Introduction

Sepsis is one of the main causes of mortality in intensive care units (ICU) and affects 19 million patients worldwide¹. To reduce mortality and morbidity² due to sepsis, the rapid and early start of suitable antibiotic therapy within the first hour of a diagnosis of sepsis or septic shock is crucial. However, the best therapeutic approach for each patient is determined by his or her particular pathophysiology³.

The class of antibiotics most commonly used in ICUs are β-lactams (BLAs), which include penicillins, cephalosporins, and carbapenems. BLAs are particularly used to combat gram-negative infections. The Spanish National Study of the Surveillance of Nosocomial Infections in Intensive Care Medicine (ENVIN-HELICS) recently published its results on the development of nosocomial infections in 205 Spanish ICUs. This report confirmed the prevalence of gram-negative bacteria (GNB) and found that Escherichia coli (13%), Pseudomonas aeruginosa (13%), and Klebsiella pneumoniae (11%) were the most frequent GNBs in ICUs⁴.

BLAs are time-dependent antibiotics⁵ and their pharmacokinetic/pharmacodynamic (PK/PD) index for bactericidal efficiency is determined by the dosing interval percentage at which the free drug concentration (fCss) is kept above the minimum inhibitory concentration (MIC) of the microorganism or suspected microorganism causing the infection (%fT_{> MIC})⁶.

However, controversy remains on the optimal dosing interval percentage and how high the fCss should be above the MIC (i.e. fCss/MIC ratio)³. Initially, the % f T_{> MIC} was established at 50% for penicillins and 40% for carbapenemics³. However, it has been argued that the target PK/PD index should be stricter in critical patients, and therefore, the fCss should be higher than the MIC over the entire (i.e. 100%) dosing interval. Other authors have proposed that the fCss should be up to four times higher than the MIC during the entire administration interval (100% f T_{≥ 4 x MIC}), thus ensuring bactericidal activity and minimizing the emergence of antibiotic resistance^7,8.

Although intermittent infusion (IIn) is the most widely used antibiotic administration method, its use could make it more difficult to attain the target PK/PD index⁹. Intermittent infusion necessarily entails plasma concentration fluctuations, which may lead to insufficient concentrations¹⁰. This risk is higher in critically ill patients, who normally have PK alterations (i.e. patients with increased renal function may have increased volume of distribution and drug clearance) and are at an increased risk of infection by microorganisms with high MICs¹¹. Previous studies have proposed that continuous infusion (CI) may be a better alternative to IIn, because CI may ensure steady state concentrations (Css) above the MIC for longer periods, thus providing better clinical outcomes^{12 13-14}.

This proposal supports the hypothesis that CI would keep the fCss above the MIC as well as constant over the entire dosing interval, thus achieving improved clinical benefit.

The main study aim was to determine the percentage of critical patients who would reach 100% fT_{≥ 4 MIC} during CI with either piperacillin/ tazobactam (PTZ) or meropenem (MER). The secondary aims were as follows: 1) to determine the percentage of patients who would reach the less demanding PK/PD target of 100% fT_{≥ MIC}; and 2) to assess possible differences in the PK/PD index reached in relation to renal function by group.

Methods

Results

Demographic and clinical data

A total of 77 patients were recruited. Sixteen patients were excluded due to nonadherence to the protocol: 12, due to missing samples to determine Css; 2, due to missing sampling time data, and 2, due to sample extraction errors. The final sample comprised 61 patients (25 MER and 36 PTZ). Table 1 shows the baseline characteristics of the patients.

Table 1. Sociodemographic and clinical characteristics of the study population 

APACHE: Acute Physiology and Chronic Health Evaluation; CNS: central nervous system; IQR: interquartile range; Md: median; MER: meropenem; MV: mechanical ventilation; PTZ: piperacillin/tazobactam; sd: standard desviation; SOFA: Sequential Organ Failure Assessment; x: median.

PK/PD data

Although all the samples were analysed, fCss alone were used in this analysis. The median fCss was 16 mg/L (11-29) for MER and 40 mg/L (21-51) for PIP (see Figure 1).

Figure 1. Free plasma concentrations of meropenem and piperacillin 24 hours after starting continuous infusion. 

Primary objective (100% f T_{≥ 4 x MIC})

Assuming theoretical MIC values for all patients, 43% (P. aeruginosa) and 72% (enterobacteriaceae) of them would reach the target PK/PD index. Overall, microbiological isolation was achieved in 36 patients and the exact MIC was determined in 35. More than 90% of these patients reached the primary PK/PD target index (95% with PTZ and 94% with MER) (see Table 2).

Table 2. Patient distribution according to the primary and secondary PK/PD targets and the calculated glomerular filtration ranges 

MER: meropenem; PIP: piperacillin; MIC: minimum inhibitory concentration; GFR: glomerular filtration rate; fCss: plasma concentration of free drug at steady state.

^*P value: chi-square.

^¥The percentages of patients who reached the target in each GFR group were calculated using all the patients belonging to each GFR group.

^°Microorganisms are listed in table 1.

Secondary objectives

Using either theoretical or real MIC values, more than 90% of patients would reach the PK/PD target of 100% fT_{≥ MIC} (Table 2). No differences were detected between the GFR groups, although some patients with a GFR ≥ 120 mL/min did not reach the PK/PD target index. Figure 2 shows the fCss/MIC ratio by GFR group.

Figura 2. fCss/MIC ratio at 100% of the dosing interval according to the antibiotic and the theoretical MIC value used: A) meropenem; B) piperacillin assuming a MIC value of 8 mg/L; and C) piperacillin assuming a MIC value of 16 mg/L. 

Discussion

The present study confirmed the results reported by other authors^9,20,21, who found that drug under-exposure in critically ill patients could be avoided by the administration of antibiotics by CI, which is a simple alternative to classic administration. Dulhunty et al.⁹ and Abdul-Aziz et al.^20,21 studied patients receiving BLAs by IIn or CI and analysed differences in the PK/PD index between groups. Their patients and our patients had similar characteristics. Dulhunty et al. used the same doses of antibiotics as those used in the present study. They found that 82% of patients in the CI group reached Css above the MIC for P. aeruginosa (100% with MER and 75% with PIP)⁹. Abdul-Aziz et al. analysed exposure after the administration of 18 g/d PTZ, 3 g/d MER, and 6 g/d cefepime. They found that 97% of the f Css in the CI group reached the target of 100% fT_{> MIC}. It is noteworthy that the use of higher PTZ doses reached a median f Css of 4 MICs in 100% of patients, in contrast to the results obtained in our study. The differences in exposure obtained with PTZ and MER could be explained by differences between the median GFR in our study population (95 mL/min/1.73 m²) and that in the population (64 mL/min) studied by Abdul-Aziz et al.²⁰. The results of the present study (43% patients reached fCss_{> 4 x MIC}) were better than those obtained in another study by Abdul-Aziz et al.²¹, in which only 31% of the patients in the CI group reached this target.

Regarding the debate on the fCss/MIC ratio and the dosing interval percentage^3,7, some in vitro and animal model studies have suggested that Css should be maintained at four to five times above the MIC to achieve an acceptable cure rate using CI regimens. The majority of these studies were conducted using resistant GNB (MIC of 64 mg/L for ceftazidime)²². For this reason, the secondary objective addressed 100% of the dosing interval instead of the 40% to 50% described in previous studies³. It is commonly argued that the higher the dosing interval percentage with Css > MIC in critical patients, the higher the probability of survival (odds ratios, 1.02 (95% CI, 1.01-1.04) for 50% fT_{> MIC} and 1.56 (95% CI, 1.15-2.13) for 100% fT_{> MIC})⁶.

A potential limitation of this study was the use of the CKD-EPI formula to determine renal function. This method was used because it was impossible to conduct 8- to 24-hour urine collection to directly measure creatinine^13,23. However, the CKD-EPI was considered to provide a more accurate calculation and prevent underestimation in those with GFR ≥ 60 mL/min¹⁵. One study²⁴ has shown that patients with increased renal function are less likely to reach the target PK/PD index. In the present study, no statistical significance was observed in relation to PK/PD target attainment in patients with a high GFR, but this result could be explained by the small sample size. Future analyses will include more patients to obtain more robust results and to be able to compare them to those of previous studies.

In this study, two analyses were performed according to the MIC value. Given that in 50% of cases no microorganism was isolated, the main and secondary objectives were analysed assuming theoretical MICs based on the breakpoint provided by EUCAST. This aspect may be another limitation, because it overestimated the number of patients who did not reach the target PK/PD index. Nevertheless, the estimation is representative of actual clinical settings, in which the MIC is rarely available at the time of antibiotic treatment. Pharmacokinetics are affected by another relevant factor, hypoalbuminemia, which is found in 40% of critical patients²⁵. Since the percentage of free drug is responsible for the bactericidal effect, measuring total plasma concentrations can lead to the under- or overestimation of the fCss. This situation applies to BLAs due to their high protein binding (more than 70% in ceftriaxone or ertapenem)²⁶, but is not considered a limitation of MER or PIP, which have very low and moderate protein binding, respectively²⁷. Nevertheless, to increase the accuracy of the results, unbound concentrations were also analysed. Finally, it should be noted that therapeutic ranges cannot be defined, because they depend on the MIC of the microorganism, the patient’s clinical/toxicity status, and pathophysiology, all of which may compromise drug exposure (i.e., high distribution volume or hyperfiltration)^24,28. Given that there is no established range for these types of antibiotics, the concentration thresholds considered to be toxic were taken as the reference maximum values: 150 mg/L for PIP (fCss 105 mg/L)²⁹ and 64 mg/L for MER (fCss 62 mg/L)³⁰.

More clinical trials are needed to establish therapeutic targets and assess their clinical impact on critical patients.

Conventional doses of MER by CI were sufficient to reach the two PK/ PD targets to treat suspected infections by microorganisms with high MICs, such as Pseudomonas aeruginosa or enterobacteriaceae, which are common in ICUs. Nevertheless, higher doses of PTZ were required to achieve a 100% fT_{≥ 4 x MIC}. However, when microbiological isolation was achieved (i.e. the MIC was known), standard doses of MER and PTZ were sufficient to obtain both target PK/PD indices. The present study found that in patients with elevated renal function there was a slight, but non significant, trend towards subtherapeutic concentrations. Pharmacotherapeutic monitoring is highly recommended to optimize treatment in critically ill patients.