiological isolates were not achieved. Nonetheless, the relatively large number of cases (145 over a 5-year period) is remarkable for a European setting, where the incidence of liver abscess is low. This reflects the role of our institution as a referral center covering a broad population area and complex cases strengthening the representativeness of our findings.

Conclusions

In conclusion, CLA and NCLA display distinct demographic, clinical and microbiological profiles. NCLA is typically associated with individuals with altered biliary tracts and a high burden of comorbidities. The wide microbiological diversity and high antimicrobial resistance rates, particularly among Enterobacterales, underscore the need for individualized clinical and antimicrobial management. In our setting, where ALA accounted for one third of CLA, amoebic aetiology should be considered in these cases, even in patients with no clearly identifiable risk factors. Empirical therapy for these patients should include active agents against E. histolytica and antimicrobial coverage tailored to the local resistance patterns.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

ALA

Amoebic liver abscess

CLA

Cryptogenic liver abscess

NCLA

Non-cryptogenic liver abscess

PLA

Pyogenic liver abscess

CT

Computed tomography

MRI

Magnetic resonance imaging

PCR

Polymerase chain reaction

ELISA

Enzyme-linked immunosorbent assay

EUCAST

European Committee on Antimicrobial Susceptibility Testing

ICD

International Classification of Diseases

IQR

Interquartile range

OR

Odds ratio

CI

Confidence interval

Author contributions

All authors contributed to the study conception and design. L.P-G, D.M-S, M.V. and P.B-N performed material preparation, data collection and analysis. L.P-G, M.V. and P.B-N. wrote the first draft of the manuscript and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Funding

The authors did not receive support from any organization for the submitted work.

Data availability

The datasets generated and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

The study protocol adhered to the ethical principles outlined in the Declaration of Helsinki (1975) and was approved by the Vall d’Hebron University Hospital Ethics Committee (Code: PR(AG)322/2022). Clinical trial number: not applicable. Informed consent was waived due to the retrospective design.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.