Low Prevalence of Colonization with Vancomycin-Resistant Enterococcus in Patients Awaiting Liver Transplantation
This study was supported in part by the Centers for Education and, Research on Therapeutics (CERTs) grant (U18-HS10399) from the Agency for Healthcare Research and Quality and the French Family Research Foundation.
* Corresponding author: Emily A. Blumberg, blumbere@ mail.med.upenn.edu
The orthotopic liver transplant (OLT) population has been particularly affected by the increase in vancomycin-resistant enterococcus (VRE) infections in recent years. Pre-transplant colonization prevalence, the role of spontaneous bacterial peritonitis (SBP) antimicrobial prophylaxis as a risk factor, and the risk of post-OLT infection in colonized patients are all unknowns. We prospectively evaluated OLT candidates at our center with the aim of answering these questions. Vancomycin-resistant enterococcus colonization status was determined by rectal culture. Data collected included illness severity, antibiotic use (including SBP prophylaxis), waiting time, previous hospitalizations, and invasive procedures. Eighty-eight patients (31 female, 57 male, median age 52 years) were enrolled. The most common diagnoses were hepatitis C (49%), primary sclerosing cholangitis (13.6%), and alcoholic liver disease. Median MELD score was 11.5 (range 7–24), and median waiting time was 551 days (range 1–2224). Vancomycin-resistant enterococcus risk factors were common in our patients: recent hospitalization in 16%, recent antibiotic exposure in 39%, and renal insufficiency in 7%. Seventeen percent were receiving SBP prophylaxis. Despite the presence of established risk factors, VRE colonization prevalence was 3.4%. Preliminary limited data showed poor correlation between screening rectal cultures and operative/peri-operative cultures. Vancomycin-resistant enterococcus colonization prevalence in an OLT candidate population with mid-level MELD scores was low, and SBP prophylaxis was not a significant risk factor.
Colonization and infection with vancomycin-resistant enterococcus (VRE) have been on the increase for more than a decade. Colonization rates in general patient populations at university hospitals range from 20 to 25% (1,2). Vancomycin-resistant enterococcus is a particularly problematic organism in the orthotopic liver transplant (OLT) population. It is one of the most common pathogens isolated after OLT, and causes significant morbidity and mortality in these patients. Invasive infection with VRE after OLT has been associated with poor outcomes (3–7). The attributable mortality of VRE infection in this population ranges from 50 to 82% (6,7)
Fecal colonization prevalence data for VRE in OLT patients is available only for transplant recipients in the postoperative period. Colonization rates in this postoperative population vary across centers, and range from 13 to 63% (4,5,7). Whether these patients are colonized before transplantation or acquire VRE postoperatively is not known. Such information would be valuable in designing risk-reduction and eradication strategies. Risk factors for VRE colonization in the general hospital population are well established, and include prior receipt of antibiotics (particularly vancomycin, third generation cephalosporins, and drugs with antianaerobic activity), duration of antibiotic therapy, prior hospitalization, duration of hospitalization, renal insufficiency, and neutropenia (8–11). To our knowledge, there are no published data regarding prevalence and risk factors specific to patients awaiting OLT.
Vancomycin-resistant enterococcus colonization in individuals and populations is associated with antibiotic use patterns. Because antibiotics are frequently prescribed to patients with end-stage liver disease (ESLD), both therapeutically and prophylactically, patients awaiting liver transplantation may be at increased risk for VRE. The use of antimicrobial prophylaxis for spontaneous bacterial peritonitis (SBP) has greatly reduced SBP incidence and is common among patients with ESLD (12,13). However, prolonged use of fluoroquinolones (FQs) in cirrhotic patients has been associated with increased risk of certain Gram-positive infections (e.g. severe staphylococcal infections) (14). Because enterococci are often resistant to FQs, it is possible that the use of these drugs for SBP prophylaxis might increase the risk of VRE gastrointestinal tract colonization.
We conducted the current study in order to determine the prevalence of VRE colonization and to identify risk factors for VRE colonization in OLT candidates. We were especially interested in determining whether pretransplant antibiotic therapy, including that used for SBP prophylaxis, is associated with VRE colonization rates.
Patients and Methods
The study was conducted at the Hospital of the University of Pennsylvania (HUP), a 725-bed, academic tertiary care center located in Philadelphia. Approximately 100 OLT procedures are performed annually at the HUP.
To calculate the prevalence of and risk factors for VRE colonization in patients awaiting OLT, we performed a cross-sectional study. All patients with ESLD followed at our institution were eligible for enrollment. Patients were enrolled between October 2001 and May 2002 at the time of their regularly scheduled outpatient liver transplant clinic appointments, or at the time of hospitalization to HUP. Enrollment of hospitalized patients was conducted only within the first 24 h of admission. Inpatients who were hospitalized for more than 24 h at the time of screening and clinic patients still under evaluation for listing were excluded.
Study investigators performed a screening questionnaire and a rectal swab culture during the enrollment visit. Additional data were obtained through review of inpatient and outpatient medical records. Data collected included the following: demographics, primary etiology of ESLD, time on waiting list, illness severity and prognosis by Model for End Stage Liver Disease (MELD) scores (15), comorbidities, antibiotic exposure, prior hospital days, invasive procedures, and any history of colonization with VRE or other resistant pathogens, such as methicillin-resistant Staphylococcus aureus (MRSA) or extended spectrum beta-lactamase (ESBL)-producing Enterobactericeae.
Screening cultures were obtained using a rectal swab culture technique. Samples were plated on selective vancomycin-containing bile-esculin agar (BEA-VA agar, Hardy Diagnostics, Santa Maria, CA) and incubated at 37 °C in a CO2 incubator. Colonies consistent with enterococcus (black, esculin-hydrolyzing colonies) were then subcultured onto blood agar plates for further identification. Confirmed enterococcus isolates were then plated onto V6 plates (a selective medium containing sodium azide and a vancomycin concentration of 6 µg/mL), and underwent susceptibility testing using MicroScan overnight Gram-positive susceptibility panels. Isolates were considered vancomycin-resistant if the vancomycin minimal inhibitory concentration (MIC) was >16 µg/µL. Vancomycin genotyping of resistant isolates was not performed.
The primary etiology of ESLD was the principle diagnosis on the original United Network for Organ Sharing (UNOS) listing form. For patients in whom ESLD was multifactorial, only the principle diagnosis was used, and the patient was counted only once. Antibiotic use was defined as receipt of ≥1 dose of antibiotics during the 3 months before enrollment. Total antibiotic days were defined as the sum of all days on which the patient received ≥1 dose of antibiotics. Prior hospitalization was defined as ≥24 h stay in a health care facility in the 3 months before enrollment. Invasive procedures included surgeries, biopsies, endoscopy, paracentesis, shunts, and tumor ablations, and were recorded if occurring in the 6 months before enrollment. Patients with creatinine >1.5 g/dL on the day of enrollment were considered to have renal insufficiency.
Given that patients awaiting OLT are likely to have many of the same risk factors for VRE colonization identified in the general hospital patient population, we hypothesized that colonization prevalence would be at least 20%. We also hypothesized that approximately 25% of patients would have received a FQ within the prior six months and that receipt of a FQ would be associated with a doubling in the likelihood of VRE colonization.
Ninety-two patients were eligible for enrollment during the study period. Enrollment of patients was discontinued when it became evident that the prevalence of VRE colonization in our study population was significantly lower than expected, and initial sample size estimates could not be realistically met within the planned study period. Two inpatients hospitalized >24 h, and two outpatients not yet officially listed, were excluded. Patient characteristics are shown in Table 1. The most common etiologies of ESLD were hepatitis C (49%), primary sclerosing cholangitis (13.6%), and alcohol abuse (12.5%). Median MELD score at enrollment was 11.5 (range 7–24). Forty-nine percent of patients had scores in the 11–18 range. Median time on the waiting list was 551 days (range 1–2224). No patient had been transplanted previously. Established and potential risk factors for VRE colonization were relatively common in our patients (Table 1). Prior antibiotic use occurred in 34 patients (39%). Of those patients who received antibiotics, 59% had >7 total days. Fourteen (16%) had at least one prior day of hospitalization, and the median hospital days in this group was 5.5 days (range 1–28). Eight patients (9%) had at least one intensive care unit (ICU) day. Six (7%) had renal insufficiency. Fifty-five (63%) had undergone invasive procedures in the previous 6 months. Fifteen patients (17%) were receiving SBP prophylaxis with levofloxacin (250–500 mg p.o. weekly) at the time of enrollment. No other SBP prophylaxis regimens were used.
Table 1. Patient characteristics
|Age (median)||52 (range 22–71)|
| Female : Male||32 : 56 (36% : 64%)|
| Caucasian||67 (76%)|
| Black||16 (18%)|
| Hispanic||2 (2%)|
| Asian||1 (1%)|
| Other||2 (2%)|
|MELD (median)||11 (range 7–24)|
| 0–10||33 (37%)|
| 11–18||43 (49%)|
| 19–24||6 (7%)|
|Time on list (median)||551 days (range 1–2224)|
| 0–24 weeks||13 (14%)|
| 25–48 weeks||16 (18%)|
| >48 weeks||61 (68%)|
|Inpatient stay in prior 3 months||14 (16%)|
|ICU stay||7 (8%)|
|Total hospital days|
| 0–7 days||9|
| 8–14 days||4|
| >14 days||1|
|Antibiotic use in prior 3 months||34 (39%)|
| 0–7 days exposure||14|
| >7 days exposure||20|
|SBP prophylaxis||15 (17%)|
|Invasive procedures||55 (63%)|
Of 88 total patients, only three (3.4%) were colonized with VRE. Thirty-two of 85 noncolonized (38%) received antibiotics other than SBP prophylaxis, and two of the three colonized (67%) received antibiotics.
Fifteen of 88 patients underwent transplantation within the first 9 months of the study, and seven of those patients had operative/perioperative cultures available. The median time between the surveillance rectal swab and the transplant was 96 days (range 0–225 days). One patient with a positive rectal swab had multiple operative cultures negative for VRE, and another patient with a negative rectal swab had a bile culture positive for VRE on postoperative day 1. The remaining five patients for whom cultures were available had negative rectal swabs and negative operative cultures.
Discussion and conclusions
Although post-transplant VRE infection is an increasingly common cause of morbidity and mortality, this study suggests that pretransplant colonization plays a minor role in this changing epidemiology. Approximately 20% of post-OLT bacterial infections at our institution are caused by VRE (Blumberg et al., unpublished data), but only 3% of the transplant candidates were colonized with VRE. Our findings raise a number of important issues, and provide direction for future research.
First, colonization with VRE appears to be very uncommon in the subset of pre-OLT outpatients with moderate illness and prognosis severity scores. This suggests that established risk factors for colonization may be applicable mainly to hospitalized OLT candidates and recipients, and that potential interventions should be aimed at these inpatient populations. It is possible that the patients captured in our study, who had largely mid-range MELD scores, may not be representative of the population of candidates in whom transplant is imminent, and that the current application of the MELD system for organ acquisition may favor a population at higher risk for VRE. However, given that 15 of 88 subjects underwent transplantation during the 9 months of the study, this group does appear to be representative of outpatient candidates who are transplanted from home. Based upon our findings, it appears that VRE colonization is not a significant issue in the outpatient awaiting transplant. Thus measures aimed at reducing VRE colonization should be directed at other subgroups of the liver transplant population: either critically ill individuals with high MELD scores or new transplant recipients.
Second, although FQ use has the potential to increase risk for Gram-positive infections (14), perhaps the net exposure from once-weekly SBP prophylaxis is not adequate to impact VRE colonization. Whether the use of a fluoroquinolone as part of a selective decontamination program may increase the risk of VRE colonization is unknown. However the absence of anaerobic coverage associated with fluoroquinolone use may decrease the likelihood of VRE colonization. The demonstrated benefits of a once weekly SBP prophylaxis with fluoroquinolone seem to outweigh the risks in this regard. While we noted a higher percentage of VRE colonized patients to have been exposed to antibiotics compared with patients not colonized, our sample size was limited and these results should be confirmed in larger studies.
Third, the discordance between rectal swab and operative cultures in two of our patients raises the question of whether rectal colonization can be equated with biliary tract colonization. If the answer is no, then preventive strategies cannot be designed around bowel decontamination. It is possible that a point prevalence study using rectal swab cultures may not accurately reflect the incidence of VRE colonization and that serial cultures may increase the yield of VRE. Although rectal swab cultures are commonly used to screen patients for VRE, one recent study questioned the sensitivity of this technique, finding an overall sensitivity of only 58% (16). This may be especially true when the agar dilution method is used for VRE detection in the setting of lower inocula of the organism in the stool. Our study design may reflect this lower sensitivity, especially given our use of the agar dilution technique. Nevertheless, the extremely low frequency of VRE colonization cannot be fully explained by technical issues alone.
In summary, we observed a low prevalence of VRE colonization among patients awaiting liver transplantation at our center, despite a significant prevalence of established risk factors for VRE colonization and infection, and despite the increasing incidence of post-OLT VRE infections. Routine exclusion of patients as OLT candidates based on preoperative rectal colonization is probably not reasonable. Continued studies aimed at the accurate identification of risk factors for VRE acquisition unique to this population, and the exact timing of VRE acquisition, are crucial in the effort to reduce the growing number of VRE infections in OLT recipients.
We gratefully acknowledge Heidi Dauter, Thomas Gay, Jean Patel, Cay Read, and Geri Libetti for their assistance in conducting this study.