Klebsiella pneumoniae (KP) is a relatively common component of the normal microbiota, having the capacity to colonize various tissues and anatomical structures (e.g., distal urethra, upper respiratory tract, or the gastrointestinal tract). It is considered an opportunistic pathogen, being incriminated in the etiology of severe infections in hospitalized patients, especially in those immunocompromised or who present severe comorbidities (1,2).

The multidrug-resistance of KP, in particular to carbapenems due to the production of carbapenemases, has increased dramatically over the past 10 years and has become an important worldwide burden for the public health systems (3,4). First recognized almost four decades ago, extended-spectrum β-lactamases (ESBL) are enzymes produced by multiple gram-negative bacteria, being significantly involved in the resistance of these bacteria to almost all β-lactam antibiotics, except for cephamycins and carbapenems (5,6).

The ESBL mechanism of action is based upon their capacity to hydrolyze the β-lactam ring of the third-generation cephalosporins and aztreonam, but are inhibited by clavulanic acid (7).

The ESBL producers can also develop co-resistance to other classes of antibiotics, such as fluoroquinolones, cotrimoxazole or aminoglycosides, which are frequently used in the treatment of urinary tract infections (UTIs), thus inducing a severe limitation to their therapeutic approach (8,9). This aspect is due to the coexistence, on large plasmids, of the genes that encode ESBL with similar genes for resistance to other antimicrobial agents, subsequently creating a multidrug-resistant phenotype that is more and more associated with ESBL-producing Enterobacteriaceae (10). As a result, the judicious choice of empirical antibiotic therapy for UTI treatment is of paramount importance as it must be approached in accordance to the risk factors for infections due to ESBL-producing K. pneumoniae (ESBL⁺ KP), such as recurrent UTIs, diabetes mellitus (DM), previous antibiotic usage, female sex or urinary catheterization (11).

The epidemiological magnitude determined by ESBL-producing KP (ESBL⁺ KP) isolates is based on their significantly increased morbidity and mortality rates, prolonged hospitalization, and important financial burden (12). Additionally, there is a major concern regarding the increasing prevalence of ESBL⁺ KP's; a multicenter study conducted over a period of 10 years revealed a worldwide prevalence of 16%, with higher rates among patients hospitalized in intensive care units (ICU) (13). A similar increasing trend over time was noted in the case of colonization with ESBL⁺ KP, with rates in ICU patients ranging from 2.6% in the US to almost 50% in India (14). However, is it likely that the real prevalence is higher than the reported numbers, due to the paucity of reported data and difficulties in laboratory assessment as many testing kits fail to detect all ESBL producers.

In this challenging context, healthcare providers aim to find the ideal, thin balance between an effective antibiotic treatment against ESBL⁺ KP and to avoid the selection of new multidrug-resistant strains, the current paradigm being focused on preventing their spread both in the community and in healthcare facilities.

A total of 75 patients met the inclusion criteria for the study. Group 1 was comprised of 34 patients (45.3%) who had an ESBL⁺ KP UTI, while group 2 was made up of 41 patients (54.7%) with an ESBL^- KP UTI. Out of the 34 ESBL⁺ KP cases, 6 (17.6%) were OXA48-producers.

Mean age of the total participants was 66±17.72 years (70.1±18.68 for group 1 vs. 62.6±7.21 for group 2; P=0.0365). The mean age of patients infected with OXA48⁺ strains was 69.3 years. We found a symmetrical sex distribution in the entire cohort (37 males and 38 females), but with a significant difference concerning the sex distribution in the two groups: (64.7% males in group 1 vs. 36.5% in group 2, P=0.087) (Table I). Out of the 6 OXA48-producing cases, 4 were males.

The baseline patient characteristics are summarized in Table II. In addition to the demographic aspects, we also found statistically significant differences between the two groups concerning the underlying conditions, such as the recurrent UTIs, chronic kidney disease, cardiovascular pathologies and malignancies, which were more commonly associated with ESBL⁺ KP. This aspect was furtherly confirmed by an increased Charlson comorbidity index among patients with ESBL⁺ KP UTI (5.7 vs. 3.5, P=0.0005). Moreover, both the presence of an indwelling urinary catheter (15 vs. 5, P=0.0011) and previous hospitalizations in the past six months (20 vs. 5, P<0.0001) or a history of antibiotic use within the last 30 days (9 vs. 4, P=0.034) were significant predictors for the presence of ESBL⁺ KP (Table II).

Another important aspect of the study was to highlight the relationship between the incidence of ESBL⁺ KP and the use of specific antibiotics. Noteworthy, we observed that patients without ESBL-producing strains were more commonly treated with one single antibiotic (P=0.046), the use of regimens consisting of multiple antibiotic molecules (≥3 antibiotics) being observed in the lot infected with ESBL⁺ KP (32.3 vs. 17%, P=0.0683). Keeping in mind the resistance profile of ESBL⁺ KP, we identified a positive correlation between the administration of ciprofloxacin, amikacin, colistin or 3rd generation cephalosporins and the absence of ESBL producers. Contrary, the presence of ESBL⁺ KP was significantly associated with the large-scale use of carbapenems, indirectly reflecting both the susceptibility profile and the severity of infection (Table III).

In terms of evolution, the patients with ESBL⁺ KP presented a severe prognosis, characterized both by prolonged hospitalization (13.1 vs. 9.4 days, P=0.0012) and by a much higher mortality rate compared to those without ESBL-producing strains (17.6 vs. 2.4%, P=0.0166; Table IV).

The differences between ESBL and non-ESBL susceptibility profiles are illustrated in Fig. 1, revealing the quasi-total resistance of ESBL⁺ KP to cephalosporins, but also to certain other classes of antibiotics, such as fluoroquinolones (ciprofloxacin) or aminoglycosides (gentamicin).

Furthermore, our study aimed to identify the risk factors for a poor outcome among the included patients. In this regard, we performed a comparative analysis between the surviving patients and the ones who were deceased. When assessing certain data from the personal history, we observed that elderly patients >70 years present a high risk for a fatal outcome (P=0.019), a similar negative evolution being also associated with previous antibiotic therapy (P=0.008) or other hospitalizations in the past six months (P=0.021). Very importantly, even if the share of severe comorbidities (except for diabetes mellitus) was higher among the deceased patients, the differences did not reach the threshold of statistical significance.

Evaluating the classic risk factors for UTIs, we found a significant association between the mortality rate and the presence of an indwelling urinary catheter (P=0.044) or a prolonged hospitalization (P=0.033). Moreover, a severe course of the infection, translated as the initiation of an aggressive antibiotic treatment (e.g., carbapenems) was significantly correlated with an increased fatality rate (P=0.018). Finally, when analyzing the ESBL production among deceased patients, we noted that 6 out of 7 fatalities presented an ESBL⁺ KP infection (P=0.016), suggesting the important negative prognostic value of these strains in the outcome of patients with UTI (Table V).

The current situation of antimicrobial resistance has reached a crucial point since 2005. In UTIs, multidrug-resistant Enterobacteriaceae, including ESBL-producing microorganisms, can be readily encountered, emerging as a growing challenge in many infectious disease clinical setups (16). In addition to the fact that patients with ESBL-producing bacteria are generally prone to a more severe course of infection compared to those without ESBL isolates, another serious issue is represented by the paucity of antibiotics that can be used for the treatment of multidrug-resistant K. pneumoniae (KP) UTIs. The persistent and often inappropriate exposure of KP strains to various β-lactam antibiotics has created the background for dynamic mutations and enhanced synthesis of β-lactamases in these bacteria, thus inducing multidrug resistance even against the extremely useful, broad-spectrum, 3^rd generation cephalosporins (17,18).

Previous studies have reported a variable, geographic-related incidence of ESBL⁺ KP. This regional variability is explained by several factors, ranging from the specific, local pattern of resistance to some socio-economic aspects, such as healthcare access or household income, with higher rates being prevalent in developing, low-income countries from Asia or Eastern Europe (19-21). Given the scarce data in the literature concerning ESBL-producing strains in the North East region of Romania (22), our study aimed to unravel the local resistance profile in an area with certain socio-epidemiological particularities (23,24).

We found a very high prevalence of ESBL-producing isolates (45.3%) among all K. pneumoniae strains detected in urinary specimens from a university hospital serving a region with ~4 million inhabitants.

Urinary catheterization is a well-known risk factor for UTIs, this aspect being confirmed by multiple previous studies, starting from Platt et al study in 1986, when about 20% of UTIs were associated with Foley catheterization, due to contamination at the insertion of the catheter (25). Additionally, Lee et al (26) found that ESBL-producing bacteria were approximately 2.4 times more frequent in patients with Foley catheterization. In line with these findings, our study revealed that 20 patients (26.6%) had an indwelling urinary catheter, which was significantly associated with the presence of ESBL⁺ KP (15 ESBL⁺ KP vs. 5 ESBL^- KP, P=0.0011).

A past history of exposure to antibiotics and hospitalizations in the last 6 months may represent another risk factor for the emergence of ESBL-producing strains. In the literature, particularly the use of cefaclor and cefminox was incriminated for the emergence of resistant strains (26). In our study, we identified a previous antibiotic exposure in 26.4% of the patients with ESBL⁺ KP, compared to only 9.7% with an ESBL^- KP (P=0.0340). Regarding recent contact with healthcare facilities, we also found a noteworthy difference among the two groups, 20 (58.8%) for the ESBL producers vs. 5 (12.1%) for the non-ESBL producers (P<0.0001) presenting a documented hospitalization in the last six months. Similarly, Yilmaz et al (27) found that 90% of the patients with an ESBL producer had a history of previous hospitalization in the past three months vs. only 16.7% for the ESBL-negative ones.

Basically, when reviewing the literature (28-30), we noted a plethora of risk factors for infection with ESBL-producing bacteria that were also identified in our research, such as recent antibiotic use, residence in long-term care facilities, recent hospitalization, old age and male sex. In our study, elderly patients (>70 years), male sex, hospitalization in the past 6 months, previous antibiotic treatment, chronic renal failure, recurrent UTIs, urinary catheterization, longer hospital stay, underlying cardiovascular pathologies, malignancies, and anemia were considerably more prevalent among the patients with ESBL⁺ KP UTIs. The latter aspects may be explained by the mutual and dynamic relationship between serious comorbidities and the occurrence and severity of KP infections. Frail patients with impaired immunity due to a severe associated pathology are more prone to develop a KP infection, and vice versa, the KP infection may negatively influence the outcome of such a patient. For example, a recent study highlighted the significant incidence and poor prognosis of carbapenemase-producing KP infections among patients with hematological malignancies or aplastic anemia (31).

Antimicrobial susceptibility was investigated for all isolates. Our results demonstrate that, among the ESBL⁺ KP patients, susceptibility rates to ciprofloxacin, gentamicin, amikacin, imipenem, meropenem and colistin were extremely low: 0, 8.9, 58.9, 67.7, 67.7 and 67.7%, respectively. On the other hand, the ESBL^- KP strains showed lower susceptibility to amoxicillin + clavulanic acid (61%), ciprofloxacin (83%) and cefuroxime (87.8%), while high susceptibility was identified for carbapenems (100%), amikacin (97.6%), gentamicin (95.1%), cefepime (95.1%), ceftazidime (90.3%), cefotaxime (95.1%), cefixime (90.3%) and cefoxitin (92.7%), these findings being in accordance with other results that have identified a similar resistance pattern, thus raising awareness both on the adequate therapeutic approach and nosocomial or community spread of ESBL⁺ KP (32). It is noteworthy to point out that the antibiotic combination between a β-lactam and an aminoglycoside is synergistic and is widely used as a first-line therapy prior to antibiogram results, especially in patients with established risk factors for multidrug-resistant infections. Given the local susceptibility profile, in addition to a β-lactam, in our setting we prefer the use of amikacin instead of gentamicin given the higher resistance rate associated with the latter.

However, despite the similarities, there are also some susceptibility-related differences, mainly due to particular local factors. Even if some of the results from a Pakistani study (33) were similar to ours concerning the susceptibility profile of ciprofloxacin, amikacin, imipenem and meropenem they also found a significantly lower susceptibility to gentamicin for the ESBL^- KP (63.1 vs. 95.1% in our study). A more recent multicentric study from the same country furtherly confirmed this regional pattern of resistance, thus bringing attention to the geographical polymorphism of ESBL⁺ KP (34). Coinciding results were found in another study performed by Müller-Schulte et al (35) regarding the differences between the susceptibility profiles depending on the ESBL production in a sub-Saharan country; for the ESBL⁺ KP the susceptibility to ciprofloxacin, gentamicin, meropenem, cefuroxime, ceftazidime and cefotaxime was 38, 21, 100, 0, 0 and 0%, respectively, while for the ESBL^- KP the susceptibility was 88, 94, 100, 100, 100 and 100%, respectively.

Numerous studies have examined whether ESBL production has an adverse effect on clinical outcomes. Although most of the reported data revealed an association between unfavorable outcome and infections with ESBL⁺ KP (36-40), other studies have failed to demonstrate such a link (41). In our patients we identified a significant association between the presence of ESBL⁺ KP and the morality rate (P=0.016). Noteworthy, of the six patients with OXA48-producing K. pneumoniae UTI, we recorded two deaths.

Furthermore, we found that the risk factors for all-cause mortality were similar to those for ESBL⁺ KP infection and included: Age >70 years (P=0.019), hospitalizations in the past 6 months (P=0.024), previous antibiotic treatment (P=0.003), urinary catheterization (P=0.044), length of hospital stay (P=0.033), carbapenem treatment based on antibiogram result (P=0.018) and ESBL production (P=0.0166).

In terms of therapeutic approach, we could not quantify accurately the superiority of any particular antibiotic regimen in the prognosis of K. pneumoniae UTI mainly due to the heterogenicity of the regimens (often consisting of at least 3 different antibiotics), but also due to some patient-related specific aspects which led to the use of only certain antibiotics. As a general algorithm, before initiating antibiotic treatment in ESBL⁺ patients, of paramount importance, is to assess the in vitro susceptibilities, the degree of the infection's source control and, finally, the clinical condition of each patient (7).

Carbapenems exhibit the broadest spectrum of β-lactam antibiotics, presenting the highest potency against Gram-negative bacteria, but being also characterized by stability to hydrolysis by the majority of β-lactamases. The majority of available data have demonstrated that carbapenem treatment is associated with improved outcomes in patients with severe ESBL⁺ KP infections and remains the ‘gold standard’ especially in critically ill patients (7,42,43). The research spectrum of ESBL⁺ KP has focused mainly on meropenem or imipenem, although a recently published multicentric study also assessed the efficacy of ertapenem (44), finding similar cure rates (90.6% with ertapenem and 75.5% with other carbapenems in empiric and 89.8 and 82.6% in targeted treatment), without significant differences concerning mortality rates. Importantly, in our study we observed a rather low ESBL⁺ KP susceptibility rate to ertapenem (44.1%), compared to imipenem or meropenem (67.6% each), thus outlining bleak future prospects when compared to a 100% susceptibility rate to both ertapenem and imipenem reported less than two decades ago (45,46).

Our study has limitations related to the relatively small size of the research group and the retrospective nature. Given the continuously increasing trend of ESBL⁺ KP prevalence, this means that our results could underestimate the actual burden of these strains, therefore requiring careful interpretation of the data. Despite these limitations, our study provides several insights on the association between the ESBL⁺ KP and the outcome of the UTI patients with this etiology, as well as the implications regarding susceptibility profiles and local risk factors from the Northeast region of Romania. In conclusion, our data may have a significant impact for further antibiotic prescription in patients with community- and/or hospital-acquired UTIs, mainly focusing on the early identification or even prevention of certain risk factors associated with ESBL-producing KP.