Ampicillin and Amoxicillin Use and the Risk of Klebsiella pneumoniae Liver Abscess in Taiwan

Journal of Infectious Diseases, Jul 2013

Background. Klebsiella pneumoniae liver abscess (KPLA) is prevalent in East Asia. Use of amoxicillin/ampicillin may lead to overgrowth of K. pneumoniae in the intestine and predispose to KPLA. We used an animal study and nationwide population-based database to investigate the association between ampicillin/amoxicillin use and KPLA in Taiwan. Methods. In an animal study, ampicillin or sterile water was administered orogastrically in serotype K1 K. pneumoniae–colonized mice and the outcome was compared. We identified 855 cases with liver abscess and selected 3420 age- and sex-matched control subjects from the National Health Insurance Research Database. Conditional logistic regression was used to estimate the adjusted odds ratios (ORs) for the association between recent use of ampicillin/amoxicillin and KPLA. Results. Ampicillin administration predisposed K. pneumoniae–colonized mice to increased bacterial burden, liver abscess and necrosis, and lethality. The population-based study showed that the adjusted OR associating the use of ampicillin/amoxicillin within the past 30 days with KPLA was 3.5 (95% confidence interval, 2.5–5.1). No association was found with use in the past 31–90 days. Conclusions. Ampicillin/amoxicillin therapy started within the past 30 days was associated with increased risk for KPLA. We should avoid the overuse of these antibiotics to prevent undesired disease in the endemic area.

A PDF file should load here. If you do not see its contents the file may be temporarily unavailable at the journal website or you do not have a PDF plug-in installed and enabled in your browser.

Alternatively, you can download the file locally and open with any standalone PDF reader:

https://jid.oxfordjournals.org/content/208/2/211.full.pdf

Ampicillin and Amoxicillin Use and the Risk of Klebsiella pneumoniae Liver Abscess in Taiwan

Yi-Tsung Lin 1 2 3 Chia-Jen Liu 0 2 6 7 Yi-Chen Yeh 2 4 5 Tzeng-Ji Chen 2 4 9 Chang-Phone Fung () 2 3 8 0 Institute of Public Health, School of Medicine, National Yang-Ming University , Taipei 1 Institute of Clinical Medicine, National Yang-Ming University , Taipei 2 Received 21 November 2012; accepted 18 January 2013; electronically published 8 April 2013. of Medicine, Taipei Veterans General Hospital , No. 201, Sec. 2, Shih-Pai Rd, 112 Taipei , Taiwan 3 Division of Infectious Diseases, Department of Medicine, Taipei Veterans General Hospital 4 Department of Medicine, School of Medicine, National Yang- Ming University , Taipei 5 Department of Pathology and Laboratory Medicine, Taipei Veterans General Hospital 6 Department of Internal Medicine, National Yang-Ming University Hospital , Yilan 7 Division of Hematology and Oncology, Department of Medicine, Taipei Veterans General Hospital 8 Institute of Emergency and Critical Care Medicine, School of Medicine, National Yang-Ming University , Taipei , Taiwan 9 Department of Family Medicine, Taipei Veterans General Hospital Background. Klebsiella pneumoniae liver abscess (KPLA) is prevalent in East Asia. Use of amoxicillin/ampicillin may lead to overgrowth of K. pneumoniae in the intestine and predispose to KPLA. We used an animal study and nationwide population-based database to investigate the association between ampicillin/amoxicillin use and KPLA in Taiwan. Methods. In an animal study, ampicillin or sterile water was administered orogastrically in serotype K1 K. pneumoniae-colonized mice and the outcome was compared. We identified 855 cases with liver abscess and selected 3420 age- and sex-matched control subjects from the National Health Insurance Research Database. Conditional logistic regression was used to estimate the adjusted odds ratios (ORs) for the association between recent use of ampicillin/amoxicillin and KPLA. Results. Ampicillin administration predisposed K. pneumoniae-colonized mice to increased bacterial burden, liver abscess and necrosis, and lethality. The population-based study showed that the adjusted OR associating the use of ampicillin/amoxicillin within the past 30 days with KPLA was 3.5 (95% confidence interval, 2.5-5.1). No association was found with use in the past 31-90 days. Conclusions. Ampicillin/amoxicillin therapy started within the past 30 days was associated with increased risk for KPLA. We should avoid the overuse of these antibiotics to prevent undesired disease in the endemic area. - Klebsiella pneumoniae is an important cause of community and nosocomial-acquired infection worldwide [1]. K. pneumoniae liver abscess (KPLA) has been reported with increasing frequency in East Asian countries in the past 3 decades, especially in Taiwan and Korea [29]. Even though reports on KPLA have also been increasing in Western countries [10, 11], it is much more common in Asian countries. K. pneumoniae, especially the virulent serotype K1 [2, 3, 9], is the dominant cause of pyogenic liver abscesses in Taiwan and Korea, and has contributed to the endemicity of the disease in Taiwan [3, 1214]. Since 1986, many researchers in Taiwan and several other areas have noted the distinctive syndrome of KPLA, complicated by metastatic endophthalmitis or central nervous system infections [4, 9, 10, 15], which leads to a poor long-term prognosis [14]. The pathogenesis of KPLA remains unclear. One animal study has demonstrated that K. pneumoniae strains have the ability to cross the intestinal barrier and cause liver abscess [16]. One recent study conducted in Korea has demonstrated that serotype K1 K. pneumoniae isolates from intestinal carriers and liver abscess patients are closely related genotypically [17]. Furthermore, we have shown that intestinal colonization by virulent K. pneumoniae is highly associated with pyogenic liver abscess [18]. The abovementioned investigations have provided evidence that KPLA is preceded by gastrointestinal colonization. One possible contributing factor to the endemic nature of KPLA is the documented overuse of amoxicillin and ampicillin in medical practice for >20 years in Taiwan [19]. K. pneumoniae isolates obtained from patients with liver abscess in the previous studies were susceptible to all antibiotics tested, except for ampicillin [2, 7, 20], which is consistent with natural resistance of K. pneumoniae. Treatment with ampicillin and amoxicillin changes the ecology of the bowel flora and may lead to overgrowth of K. pneumoniae. However, direct evidence that amoxicillin and ampicillin have contributed to the prevalence of KPLA in Taiwan is lacking. We used an animal model and nationwide population-based case-control study to confirm the possible association between ampicillin/amoxicillin use and KPLA over a 10-year period in Taiwan. Mice C57BL/6 (B6) mice were purchased from the National Laboratory Animal Center in Taiwan. The animals were housed at 4 mice per cage and maintained on standard laboratory chow and water ad libitum in the animal center of National Yang Ming University. The animals were raised and cared for in accordance with guidelines established by the National Science Council of Taiwan. All procedures, care, and handling of the animals were reviewed and approved by the Institutional Animal Care and Use Committees of National Yang Ming University. Six- to 8-week-old and sex-matched mice were used in all experiments. Effect of Oral Ampicillin on Predisposition to KPLA K. pneumoniae serotype K1 strain (STL43) described in our previous study [21] was used in all the experiments in this study. Mice were starved of food for 16 hours before inoculation. One hundred microliters of bacterial suspension containing 104 colony forming units (CFUs) of mid-log-phase K. pneumoniae was orally inoculated into mice by using a 21gauge feeding needle to establish intestinal colonization. The selective medium recommended by Wong et al [22] was used for the isolation and enumeration of K. pneumoniae in stool samples. Forty-eight hours later, the mice were divided into two groups of 4; one inoculated orally with 100 L sterile water daily for 5 days, and the other with 100 L sterile ampicillin solution (500 mg/kg, approximately 10 mg per mouse) daily for 5 days. The mice were sacrificed at indicative times after infection, and blood and liver samples were retrieved. Serial dilutions of the liver tissue homogenates were cultured to determine bacterial counts. For histological examination, livers were fixed in 10% formaldehyde solution and processed for paraffin embedding, and 5-m-thick sections were prepared and stained with hematoxylin and eosin. The degree of liver inflammation was determined by a blinded histopathology score modified from the previous study [23]. A score of 1 indicates that the number of microabscesses on each liver section was <10 and that no necrosis region was found. A score of 2 indicates that the number of microabscesses on each liver section was more than 10 but less than 20 and that no necrosis region was found. A score of 3 indicates that the number of microabscesses on each liver section was more than 20 but less than 30 and that no necrosis region was found. A score of 4 indicates that the number of microabscesses on each liver section was more than 30 and that no necrosis region was found. A score of 5 indicates that the number of necrosis region was <5. A score of 6 indicates that the number of necrosis region was more than 5 but less than 10. A score of 7 indicates that the number of necrosis region was more than 10 but less than 15. A score of 8 indicates that the number of necrosis region was more than 15. The average score in each group was generated by the examination of liver sections from 9 mice. The serum samples of the mice were further examined for cytokines, including tumor necrosis factor (TNF), interleukin (IL)6, and IL-1, as described previously [21]. Survival of the infected mice was monitored daily for 14 days. Human Data Sources This study was based on data from the National Health Insurance (NHI) research database. The NHI program provides compulsory universal health insurance, implemented on 1 March 1995, that covers all forms of health care services in 98% of the entire population of Taiwan. Using a systematic sampling method for research purposes, the National Health Research Institute randomly sampled a representative database of 1 000 000 patients enrolled in the NHI program in the year 2000. There were no statistically significant differences in age, sex, or health care costs between the sampled group and all enrollees, as reported by the National Health Research Institute. The database includes comprehensive information on insured subjects, including demographic data, dates of clinical visits, diagnostic codes, details of prescriptions, expenditure levels, and date of enrollment and withdrawal between January 1996 and December 2009. To help ensure the precision of the claims data, the Bureau of National Health Insurance performs quarterly expert reviews on a random sample of every 50100 ambulatory and inpatient claims in each hospital and clinic. The highly trustworthy NHI Research Database (NHIRD) has been extensively used in the research of pyogenic liver abscess published in peer-reviewed journals [1214]. The International Classification of Diseases, Ninth Revision (ICD9) codes were used to define diseases during the study period. The dataset used in this study consisted of de-identified secondary data released to the public for research purposes, and this study was approved by the Institutional Review Board of Taipei Veterans General Hospital. Study Population We conducted a retrospective cohort study from 1 January 2000 to 31 December 2009. Patients who were hospitalized with a primary diagnosis of pyogenic liver abscess (ICD9, Clinical Modification code 572.0) for the first time between 1 January 2000 and 31 December 2009 were recruited, whereas patients with precedent diagnosis of pyogenic liver abscess between 1 January 1996 and 31 December 1999 were excluded. We identified 1167 patients with newly diagnosed pyogenic liver abscess during the study period. Patients under the age of 20 years (n = 9) were excluded. The etiology of cryptogenic liver abscess was nearly always K. pneumoniae [24, 25]. Therefore, we further excluded patients with a history of hepatobiliary malignancy, cholangitis, common bile duct stones, or pyogenic liver abscess as the initial manifestation of underlying hepatocellular carcinoma (n = 303) in order to limit our sample to KPLA. The final study cohort included 855 patients. The comparison cohort was extracted from the same database. We randomly selected 4 patients without a history of liver abscess for every cryptogenic liver abscess case, matched for age, sex, and enrollment date within the same observation period. The same exclusion criteria were also applied to the comparison cohort. A total of 3420 patients were chosen for the comparison group. Exposure Definition The main exposure of interest was use of ampicillin and/or amoxicillin in oral or intravenous form. The defined daily dose (DDD) recommended by the WHO is a unit for measuring a prescribed amount of drug; it is the assumed average maintenance dose per day of a drug consumed for its main indication in adults [26]. Cumulative DDD (cDDD), which indicates exposure duration, was estimated as the sum of dispensed DDD of antibiotics to compare their use to the risk of liver abscess. Patients were considered exposed to ampicillin or amoxicillin if the cDDD 3 during the past 90 days before diagnosis of KPLA. Patients who had redeemed a prescription for ampicillin or amoxicillin during the 30 days and 3190 days before diagnosis of cryptogenic liver abscess were classified as recent and past users, respectively. Data Analysis For animal study, the Student t test was used to analyze the statistical significance of differences using the Prism software package (GraphPad), and P < .05 was considered statistically significant. The survival rate was determined by KaplanMeier analysis with a log-rank test, and statistical significance was accepted at P < .05. For the human study, the crude and adjusted odds ratios (ORs) with 95% confidence intervals (CIs) of exposure for liver abscess cases compared with controls were estimated using conditional logistic regression. In the multivariate analysis, we adjusted for the underlying diseases that predisposed to liver abscess. Results Animal Study To determine whether ampicillin predisposed mice colonized with K. pneumoniae to development of liver abscess and death, we challenged mice with 104 CFU K. pneumoniae on day 0 for the establishment of intestinal colonization, and treated them with ampicillin or sterile water daily for 5 days from day 2. The amount of K. pneumoniae in stools did not differ between the groups treated with ampicillin or sterile water on day 2. As shown in Figure 1, ampicillin administration markedly predisposed K. pneumoniaecolonized mice to death, with 82% mortality in mice that received ampicillin, compared to 23% in sterile watertreated mice (P = .035). All mice in control groups, which did not receive oral inoculation with K. pneumoniae, survived for 14 days after treatment with ampicillin or water. We further performed the experiments using gentamicin, which can kill K. pneumoniae, in our model and no mortality was found for 14 days throughout the experiments. We next investigated whether ampicillin mediated lethality in K. pneumoniaecolonized mice by increasing the severity of diseases than were the controls. Among the 855 cases and 3420 controls, 69 (8.1%) and 76 (2.2%) were recent users of amoxicillin or ampicillin (within 30 days before diagnosis of KPLA). We found the significance of associating recent use of amoxicillin or ampicillin with KPLA (adjusted OR = 3.5, 95% CI, 2.5 5.1). No definite association was found with past users of amoxicillin or ampicillin (within 3190 days before diagnosis of KPLA) (OR = 1.1; 95% CI, .71.7; Table 2). A doseresponse Table 1. Characteristics of Cases With Cryptogenic Liver Abscess and Control Subjects Controls (n = 3420) No. 2280 Figure 2. The degree of inflammation in liver tissues quantitated by histological examinations in K. pneumoniaecolonized group treated with ampicillin or sterile water (*P = .027; n = 9, 5 male and 4 female mice for each group). The control group did not receive inoculation with K. pneumoniae (n = 5, 3 male and 2 female mice for each group). Abbreviation: KP, K. pneumoniae. KPLA and systemic inflammation. We sacrificed animals on day 7 (24 hours following the last dose of ampicillin) and obtained samples of liver and serum. The histological examination showed that microabscess formation and necrosis of hepatocytes appeared in the group treated with ampicillin after K. pneumoniae inoculation. However, small amounts of abscess formation were found in the group treated with sterile water after K. pneumoniae inoculation. The degree of inflammation in liver tissues quantitated by histological examination was more severe in the group treated with ampicillin (Figure 2). Compared with sterile watertreated control mice (without K. pneumoniae inoculation), no significant histological difference was found in the liver tissue of control mice treated with ampicillin alone. In the mice treated with sterile water after K. pneumoniae inoculation, only small numbers of K. pneumoniae were identified in the liver. In contrast, treatment with ampicillin significantly increased the bacterial loads of K. pneumoniae in the liver (Figure 3). No bacteria were found in the mice without K. pneumoniae inoculation treated with sterile water and ampicillin. The expression of cytokines, including TNF-, IL-6, and IL-1, in the blood was higher in the group treated with ampicillin than sterile water (data not shown). These experiments were reproducible and repeated 3 times. These results suggest that ampicillin increased the bacterial burden, liver abscess formation and injury, systemic inflammation, and lethality in K. pneumoniaecolonized mice. Human Population-Based Study The characteristics of cases and controls are presented in Table 1. Cases were generally more burdened by chronic Characteristics Male Age, mean SD History of: Malignancy Chronic RF CHF Stroke Abbreviations: CHF, congestive heart failure; COPD, chronic obstructive pulmonary disease; DM, diabetes mellitus; RF, renal failure. 3.7 .380 Exposure to Amoxicillin or Ampicillin Nonrecipient Cases, n (%) Controls, n (%) Abbreviations: cDDD, cumulative defined daily dose; CI, confidence interval; OR, odds ratio. 3.4 (2.34.9); <.001 6.2 (3.012.9); <.001 3.1 (2.04.6); <.001 6.1 (2.713.6); <.001 relationship was found in cumulative dose (OR = 3.1 and 6.1) within the past 30 days (Table 2). Table 3 shows the crude and adjusted stratum-specific ORs for various subgroups of patients. All groups showed ORs above unity, although not all had a sufficient sample size to show statistical significance. DISCUSSION We demonstrated that ampicillin that inhibited the intestinal microflora but had no activity against K. pneumoniae promoted liver abscess and injury, systemic inflammation, and mortality in mice colonized with this pathogen. We further provided evidence that exposure to ampicillin/amoxicillin within the past 30 days predisposed to KPLA in a nationwide populationbased study in Taiwan. The doseresponse relationship was evident in the current study. Our findings imply that an intestinal reservoir of virulent K. pneumoniae strains can lead to liver abscess, which is precipitated by the use of amoxicillin or ampicillin. To the best of our knowledge, this is the first study to investigate the effect of antibiotics on KPLA using both animaland population-based studies. Various factors, including ethnicity, host susceptibility to infection, difference in carriage rates, and environmental factors, might contribute to the distinct epidemiology of KPLA [17, 27]. We have shown previously that KPLA is preceded by colonization of the gastrointestinal tract by virulent K. pneumoniae [18], and the high prevalence of virulent K. pneumoniae strains in patients of Asian descent is probably why the prevalence of KPLA is so high in this population [17, 27, 28]. Whether ethnicity or geographical residence, or both, are risk Cases, n (%) Controls, n (%) Abbreviations: CI, confidence interval; CHF, congestive heart failure; COPD, chronic obstructive pulmonary disease; DM, diabetes mellitus; OR, odds ratio; RF, renal failure. a Crude and adjusted ORs with 95% CIs of exposure for liver abscess cases compared with control subjects were estimated using unconditional logistic regression, except in the subgroup stratified by sex and age. Crude OR (95 % CI) 3.0 (2.04.5) 6.2 (3.511.0) 4.5 (1.811.6) 3.6 (2.25.8) 4.1 (2.37.2) Adjusted OR (95 % CI) 2.8 (1.84.5) 5.4 (2.910.1) 5.4 (1.915.4) 3.4 (1.95.8) 3.7 (2.06.6) Exposure Malignancy Chronic RF CHF Stroke factors for gastrointestinal colonization of K. pneumoniae remains unresolved [17, 27]. Our findings suggest that, in addition to high carriage rates of virulent K. pneumoniae strains in the intestinal tract in Taiwan [27], subsequent use of amoxicillin or ampicillin selects for these virulent strains and eliminates competing flora, therefore nullifying colonization resistance, which facilitates intestinal overgrowth and formation of liver abscess. These findings may partially explain the geographical difference in the epidemiology of KPLA. The introduction of potent antibiotics to medical and agricultural practice has inadvertently disrupted intestinal microbial populations with untoward effects that are becoming increasingly apparent. Disruption of normal barriers, such as gastric acidity and the indigenous microflora of the colon, facilitates overgrowth of pathogens that reside in the intestine. Selective pressure exerted by antibiotics plays a particularly important role in pathogen colonization, and adverse effects associated with these agents often persist beyond the period of treatment [29]. Vancomycin-resistant Enterococcus and Clostridium difficile are well-known examples that treatment with broad-spectrum antibiotics leads to increased susceptibility to bacterial infections [3032]. One recent animal study also has suggested that antibiotics that disturb the intestinal anaerobic microflora and lack significant activity against carbapenemaseproducing K. pneumoniae promote colonization by this organism [33]. In the current study, we demonstrated that antibiotic use predisposed to wild-type K. pneumoniae infection, in addition to other well-known nosocomial pathogens. In contrast to the previous study that focused on innocuous but highly antibiotic-resistant bacteria, K. pneumoniae (serotype K1) used in our animal study was resistant to ampicillin only. This finding suggested that overuse of antibiotics in the community can also have a harmful outcome. In the current animal study, we demonstrated that the use of ampicillin can increase the risk of subsequent KPLA. We may simply attribute antibiotic-induced invasion of K. pneumoniae to the loss of colonization resistance to K. pneumoniae [34]. However, the detailed mechanisms by which antibiotics render the host susceptible to KPLA are not well defined. The consequences of an altered intestinal microbiota after antibiotic therapy are not fully understood, but include increased carriage of antibiotic resistance genes, increased enteric infection, and an altered inflammatory response [35]. We did not explore how the indigenous microbiota acts in concert with the host to prevent colonization and virulence of K. pneumoniae and how antibiotic administration disturbs hostmicrobiota homeostasis, leading to KPLA. The implication of this study is that the overuse of antibiotics, even in the community, can lead to KPLA. It also poses an opportunity for intervention, if established intestinal colonization can be eliminated. Strategies to enhance mucosal innate immune defenses to inhibit pathogen colonization of the intestinal lumen and/or prevent pathogen translocation across mucosal barriers may provide opportunities for vaccine development. The major limitation of our study was that our findings from the NHIRD must be viewed with caution because our exposure and outcome measures were subject to the limitations inherent in database-dependent observational studies. We presumed that all prescribed medications were actually taken by patients as prescribed, which may have overestimated the actual ingested dosage, because some degree of noncompliance is always expected. In conclusion, our animal study suggests that oral ampicillin that disrupts the intestinal microflora but lacks activity against serotype K1 K. pneumoniae may promote KPLA in K. pneumoniae colonized mice. The translation of these insights into the clinical arena demonstrated that exposure to ampicillin or amoxicillin within the previous 30 days predisposed to KPLA, and the doseresponse relationship was evident in the population-based study. We should avoid the overuse of these antibiotics in general practice, and the refinement of antibiotic stewardship is essential to prevent the undesired KPLA in the endemic area. Financial support. This work was supported by Taipei Veterans General Hospital (V101C-017) and Yen Tjing Ling Medical Foundation (CI-100-30). Potential conflicts of interest. All authors: No reported conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.


This is a preview of a remote PDF: https://jid.oxfordjournals.org/content/208/2/211.full.pdf

Yi-Tsung Lin, Chia-Jen Liu, Yi-Chen Yeh, Tzeng-Ji Chen, Chang-Phone Fung. Ampicillin and Amoxicillin Use and the Risk of Klebsiella pneumoniae Liver Abscess in Taiwan, Journal of Infectious Diseases, 2013, 211-217, DOI: 10.1093/infdis/jit157