Our study confirms previous reports that the incidence of sepsis in the population of KTRs is substantially higher than in the general population [18]. Furthermore, UTI was the dominant type of infection among KTRs admitted to the hospital and accounted for 67% of all infection cases in our study group. Although the mortality related to urosepsis among KTRs was relatively low and did not exceed 10%, it was still twice as high as in the general population [19]. Furthermore, we found that US poses a common insult to kidney graft and contributes significantly to its failure and subsequent loss. A microbiological map of KTRs admitted to our center due to US and UTI revealed that the bacterial pathogens responsible for infection resemble a highly resistant hospital microbiome – which is in line with observations from the general patient population [10].

The collected data also enabled us to characterize local antibiotic practices. First, the median time of antibiotic therapy in patients with US (median of 13 days) was substantially longer than the mean antibiotic course that is recommended by EAU (7–10 days) and SSC guidelines in the general population (5–10 days with recommendation towards shorter courses) [7, 8]. Conversely, the guidelines of the American Society of Transplantation recommend even longer 14–21-day courses of targeted narrow spectrum antibiotic therapy for complicated pyelonephritis or urosepsis [9].

We believe that such long courses seem excessive, especially in the light of emerging evidence that in the case of adequate source control, antibiotic treatment can be safely discontinued at day 4 to 5 [20, 21]. Although in our study, an appropriate choice of antimicrobial agent for the treatment of US (as per EAU guidelines) was observed in the majority of study cases, the frequency of inappropriate empiric treatment was 46% [8]. This partially explains the longer median times of antibiotic treatment, which even with this prolongation, can be considered as “longer antibiotic courses” as per SSC nomenclature [7].

The main contributory organisms identified in our cohort were similar to those found in the general population, with Escherichia coli being the most common microorganism, followed by Klebsiella pneumoniae and other Gram-negative pathogens, irrespectively of septic status. In addition, the frequency of MDR was not associated with infection phenotype (US vs. UTI), but rather with the microorganism species, with the most common being among Klebsiella pneumoniae isolates.

Kaplan Meier curve for readmission due to UTI in both study groups during one year follow-up

Our study also indicates that in settings where ciprofloxacin and TMP/SMX were used very frequently (for UTI and Pneumocystis jiroveci prophylaxis in the early post-transplant period), very high resistance rates were observed which reached 85–90% of all cultures (Table 2). Therefore, in the settings described these two antibiotics along with amoxicillin should not be used for the treatment of US or complicated pyelonephritis, as indicated by EAU and AST [8, 9]. The recent update of EAU guidelines recommends against the use of ciprofloxacin in any patient with pyelonephritis who is being treated in as an inpatient, has a history of previous ciprofloxacin use within the previous 6 months and when anticipated resistance rates of the causative pathogens exceed 10% [22]. On the other hand, IDSA guidelines regarding uncomplicated pyelonephritis refrain from defining the threshold of local ciprofloxacin resistance, which should trigger the use of higher tier antibiotics [23]. It therefore seems that ciprofloxacin and TMP/SMX should not be used for the empiric treatment of any form of UTI in kidney transplant patients. However, in cases with proven microbial susceptibility, they could be used as targeted therapy.

Our results also clearly highlight the association between a delay in appropriate antibiotic therapy and detrimental patient and kidney graft outcomes. The need for antibiotic escalation due to empiric agent failure was associated with an increased risk of death, the need for renal replacement therapy and non-recovery from AKI at one month. Recent insights from the SERPENS study, as well as other smaller reports, confirm the association between inappropriate empiric antibiotic therapy and mortality in US in the general population, despite a generally favorable prognosis with low mortality risk compared to other sources of infection [19, 24]. To date, we are not aware of any reports that analyze the outcomes of US in KTRs, in terms of antibiotic practices.

The presented association between graft function and inappropriate antibiotic selection seems to be in line with common knowledge. Therefore, in cases of evident US or severe UTI, physicians should choose antibiotics with proven activity against local microflora (in our population, third generation cephalosporins, piperacillin with tazobactam or carbapenems). In patients presenting with shock or rapidly deteriorating, the empiric use of carbapenems seems to be justified as it presented almost 100% activity in the harvested Gram-negative cultures. It is also worth mentioning that in only one patient in the US group, did the microbiologic cultures justify the use of vancomycin (with methicillin-resistant Staphylococcus aureus). The microbiologic map showed that the absolute majority of microorganisms (both Gram-positive and negative) were susceptible to some form of beta-lactams and other non-glycopeptide antibiotics. We therefore suggest that dual empiric vancomycin and beta-lactam therapy should only be used in KTRs presenting with US and septic shock. Excessive and unjustified use of vancomycin may promote the development of antimicrobial resistance, increase the rate of Clostridioides difficile infections and lead to additional kidney graft injury [25]. A growing body of evidence indicates that the concomitant use of vancomycin and piperacillin with tazobactam may further increase the risk of acute kidney injury [26, 27].

Finally, we also analyzed the frequency of prophylactic antibiotic therapy, which was prescribed to KTRs after hospital discharge, after a US or UTI episode. Approximately 60% of our study population received a prophylactic antibiotic course, both in subjects within and after the first 6 months post-transplantation. Its use did not reduce UTI recurrence, either due to high resistance rates against the antibiotics prescribed or due to the futility of such an intervention. In a multivariable analysis, the only modifiable risk factor contributing to the recurrence of UTI was the presence of urine outflow obstruction (both at the lower- and upper urinary tract level). In selected cases, the obstruction was treated by bladder catheterization or ureteral stent implantation. It is worth mentioning that our model presented in Table S6 reached a low R2 of 0.4851, which indicates that there is still a large portion of variance that needs to be explained by other measurable or non-measurable factors. A meta-analysis conducted by Green et al., showed a neutral effect of antibiotic UTI-prophylaxis in the first 6 months after kidney transplantation on mortality and graft loss [12]. The authors also reported that antibiotic prophylaxis reduced the incidence of urosepsis and bacteriuria but significantly increased the proportion of UTIs with pathogens that were identified as TMP/SMX resistant. We hypothesize that in our study, the lack of prophylaxis efficiency was associated with widespread resistance against the most commonly used antibiotics. Therefore, as AST guidelines indicate, antibiotic prophylaxis of UTI should only be introduced in highly selected cases and should be based on actual urine culture with antibiogram [9].

Based on the experience from our center as well as analyses of the literature, we propose that in every kidney transplantation ward, a local microbiological resistance map should be created and regularly updated to guide rational antibiotic therapy, reduce the development of antimicrobial resistance, and minimize unnecessary exposure to potentially nephrotoxic drugs. The use of rapid multiplex polymerase chain reaction-based (PCR) diagnostic testing may also improve antibiotic use and reduce the time to commencement of appropriate antimicrobial therapy [28, 29]. In the absence of guidelines for the treatment of US and sepsis in KTRs, there is a significant need for randomized clinical trials in order to develop strategies to preserve graft function, especially when there is a donor shortage.

The main limitation of our study is its retrospective nature and relatively small sample size of patients recruited. Although the low case volume of US did not enable us to create a microbiological map with an optimal single isolate sample size of n = 100 (as indicated by CLSI), a minimal sample size of 30 isolates for Escherichia coli and Klebsiella pneumoniae was achieved [16, 17]. Another drawback is the lack of reliable data on antibiotic dosage. We were therefore unable to comment on the efficiency of maximal dosing or GFR-adjusted dosing strategies or continuous beta-lactam infusion regimens in the study setting. Despite these limitations, our study highlights the importance of creating microbiological maps and antibiotic stewardship, even in such small populations as kidney transplant recipients.