Urinary Tract Infection Caused By Enterococcus Spp: Risk Factors And Mortality. An Observational Study.

A retrospective in-patients study was conducted with bacteriemic UTIs caused by Enterococcus spp. We compared bacteriemic UTIs caused by Enterococcus spp. vs. a random sample of 100 in-patients with bacteriemic UTIs caused by others enterobacteria.


Backgroud
The Enterococcus genus is included in the Enterococcaceae family, in which there are more than 30 recognized species, although Enterococcus faecalis, Enterococcus faecium and Enterococcus durans are the three main species that affect humans(1) E. fecalis and E. faecium are rare causes of urinary tract infections (UTIs) in young women. Nevertheless, they are frequently involved in complicated infections. Thus, recent epidemiological studies have detected that these species are the second cause of complicated UTIs, followed by Escherichia coli, reaching 7-25% of the total number of cases of UTIs (2,3).
Despite the high frequency of UTIs caused by Enterococcus spp., there is a low rate of bacteremia associated with infection (2,4). Moreover, in mixed UTIs, where other uropathogens are involved, isolation of Enterococcus spp. is considered a commensal pathogen and is usually not treated (5). On the one hand, Enterococci are tolerant or intrinsically resistant (particularly E. faecium) to several antimicrobials, such as broad spectrum penicillins, cephalosporins and carbapenems, that are typically used in complicated UTIs. Guidelines Panel, the empiric treatment of the enterococcus species has been suggested for patients with classic risk factors for this infection (6)(7)(8).
Although risk factors for overall bacteremia by Enterococcus spp. have been de ned in the literature, the risk factors for UTI have not been well de ned, and the consequences of a delay in optimal treatment have also not been well demonstrated.
Thus, the aim of this work was to describe the UTIs caused by Enterococcus spp., identifying the risk factors involved in hospitalized patients, and the overall mortality and consequences of non-optimal empiric treatment.

Patients
First a retrospective, observational study during the years 2012-2017 was carried out in two centers of Castilla y León: Complejo Asistencial de Palencia (CAUPA) and Complejo Asistencial Universitario de Salamanca (CAUSA). We reviewed the records of in-patients with a diagnosis of urinary tract infection caused by Enterococcus spp from January 2012 to december 2017 as shows in Fig. 1. We included to patients admitted to any services at these centers who needed hospitalization with clinical manifestations compatible with urine tract infection (UTI). UTI is considered if the patient presents cystic syndrome (dysuria, frequency and/or nocturia) or acute pyelonephritis (if there was pain in the renal fossa, fever > 37.7ºC, involvement of the parenchyma during imaging techniques or criteria presented for systemic in ammatory response syndrome (SIRS) ); or ancient patients (> 65 years) with confusional syndrome and fever > 37,7 without other infectious foci. Sepsis or septic shock was considered if the recommendations of the Third International Consensus De nitions for Sepsis and Septic Shock were met (9). The Sequential Organ Failure Assessment (SOFA) and Pitt Bacteremia Score (PBS) were systematically collected when samples were taken. The PBS was calculated for each patient at baseline when samples were taken (10). Hypotension, mechanical ventilation, mental status, and maximum temperature parameters of the PBS were measured for the baseline date. We used the Friedman criteria for the classi cation of UTIs as nosocomial, associated with care or community-acquired (11).
Microbiological criteria for urinary infection caused by Enterococcus were con rmed by isolation of E.
faecalis or E. faecium in urine cultures (10 5 CFU/ml). We selected as cases, those patients with concomitant urine and blood cultures, and a random sample of patients with UTIs caused by Enterococcus with blood culture done without bacteremia.
Second, we compared bacteriemic UTIs caused by Enterococcus spp. vs. a random sample of 100 inpatients with bacteriemic UTI caused by others enterobacteria. We used the same clinical and microbiological diagnostic criteria for enterobacteria as for Enterococcus spp.
The exclusion criteria for the study were as follows: i) patients who presented other concomitant infectious foci and ii) patients who ful lled the de nition of asymptomatic bacteriuria. All patients included in the study participated in a protocol that included the collection of demographics, epidemiological and clinical data. The empirical and directed antibiotic treatment and the carrying out of blood cultures and imaging studies were left to the clinician's discretion. The usual recommendations for taking samples according to the protocol were applied in the two centers. Urine specimens collected by clean catch were plated on agar blood (bioMérieux, Spain) and agar MacConkey (bioMérieux, Spain) using a 0.001-ml calibrated loop. Samples collected from indwelling catheters or transurethral catheterization were also plated on CLED agar (bioMérieux, Spain). All plates were incubated for 24 hours at 35-37 ºC. A colony count of 10 5 cfu/mL was considered positive. Identi cation and susceptibility were assessed with an automated MicroScan WalkAway system (Beckman Coulter, Spain). MIC was interpreted according to the cut-off points of the European Committee for Antimicrobial Susceptibility Testing (EUCAST). Matrix-assisted laser desorption/ionization time-of-ight mass spectrometry (MALDI-TOF MS) was used for the identi cation of species of Enterococcus when the results of biochemical methods were doubtful.
Finally, the presence of endocarditis was analyzed collecting those patients with bacteremia in which echocardiography was performed.

Data analysis
The results obtained were expressed as the mean ± Standard Deviation (SD) and percentage. The risk factors were expressed as odds ratios (ORs) with 95% con dence intervals (CIs). The X 2 test was used to evaluate the risk factors for urinary tract infections (UTIs) caused by Enterococcus. Moreover, we used the X 2 test for the analyses of mortality. A multivariate analysis was carried out, introducing only the variables with statistical signi cance (p < 0.05) in the univariate analysis. Receiver operating characteristic (ROC) curves were performed to evaluate and compare the validity of the PBS and SOFA as predictors of mortality. Binary logistic regression was carried out by introducing only the variables that were signi cant in the bivariant analysis. The odds ratio with 95% CI was assessed in all signi cant variables. Data analysis was performed using SPSS 25.0 (Statistical Package for the Social Sciences).

Results
In this study, we included one hundred and six patients with UTIs caused by Enterococcus spp. The main characteristics of the patients are shown in Table 1. Seventy-three (68,9%) patients were male, with an average age of 73.8 ± 19 years and an average Charlson comorbidity index of 5.9 ± 2.9. A total of 87.6% of the patients had a Charlson index ≥ 3. Thirty-three (31.1%) patients had community-acquired UTIs, 23 (   The overall in-patient mortality for UTIs caused by Enterococcus spp. was 17(16,5%). The predictive risk factors for mortality are shown in  Among patients with bacteremia, the SOFA score was a better predictor of mortality than the PBS (AUC-ROC 0.8 vs 0.5), as shown in Fig. 2.
We did not detect an association between bacteremia and mortality (p > 0.05). A total of 49.4% of patients with UTIs caused by Enterococcus spp. achieved correct, empiric antibiotic therapy according to the nal antibiogram, although we did not detect a lower mortality rate in this group of patients (p < 0.05).

Discussion
Enterococci species are part of the healthy human gut microbiota (12) and sometimes can also colonize in the urinary tract, causing different clinical settings from asymptomatic bacteriuria (AB) to UTIs.
Bacteriuria caused by enterococci has a deleterious effect on the urinary tract, which promotes innate immune suppression and increases the risk of infection by other uropathogens (13), suggesting that the treatment of AB can be successful in terms of the prevention of the recurrence of urinary tract infection. Nevertheless, paradoxically, it has been shown that antibiotic treatment of AB caused by Enterococcus, in patients with recurrent infection, increases the risk for recurrent urinary infection by other uropathogens (14). Regarding UTIs, Enterococcus spp. is the cause of cystitis, prostatitis and epididymitis (1). Experimental studies in mice have shown that enterococci can also affect the higher urinary tract, thus causing pyelonephritis (15), which is one of the most frequent pathogens involved in complicated UTIs. In a review, enterococci was the cause in 7-25% of patients with severe urinary sepsis and UTIs (3). Despite these data, pyelonephritis caused by Enterococcus spp. is not usually associated with bacteremia (2,4) In this sense, empiric treatment against enterococci is not included in guidelines of UTIs management or is reserved only for patients at risk (6)(7)(8). Thus, the rst aim of this work was to determine the risk factors associated with UTIs caused by enterococci. Although there are several studies that have analyzed the risk factors for bacteremia, there have been few studies that have focused on urinary infections and how these risk factors affect infection. Therefore, we reviewed the records from more than one hundred inpatient cases from different hospitals in Castilla y Leon (region of northwestern Spain) with a diagnosis of UTIs caused by Enterococcus, half of whom with positive blood cultures and the rest with negative blood cultures. In our cohort of patients, the main species that was detected was E. faecalis, whereas only fteen percent of cases were due to E. faecium. This distribution, with a predominance of E. faecalis, also have been found in both a European (15) and North-American series of complicated UTIs cases (16).
Regarding demographic characteristics, male sex has been associated with Enterococcus spp. Thus, in a cohort of 700 patients with bacteremia, which included more than 180 patients with a urinary focus, there were twice as many males as females (15). Our results also support a predominance for male sex in complicated UTI cases, with or without bacteremia. In our work, nearly half of the cases of UTI caused by Enterococcus were acquired in a hospital. In a previous series of enterococcus infections, other authors have shown similar results (15) and have supported the fact that Enterococcus spp. must be considered, similar to Pseudomonas aeruginosa, especially in cases of nosocomial UTIs. Enterococcus spp. are frequently associated with bio lms of both blood and urinary catheters. In a recent study, patients with an indwelling catheter had twice the risk for Enterococcus spp. than other patients with community-acquired UTIs (17). In our cohort, the patients frequently had bladder, nephrostomy or double J catheters. It is possible, as is the case in other works, that the presence of a nephrostomy catheter is speci cally involved in Enterococcus spp infection (18).
The population in this study had a high comorbidity Charlson index. Among the different comorbidities demonstrated by the patients, the most important were urinary disorders or severe systemic diseases. Among urological diseases, urological cancer has been described as a risk factor for Enterococcus UTIs (19,20). It is possible, as is the case in bacteremia associated with colorectal cancer (20), that lesions due to urinary epithelial colonization can be one of the most important predictive risk for urinary infections caused by Enterococcus.
Enterococcus species are considered by many authors an opportunistic pathogen (21), and factors such as immunosuppression, neutropenia or hemopathy are frequently detected as risk factors for bacteremia (16). The results of our work suggest that in this epidemiological setting, Enterococcus can be considered one of the most probable causes of UTI.
Finally, another risk factor detected in our work was previous antibiotic treatment. This nding is in accordance with other works in which patients treated with ceftriaxone were associated with an increased risk for reinfection by Enterococcus spp (22). The fact that more than one-half of our patients had been treated in the previous month with 3rd -generation cephalosporins may explain these results.
Another aim of this study was to evaluate the variables associated with bacteremia. Although there are several studies that have analyzed bacteremia caused by enterococcus and its risk factors, they included different types of infections, and the results were not focused on UTIs (16,20). In our study, we selected only patients with a clinical setting of UTIs and an isolation of enterococcus in urine culture, and we compared the risk factors in the two groups with and without bacteremia. Our results are consistent with other ndings, and we showed that neutropenia, solid organ transplant, bone medullary transplant, and immunosuppression were also risk factors for bacteremia in patients with tract urinary infection caused by Enterococcus spp. (16,20). These are the same classic risk factors in patients with bacteremia, regardless of other origins (4). Moreover, we detected that urinary cancer was associated with bacteremia in patients with UTIs. In a recent study, urological cancer was also associated with bacteremia (20). These results support that regardless of E. faecalis and E. faecium colonizing urinary tract infections, it is di cult for these pathogens to cross the urinary epithelium when innate immunity is preserved, but when it exists, rupture of the epithelial barrier can result in bacteremia.
Another objective of this work was to study the mortality and morbidity following UTIs caused by Enterococcus. The mortality rates seen with this infection in our study were higher than those in other series of in-patients with UTIs (23,24) and were similar to those shown by Billinton et al. (16) in a subgroup of patients with urosepsis caused by Enterococcus. In our study, mortality was not different between patients with or without bacteremia. Factors that depend on the clinical setting, such as a high SOFA score and severe comorbidity (such as malignant hemopathy and immunosuppression), and factors that depend on the species involved (E. faecium vs E. faecalis or its pattern of resistance to ampicillin and vancomycin) were the only factors associated with high mortality. Similar data have been reported by other authors regarding bacteremia (15). We did not detect a lower mortality rate for patients with correct empiric treatment for Enterococcus spp. This result could be due to the small number of patients included in the study and to the heterogenicity in the empirical treatments performed. However, because our study was retrospective, it is also possible that there was a selection bias, so patients with optimal empirical treatment according to the last antibiogram could also be the most critical.
Finally, we evaluated infectious endocarditis, a frequent morbidity detected in patients with bacteremia caused by Enterococcus spp. In a Danish cohort, 6% of patients with nosocomial infections and 25% of patients with community-acquired infections had infectious endocarditis. In our study, we found endocarditis in 5.8% of patients with UTIs and bacteremia with a urinary origin (2.8% of the total patients included in the study). These data support that in patients with bacteremia and a urinary focus, echocardiography must be performed to diagnose this severe complication (15) We believe that our work can help to better de ne a group of patients with urinary infection that have a high mortality, thus requiring a differentiated antimicrobial treatment from the one included in the current guidelines. Nevertheless, we also recognize that our work has some limitations. It was a retrospective study of in-patients, with a relatively low number of patients and with selection bias, so risk factors or mortality may not be similar in other populations of patients.

Conclusion
Enterococcus is one of the most frequent causes of UTIs complicated by a pro le of risk factors, in which male sex, a high Charlson index, urinary catheter, previous antibiotic treatment, urological cancer or several types of immunosuppression are the main predictive factors. These last two factors are also risk factors for bacteremia. High mortality secondary to a severe clinical setting and high comorbidity may be su cient reasons for implementing empiric treatment of patients at risk, although we did not show a higher survival rate in patients with this treatment strategy. Figure 1 Participant pro le. Area under the ROC curve (AUC) of the SOFA score and PBS as predictors of mortality.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.