Beta‐lactam antibiotic concentrations in critically ill patients with standard and adjusted dosages: A prospective observational study

Antibiotic concentration target attainment is known to be poor in critically ill patients. Dose adjustment is recommended in patients with altered clearance, obesity and those with bacterial species with intermediate susceptibility. The aim of this study was to investigate the variation of antibiotic concentration in critically ill patients with standard or adjusted dosing regimens.

Conclusions: Beta-lactam antibiotics concentration vary widely in critically ill patients.The current standard dosing regimens employed during the study were not sufficient to reach 100% ƒT > MIC in approximately a quarter of the patients.In patients where dose adjustment was performed, the group with increased dose also had low target attainment, as opposed to patients with dose reduction, who all reached target.This suggests the need for further individualization of dosing where therapeutic drug monitoring can be an alternative to further increase target attainment.

Editorial Comment
This findings in this propective observational study support the interpretation that adequate beta-lactam concentrations are difficult to achive.Therapeutic drug monitoring can be useful.

| INTRODUCTION
[3] Timely and appropriate administration of antibiotics improves outcomes, with beta-lactams being the most frequently used antibiotic class, in empirical as well as in definitive therapy. 4However, beta-lactam concentrations may vary substantially, due to variations in in body mass index (BMI) 5,6 and altered physiology in critical illness, 7 such as larger volume of distribution 8 and augmented renal clearance. 9fotaxime, piperacillin-tazobactam and meropenem are frequently used beta-lactam antibiotics in the ICU.Their bactericidal effect depends on the amount of time (T), the free (unbound) antibiotic concentration (ƒ) is above the bacteria's minimal inhibitory concentration (MIC), commonly referred to a ƒT > MIC. 10 The optimal concentration is still unclear but recent studies suggest that maintaining an antibiotic concentration above MIC during the whole dosing interval (100%ƒT > MIC), is associated with positive clinical outcome. 11An antibiotic concentration four times the MIC during the whole dosing interval (100%ƒT > 4MIC) has been suggested to have even further bactericidal effect. 12,13rrent beta-lactam dosing regimens are derived from healthy individuals whose physiology do not resemble that of critically ill patients. 14

| Study design and setting
This was a prospective, single-center, observational study performed at the ICU of Skåne University Hospital, Malmö, Sweden, which is a mixed medical and surgical ICU with 10 beds with $800 admissions per year.

| Participants
Adult patients admitted to the ICU with suspected or confirmed infection that had received at least four doses of cefotaxime, piperacillintazobactam, or meropenem were eligible for participation.Patients on prophylactic treatments and receiving palliative care were not included.The inclusion used convenience sampling depending on research staff availability.

| Antibiotic treatment
Antibiotic treatment initiation, choice of agent, dose, and potential adjustment was determined by the treating intensive care physician together with an infectious disease consultant, according to clinical routine, independent of the study.Dosing was evaluated on daily rounds and adjusted according to local protocol.Dose reduction was generally recommended with renal clearance <20 mL/min but could be individualized.Patients on continuous renal replacement therapy (CRRT) were recommended to stay on standard dose according to national guidelines.Increased dosing was applied, amongst others to microorganisms pertaining to the European Committee on Antimicrobial Susceptibility Testing (EUCAST) "I" classification. 15TDM was not used in clinical routine and the results of the antibiotic concentrations were not available for the treating physicians and were thus not part of any adjustments of antibiotic dosing.During the study, standard dosages were: 4 g q8h for piperacillin/tazobactam and 1 g q8h for meropenem and cefotaxime.Antibiotic dosage was classified as reduced if lower, and increased if higher, than these dosages.

| Blood sampling
All patients had received a minimum of four doses of the studied antibiotics before sampling, to approximate steady state.Five milliliter of blood was sampled in serum tubes by qualified staff at two time points; midway in a dosing interval (mid) and directly prior to the next antibiotic dose (trough).The blood sample was immediately put on ice before transferred to the clinical biochemistry department where the blood was left on ice to congeal for 30 min and then centrifuged at 4 C for 10 min at 2000g, aliquoted and stored frozen in À80 C until further analysis.Samples were analyzed in batches and results were not available to the clinical staff.

| Antibiotic concentration analysis
The determination of cefotaxime, piperacillin and meropenem was performed by liquid chromatography tandem mass spectrometry (LC-MS/MS) validated at the department of clinical chemistry, Skåne University Hospital, Lund, using deuterated internal standards for each analyte.All samples were protein precipitated before analysis and monitored by electrospray in positive mode.The method was developed and validated using the FDA criteria for bioanalysis.The limit of quantitation (LOQ) was 1.0 mg/L and the upper range 256 mg/L.Precision at LOQ was 7% for cefotaxime and meropenem and 11% for piperacillin.Due to the lower LOQ being 1.0 mg/L, test results under this level were reanalyzed with a lower detection level of 0.10 mg/L.

| Primary endpoint-MIC breakpoint target
Non-species related MIC breakpoints for cefotaxime, piperacillintazobactam and meropenem were obtained from the EUCAST website.The MIC breakpoint for susceptibility was ≤1 mg/L for cefotaxime, ≤8 mg/L for piperacillin-tazobactam, and ≤2 mg/L for meropenem. 19To facilitate comparison, these targets were converted to pharmacokinetic/pharmacodynamic (PK/PD) ratios by dividing the free concentration of antibiotic with the MIC breakpoint for the respective antibiotic.Thus, a free concentration of meropenem of 4 mg/L would entail a PK/PD ratio of 2 (4/2).The primary endpoint was the proportion of patients achieving a concentration above the MIC during the whole dosing interval, 100%ƒT > MIC, meaning a PK/PD ratio >1 at the trough measurement.

| Clinical routine data
Study data were collected prospectively during the inclusion period and managed using REDCap electronic data capture tools hosted at Lund University. 20,21This included demographic data on sex, age, BMI, comorbidities (Charlson comorbidity index), and site of infection.Estimated glomerular filtration rate, eGFR, was calculated using the Cockcroft-Gault equation.Patients on CRRT were considered to have an eGFR of 30 mL/min according to local settings.
Organ dysfunction and disease severity were assessed trough the sequential organ failure assessment and simplified acute physiology score 3 (SAPS 3).

| Statistical analysis
Patient data are presented as median with interquartile range or percentages for categorical variables.For comparison between those

| RESULTS
We included 102 patients (38 treated with cefotaxime, 30 treated with piperacillin and 34 treated with meropenem) between April 2020 and January 2022.The median age was 66 years, 73% were males, 40% were obese and the median SAPS 3 score was 63 points.The most common site of infection was lung (48%), followed by abdominal (22%) and unknown (16%).Positive microbiological cultures were found in 62 patients (61%), with no cases of multidrug resistant bacteria.Patient demographics are presented in Table 1 and microbiological cultures in Table S1.
For the trough concentration, the variation was largest in patients treated with cefotaxime and lowest in patients treated with meropenem (Figure 1).
T A B L E 2 Antibiotic data for pharmacokinetic/pharmacodynamic targets.Patients that did not attain target were younger with higher eGFR and lower SAPS 3.There was no difference between gender and comorbidity between the two groups.Those patients not achieving target tended to be more obese.See Table 3 for baseline characteristics between primary endpoint groups.
In patients with dose adjustments, renal clearance varied between the groups (reduced, standard, increased) which however did not reach statistical difference ( p = .08),whereas they were similar in age and BMI (Table 4).
Patients on CRRT (N = 26) were estimated to have an eGFR of 30 mL/min, according to unit specific dialysis protocol.All patients on CRRT attained 100%ƒT > MIC.Of those, 18 patients had normal dose, 5 patients had reduced dose and 3 patients had increased dose.

| Maximum concentrations
For cefotaxime, the two patients with the highest trough concen- In this single-center prospective study, we found that beta-lactam antibiotic concentrations in critically ill patients varied widely, and that suggested target attainment was not achieved in approximately a quarter of the patients.We also found that in patients where a clinical decision was made to reduce antibiotic dosing, all met the target of 100%ƒT > MIC whereas in patients who had received an increased dose, target was not attained in about a fifth of the patients.
Our study shows that many critically ill patients are still underdosed, with a quarter of patients not achieving what could now be considered the most conservative target of 100%ƒT > MIC. 11This is not the first study to show that standard dosing may be insufficient; 40% of patients in the DALI study did not achieve target levels 7 and in a recent Swedish study those numbers were 45%. 22We decided to use standard non-species specific breakpoints, as opposed to previous studies, that often have used higher breakpoints to include, what is often referred to as "worst case scenario." 23,24If higher breakpoints would have been used as targets, the proportion not reaching 100% ƒT > MIC would be even higher in the present study.
Conversely, if a more liberal target with 50%ƒT > MIC would have been used, a majority of patients would have reached sufficient concentrations (as seen in Figure 1).Even though 50%ƒT > MIC has been shown to be the minimal target for effect by beta-lactam antibiotics, it is by most experts not deemed a sufficient target for critically ill patients, due to the small margin of error, resulting in a risk of ineffective treatment.
Even though an empirical treatment could be argued, also should cover these difficult to treat bacteria, the study was performed in a low resistance setting with low prevalence of difficult to treat microorganisms, where these lower breakpoints better reflect standard clinical practice.In addition, we chose to sample patients at a steady state level, defined as after the fourth dose or more and can therefore not draw any conclusions of target attainment in the earlier phases of the treatment period.
Renal function was a major determinant of serum concentrations of beta-lactam antibiotics, as also shown in previous studies. 8Patients with a higher eGFR were more likely to not attain target concentrations.Patient who did not attain target were also younger which is likely to reflect a better underlying renal function, that is, an augmented renal clearance. 9Obesity had a non-significant association to impact target attainment.In previous studies, obesity has been more associated with time to achieve steady state antibiotic concentration but not the actual level. 6l patients in whom clinicians decided that dosing should be reduced attained target concentrations.At the same time, only 79% of patients with a higher dose reached targets.These patients were younger males with higher eGFR and BMI, suggesting that dosing may need to be further individualized in this patient category.
The literature on beta-lactam concentrations and toxicity is heterogeneous, without clearly defined thresholds associated with toxicity but upper limits has been suggested as $35 mg/L for cephalosporins, 130 mg/L for piperacillin/tazobactam and 44 mg/L for meropenem. 25,26A few patients in our study reached concentrations toward those levels but did not surpass to any larger extent.These patients were given standard or high dosage despite low eGFR and no CRRT, suggesting that increased dosage may be excessive when eGFR is low.
TDM could also be beneficial to reduce toxicity in this group.Only one CRRT patient had a beta-lactam concentration near suggested toxic levels indicating that the recommendation to remain on standard dose during CRRT seems to be safe.This is also in accordance with current literature, although large variations still exist. 27,28ere are several limitations to our study.As a single-center study, the extrapolation of results to other ICUs could be limited.
However, the three antibiotics studied are used world-wide and it is reasonable to assume that patients´physiology are similar across ICUs.As mentioned above we used non-species specific breakpoints from EUCAST, which may affect generalizability.In addition, we did not record the reason for dose adjustment, which limits the interpretability of these results.Furthermore, symptoms of betalactam toxicity were not evaluated.We cannot draw any conclusions regarding patient-centered outcomes such as mortality in the present study due to lack of power and that many patients had concomitant antibiotics during their ICU stay.estimates on each antibiotic's protein binding abilities.This most likely simplifies the complexity of protein-binding, and a discrepancy between predicted and true values is to expect.Our standard dosing, especially for piperacillin/tazobactam (4 g q8h) could be considered low.This antibiotic had the lowest target attainment and higher doses could have increased the proportion of patients reaching 100%ƒT > MIC.
We also used eGFR and not timed urine collections for estimation of renal clearance which likely is less accurate at higher creatinine clearances.
No firm evidence exists with regards to optimal beta-lactam PK/PD ratio to optimize patient centered outcomes but a target of 100%ƒT > MIC has generally been recommended over the past several years. 11,25As such, one of the major conclusions of this study is that there is still room for improvement with further individualization of antibiotic dosing.TDM has been shown to achieve improved target attainment but the clinical benefits such as improved cure and mortality are still to be proven as shown in a recent study where there was no benefit of TDM on ICU length of stay and even no difference in target attainment. 23,29 this study, meropenem displayed a much less variation in concentrations among treated patients compared to cefotaxime and piperacillin/tazobactam.The reason for this variation is not clear from the results in this study but suggests that some antibiotics will be in higher need of individualization compared to others to show a difference in outcome.All our patients were on intermittent antibiotic administration and not on continuous infusion, due to local protocol.
It has been stated before that intermittent infusion is a risk for not attaining target concentration 30 and therefore continuous infusion is superior. 31,32Other studies such as the BLING III trial will show if continuous infusion has a better target attainment and outcome than intermittent infusion.

| CONCLUSION
In conclusion, this study shows that standard dosages were not sufficient to reach the recommended antibiotic concentrations in a large portion of critically ill patients and excessively high concentrations were rare.Dose reductions according to the current clinical routine seems adequate, but further research on when and how to increase dosages is needed.Standard dosing, and not reduced, seem to be appropriate in patients with CRRT.
Dose adjustment recommendations are based on clinical factors, such as reduction in renal impairment or higher doses in obesity.However, measuring beta-lactam concentrations in the blood through therapeutic drug monitoring (TDM) is becoming more accessible and has been recommended to improve the attainment of antibiotic target concentration. 11Yet, many issues still exist regarding TDM, such as what patients should be sampled and recommendations for dose adjustments.Our primary aim was to evaluate what proportion of patients achieved concentrations 100%ƒT > MIC and 100%ƒT > 4MIC using a dosing regimen according to standard clinical routine (normal dosage = 4 g q8h for piperacillin/tazobactam, 1 g q8h for meropenem and cefotaxime), including routinely used dosing adjustments.
Demographics of all patients.vs not reaching the primary endpoint, Mann Whitney and Kruskal-Wallis test was used for continuous-and Chi-squared test for categorical variables.A p value of <.05 was considered significant.Analysis and figures were done in SPSS (IBM SPSS Statistics Version 27) or in R Statistical Software (v4.2.2, R Core Team (2023).
Note: Data are presented as median (interquartile range) unless otherwise specified.Abbreviations: BMI, Body mass index; SOFA, Sequential organ failure assessment; SAPS 3, Simplified acute physiology score; eGFR, estimated glomerular filtration rate; CRRT, Continuous renal replacement therapy; CCI, Charlson comorbidity index; SOFA registered on initiation of antibiotic treatment or, if already on treatment, at admission to ICU. reaching Differences in baseline characteristics between primary endpoint groups.Characteristics and outcomes of patients receiving low, standard, and high dose.
T A B L E 4Note: Data are presented as number (%) or median (interquartile range) unless otherwise specified.Abbreviations: BMI, body mass index; eGFR, estimated glomerular filtration rate.
PK/PD ratio by dosage and estimated GFR. PK/PD ratio = the ratio of free serum concentration/non-species related MIC breakpoints for each antibiotic (trough value).Normal dosage = 4 g q8 h for piperacillin/tazobactam, 1 g q8 h for Meropenem and Cefotaxime.Low and high dosage represent any dosage below or above normal.GFR = glomerular filtration rate in mL/min, estimated by the Cockcroft-Gault formula.
The bioanalysis could only provide total antibiotic serum concentration.Unbound antibiotic serum concentrations were thereafter predicted from