Comparison of Clinical Features and Treatment Outcomes of Pyogenic Liver Abscess in Diabetic and Non-Diabetic Patients in a Chinese Hospital

Introduction

Pyogenic liver abscess (PLA) is a common life-threatening liver infection associated with bacterial infections, diabetes mellitus, tumors, biliary tract disorders, and immune dysfunction.¹ An annual increase in liver abscess cases has been observed worldwide in recent years, with incidence rates ranging from 1.0–4.1 per 100,000 in Western countries to 1.1–5.4 per 100,000 in mainland China, and reaching 12–18 per 100,000 in certain Asian nations.^2–5 Furthermore, diabetic patients exhibit substantially higher morbidity and mortality compared to their non-diabetic counterparts.^6,7 Fever, pain in the right upper quadrant of the abdomen, and jaundice are typical clinical features of a liver abscess.⁸ Klebsiella pneumoniae, Escherichia coli, Enterococcus, and Streptococcus are the key agents causing PLA.^9,10 Although advances in antibiotic therapy and percutaneous drainage treatment over the last few decades have helped reduce the mortality rate of PLA from 70% to 6–14%, several challenges remain in the management of liver abscesses.^11,12 Conventionally, the management of large liver abscesses involves antibiotic therapy combined with percutaneous drainage, with surgical intervention reserved for cases of drainage failure. However, controversy persists regarding how to select the optimal individualized treatment strategy at the initial presentation.¹¹

Diabetes mellitus (DM) is considered an independent risk factor for PLA, likely due to increased susceptibility to invasive bacterial infection resulting from hyperglycemia-induced impairment of neutrophil chemotaxis and bactericidal activity, diminished mononuclear phagocyte activation, and compromised intestinal mucosal barrier function.⁵ Although numerous previous studies have comprehensively investigated pyogenic liver abscess (PLA), the differences in clinical features, microbiological profiles, and risk of treatment failure between PLA patients with and without diabetes mellitus within the Chinese population have not been fully elucidated.^13,14 Furthermore, the efficacy of different treatments for liver abscesses among diabetic patients, or between diabetic and non-diabetic patients, also needs to be urgently clarified. This study aims to compare the clinical characteristics of PLA patients with and without concomitant DM and to investigate whether different treatment modalities affect the prognosis of these patients.

Methods

Study Population

This study complies with the Declaration of Helsinki and was by the Clinical Research Ethics Committee of the First Affiliated Hospital, Zhejiang University School of Medicine (No. 2023–0115). We retrospectively reviewed all of the inpatients diagnosed with PLA and treated at the First Affiliated Hospital, Zhejiang University School of Medicine, from January 2017 to December 2019.

PLA was diagnosed using clinical symptoms, etiological tests, and imaging studies. Diabetes was defined based on the criteria established by the American Diabetes Association.¹⁵ The study included only patients with newly diagnosed PLA; those with parasitic or fungal abscesses were excluded.

The following patient data were retrieved from the hospital database: demographic characteristics, clinical features, laboratory results, abscess location and size, microbiological results, diagnostic and treatment approaches, and the response and outcome of treatments.

The initial treatment for PLA involved empirical treatment with antibiotics. If the antibiotics proved to be ineffective, they were changed based on culture results, or empirically if the culture results were negative. Standard diagnostic methods were used to isolate blood or pus cultures for both aerobic and anaerobic organisms. As discussed in previous studies,^9,11 all percutaneous aspiration or drainage was performed under ultrasound guidance, and the ultrasound results were periodically reviewed to assess the efficacy.

Complete recovery was characterized by notable alleviation of clinical symptoms and marked improvement in imaging within 28 days after concluding the initial appropriate antibiotic treatment and/or drainage. The patients were regarded as having primary treatment failure if their treatment was not deemed curative.

Statistical Analysis

Continuous variables were expressed as mean ± standard deviation or median + interquartile range, and categorical variables were expressed as number (percentage). The variables between the DM and non-DM groups were compared using standard statistical tests. Differences in continuous variables were compared using Mann–Whitney U-tests and Student’s t-tests. We compared categorical variables with Chi-square tests. To identify the risk factors for treatment failure in PLA patients, a multivariate logistic regression was carried out. SPSS 22.0 software was utilized for statistical analysis, and two-sided P-values of less than 0.05 were regarded as statistically significant.

Results

Demographic Characteristics

In total, 712 patients diagnosed with PLA were included in this study. Of these 712 patients, 239 (33.6%) had diabetes. The demographic characteristics and clinical features of these 712 patients are illustrated in Table 1. Male patients constituted the majority (67.7%). In addition, 51.4% of the 712 patients were not less than 60 years of age; 94.4% of patients presented with definite fever during the course of the disease; and 12.2% had a history of surgery. The overall median length of hospitalization was 12 days and the median cost was RMB 21,688.02.

Table 1 Clinical Characteristics of 712 PLA Patients with Diabetes Mellitus (DM) or Non-DM

Among the PLA patients, the highest percentage of the underlying disease was cholelithiasis (27.7%), followed by malignant tumors (17.6%) and fatty liver (7.4%). Among patients without diabetes, cholelithiasis, tumors, and cirrhosis were identified as major disorders (P < 0.05). In contrast, fatty liver was more common in the DM group (P < 0.05). Among the laboratory parameters, elevated CRP and hypoalbuminemia were more frequently observed in the DM group (P < 0.05). In the non-DM group, the number of patients with increased TB levels was larger (P = 0.042). The two groups did not show any statistically significant differences in related indexes such as white blood cell count, ALT, AST, and Cr.

Imaging of Liver Abscesses

All patients underwent abdominal ultrasonography. Most of the lesions (474/712, 66.6%) were found in the right lobe of the liver (67.8% in the DM group and 66.0% in the non-DM group). Notably, 52.7% of these lesions had a single abscess (53.1% in the DM group and 52.4% in the non-DM group) (Table 2). The mean diameter of the abscesses was in the range of 6.2 ± 2.8 cm. Most of the abscesses (55.3%) were in the range interval of 5–10 cm, 34.8% of the abscesses were less than 5 cm, and the remaining 9.8% were more than 10 cm. However, the difference was not statistically significant for either abscess lesion number or size between the DM and non-DM groups (P > 0.05).

Table 2 Image Characteristics of 712 Patients with Pyogenic Liver Abscess

Etiology

Pus cultures were performed in 470 patients, 445 of whom were treated with percutaneous drainage and 25 with surgery. In total, 293 bacterial strains were isolated (293/470, 62.3%), including 114 (114/167, 68.2%) in the DM group and 179 (179/303, 59.1%) in the non-DM group. Blood cultures were performed in all febrile patients, 83 of whom were positive (83/672, 12.4%). Of these 83 patients, 36 (36/225, 16.0%) were in the DM group and 47 (47/447, 10.5%) were in the non-DM group. K. pneumonia was the predominant bacterium in both pus and blood cultures, followed by E. coli. Other isolated pathogens accounted for less than 5% of all cases (Table 3).

Table 3 Microbiological Data of 712 Patients with Pyogenic Liver Abscess

Among the 209 K. pneumoniae isolates collected, antimicrobial susceptibility testing revealed that 44.0% (92/209) were pan-susceptible to all tested antibiotics. The proportion of pan-susceptible strains was comparable between patients with diabetes mellitus (DM) and those without (non-DM), with rates of 46.6% (48/103) and 41.5% (44/106), respectively (P = 0.45). The overall prevalence of multidrug-resistant (MDR) strains, defined as resistance to at least three antibiotic classes, was low at 7.6% (16/209). Notably, there was no significant difference in the distribution of MDR isolates between the DM group (7.8%, 8/103) and the non-DM group (7.5%, 8/106) (P = 0.93). Similarly, the rate of carbapenem-resistant klebsiella pneumoniae (CRKP) was 5.7% (12/210), with an identical distribution observed in both groups (DM: 5.8% [6/103] vs. non-DM: 5.7% [6/106]; P = 0.95). The remaining isolates exhibited intermediate resistance patterns (resistant to one or two classes) without meeting the MDR criteria. Detailed susceptibility data stratified by diabetes status are presented in Figure 1.

Figure 1 Sensitivity of Klebsiella pneumoniae isolates to different antibiotics in the diabetes mellitus (DM) group and Non-DM group.

Treatments and Outcomes

Intravenous anti-infective therapy was administered to all 712 patients. The most commonly used antibiotics were carbapenems (67.2%, 479/712) and third-generation cephalosporins (28.7%, 204/712). The remaining 4.1% (29/712) received other regimens, including piperacillin-tazobactam or fluoroquinolones. Comparison between the DM and non-DM groups revealed no significant disparity in the choice of antibiotics. Carbapenems were administered to 67.8% (162/239) of DM patients and 67.0% (317/473) of non-DM patients (P > 0.05). Similarly, the usage rates of third-generation cephalosporins were comparable between the two groups (26.4% vs. 29.8%, P > 0.05)).

Among the 712 patients, 445 (62.5%) received concomitant puncture therapy, 25 (3.5%) received surgical treatment, and the remaining 242 (34.0%) received antimicrobials only (Table 1). During the study period, 23 patients (3.2%) died. The DM group experienced a reduced rate of primary treatment failure (P < 0.05). In addition, fewer patients of the DM group required re-hospitalization or experienced death compared to those in the non-DM group. The two groups did not show any statistically significant differences in terms of the time needed to improve febrile symptoms, mean length of hospitalization, and hospitalization costs (Table 1).

We compared the differences between the antibiotic-only treatment group (conservative treatment group) and the puncture treatment group (non-conservative treatment group) and found that the former had 72 patients with diabetes and the latter had 164 such cases (Table 4). In the non-conservative treatment group, the majority of patients reported increased WBCs, higher CRP, and abscesses with larger diameters and also faced longer hospitalization time (P < 0.05). However, no notable differences were observed regarding fever-resolution time, prognosis, or hospitalization costs between the two groups (P > 0.05).

Table 4 Comparison of Conservative and Non-Conservative Treatment in PLA Patients with DM

The conservative treatment group had 170 patients without diabetes, while the non-conservative treatment group had 281 such cases (Supplemental Table 1). Similarly, leukocytosis, high CRP, and larger abscess diameters were more common in the non-conservative treatment group (P < 0.05). Both groups showed no notable differences in prognosis and hospital costs (P > 0.05), although the fever-resolution time and hospitalization time were longer in the non-conservative group (P < 0.05).

Supplemental Table 2 presents a comparison of non-conservative treatment options for the DM and non-DM groups. Among the 445 patients who received non-conservative treatment, 164 belonged to the DM group and 281 to the non-DM group. There was no significant difference in inflammation-related laboratory parameters such as WBC and CRP, length of hospitalization, and hospitalization costs between the two groups (P > 0.05). However, the prognosis for the DM group was notably better compared to that for the non-DM group (P < 0.05).

Supplemental Table 3 presents a comparison of the differences regarding conservative treatment modalities between the diabetic and non-diabetic groups. There were 242 patients who underwent conservative treatment, 72 of whom belonged to the DM group and 179 to the non-DM group. No statistically significant differences regarding inflammatory markers, prognosis, length of hospitalization, and hospitalization costs between the two groups were observed.

Risk Factors for Primary Treatment Failure Among Patients with and Without Diabetes

Eighteen patients (18/239, 7.6%) in the DM group experienced treatment failure (Table 5). Of these 18 patients, 5 (5/239, 2.1%) died and 13 (13/239, 5.4%) required re-treatment. A multivariate analysis of risk factors for treatment failure in the DM group is presented in Table 5. According to the multivariate logistic regression, malignant tumors (OR: 3.65, 95% CI: 1.022–13.053; P = 0.046), TB > 17 μmol/L (OR: 2.95, 95% CI: 1.021–8.514; P = 0.046), and time for temperature normalization (OR: 1.050, 95% CI: 1.015–1.087; P = 0.005) were strongly associated with unfavorable outcomes in DM patients with PLA.

Table 5 Risk Factors for Primary Treatment Failure in 239 Patients with DM by Multivariate Logistic Regression

Moreover, in the non-DM group, 59 patients (59/473, 12.5%) were identified as experiencing primary treatment failure, 18 of whom (18/473, 3.8%) died during their hospitalization and 41 (41/473, 8.7%) needed further medical intervention (Table 6). The multivariate analysis identified malignant tumors (OR: 2.74, 95% CI: 1.404–5.328, P = 0.003), previous surgery (OR: 2.15, 95% CI 1.022–4.514, P = 0.044), TB > 17 μmol/L (OR: 2.02, 95% CI 1.111–3.659, P = 0.021), and blood culture results (P < 0.001) as independent risk factors for primary treatment failure among non-DM patients with PLA. Further stratification of the blood culture results revealed that this risk was driven primarily by non-Klebsiella pathogens. Infection with other organisms markedly increased failure risk versus culture-negative cases (OR: 8.45, 95% CI: 3.49–20.46, P < 0.001), whereas K. pneumoniae showed no significant association (OR: 1.53, 95% CI: 0.42–5.63, P > 0.05).

Table 6 Risk Factors for Primary Treatment Failure in 473 Patients with Non-DM Analyzed by Multivariate Logistic Regression

Discussion

DM is one of the risk factors for PLA and a common comorbidity in PLA patients, with a 3.6–9 times greater risk rate than that for the general population.¹¹ In our study, 33.6% of patients with PLA had DM as a comorbidity. Similar results have also been reported from other regions of China.^5,13 Uncontrolled hyperglycemia may impair the immune function, aggravate metabolic disorders, and promote bacterial growth in tissues, which may lead to the formation of liver abscesses.¹⁶ In our study, after DM, cholelithiasis was identified as the second major underlying disease that may lead to biliary tract infections and ultimately predispose the patients to PLA. Hence, cholelithiasis is a major risk factor for PLA.¹⁷ PLA patients without DM have a higher prevalence of cholelithiasis, in line with the previous study results.¹⁴ Additionally, patients with DM showed a higher occurrence of fatty liver, indicating that PLA patients with DM were also more likely to have liver metabolic disorders.

Due to severe systemic infection, fever is the most common clinical symptom. Most patients have elevated leukocytes and CRP.¹⁸ However, some severe cases may even experience myelosuppression leading to leukopenia, which requires increased vigilance from physicians.¹⁹ The two groups did not show any statistically significant differences regarding liver function indicators such as ALT, AST, and ALP. The non-DM group had more patients with elevated total bilirubin, probably because this group had more patients with cholelithiasis.²⁰ In contrast, hypoproteinemia is the most common condition in the DM group, probably because protein synthesis and metabolism in these patients are impacted by insulin deficiency.²¹

Imaging is crucial for the diagnosis of PLA. Routine abdominal CT and ultrasound are recommended because they are accurate in up to 90% of patients with PLA.¹³ In this study, most of the abscesses were found in the right lobe and ranged in size from 5 to 10 cm, and more than half of them were singular. Similar results have been reported in previous studies.⁵ This may be because the right liver lobe receives most of the portal blood because of its large size.¹¹

The incidence of positive pus culture was significantly higher than blood culture in both groups, probably because the pathogen was confined to the liver only. The use of empirical antibiotics reduced the positivity of blood cultures. Similar to previous studies, the present study found Gram-negative bacteria to be the most prevalent, indicating that the empirical antibiotic treatment should focus on these bacteria.⁵ Consistent with the previous studies on the topic, K. pneumoniae was found to be the predominant pathogen in this study, notably in the DM group.^22–24 This could be because the predilection of K. pneumoniae is associated with tissue hyperglycemia.¹ It has also been suggested that the infection of K. pneumoniae may be related to vascular intimal vascular defects.¹⁴ E. coli was the second most commonly found pathogen, especially in the non-DM group, probably because a greater proportion of patients in this group have cholelithiasis and a recent surgical history. In addition, E. coli is the primary strain isolated from the bile ducts of patients with biliary tract issues.^25,26 In China, hypervirulent K. pneumoniae (hvKP), characterized by specific capsular serotypes (eg, K1, K2) and a hypermucoviscous phenotype, is strongly associated with invasive syndromes and metastatic complications (eg, endophthalmitis, meningitis), predominantly in hosts with diabetes or immunocompromise.^27,28 While K. pneumoniae liver abscesses currently show low mortality, the emergence of strains combining multidrug resistance and hypervirulence poses a shifting global threat.⁹ Our findings suggest that other pathogen-specific factors may play a more critical role in determining treatment failure in non-DM patients. For instance, Escherichia coli infections often involve the biliary tract, complicating source control, while Enterococcus virulence factors (eg, cytolysins, proteases, hyaluronidases) can evade phagocytosis and promote dissemination.^9,29 Crucially, these non-Klebsiella infections frequently occur in patients with complex comorbidities like malignancy and malnutrition.^9,17,30 This convergence of aggressive pathogen mechanisms and compromised host immunity likely underpins the poor prognosis observed in the non-K. pneumoniae group.

Our previous study demonstrated that percutaneous drainage was an effective method for the treatment of PLA, with a cure rate of 90.1%.¹¹ In the present study, 62.5% of patients underwent percutaneous drainage and 34% opted for conservative treatment only. Therefore, a comprehensive comparative analysis of the choice between conservative and non-conservative treatments was performed in the two groups. We found that in both the DM and non-DM groups, the diameter of the abscess was significantly larger in the patients treated non-conservatively than it was in those treated conservatively, which may explain why these patients required different treatment methods. The non-conservative group needed a longer duration of hospitalization and a longer time for temperature normalization, chiefly because this group had larger abscesses. The cure rate remained the same regardless of the chosen modality, confirming the safety and effectiveness of percutaneous drainage.

Among patients undergoing non-conservative treatment, the cure rate was better for the DM patients than it was for the non-DM patients. Additionally, we noticed that, compared to the DM group, the non-DM group had a significantly higher proportion of tumors and a recent surgical history. More severe underlying disease might be one of the reasons for treatment failure.^9,31 Multifactorial analysis revealed that tumors, elevated bilirubin, and time to normal temperature control were the major independent risk factors for primary treatment failure in the DM group. In contrast, tumors, elevated bilirubin, recent surgical history, and positive blood cultures were identified as independent risk factors in the non-DM group. Tumors and recent surgical history often predispose patients to a greater burden from underlying diseases, leading to a worse prognosis.³² Individuals with positive blood cultures are at a risk of infection spreading to various organs and the progression of invasive liver abscess syndrome, leading to higher mortality in PLA patients.^33,34 Poorly controlled infection usually leads to persistent fever. Therefore, the duration of fever reduction is also closely related to the prognosis of the patient. Infection with drug-resistant bacteria, size and location of the PLA, and underlying disease are all potential causes of poor infection control.³⁵

This study has certain limitations, too. The investigation is a single-center retrospective analysis. As a result, the study findings may not be generalizable. Multicenter randomized controlled trials (RCTs) are needed to confirm these findings. Glycemic control plays a critical role in the development of PLA. Although all diabetic patients received standardized glucose-lowering therapy under endocrinological guidance upon admission, the absence of baseline glycated hemoglobin and pre-admission glycemic data limited our ability to fully assess the impact of chronic hyperglycemia. Furthermore, the lack of molecular virulence profiling prevented us from correlating specific hypervirulent clones with diabetes status or glycemic levels. Future prospective studies integrating detailed metabolic profiles with genomic virulence typing are needed to address these gaps. Nevertheless, our study findings are derived from numerous cases and should be beneficial to health care professionals.

Conclusion

In conclusion, DM is a common comorbidity of PLA, with K. pneumoniae as the predominant etiologic agent, predominantly on the right side of the liver and single lesions. Percutaneous drainage is proven to be an effective and safe treatment modality. Conservative treatment is more appropriate for patients with mild symptoms or small abscesses. Patients with severe conditions or those with large abscesses are more likely to be treated through percutaneous drainage. Primary treatment failure is more prevalent in the non-DM group, potentially attributed to a higher burden of severe underlying comorbidities. In addition, tumors, elevated bilirubin, duration of temperature control, surgical history, and positive blood cultures independently contribute to primary treatment failure and hence deserve more attention from clinicians. To address the limitations of this retrospective single-center study, we plan prospective RCTs to provide definitive evidence and optimize PLA treatment strategies.