According to the National Healthcare Safety Network, Pseudomonas aeruginosa is the fifth most common cause of hospital-acquired infections,1 it is a ubiquitous Gram-negative microorganism associated with a high mortality. In fact, P. aeruginosa is included in the group ESKAPE, acronym introduced by Rice2 in 2008 to designate a group of bacteria who escape the lethal action of antibiotics: Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, P. aeruginosa and Enterobacter species. These bacteria are increasingly prevalent in our hospitals and increasingly resistant to many of the antimicrobial agents,3 which is a serious health problem that needs to be addressed urgently.

Alert to antimicrobial resistance crisis, the May 2015 World Health Assembly adopted a global action plan on this issue.4 The main goal of this plan is to ensure, for as long as possible, continuity of successful treatment and prevention of infectious diseases with effective and safe medicines that are quality-assured, used in a responsible way, and accessible to all who need them.

The therapeutic approaches against P. aeruginosa infections are particularly challenging due to its intrinsically resistance to the majority of antimicrobial agents, and its ability to become resistant in the course of the antibiotic treatment, which greatly complicates the selection of appropriate treatment, and subsequently, increases the morbidity and mortality.5 In this sense, the inadequate empirical therapy, suboptimal dosing and delays in the initiation of appropriate treatment are associated with high mortality rates and contribute to an increased length of hospital stay.6

Surveillance has been recognized as a fundamental component in the control of organisms with resistance to antimicrobial agents. Surveillance of antimicrobial susceptibility enables the assessment of the burden of disease, determination of risk factors, and identification of temporal trends in occurrence and resistance patterns of infectious diseases. Such information may be used to establish empirical antimicrobial therapy recommendations,7 although it would imply that clinicians should be familiarized with the local epidemiologic surveillance programs to choose wisely the most appropriate empirical therapy against pseudomonal infections.6

Another useful tool to guide antipseudomonal therapy is the pharmacokinetic/pharmacodynamic (PK/PD) analysis with Monte Carlo simulation, which provides a reasonable prediction of the probability of success for a treatment, incorporating the variability of the pharmacokinetic parameters and the bacterial population MIC distribution (local MIC distributions).6,8 In a recent surveillance study carried out with P. aeruginosa isolates from an intensive care unit (ICU), we concluded that both, susceptibility rates and the success probability associated to the activity of the antimicrobial agent are complementary tools, and they have to be considered together to optimize the antimicrobial dose regimen for clinical making-decisions.9

Therefore, considering the alarming increase of P. aeruginosa resistance against several antimicrobials and the consequent loss of treatment options that this fact entails, the aim of the present study was to evaluate the adequacy of the antimicrobial therapy against P. aeruginosa, in admitted patients in a tertiary Spanish Hospital, excluding those at Intensive Care Units (ICU). With this goal in mind, firstly, the changes in the susceptibility of P. aeruginosa strains to the antimicrobials in an 18 years period (2000–2017) were analyzed. Moreover, the therapy success probability over the 18 years was also evaluated by applying a PK/PD modeling approach as a microbiological surveillance tool, by using PK/PD indexes as surrogate markers of efficacy.

The percentage of susceptible strains and the probability that PK/PD indices reach the target over time, were calculated over a 18 years period. The annual rates were compared by linear regression for trends. All statistical analyses were performed with IBM® SPSS®, Statistics for Windows, Version 24 (IBM). According to Friedrich et al., 22 an appropriate degree of fit was considered with a coefficient of determination (r2) of at least 0.5 (corresponding to a correlation coefficient of ≥0.7). A p value <0.05 was considered statistically significant.

Table 2 features the antimicrobial year-by-year susceptibility to amikacin, cefepime, ceftazidime, ciprofloxacin, gentamicin, imipenem, levofloxacin, meropenem, piperacillin/tazobactam and tobramycin of P. aeruginosa isolates from the hospitalized patients. As it is shown, the last year evaluated (2017) P. aeruginosa displayed a susceptibility to amikacin, penicillins and cephalosporins equal or higher than 85%; for tobramycin and meropenem, it was 76% and 75%, respectively. Susceptibility to the other antimicrobials was under 70%.

The trend analyses of antimicrobial susceptibility of P. aeruginosa over time are summarized in Table 3. The susceptibility rates to amikacin and betalactams were stable during the evaluated period. In contrast, the susceptibility to quinolones, gentamicin and tobramycin decreased significantly over time.

Figs. 1 and 2 provide an overview of the probability of PK/PD target attainment according to MIC distributions (CFR values) for the antimicrobials studied at the selected dosing regimens. In summary, in 2017, the last year evaluated (Fig. 1), only meropenem 1g every 6h (q6h) administered as extended infusion was able to attain CFR >90% (92%). CFRs between 80–90% were attained with ceftazidime 2g q8h (88%), piperacillin/tazobactam 4.5g q6h (83%), meropenem 1g q8h (86%), and imipenem 1g q6h (82%). The CFRs for the other treatments evaluated were always lower than 80%. Fluoroquinolones and aminoglycosides (Fig. 2) achieved the lowest CFR values (<35%). Supplementary Table 1 shows the CFR values obtained annually of each antimicrobial evaluated. CFR values for amikacin have not been included because when antibiograms were carried out, amikacin concentrations tested ranged from 4mg/L to 32mg/L and, therefore, MICs lower than 4mg/L were not available and the CFR could not be adequately calculated.

Fig. 1.

Cumulative fraction of response (CFR) for standard dosage regimens of penicillins, cephalosporins and carbapenems.

(0.58MB).

Fig. 2.

Cumulative fraction of response (CFR) for standard dosage regimens of aminoglycosides and fluoroquinolones.

(0.38MB).

Table 4 presents the linear regression studies for CFR values. A statistically significant trend over time was observed for fluoroquinolones and imipenem. In the case of levofloxacin and ciprofloxacin, a significant decrease was observed in CFR values from 40–59% in 2000 to 0–29% in 2017, depending on the dose but the activity was low from the start of the evaluated period. Imipenem showed a decrease in antimicrobial activity in all dosing regimens except for 500mg q12h, for which CFR values were always under 70%; for 1g q6h antimicrobial activity decreased from 95% to 82%, maintaining moderate activity. For other regimens, CFR values decreased from high probabilities to target attainment (>90%) to low probabilities (CFR<70%).

Two microbiological surveillance tools have been used to evaluate the antimicrobial therapy against P. aeruginosa: (i) the analysis of susceptibility changes of the antimicrobials over time, and (ii) the analysis of the activity of the empirical antipseudomonal treatments by using a PK/PD modeling approach. After the evaluation over 18 years of the adequacy of the therapy, differences in the expected efficacy of the antimicrobials, in terms of susceptibilities or CFRs, were observed depending on the tool used.

P. aeruginosa is intrinsically resistant to several antimicrobial agents. Recently, Mensa et al.23 have published the estimated prevalence of this bacteria in Spanish hospitals, and overall, resistance rates are over 20–30% for most antipseudomonal antimicrobials, except for amikacin, colistin and the recently introduced ceftolozane-tazobactam with values over 5%. In our study, in 2017 only amikacin and the betalactams ceftazidime, cefepime and piperacillin/tazobactam, showed susceptibilities higher than 85% (Table 2). Over the 18 years evaluated, a decrease in the susceptibility was observed only for gentamicin, tobramycin, and fluoroquinolones (Table 3); but this reduction showed a poor relationship between both variables (r2<0.5) susceptibility and time, despite statistically significant (p<0.05).

Different national action plans on antimicrobial resistance, which have been implemented in the HUA hospital to control the emergence of resistances, have probably influenced substantially the obtained results, that is, the susceptibility rates and their evolution over the time. Bacteremia Zero24 project (2009) and “Zero-VAP” bundle25 (2011), consisting of the implantation of measures to prevent central venous catheter-related bacteremia and a bundle of ventilator-associated pneumonia (VAP) prevention measures, respectively. Although these programs have been implemented in the ICU, they can have an important role in the prevention of the intra-hospital dissemination of resistances. Programs for optimizing the use of antibiotics in hospitals (called PROA), implemented in the HUA hospital since 2015, are promoted by the Ministry of Health, Social Services and Equality of Spain as a Strategic and Action Plan to Reduce the Risk of Selection and Dissemination of Resistance to Antibiotics 2014–2018.26 The main objectives27 of these programs are (i) to improve the clinical results of patients with infections, (ii) to minimize the adverse effects associated with the use of antimicrobials (including the appearance and dissemination of the resistance), and (iii) to ensure the use of cost-effective treatments.

PK/PD analysis is outlined as a needed strategy to wisely optimize dosing regimens of the antimicrobial agents in order to conserve their therapeutic value,8 moreover, its incorporation to clinical routine would contribute to reach the main objectives of the surveillance programs.

In this study, Monte Carlo simulation, pharmacokinetic modeling and institution-specific MIC determination have been used to evaluate antimicrobial dosing regimen for the empirical treatment of P. aeruginosa. It is important to remark that the expected probability of success estimated by applying Monte Carlo simulation for the evaluated antimicrobials do not match their susceptibilities. Meropenem (susceptibility of 75%) administered as 1g q6 or 8h was able to attain high and moderate probabilities of success (92% and 86%, respectively). Ceftazidime and piperacillin/tazobactam (susceptibilities of 91 and 88%, respectively), administered at the highest dose, showed only moderate probabilities to attain the PK/PD target, and cefepime, despite its high susceptibility (85%) provided CFRs under 70% for all dosage regimens evaluated. On the contrary and surprisingly, imipenem at highest dose was able to achieve moderate probability of success (CFR 82%) with only a 67% of susceptible isolates.

In the case of cefepime there is a controversy on its susceptibility breakpoint that could explain the observed differences. According to the CLSI criteria,19 the cefepime MIC breakpoint was ≤8mg/L. However, Bhat et al.28 based on pharmacodynamic and clinical grounds, have suggested to lower the breakpoints for cefepime in countries where the cefepime dosage of 1 to 2g every 12h is the licensed therapy for serious infections, so that organisms with a cefepime MIC of 8μg/ml should be no longer regarded as susceptible to the antibiotic. More recently, Sue et al.29 also demonstrated worse outcomes related to high cefepime MICs for P. aeruginosa, despite being “susceptible” in the currently range, and proposed that the current CLSI criteria for cefepime susceptibility did not predict clinical outcomes appropriately and that the breakpoint of 8mg/L is too high. In our study, if we calculate the susceptibility rate of P. aureginosa against cefepime considering 8mg/L as resistant, the susceptibility rate decreases (for instance, in 2017 from 85% to 53%), and in this case, CFR values would match better their susceptibilities.

The concentration-dependent antimicrobials, fluoroquinolones, gentamicin and tobramycin, showed CFRs under 35%, although susceptibilities ranged from 67 to 76%. Unfortunately, it has not been possible to calculate CFR values for amikacin, because of MIC concentrations tested in the hospital only ranged from 4mg/L to 32mg/L (susceptibility breakpoint MIC ≤16mg/L). This concentration range is adequate to categorize the strains as susceptible or resistant, but it is not useful to estimate the CFR properly, whose value depends on the knowledge of the MIC values corresponding to a wide distribution.

Estimation of CFR is also useful to determine not only which antimicrobial, but also the dose regimen with the best likelihood of success. An increase of the antimicrobial doses not always implies relevant changes in CFR values, for example, increases in ciprofloxacin dose did not significantly improve the probability of target attainment (400mg q12h and 400mg q8h, CFR 3–28% respectively); on the contrary, for piperacillin/tazobactam CFR ranged between 49% (4.5g q6h8) to 83% (4.5g q6h). Moreover, considering the time-dependent activity of all betalactams, the efficacy probably would improve by administering them through extended or continuous infusion, as shown with meropenem, although stability should be considered, as in the case of imipenem, with poor stability at room temperature.

Fluoroquinolones, and also imipenem, showed a statistically significant decreasing trend in the values of CFR over the 18 years evaluated, however, only in the case of ciprofloxacin and levofloxacin the coefficient of determination r2 was higher than 0.5, indicating good correlation.

In order to evaluate properly the results obtained in this theoretical PK/PD analysis, some limitations must be considered. (i) PK information from the patients from whom P. aeruginosa was isolated was not available. Therefore, it was extracted from prospective studies carried out in patients with infections, excluding patients in ICUs and the PK/PD analysis was carried out by using the mean PK parameters and their variability; (ii) PK/PD analysis is conditioned by the place of infection since the CFR would vary according to location, preferably in the case of antibiotics with wide urinary excretion; (iii) colistin has not been evaluated because it is hardly used in the patients admitted at the HUA, with the exception of ICU patients (not included in the study). Additionally, colistin's MIC distribution has not been tested until recent years, and this antimicrobial does not have a widely accepted PK/PD index; (iv) extended infusion of beta-lactams has not been evaluated, except for meropenem; (v) in this study, only the CLSI breakpoints have been considered. EUCAST30 (European Committee on Antimicrobial Susceptibility Testing) and CLSI10 susceptibility breakpoints agree on six of the 10 antimicrobials studied, and the breakpoints of amikacin, imipenem, ciprofloxacin and levofloxacin differ only in one dilution. In 2017, amikacin presented a susceptibility value of 22 percentage points lower by using EUCAST criterion instead CLSI; the difference for the other three affected antimicrobials was less than 10 percentage points.

In brief, empirical antipseudomonal therapy would vary considerably if, in addition to susceptibility data, PK/PD analysis is considered. Based only on susceptibility, amikacin, ceftazidime, piperacillin/tazobactam but also cefepime would be the best therapeutic options. PK/PD analyses was able to identify changes in antimicrobial activity not detected by simply assessing MIC indices, meropenem provided high probabilities to achieve the PK/PD target, followed by ceftazidime, piperacillin/tazobactam and imipenem, with moderate probabilities, all of them administered at the highest doses. In conclusion, PK/PD approach has allowed to preserve the therapeutic value of antimicrobials with low susceptible values, such as carbapenems, and the selection of the most efficacy antimicrobials among those with high rates of sensible isolates. Both microbiological surveillance tools, analysis of susceptibility and PK/PD modeling, should be considering together in the clinical routine to determine the most appropriate antimicrobial drug and its dose regimen, contributing in this way to decrease the risk of treatment failure and resistances development.

Not applicable.

Nothing to declare.