Efficacy of β‐lactam/β‐lactamase inhibitors to treat extended‐spectrum beta‐lactamase‐producing Enterobacterales bacteremia secondary to urinary tract infection in kidney transplant recipients (INCREMENT‐SOT Project)

Whether active therapy with β‐lactam/β‐lactamase inhibitors (BLBLI) is as affective as carbapenems for extended‐spectrum β‐lactamase‐producing Enterobacterales (ESBL‐E) bloodstream infection (BSI) secondary to urinary tract infection (UTI) in kidney transplant recipients (KTRs) remains unclear.


| INTRODUC TI ON
Bloodstream infections (BSI) represent a common complication after solid organ transplant (SOT), with an incidence higher than that expected in the general population. 1 Urinary tract infection (UTI) is the most common source of BSI in kidney transplant recipients (KTRs), 2-4 mainly as a result of the combined impact of invasive procedures on the urinary tract and underlying immunosuppression. 2,5 The increasing prevalence of infections caused by multidrug-resistant (MDR) gram-negative bacilli, such as extended-spectrum β-lactamase (ESBL)-producing Enterobacterales (ESBL-E), is of particular concern in the SOT setting. [6][7][8][9] Approximately 10% of KTR will develop an UTI caused by ESBL-E within the first year, 10 and these patients face a three times higher risk of recurrence compared to those infected with non-MDR bacteria. 10,11 The management of infections caused by ESBL-E remains challenging, with limited antimicrobials available and scarce supporting evidence. Carbapenems have been considered as the front-line therapy both in the general population 12 and in immunocompromised patients, including KTR. 13 Observational studies conducted in the general population-such as the multinational INCREMENT cohort (ClinicalTrials.gov identifier: NCT01764490)-have shown that, for organisms showing in vitro susceptibility, β-lactam/β-lactamase inhibitors (BLBLI) may be a good alternative to carbapenems for the treatment of BSI caused by ESBL-E, particularly among non-critically ill patients with UTI. [14][15][16][17] On the contrary, other studies, including a recently published randomized trial, have reported a difference in mortality favoring carbapenems. [18][19][20] Interpretation of previous studies is further complicated by the lower reliability and reproducibility of in vitro susceptibility testing to piperacillin-tazobactam as compared to carbapenems when gradient methods, such as E-test, are used. 21 Whether these findings can be extrapolated to the SOT population remains to be assessed. The aim of the present study was to compare the impact of therapeutic regimens based on carbapenems versus BLBLI on the clinical outcome in a large multinational cohort of KTR with ESBL-E BSI secondary to UTI.

| Study population and setting
The INCREMENT-SOT project (ClinicalTrials.gov identifier NCT02852902) comprised a retrospective international cohort of SOT recipients diagnosed with clinically significant (ie, meeting criteria for systemic inflammatory response syndrome) BSI caused by ESBL-E or carbapenemase-producing Enterobacterales admitted to 40 tertiary hospitals in 16 countries from January 2004 to October 2016. For the present analysis, KTRs with monomicrobial ESBL-E BSI secondary to UTI were eligible. Patient data were collected at each site by review of microbiology reports and patients' charts until day 30 after incident blood cultures (BCs) were taken. Exclusion criteria were key missing data regarding therapeutic regimens and/or outcomes, death earlier than 24 hours after the index date (ie, that of BSI onset), and the administration of active therapy for at least 2 days prior to BC sampling. The study protocol was approved by the Spanish Agency of Medicines (code FIB-COL-2015-01) and by the Ethics Committee of the Hospital Universitario Reina Sofía (Act 243, code 2907), which waived the need to obtain written informed consent. Approval was also gained at participating centers according to local requirements.

| Study outcomes and definitions
The primary study outcome was therapeutic failure, defined as the lack of cure or clinical improvement (ie, persistence or worsening mostly piperacillin-tazobactam). Propensity score (PS)-adjusted models revealed no significant impact of the choice of active therapy (carbapenem-containing vs any other regimen, BLBLI-vs carbapenem-based monotherapy) within the first 72 hours on any of the study outcomes.

Conclusions:
Our data suggest that active therapy based on BLBLI may be as effective as carbapenem-containing regimens for ESBL-E BSI secondary to UTI in the specific population of KTR. Potential residual confounding and unpowered sample size cannot be excluded (ClinicalTrials.gov identifier: NCT02852902).

K E Y W O R D S
bloodstream infection, carbapenem-sparing regimen, extended-spectrum β-lactamaseproducing Enterobacterales, kidney transplantation, outcomes, urinary tract infection of fever, leukocytosis or other signs of infection, and/or persistently positive BC for the same microorganism), and/or death from any cause at day 7 from the onset of BSI. Therapeutic failure at day 30 was considered as secondary outcome. The main explanatory variable was the type of active therapy (according to the categories defined below) administered within the first 72 hours from BSI onset. Sensitivity analyses were also performed based on the regimen used during the first 24 hours and 7 days. The tested hypothesis (BLBLI are not associated with worse outcomes than carbapenem-containing regimens after controlling for potential confounders) was specified a priori in the study protocol. Because of the exploratory nature of the study and the expected low proportion of patients treated with BLBLI across participating institutions, no sample size estimation on the basis of the anticipated incidence of study outcomes was performed. In addition, the statistical analysis was not formally modeled on a non-inferiority assumption. Active therapy was defined as administration of at least one antimicrobial agent to which the isolate showed susceptibility in vitro, at the standard dose and frequency. 12 Specifically, standard intravenous dosing regimens for the most common antimicrobials administered were as follows: piperacillin-tazobactam, 3/0.375 g to 4/0.5 g every 6-8 hours; meropenem, 1-2 g every 8 hours; ertapenem, 1 g every 24 hours; and imipenem-cilastatin, 500/500 mg to 1/1 g every 6-8 hours. All doses were adjusted to renal function. The therapy was considered to be inactive if the isolate was non-susceptible to the agent(s) administered or the dosing was inappropriate. Monotherapy was defined as the administration of a single active drug for at least 48 hours (except for patients who died in less than 48 hours, and who were included if they received at least one complete day of therapy).

| Statistical analysis
Continuous variables were presented as the mean ± standard deviation (SD) or the median with interquartile range (IQR). Categorical data were expressed as absolute and relative frequencies. The χ 2 test or Fisher's exact test were used to compare categorical variables, as appropriate. The Student's t-test or Mann-Whitney U test were applied for continuous variables. Univariate and multivariable logistic regression models were applied to identify factors predicting therapeutic failure. For analysis of therapeutic failure at days 7 and 30 (primary and secondary outcomes), we explored the impact of the antibiotic regimen administered within the first 72 hours from the onset of BSI. Further sensitivity analyses were performed according to the regimen used during the first 24 hours (for primary and secondary outcomes) and 7 days (for the secondary outcome only). At each of these windows, therapeutic regimens were classified into one of the following mutually exclusive categories: active versus inactive therapy; combination therapy versus monotherapy; carbapenem-containing versus other active regimens; and carbapenem versus BLBLI monotherapy. Absolute risk differences with 95% confidence intervals (CIs) were determined with the allegedly more effective regimen (ie, combination therapy, carbapenem-containing regimen, and carbapenem monotherapy) as the reference.
Associations were given as odds ratios (ORs) and 95% CIs.
Multicollinearity among explanatory variables was analyzed using the variance inflation factor (VIF). The Hosmer-Lemeshow test was used to assess the goodness-of-fit of the models. Thirty-day survival curves were plotted by the Kaplan-Meier method and differences related to therapeutic regimens were compared with the log-rank test.
To partially overcome the limitation posed by the non-randomized design of the study, we calculated the propensity scores (PS) for receiving either the carbapenem-containing therapy (vs any other active regimen) or the BLBLI-based (vs carbapenem-based) monotherapy, within the first 72 hours and given the patient's baseline characteristics and the clinical features at BSI presentation. Both scores were estimated by means of backward stepwise logistic regression models including variables with P-values < 0.1 in the univariate analysis (Tables S1 and S2), and the fit of the resulting models was assessed by means of the area under the receiving operator characteristics curve (auROC). PS were entered as a covariate in multivariable models to adjust for potential confounding by factors influencing the choice of therapy.
Statistical analysis was performed with SPSS version 20.0 (IBM Corp.) and graphs were generated with Prism version 6.0 (GraphPad Software Inc).

| Risk factors for therapeutic failure
Univariate and multivariable analyses of factors predicting therapeutic failure at day 7 (primary outcome) are shown in Table 3.
At the univariate level, recipient gender, time interval from transplantation to BSI onset, use of trimethoprim-sulfamethoxazole prophylaxis, presence of urinary stenosis, hospital-acquired infection, acute rejection within the prior month, Pitt bacteremia score, and the degree of sepsis severity were associated with this outcome. Since the Pitt score and the presence of septic shock exhibited significant multicollinearity (VIF values > 1.5), only the former variable was included into the logistic regression model.

| Impact of different therapeutic regimes on study outcomes
The impact on study outcomes of different regimens was next investigated within the subgroup of participants who received active therapy. First, we compared the incidence of therapeutic failure at day 7 (primary outcome) in patients receiving combination therapy versus monotherapy during the first 72 hours from the onset of BSI, with no significant differences found between both groups (8.3% [1/12]  ties. In addition, there were no significant differences in hospital stay between any of these therapeutic regimens (Table S3).  Figure S1).
This methodological approach was also applied to compare

| Sensitivity analysis
Finally, to evaluate the consistency of these findings, we investigated the impact of therapy administered during time periods other than the 72-hour window. There were no significant differences in the incidence of 7-day and 30-day therapeutic failure among different therapeutic regimens administered within the first 24 hours from BSI ( Figures S3 and S4, Table S4). No significant differences were found in 30-day therapeutic failure according to the type of therapy used within the first 7 days either ( Figure S5, Table S4).

| D ISCUSS I ON
In the present study, we were not able to detect significant dif-  Carbapenem monotherapy (primarily meropenem) was the most frequent active therapy used, followed by BLBLI (mostly piperacillin-tazobactam). To overcome the limitation posed by the non-randomized retrospective design, PS-adjusted analyses for receiving the front-line and intuitively "more potent" therapy (carbapenem-containing or carbapenem-based regimens) versus the "alternative" less  Note: BLBLI, β-lactam/β-lactamase inhibitor; BSI, bloodstream infection. a Piperacillin-tazobactam was administered at the following doses: 4/0.5 g every 8 hours (n = 14), 2/0.25 g every 8 hours (n = 6), 2/0.5 g every 6 hours (n = 3), 4/0.5 g every 12 hours (n = 2), 3/0.375 g every 6 hours (n = 2), 4/0.5 g every 24 hours (n = 1), unknown (n = 2). b Other monotherapy regimens used within the first 24 hours included cefepime (n = 3), trimethoprim-sulfamethoxazole (n = 2), and tigecycline (n = 1). c Other combination regimens used within the first 24 hours included BLBLI plus aminoglycoside (n = 1) or quinolone (n = 1), and ceftazidime plus quinolone (n = 1).   while we considered data regarding BLBLI dose, frequency of administration, and duration of treatment in order to assess the adequacy of therapy, the low number of patients precluded any further analyses regarding the potential impact of the different treatment schemes used. High-dose and/or continuous infusion regimens have been associated with higher probability of therapeutic success. 15,44 Fifthly, no specific information on the differential impact of the therapeutic regimens analyzed on graft function was collected. Finally, In conclusion, although preliminary in nature, our results would support previous evidence from non-immunocompromised patients suggesting that BLBLI (namely piperacillin-tazobactam) may be as effective as carbapenem-containing regimens to treat ESBL-E BSI secondary to UTI in KTRs, provided the isolate is susceptible in vitro. The present findings can inform that the design of pragmatic, non-inferiority randomized clinical trials confirm the role of carbapenem-sparing approaches in the specific KTR population.

ACK N OWLED G EM ENTS
Study group. We acknowledge the work of the following members of