The role of infection models and PK/PD modelling for optimising care of critically ill patients with severe infections

Abstract

Critically ill patients with severe infections are at high risk of suboptimal antimicrobial dosing. The pharmacokinetics (PK) and pharmacodynamics (PD) of antimicrobials in these patients differ significantly from the patient groups from whose data the conventional dosing regimens were developed. Use of such regimens often results in inadequate antimicrobial concentrations at the site of infection and is associated with poor patient outcomes. In this article, we describe the potential of in vitro and in vivo infection models, clinical pharmacokinetic data and pharmacokinetic/pharmacodynamic models to guide the design of more effective antimicrobial dosing regimens. Individualised dosing, based on population PK models and patient factors (e.g. renal function and weight) known to influence antimicrobial PK, increases the probability of achieving therapeutic drug exposures while at the same time avoiding toxic concentrations. When therapeutic drug monitoring (TDM) is applied, early dose adaptation to the needs of the individual patient is possible. TDM is likely to be of particular importance for infected critically ill patients, where profound PK changes are present and prompt appropriate antibiotic therapy is crucial. In the light of the continued high mortality rates in critically ill patients with severe infections, a paradigm shift to refined dosing strategies for antimicrobials is warranted to enhance the probability of achieving drug concentrations that increase the likelihood of clinical success.

Introduction

Treatment outcomes for severe infections remain poor, with critically ill patients having high mortality rates [1, 2]. The treatment of infections in critically ill patients is very challenging, as infection paradigms are mostly based on infection models and clinical data that do not specifically account for the antimicrobial pharmacokinetics (PK) and severity of illness of these patients. PK changes to antimicrobial volume of distribution, clearance, protein binding and tissue penetration can be significantly different to that observed in other patient groups [3] often resulting in insufficient antimicrobial concentrations in plasma and at the site of infection. Generally accepted PK/pharmacodynamic (PK/PD) targets may not apply for these patients [4]. Further, infections in intensive care units (ICUs) are commonly caused by multidrug-resistant pathogens which are associated with even worse outcomes [5].

The current environment of ICUs requires existing antimicrobial agents to be used more judiciously and optimally to respond to the challenges of infection in critical illness and increasing antimicrobial resistance. In view of the heterogeneity of critically ill patients, a move towards individualised dosing is required. Obtaining the data to support the required novel dosing approaches can be gained from models that quantify antimicrobial exposure–effect relationships relating to both microbial killing and suppression of resistance [6]. In this article, we review the potential roles of various infection models, clinical PK and PK/PD modelling for the development of optimised dosing regimens for use in critically ill patients.

Preclinical data

Laboratory-based in vitro and animal data are used to define antimicrobial PD, which supports the initial design of dosing regimens.

Describing pathogen susceptibility to antimicrobials

The PD characteristics of an antimicrobial relate drug concentrations to its ability to kill the pathogen and suppress the emergence of resistance. The most widely used PD measurement for describing the potency of an antimicrobial is the minimum inhibitory concentration (MIC), i.e. the lowest antimicrobial concentration required to inhibit visible growth of the organism in vitro. In practice, determination of the MIC enables an organism to be deemed ‘susceptible’, ‘intermediate’ or ‘resistant’ to an antimicrobial according to clinical breakpoints [79] taking into account PK properties of the antimicrobial and PD targets. These data define suitable antimicrobials that can be used in a clinical scenario, but do not define the optimal dose for individual patients. Other concepts may be of relevance for suppression of resistance, e.g. the mutant prevention concentration (MPC) [10], but are not used routinely.

Antimicrobial pharmacodynamics

The shape of the antimicrobial concentration–time curve and microbial susceptibility, measured as MIC, are typically combined into an index variable related to antimicrobial response. The time that the non-protein-bound or free (f) concentration of the drug is above the MIC (% fT > MIC) is suggested to be the best index for some antimicrobials (e.g. beta-lactams). For other antimicrobials (e.g. aminoglycosides and quinolones) the peak concentration achieved (C max/MIC) or the area under the concentration–time curve relative to the MIC (fAUC/MIC) may be more important in predicting therapeutic success [1114].

Knowledge of these PK/PD indices enables the design of dosing strategies that aim to achieve PD targets associated with adequate antimicrobial effects for any specific drug/organism combination. These indices may also be used to inform dose modulations that may be required in different clinical scenarios. For instance, knowledge that C max/MIC is of high importance for an antimicrobial means that when a patient manifests reduced drug clearance the frequency of administration should be reduced (e.g. 24-hourly to 36- or 48-hourly), rather than the drug dose.

In vitro pharmacodynamic models

Although determination of MIC is often sufficient in clinical practice, data on the effects of changing antimicrobial concentrations on the time-course of microbial killing may be important in the development of dosing recommendations. This information can be gained using static time-kill models in which the pathogen is exposed to fixed antimicrobial concentrations and changes in pathogen concentrations are assessed with repeated sampling [15]. Static in vitro time-kill studies enable exploration of the dynamics of bacterial growth and killing, the effects of the bacterial starting inocula on pathogen killing and the probability of emerging resistance during antimicrobial exposure [16]. However, static concentrations do not represent the clinical scenario except when continuous infusion is used. Therefore, the provided data is of moderate relevance for predicting optimal intermittent dosing regimens in individual patients unless a PK/PD model is developed, from which the effects of different concentration–time profiles can be predicted.

Dynamic in vitro models allow the drug concentrations to be continually adjusted to mimic any PK profile that may occur in vivo [17]. Microbes are exposed to antimicrobials in reaction vessels connected in series or parallel and perfused with fresh culture media and can usually be sampled repeatedly during experiments, enabling kill-kinetics, drug concentrations and emergence of resistance to be monitored over several days. Hollow-fibre systems are frequently employed [18], whereby microbes are trapped in a mesh of capillaries over which drugs, nutrients and waste products can diffuse [1921]. More research is warranted to validate the clinical relevance of PK/PD relationships observed in vitro. Further, a lack of immune system in most models provides a caveat in the interpretation of results, although it also means that a drug regimen which is active in vitro is more likely to be effective in critically ill patients.

Animal infection models

Animal infection models provide the opportunity to study drug efficacy while accounting for the host immune response, virulence of the pathogen, antimicrobial susceptibility and bacterial load. Further, they allow for measurements of the antimicrobial effects at the site of infection. Most PK/PD studies in animals have been performed in rodent (murine, rat), rabbit or more recently porcine hosts [22]. Studies can be designed to mimic infection syndromes that occur in critically ill patients (e.g. pneumonia, intra-abdominal infection, bacteraemia or meningitis). Establishment of infection may require the animals to be rendered neutropenic by prior administration of an immunosuppressant, which is therefore likely to provide antimicrobial activity data that differ to that observed in an immunocompetent patient.

A challenge of small animal models is that the antimicrobial PK profile can be very different to that observed in humans as a result of much faster drug clearances and therefore requires careful design of the dosing strategy. Few studies to date have been designed to reproduce the altered PK observed in critically ill patients, which is an important field of future research. Biological models allow for assessing multiple endpoints including mortality, quantitative bacterial counts in tissues and description of inflammatory responses (e.g. cytokine measurements). Still, differences between species, e.g. in the host immune response, may be present that can affect the clinical relevance of animal data.

Antimicrobial PK alterations in critically ill patients

Collecting concentration–time data of antimicrobials in critically ill patients is essential for developing population PK models which can be progressed into dosing regimens that are appropriate given the profoundly altered concentrations that can manifest in these patients [4]. The volume of distribution of antimicrobials may be up to a 5-fold increase in the critically ill as compared to other patient groups for which the recommended dosing regimens have been developed [23]. This typically occurs with hydrophilic drugs such as beta-lactams as a result of increased body weight, capillary leakage and fluid resuscitation and may be magnified in the presence of hypoalbuminemia which can affect the unbound (pharmacologically active) drug concentration.

However, changes in clearance of antimicrobials probably have the biggest impact on the observed PK alterations. In this context, the kidney is the most important organ as many of the commonly used antimicrobials are cleared renally. Whereas impaired renal function is a well-known clinical problem, it has become clear in recent years that significantly elevated creatinine clearance—a phenomenon described as augmented renal clearance (ARC)—is also important and may result in antimicrobial underdosing in infected critically ill patients, therapeutic failure and emergence of resistant pathogens [24]. Burns represent a subgroup of patients that often has a profound day-to-day variability in PK of antimicrobials due to the factors described above (e.g. ARC and dramatic fluid shifts) [25].

Antimicrobial dosing in renal replacement therapy (RRT) and extracorporeal membrane oxygenation (ECMO) is a challenging clinical scenario for dosing of many antimicrobials and requires careful consideration by the clinician (Table 1) [26, 27]. Dosing in the presence of RRT (intermittent, prolonged or continuous) requires knowledge of sources of drug adsorption (filter material) and altered drug clearance (changes in blood dialysate and ultrafiltration flow rate). Once this is known, PK data can be entered into a dosing nomogram, as proposed by Choi et al. [26], to generate a patient-specific antimicrobial dose. For ECMO, far less is known and despite initial concerns about drug adsorption, this does not appear to be a clinically significant effect in the absence of high protein binding or lipophilicity [27]. As a result of renal and hepatic organ dysfunction present in many ECMO patients, lower antimicrobial doses may be required, at least in the maintenance phase.

Table 1 Recommendations for maintenance dosing in critically ill patients with different creatinine clearances, in continuous renal replacement therapy (CRRT) or extracorporeal membrane oxygenation (ECMO)

The mechanisms mentioned above will primarily affect plasma PK. Infection site antimicrobial concentrations, which are likely to be more important in the treatment of most infections, are even less predictable and are affected by drug, tissue type and pathology (e.g. presence of shock). To date, there is a general shortage of robust studies describing both plasma and infection site PK and careful methodological and statistical power considerations are required in designing future studies to gain more informative data.

Clinical PK studies

Given the significant PK variations observed in critically ill patients it is important that the PK of antimicrobials is well characterised in clinical studies targeting the intended patient group and type of infection.

Importance of robust study design and optimal sampling

Rapidly changing physiology in critically ill patients may be captured through repeated sampling and the collection of relevant physiological parameters. Such data may allow for an understanding of how disease/patient factors such as renal function, presence of extracorporeal therapies and sickness severity scores can define altered dosing needs. For instance, cefazolin daily dosing requirements that are specific for critically ill patients were found to vary by up to 400% between and within subjects dependent on serum albumin concentration, renal function and body weight in one study [28]. An improved understanding of the effect of disease on drug disposition allows for more specific, effective and safe dosing regimens to be developed [29, 30].

In some critically ill patients, intensive sampling for determination of antibiotic concentrations may be unfeasible and difficult to justify. Critically ill neonates and infants present unique demands for the conduct of high quality clinical trials because of the limitations of blood sampling volumes. Statistical methods (such as D-optimal design theory) can be used to define the most informative sampling times while minimising the number of samples, and potentially number of patients, needed [31].

Sampling site considerations and tissue penetration

Serum or plasma concentrations are commonly used to describe antimicrobial PK. However, because most infections occur in the extravascular space [32], antimicrobial concentrations measured in tissues, e.g. sampling epithelial lining fluid in pneumonia [33], provide additional information of drug distribution and likely effects for infections in specific body sites. In fact, insufficient antimicrobial concentrations at the site of infection have been postulated as a possible reason for persisting poor infection outcomes in critically ill patients [34]. Traditionally, tissue homogenates obtained from biopsies have been used to describe antimicrobial tissue distribution. However, many pathogens are located in the interstitial fluid (ISF) of tissues and as homogenate concentrations represent an average of intra- and extracellular concentrations they do not provide accurate infection-site data [35].

Tissue ISF concentrations of unbound antimicrobial, which can be measured using microdialysis, are considered to provide the most informative data for extracellular infections [36]. When tissue distribution occurs by passive diffusion exclusively, as is likely the case for most small molecules including beta-lactams and aminoglycosides, plasma and tissue ISF antimicrobial AUCs are similar. However, the shape of the unbound concentration–time profile may differ because of delayed distribution, resulting in a lower and later C max in the tissue. Further, variations between plasma and tissue ISF concentration profiles can be caused by the presence of active efflux transport [37], degradation of antimicrobials [38] or reduced perfusion into peripheral tissues as occurs in septic shock [34]. The observed differences between plasma and tissue concentrations may have implications for dosing regimens in the treatment of severe infections (Table 2) [3948].

Table 2 Reported tissue penetration of selected antimicrobials and potential implications for the treatment of pulmonary infections in critically ill patients

Population PK/PD modelling

Mathematical modelling to maximise use of clinical data and more accurately describe PK/PD relationships

Data on antimicrobial concentrations is most robustly analysed using population PK models, which describe both the typical drug concentrations over time in a population and the inter- and intra-subject variability. Population analysis allows for sparse sampling from each individual because information is shared within the population [49]. There are a number of software packages able to perform population PK analysis, which are discussed elsewhere [50]. In a PK model the simplest structural model includes one compartment and two parameters, typically the volume of distribution and clearance. However, more complex structural models, e.g. including two compartments and four parameters, are often required to accurately describe changes in drug concentrations over time.

Once a population PK model has been developed, Monte Carlo simulations, where concentration–time profiles for a large population are generated to compute the proportion of patients meeting a certain PD target [7, 51], are the most common strategy to define an appropriate dosage of an antimicrobial. However, the use of a fixed PD target has limitations since the time course of bacterial killing and emergence of resistance is not considered, and the target’s dependence on PK is ignored. Meropenem is an example of an antimicrobial where the PD target determined from non-clinical studies has been questioned for extrapolation to the clinical setting [52, 53]. By coupling a PK/PD model developed on the basis of in vitro data to clinical concentration–time profiles it can be shown that prolonged infusions are preferable for patients with augmented clearances [52, 54], which has also been suggested for critically ill patients [55].

For antimicrobials, much of the knowledge on PK/PD relationships has been generated using non-clinical data [22]. This is primarily because quantitative measurements on drug effects are difficult to achieve in patients. Dichotomous outcome data obtained from clinical studies such as cure/no cure provide little information on the impact of the ideal concentration–time course to achieve fast and extensive microbial killing. The clinical PK/PD analysis is therefore limited to analyses that identify exposure thresholds that best separate cure/no cure or mortality/survival (e.g. Classification and Regression Tree (CART) analysis) [56, 57]. Implementation of CART results to individualise antimicrobial dosing has been proposed [58, 59]. However, as for all analyses using PD targets that ignore the time course of effects, exploration of different dosing regimens (e.g. bolus doses and extended or continuous infusions) should be made with caution.

Application of PK/PD models to develop dosing recommendations and to guide individualised dosing

PK/PD models to predict dosing efficacy

As discussed above, PK/PD models can combine available knowledge of PD, based on in vitro, animal or clinical data, with clinical PK. Because of the PK variability in critically ill patients, dosing simulations for different clinical scenarios are required to propose dosing regimens suitable for individual patients. These dosing simulations can incorporate relevant patient characteristics known to affect PK, e.g. body size and organ function. Attainment of PD targets can be predicted for different combinations of antimicrobial doses and pathogen susceptibilities, and optimised dosing regimens can be suggested for specific scenarios (e.g. depending on infection type, severity of illness or immunosuppression). Still, the predictions may be imperfect given the uncertainty in PK and PD present [60] and should be validated in clinical studies when relevant and appropriate.

Dosing guided by therapeutic drug monitoring

PK models coupled to a PD target can be used to design nomograms in which one or more patient characteristics that drive the PK variability are required to determine optimal dosing [61]. Dosing based on validated nomograms increases the probability of achieving PD targets as compared to conventional dosing [62]. A drawback is that deviations from dose times and sampling cannot be handled appropriately. A more time-intensive, but likely more accurate approach is entering the models into dosing software and then using relevant patient data (e.g. age, renal function and body weight) to enable the software to calculate a patient-specific antimicrobial dose [63]. Software packages also allow adaptive feedback (input of concentration data to enable accurate calculation of antimicrobial PK in the individual patient) to further individualise dosing (Fig. 1).

Fig. 1
figure1

Schematic for antimicrobial dosing where dosing is based only on Product Information (orange section), empirical dosing is individualised using relevant patient factors (e.g. renal function or weight; green section) or where dosing is based on collection of therapeutic drug monitoring data (grey sections) and is adapted on the basis of dosing nomograms (top) or adaptive feedback (bottom)

Therapeutic drug monitoring (TDM) involves the measurement of drug concentrations and dose adjustment based on the observed concentration in relation to a target drug exposure. Traditionally, antimicrobial TDM has been used to minimise drug toxicity, but it is increasingly used to maximise drug efficacy (Table 3). This has been made possible by the exposure–effect relationships that have been established for antimicrobials. In the clinical setting, numerous reports have now described associations between plasma concentrations and efficacy or toxicity for selected drugs such as aminoglycosides, glycopeptides, azole antifungal drugs, beta-lactams and oxazolidinones [64].

Table 3 Dosage and targets for therapeutic drug monitoring in critically ill patients and strategies to reduce dosages in patients with normal renal function and moderate to severe renal impairment

Potential of PK models incorporating antimicrobial toxicity and biomarker response

Use of PK/toxicodynamic models to balance efficacy and risk of side effects

As a result of the global escalation of multidrug-resistant pathogens and the lack of effective antimicrobials for monotherapy, physicians have been forced to use antimicrobial regimens for which safety and efficacy have not been established [65]. These treatment options include regimens with doses higher than usually recommended throughout or at initiation of therapy (i.e. front-loaded regimens) and combinations of two or more antimicrobials [66, 67]. Since safety is typically evaluated for monotherapy in healthy volunteers and non-critically ill patients during drug development, limited clinical data is available for combinations.

CART analyses can be used to suggest a drug exposure threshold (e.g. a trough concentration) which is associated with antimicrobial toxicity and describe the antimicrobial effects of monotherapy at a defined dosing regimen [68, 69]. Other statistical analyses, such as logistic regression, can be used to define the increased risk of drug toxicity at higher drug exposures in large patient cohorts [70]. However, these approaches are difficult to apply to regimens with changing doses over the course of therapy and combinations. In this context, mechanism-based toxicodynamic models might be useful to provide a more thorough understanding and predict the time course of antimicrobial toxicity and safety [7173].

Biomarker modelling for early identification of effective therapy

The use of biomarkers (e.g. C-reactive protein, procalcitonin), as surrogates of microbial infection and magnitude of the infective burden, is widely employed in clinical practice to guide initiation and discontinuation of antimicrobial therapy [74]. Its incorporation as PD data in PK/PD modelling has been described recently [75]. The use of PK/PD models for biomarkers has the potential to provide valuable information on the appropriateness of antimicrobial therapy, especially if there is an established correlation between early biomarker response and clinical outcome. However, further evaluation is needed to determine the potential benefits of biomarker modelling in the management of critically ill patients. Importantly, biomarker changes might not be highly specific for the treated infection and there needs to be confidence that its dynamics reflect the interaction between drug, pathogen and host response. Existing additional factors (e.g. surgery and tissue damage) affecting the biomarker response need to be considered in a more advanced modelling framework.

Future directions to improve use of PK/PD for optimal care of critically ill patients

In light of the altered PK in critically ill patients it is clear that a shift towards individualised dosing is required to optimise the efficacy of antimicrobials while reducing the risk of toxicity. This is becoming increasingly important in an era of emerging antimicrobial resistance. However, evidence to support appropriate dosing regimens for specific patient subgroups and infection types is difficult to obtain from clinical trials with these patients. Therefore, preclinical in vitro and animal data and PK/PD models will remain important to support dosing of critically ill patients with severe infections (Fig. 2).

Fig. 2
figure2

Schematic of the role of preclinical infection models, PK/PD modelling and clinical PK studies for optimising antimicrobial dosing and therapeutic drug monitoring in critically ill patients with severe infections. Ideally, future studies with this patient group should assess drug concentrations achieved at the infection site, in addition to plasma, and the associations between measured PK and clinical outcome as outlined in this figure

Further studies are suggested to confirm whether these models can provide precise guidance for treating critically ill patients. Also, efforts are needed to generate more precise data reflecting the dynamics of PK and PD in these patients. For example, dynamic in vitro studies and PK/PD modelling can be used to better describe shifts in appropriate PK/PD targets by evaluating the effects of clinically relevant PK profiles that may occur in treated patients. The human immune system and the bacterial inoculum size are other important determinants of treatment success that may be difficult to measure in critically ill patients but can be explored in vitro and in animal infection models. Better tools need to be developed to quantify the impact of these factors on antimicrobial PD, which can then be translated to clinical practice.

Clinical studies on antimicrobial PK and infections in the critically ill can also be improved to provide maximally informative data to be incorporated in PK/PD models. Ideally, future clinical studies should assess drug concentrations at the infection site, in addition to plasma, susceptibility of the causative pathogen and clinical outcome to determine relevant PK/PD targets for the intended infection site and patient population. Currently, microdialysis is preferred for sampling ISF; however, the development of less complex and more affordable methods would be of great value to improve the feasibility of such studies. The obtained data could then be used for example to understand the mechanism underlying advantages of one dosing approach over another, e.g. systemic versus aerosolised antimicrobial administration for treatment of pneumonia caused by multidrug-resistant pathogens.

Knowledge of the drug exposures associated with optimal antimicrobial effects is important, as is the rapid attainment of these therapeutic exposures in individual patients. To this end, bedside technology that can measure antimicrobial concentrations and microbial susceptibility with rapid reporting is required. In addition, a wider range of routine drug assays available in clinical laboratories is highly desirable to facilitate point-of-care TDM.

Incorporation of the relevant patient-specific data into dosing software that contains robust PK/PD relationships developed with relevant PD data from in vitro, animal and clinical sources and PK data from critically ill patients is likely to improve patient outcomes. A PK/PD-based approach may lead to altered dosing requirements for different infections types and patient populations, but also to making a more refined approach to antimicrobial susceptibility testing. Characterising the impact of intra-individual PK variability in the critically ill on the efficiency of TDM should be further evaluated. If high value of TDM is shown, then an important future step would be to assess the cost-effectiveness of PK/PD-optimised therapy in relevant critically ill patient subgroups and infection pathologies through comprehensive and generalisable prospective clinical studies.

Conclusion

Use of data from preclinical infection models coupled with clinical PK and PD data as well as TDM-guided dosing regimens can maximise the utility of antimicrobial regimens. However, clinical outcome and cost-effectiveness studies quantifying the impact of such an intensive intervention are required.

References

  1. 1.

    Gohil SK, Cao C, Phelan M, Tjoa T, Rhee C, Platt R, Huang SS, Centers for Disease Control and Prevention Epicenters Program (2016) Impact of policies on the rise in sepsis incidence, 2000–2010. Clin Infect Dis 62:695–703

    Article  PubMed  PubMed Central  Google Scholar 

  2. 2.

    Vincent JL, Rello J, Marshall J, Silva E, Anzueto A, Martin CD, Moreno R, Lipman J, Gomersall C, Sakr Y, Reinhart K, EPIC II Group of Investigators (2009) International study of the prevalence and outcomes of infection in intensive care units. JAMA 302:2323–3239

    CAS  Article  PubMed  Google Scholar 

  3. 3.

    Udy AA, Roberts JA, Lipman J (2013) Clinical implications of antibiotic pharmacokinetic principles in the critically ill. Intensive Care Med 39:2070–2082

    CAS  Article  PubMed  Google Scholar 

  4. 4.

    Roberts JA, Abdul-Aziz MH, Lipman J, Mouton JW, Vinks AA, Felton TW, Hope WW, Farkas A, Neely MN, Schentag JJ, Drusano G, Frey OR, Theuretzbacher U, Kuti JL, International Society of Anti-Infective Pharmacology and the Pharmacokinetics and Pharmacodynamics Study Group of the European Society of Clinical Microbiology and Infectious Diseases (2014) Individualised antibiotic dosing for patients who are critically ill: challenges and potential solutions. Lancet Infect Dis 14:498–509

    Article  PubMed  PubMed Central  Google Scholar 

  5. 5.

    Kwa AL, Low JG, Lee E, Kurup A, Chee HL, Tam VH (2007) The impact of multidrug resistance on the outcomes of critically ill patients with Gram-negative bacterial pneumonia. Diagn Microbiol Infect Dis 58:99–104

    CAS  Article  PubMed  Google Scholar 

  6. 6.

    Muller AE, Theuretzbacher U, Mouton JW (2015) Use of old antibiotics now and in the future from a pharmacokinetic/pharmacodynamic perspective. Clin Microbiol Infect 21:881–885

    CAS  Article  PubMed  Google Scholar 

  7. 7.

    Mouton JW, Brown DF, Apfalter P, Cantón R, Giske CG, Ivanova M, MacGowan AP, Rodloff A, Soussy CJ, Steinbakk M, Kahlmeter G (2012) The role of pharmacokinetics/pharmacodynamics in setting clinical MIC breakpoints: the EUCAST approach. Clin Microbiol Infect 18:e37–e45

    CAS  Article  PubMed  Google Scholar 

  8. 8.

    EUCAST. Antimicrobial susceptibility testing. http://www.eucast.org/ast_of_bacteria/. Accessed 15 Jan 2017

  9. 9.

    Clinical and Laboratory Standards Institute (CLSI) (2015) M100-S25: Performance standards for antimicrobial susceptibility testing; 25th informational supplement. CLSI, Wayne

  10. 10.

    Zhao X, Drlica K (2001) Restricting the selection of antibiotic-resistant mutants: a general strategy derived from fluoroquinolone studies. Clin Infect Dis 33:S147–S156

    CAS  Article  PubMed  Google Scholar 

  11. 11.

    Craig WA (1998) Pharmacokinetic/pharmacodynamic parameters: rationale for antibacterial dosing of mice and men. Clin Infect Dis 26:1–10

    CAS  Article  PubMed  Google Scholar 

  12. 12.

    Drusano GL (2004) Antimicrobial pharmacodynamics: critical interactions of bug and drug. Nat Rev Microbiol 2:289–300

    CAS  Article  PubMed  Google Scholar 

  13. 13.

    Barger A, Fuhst C, Wiedemann B (2003) Pharmacological indices in antibiotic therapy. J Antimicrob Chemother 52:893–898

    CAS  Article  PubMed  Google Scholar 

  14. 14.

    Mouton JW, Dudley MN, Cars O, Derendorf H, Drusano GL (2005) Standardization of pharmacokinetic/pharmacodynamic (PK/PD) terminology for anti-infective drugs: an update. J Antimicrob Chemother 55:601–607

    CAS  Article  PubMed  Google Scholar 

  15. 15.

    National Committee for Clinical Laboratory Standards (NCCLS) (1999) M26-A: Methods for determining bactericidal activity of antimicrobial agents; approved guideline. NCCLS, Wayne

  16. 16.

    Bergen PJ, Forrest A, Bulitta JB, Tsuji BT, Sidjabat HE, Paterson DL, Li J, Nation RL (2011) Clinically relevant plasma concentrations of colistin in combination with imipenem enhance pharmacodynamic activity against multidrug-resistant Pseudomonas aeruginosa at multiple inocula. Antimicrob Agents Chemother 55:5134–5142

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  17. 17.

    Gloede J, Scheerans C, Derendorf H, Kloft C (2010) In vitro pharmacodynamic models to determine the effect of antibacterial drugs. J Antimicrob Chemother 65:186–201

    CAS  Article  PubMed  Google Scholar 

  18. 18.

    Tam VH, Louie A, Deziel MR, Liu W, Drusano GL (2007) The relationship between quinolone exposures and resistance amplification is characterized by an inverted U: a new paradigm for optimizing pharmacodynamics to counterselect resistance. Antimicrob Agents Chemother 51:744–747

    CAS  Article  PubMed  Google Scholar 

  19. 19.

    Zinner SH, Husson M, Klastersky J (1981) An artificial capillary in vitro kinetic model of antibiotic bactericidal activity. J Infect Dis 144:583–587

    CAS  Article  PubMed  Google Scholar 

  20. 20.

    Tsuji BT, Bulitta JB, Brown T, Forrest A, Kelchlin PA, Holden PN, Peloquin CA, Skerlos L, Hanna D (2012) Pharmacodynamics of early, high-dose linezolid against vancomycin-resistant enterococci with elevated MICs and pre-existing genetic mutations. J Antimicrob Chemother 67:2182–2190

    CAS  Article  PubMed  Google Scholar 

  21. 21.

    Bergen PJ, Bulitta JB, Kirkpatrick CM, Rogers KE, McGregor MJ, Wallis SC, Paterson DL, Lipman J, Roberts JA, Landersdorfer CB (2016) Effect of different renal function on antibacterial effects of piperacillin against Pseudomonas aeruginosa evaluated via the hollow-fibre infection model and mechanism-based modelling. J Antimicrob Chemother 71:2509–2520

    CAS  Article  PubMed  Google Scholar 

  22. 22.

    Nielsen EI, Friberg LE (2013) Pharmacokinetic-pharmacodynamic modeling of antibacterial drugs. Pharmacol Rev 26:1053–1090. doi:10.1124/pr.111.005769

    Article  Google Scholar 

  23. 23.

    Gonçalves-Pereira J, Póvoa P (2011) Antibiotics in critically ill patients: a systematic review of the pharmacokinetics of β-lactams. Crit Care 15:R206

    Article  PubMed  PubMed Central  Google Scholar 

  24. 24.

    Udy AA, Roberts JA, Boots RJ, Paterson DL, Lipman J (2010) Augmented renal clearance: implications for antibacterial dosing in the critically ill. Clin Pharmacokinet 49:1–16

    CAS  Article  PubMed  Google Scholar 

  25. 25.

    Jamal JA, Economou CJ, Lipman J, Roberts JA (2012) Improving antibiotic dosing in special situations in the ICU: burns, renal replacement therapy and extracorporeal membrane oxygenation. Curr Opin Crit Care 18:460–471

    Article  PubMed  Google Scholar 

  26. 26.

    Choi G, Gomersall CD, Tian Q, Joynt GM, Freebairn R, Lipman J (2009) Principles of antibacterial dosing in continuous renal replacement therapy. Crit Care Med 37:2268–2282

    CAS  Article  PubMed  Google Scholar 

  27. 27.

    Shekar K, Fraser JF, Smith MT, Roberts JA (2012) Pharmacokinetic changes in patients receiving extracorporeal membrane oxygenation. J Crit Care 27:741.e9–741.e18

    CAS  Article  Google Scholar 

  28. 28.

    Roberts JA, Udy AA, Jarrett P, Wallis SC, Hope WW, Sharma R, Kirkpatrick CM, Kruger PS, Roberts MS, Lipman J (2015) Plasma and target-site subcutaneous tissue population pharmacokinetics and dosing simulations of cefazolin in post-trauma critically ill patients. J Antimicrob Chemother 70:1495–1502

    CAS  Article  PubMed  Google Scholar 

  29. 29.

    Hope WW, Drusano GL (2009) Antifungal pharmacokinetics and pharmacodynamics: bridging from the bench to bedside. Clin Microbiol Infect 15:602–612

    CAS  Article  PubMed  Google Scholar 

  30. 30.

    Ramos-Martín V, Paulus S, Siner S, Scott E, Padmore K, Newland P, Drew RJ, Felton TW, Docobo-Pérez F, Pizer B, Pea F, Peak M, Turner MA, Beresford MW, Hope WW (2014) Population pharmacokinetics of teicoplanin in children. Antimicrob Agents Chemother 58:6920–6927

    Article  PubMed  PubMed Central  Google Scholar 

  31. 31.

    Tam VH, Preston SL, Drusano GL (2003) Optimal sampling schedule design for populations of patients. Antimicrob Agents Chemother 47:2888–2891

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  32. 32.

    Ryan DM, Cars O (1983) A problem in the interpretation of β-lactam antibiotic levels in tissues. J Antimicrob Chemother 12:281–284

    CAS  Article  PubMed  Google Scholar 

  33. 33.

    Müller M, dela Peña A, Derendorf H (2004) Issues in pharmacokinetics and pharmacodynamics of anti-infective agents: distribution in tissue. Antimicrob Agents Chemother 48:1441–1453

    Article  PubMed  PubMed Central  Google Scholar 

  34. 34.

    Joukhadar C, Frossard M, Mayer BX, Brunner M, Klein N, Siostrzonek P, Eichler HG, Müller M (2001) Impaired target site penetration of β-lactams may account for therapeutic failure in patients with septic shock. Crit Care Med 29:385–391

    CAS  Article  PubMed  Google Scholar 

  35. 35.

    Mouton JW, Theuretzbacher U, Craig WA, Tulkens PM, Derendorf H, Cars O (2008) Tissue concentrations: do we ever learn? J Antimicrob Chemother 61:235–237

    CAS  Article  PubMed  Google Scholar 

  36. 36.

    Marchand S, Chauzy A, Dahyot-Fizelier C, Couet W (2016) Microdialysis as a way to measure antibiotics concentration in tissues. Pharmacol Res 111:201–207

    CAS  Article  PubMed  Google Scholar 

  37. 37.

    Dahyot-Fizelier C, Frasca D, Gregoire N, Adier C, Mimoz O, Debaene B, Couet W, Marchand S (2013) Microdialysis study of cefotaxime cerebral distribution in patients with acute brain injury. Antimicrob Agents Chemother 57:2738–2742

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  38. 38.

    Dahyot-Fizelier C, Lefeuvre S, Laksiri L, Marchand S, Sawchuk RJ, Couet W, Mimoz O (2010) Kinetics of imipenem distribution into the peritoneal fluid of patients with severe peritonitis studied by microdialysis. Clin Pharmacokinet 49:323–334

    CAS  Article  PubMed  Google Scholar 

  39. 39.

    Panidis D, Markantonis SL, Boutzouka E, Karatzas S, Baltopoulos G (2005) Penetration of gentamicin into the alveolar lining fluid of critically ill patients with ventilator-associated pneumonia. Chest 128:545–552

    CAS  Article  PubMed  Google Scholar 

  40. 40.

    Boselli E, Breilh D, Rimmelé T, Poupelin JC, Saux MC, Chassard D, Allaouchiche B (2004) Plasma and lung concentrations of ceftazidime administered in continuous infusion to critically ill patients with severe nosocomial pneumonia. Intensive Care Med 30:989–991

    Article  PubMed  Google Scholar 

  41. 41.

    Boselli E, Breilh D, Duflo F, Saux MC, Debon R, Chassard D, Allaouchiche B (2003) Steady-state plasma and intrapulmonary concentrations of cefepime administered in continuous infusion in critically ill patients with severe nosocomial pneumonia. Crit Care Med 31:2102–2106

    CAS  Article  PubMed  Google Scholar 

  42. 42.

    Tomaselli F, Maier A, Matzi V, Smolle-Jüttner FM, Dittrich P (2004) Penetration of meropenem into pneumonic human lung tissue as measured by in vivo microdialysis. Antimicrob Agents Chemother 48:2228–2232

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  43. 43.

    Tomaselli F, Dittrich P, Maier A, Woltsche M, Matzi V, Pinter J, Nuhsbaumer S, Pinter H, Smolle J, Smolle-Jüttner FM (2003) Penetration of piperacillin and tazobactam into pneumonic human lung tissue measured by in vivo microdialysis. Br J Clin Pharmacol 55:620–624

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  44. 44.

    Zeitlinger MA, Traunmüller F, Abrahim A, Müller MR, Erdogan Z, Müller M, Joukhadar C (2007) A pilot study testing whether concentrations of levofloxacin in interstitial space fluid of soft tissues may serve as a surrogate for predicting its pharmacokinetics in lung. Int J Antimicrob Agents 29:44–50

    CAS  Article  PubMed  Google Scholar 

  45. 45.

    Lamer C, de Beco V, Soler P, Calvat S, Fagon JY, Dombret MC, Farinotti R, Chastre J, Gibert C (1993) Analysis of vancomycin entry into pulmonary lining fluid by bronchoalveolar lavage in critically ill patients. Antimicrob Agents Chemother 37:281–286

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  46. 46.

    Bulik CC, Wiskirchen DE, Shepard A, Sutherland CA, Kuti JL, Nicolau DP (2010) Tissue penetration and pharmacokinetics of tigecycline in diabetic patients with chronic wound infections described by using in vivo microdialysis. Antimicrob Agents Chemother 54:5209–5213

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  47. 47.

    Kim A, Suecof LA, Sutherland CA, Gao L, Kuti JL, Nicolau DP (2008) In vivo microdialysis study of the penetration of daptomycin into soft tissues in diabetic versus healthy volunteers. Antimicrob Agents Chemother 52:3941–3946

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  48. 48.

    Boselli E, Breilh D, Rimmelé T, Djabarouti S, Toutain J, Chassard D, Saux MC, Allaouchiche B (2005) Pharmacokinetics and intrapulmonary concentrations of linezolid administered to critically ill patients with ventilator-associated pneumonia. Crit Care Med 33:1529–1533

    CAS  Article  PubMed  Google Scholar 

  49. 49.

    Duffull SB, Wright DF, Winter HR (2011) Interpreting population pharmacokinetic-pharmacodynamic analyses—a clinical viewpoint. Br J Clin Pharmacol 71:807–814

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  50. 50.

    Kiang TK, Sherwin CM, Spigarelli MG, Ensom MH (2012) Fundamentals of population pharmacokinetic modelling: modelling and software. Clin Pharmacokinet 51:515–525

    CAS  Article  Google Scholar 

  51. 51.

    Drusano GL, Preston SL, Hardalo C, Hare R, Banfield C, Andes D, Vesga O, Craig WA (2001) Use of preclinical data for selection of a phase II/III dose for evernimicin and identification of a preclinical MIC breakpoint. Antimicrob Agents Chemother 45:13–22

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  52. 52.

    Kristoffersson AN, David-Pierson P, Parrott NJ, Kuhlmann O, Lave T, Friberg LE, Nielsen EI (2016) Simulation-based evaluation of PK/PD indices for meropenem across patient groups and experimental designs. Pharm Res 33:1115–1125

    CAS  Article  PubMed  Google Scholar 

  53. 53.

    Li C, Du X, Kuti JL, Nicolau DP (2007) Clinical pharmacodynamics of meropenem in patients with lower respiratory infections. Antimicrob Agents Chemother 51:1725–1730

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  54. 54.

    Lodise TP, Sorgel F, Melnick D, Mason B, Kinzig M, Drusano GL (2011) Penetration of meropenem into epithelial lining fluid of patients with ventilator-associated pneumonia. Antimicrob Agents Chemother 55:1606–1610

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  55. 55.

    De Waele J, Carlier M, Hoste E, Depuydt P, Decruyenaere J, Wallis SC, Lipman J, Roberts JA (2014) Extended versus bolus infusion of meropenem and piperacillin: a pharmacokinetic analysis. Minerva Anestesiol 80:1302–1309

    PubMed  Google Scholar 

  56. 56.

    Rhodes NJ, O’Donnell JN, Lizza BD, McLaughlin MM, Esterly JS, Scheetz MH (2015) Tree-based models for predicting mortality in gram-negative bacteremia: avoid putting the CART before the horse. Antimicrob Agents Chemother 23:838–844

    Google Scholar 

  57. 57.

    Aitken SL, Altshuler J, Guervil DJ, Hirsch EB, Ostrosky-Zeichner LL, Ericsson CD, Tam VH (2015) Cefepime free minimum concentration to minimum inhibitory concentration (fCmin/MIC) ratio predicts clinical failure in patients with Gram-negative bacterial pneumonia. Int J Antimicrob Agents 45:541–544

    CAS  Article  PubMed  Google Scholar 

  58. 58.

    Lestner JM, Roberts SA, Moore CB, Howard SJ, Denning DW, Hope WW (2009) Toxicodynamics of itraconazole: implications for therapeutic drug monitoring. Clin Infect Dis 49:928–930

    CAS  Article  PubMed  Google Scholar 

  59. 59.

    Neely MN, Youn G, Jones B, Jelliffe RW, Drusano GL, Rodvold KA, Lodise TP (2014) Are vancomycin trough concentrations adequate for optimal dosing? Antimicrob Agents Chemother 58:309–316

    Article  PubMed  PubMed Central  Google Scholar 

  60. 60.

    Colin P, Eleveld DJ, Jonckheere S, Van Bocxlaer J, De Waele J, Vermeulen A (2016) What about confidence intervals? A word of caution when interpreting PTA simulations. J Antimicrob Chemother 71:2502–2508

    CAS  Article  PubMed  Google Scholar 

  61. 61.

    Roberts JA, Taccone FS, Udy AA, Vincent JL, Jacobs F, Lipman J (2011) Vancomycin dosing in critically ill patients: robust methods for improved continuous-infusion regimens. Antimicrob Agents Chemother 55:2704–2709

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  62. 62.

    Cristallini S, Hites M, Kabtouri H, Roberts JA, Beumier M, Cotton F, Lipman J, Jacobs F, Vincent JL, Creteur J, Taccone FS (2016) New regimen for continuous infusion of vancomycin in critically ill patients. Antimicrob Agents Chemother 60:4750–4756

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  63. 63.

    Wong G, Farkas A, Sussman R, Daroczi G, Hope WW, Lipman J, Roberts JA (2015) Comparison of the accuracy and precision of pharmacokinetic equations to predict free meropenem concentrations in critically ill patients. Antimicrob Agents Chemother 59:1411–1417

    Article  PubMed  PubMed Central  Google Scholar 

  64. 64.

    Jager NG, van Hest RM, Lipman J, Taccone FS, Roberts JA (2016) Therapeutic drug monitoring of anti-infective agents in critically ill patients. Expert Rev Clin Pharmacol 9:961–979

    CAS  Article  PubMed  Google Scholar 

  65. 65.

    Zavascki AP, Bulitta JB, Landersdorfer CB (2013) Combination therapy for carbapenem-resistant Gram-negative bacteria. Expert Rev Anti Infect Ther 11:1333–1353

    CAS  Article  PubMed  Google Scholar 

  66. 66.

    Tsuji BT, Brown T, Parasrampuria R, Brazeau DA, Forrest A, Kelchlin PA, Holden PN, Peloquin CA, Hanna D, Bulitta JB (2012) Front-loaded linezolid regimens result in increased killing and suppression of the accessory gene regulator system of Staphylococcus aureus. Antimicrob Agents Chemother 56:3712–3719

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  67. 67.

    Bulitta JB, Okusanya OO, Forrest A, Bhavnani SM, Clark K, Still JG, Fernandes P, Ambrose PG (2013) Population pharmacokinetics of fusidic acid: rationale for front-loaded dosing regimens due to autoinhibition of clearance. Antimicrob Agents Chemother 57:498–507

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  68. 68.

    Lodise TP, Patel N, Lomaestro BM, Rodvold KA, Drusano GL (2009) Relationship between initial vancomycin concentration–time profile and nephrotoxicity among hospitalized patients. Clin Infect Dis 49:507–514

    CAS  Article  PubMed  Google Scholar 

  69. 69.

    Bhavnani SM, Rubino CM, Ambrose PG, Drusano GL (2010) Daptomycin exposure and the probability of elevations in the creatine phosphokinase level: data from a randomized trial of patients with bacteremia and endocarditis. Clin Infect Dis 50:1568–1574

    CAS  Article  PubMed  Google Scholar 

  70. 70.

    Hanrahan TP, Harlow G, Hutchinson J, Dulhunty JM, Lipman J, Whitehouse T, Roberts JA (2014) Vancomycin-associated nephrotoxicity in the critically ill: a retrospective multivariate regression analysis. Crit Care Med 42:2527–2536

    CAS  Article  PubMed  Google Scholar 

  71. 71.

    Boak LM, Rayner CR, Grayson ML, Paterson DL, Spelman D, Khumra S, Capitano B, Forrest A, Li J, Nation RL, Bulitta JB (2014) Clinical population pharmacokinetics and toxicodynamics of linezolid. Antimicrob Agents Chemother 58:2334–2343

    Article  PubMed  PubMed Central  Google Scholar 

  72. 72.

    Croes S, Koop AH, van Gils SA, Neef C (2012) Efficacy, nephrotoxicity and ototoxicity of aminoglycosides, mathematically modelled for modelling-supported therapeutic drug monitoring. Eur J Pharm Sci 45:90–100

    CAS  Article  PubMed  Google Scholar 

  73. 73.

    Sasaki T, Takane H, Ogawa K, Isagawa S, Hirota T, Higuchi S, Horii T, Otsubo K, Ieiri I (2011) Population pharmacokinetic and pharmacodynamic analysis of linezolid and a hematologic side effect, thrombocytopenia, in Japanese patients. Antimicrob Agents Chemother 55:1867–1873

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  74. 74.

    Bomela HN, Ballot DE, Cory BJ, Cooper PA (2000) Use of C-reactive protein to guide duration of empiric antibiotic therapy in suspected early neonatal sepsis. Pediatr Infect Dis J 19:531–535

    CAS  Article  PubMed  Google Scholar 

  75. 75.

    Ramos-Martín V, Neely MN, McGowan P, Siner S, Padmore K, Peak M, Beresford MW, Turner MA, Paulus S, Hope WW (2016) Population pharmacokinetics and pharmacodynamics of teicoplanin in neonates: making better use of C-reactive protein to deliver individualized therapy. J Antimicrob Chemother 71:3168–3178

    Article  PubMed  Google Scholar 

Download references

Acknowledgements

Jason A. Roberts is funded by a Practitioner Fellowship from the National Health and Medical Research Council of Australia (APP1117065) and would like to acknowledge funding from the Australian National Health and Medical Research Council for Centre of Research Excellence (APP1099452).

Author information

Affiliations

Authors

Consortia

Corresponding author

Correspondence to Jason A. Roberts.

Ethics declarations

Conflicts of interest

JA Roberts has received grants from MSD and Cardeas Pharma; JAR has consulted for Astellas, bioMerieux, MSD and Infectopharm.

Additional information

Take-home message: The altered pharmacokinetics of antimicrobials in infected critically ill patients is associated with poor clinical outcomes. We review the role of in vitro and animal infection models, clinical PK data and PK/PD modelling as tools to enable individualised antimicrobial dosing regimens that enhance the probability of better clinical outcomes for critically ill patients.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Tängdén, T., Ramos Martín, V., Felton, T.W. et al. The role of infection models and PK/PD modelling for optimising care of critically ill patients with severe infections. Intensive Care Med 43, 1021–1032 (2017). https://doi.org/10.1007/s00134-017-4780-6

Download citation

Keywords

  • Individualised dosing
  • Antibiotics
  • Pharmacokinetics
  • Pharmacodynamics
  • Mathematical modelling