Actual State-of-the-Art

Most HAP/VAP treatments are based on the “best guess” approach

Despite many advances in antimicrobial therapy, successful treatment of patients with hospital-acquired pneumonia (HAP), including ventilator-associated pneumonia (VAP) and, more generally, healthcare-associated pneumonia (HCAP), remains a difficult and complex undertaking. Failure to initiate prompt and adequate therapy (i.e. selecting agent(s) to which the etiologic organism is susceptible, applying the optimal dose and schedule, and using the correct route of administration) has been a consistent factor associated with increased mortality and emergence of resistance.

Beyond methicillin-resistant S. aureus (MRSA), the most frequently encountered etiological organisms are Gram-negative bacteria. Accordingly, therapeutic regimens almost always include a b-lactam antibiotic with demonstrated anti Gram-negative activity. Unfortunately, some of these bacteria (e.g., P  aeruginosa) are intrinsically poorly susceptible to first line agents and/or have developed over time an array of low level mechanisms of resistance that decreases their susceptibility to these agents.

Infections due to organisms with reduced susceptibility can be successfully treated if the free antibiotic blood concentrations are carefully adjusted to meet the required PK/PD index. Unfortunately, and unless some specific and rare situations, the clinician has no easy access to actual b-lactam serum concentrations, and cannot, therefore, rationally adjust the drug dosage. As a result, patients at risk for infection with these organisms usually receive a combination of different antibiotics to provide adequate coverage and, if possible, obtain a synergic effect.

Several observational studies have now confirmed that the use of a regimen that combines initially a broad-spectrum b-lactam potentially active against poorly-susceptible pathogens (i.e., piperacillin-tazobactam, ceftazidime, cefepime, imipenem-cilastatin, meropenem; see 2005 ATS/IDSA guidelines and the corresponding European consensus) with an aminoglycoside increases the proportion of patients appropriately treated as compared to monotherapy or to a regimen combining a b-lactam with a fluoroquinolone.

However, when combination therapy was evaluated in randomized controlled studies, its benefit was inconsistent or null after the first 3-5 days of treatment, even when the results were pooled in meta-analyses or the analysis was restricted to patients infected by P. aeruginosa. Furthermore, no beneficial impact on the emergence of resistant microorganisms could be demonstrated. Based on these data, therapy could be safely switched to monotherapy in most patients after 3-5 days, provided that initial therapy was appropriate, clinical course appears favourable, and that microbiological data do not suggest a very difficult-to-treat microorganism, with a very high in vitro minimal inhibitory concentration (MIC), as it can be observed with some nonfermenting-Gram(-) bacteria.

The current choice of a specific b-lactam (before culture results are available) relies on local epidemiological data (anticipated intrinsic antibacterial activity), pharmacokinetic considerations, and knowledge of which therapies the patient may have recently received (within the past 2 weeks, striving not to repeat the same antimicrobial class). Therapy is re-adjusted once culture results are available, depending on the specific susceptibility pattern of the responsible pathogens.
Current therapies are always intravenous. With respect to the above-mentioned b-lactams, adult doses of antibiotics for therapy of HAP/VAP/HCAP in patients with late-onset disease or risk factors for multidrug-resistant pathogens (including prolonged duration of hospitalization [> 5 days], admission from a healthcare related facility, recent prolonged antibiotic therapy, and specific local epidemiological data), as recommended by the 2005 ATS/IDSA guidelines.

The administration by continuous infusion (CI) or extended infusion (EI) has been developed and promoted to increase the time during which the b-lactam concentration can be maintained above the MIC but these approaches as such remain open to clinical controversies with respect to efficacy. A limit is also the lack of stability of the b-lactams when maintained in concentrated solution for several hours.


Dosage adjustments based on population PK-PD targets is the current “best approach”

b-lactams are described as ‘time-dependent’ antibiotics and bacterial eradication is related to the time the free (unbound to proteins) b-lactam concentration remains above the MIC of the offending micro-organisms (fT>MIC). This is now accepted not only for pre-clinical and clinical assessment of novel b-lactams (or the reassessment of older ones) but also for registration of b-lactams by regulatory authorities and used as guidance for setting the clinical breakpoints (i.e. defining up to which level the MIC can rise while ensuring a sufficient likelihood of clinical success) in Europe.

In case of severe infection such as HAP/VAP/HCAP, fT>MIC should be maximized, close to 100% of the dosing interval. Observational studies have demonstrated that this target was achieved only in a low proportion of patients when using discontinuous administration (29-58%). These findings confirm the need of stratification according to individual patient’s pharmacokinetics and bacterial pathogens’ susceptibilities, all so far since the pharmacokinetic of antimicrobials is totally unpredictable in many ICU patients, particularly in case of severe sepsis. Altered pharmacokinetics can result in insufficient serum b-lactam concentrations, or inversely, in increased toxicity in case of renal dysfunction, when standard dosages are administered. This emphasizes the need to carefully monitor trough levels when treating severe infections. A recent interventional study showed that dose readjustment was needed in about 75 % of critically-ill patients in order to reach PK/PD targets, demonstrating a clear interest for TDM.

Several libraries of b-lactams pharmacokinetics (population-based PK-PD targets) are available in the literature as well as from the data submitted for registration by the suppliers of the b-lactams. These population based PK-PD target mainly efficacy. Dosage adjustment based on the population-based PK-PD principles is the current “best approach” for optimizing therapy and is now used in ICU (eg. Bayesian-based dosing software; non-linear regression approaches; monograms based dosing …). However, they do not really allow tailoring b-lactam regimens to the need of individual patients especially if their physiopathological situations change rapidly as it is the case in severe infections.

The problem is still more complex in paediatric populations, especially in prematures and neonates, where dose adaptations are currently made based on weight/size because of lack of data integrating other physiological co-variables. An ongoing open-label clinical trial involving UTARTU (EudraCT 2011-001515-31) is recruiting septic neonates with the aim of determining clinical success of meropenem vs. current standard of care and of determining its PK profile in this specific population.


However this approach still raises concerns that stimulate further improvements

The variability in b-lactam blood levels beyond what can be predicted from PK-PD models has recently prompted the development of b-lactam TDM

As a matter of fact, dosage adjustment based on population-based PK-PD targets does not address the issues related to actual variability of free (unbound to proteins) b-lactam blood concentrations beyond what can be predicted from PK-PD models. Some studies extrapolate free concentrations based on published protein binding values, and studies in which free b-lactam serum levels have been measured have actually shown much larger than predicted inter-patients and intra-patient variations and/or divergences from the models, leading most often to suboptimal therapies (a large non interventional study involving 70 EU centers is presently recruiting ICU patients with the aim of characterizing their individual PK/PD profile and determining how it correlates with clinical outcome). Accordingly, patient-specific and difficult to control parameters are probably most critical. Therefore, therapeutic drug monitoring (TDM) which has traditionally used to minimize drug toxicity, appears nowadays attractive for optimizing the drug exposure at the individual level.

Unfortunately b-lactams represent typically a class of drugs erroneously neglected by TDM as a consequence of the relative safety of use and the absence of automated analytical analysis. Thus, (i) the unpredictable and patient-specific alterations of pharmacokinetics in critically ill patients (altered volumes of distribution, fluctuating renal function, altered protein binding making the free concentrations unpredictable) and (ii) the problem of decreased susceptibility have stimulated the development of b-lactam TDM. However, better b-lactam TDM is still needed to optimize b-lactam exposure and to improve new dosing regimens in individual patients.

Unfortunately the main bottleneck to the expansion of b-lactam TDM in clinical practice is still the absence of appropriate rapid free b-lactam monitoring methodologies

Routine dosages of b-lactams are currently difficult to perform, and/or their results are not available at the bed-side (ie. early enough) to positively impact the therapy. The current techniques for measuring free b-lactam blood levels involve a solid-phase or liquid extraction followed by HPLC associated with UV detection or tandem mass spectrometry (MS-MS).

Major drawbacks are:

  • current methods all require dedicated instruments that are costly, can only be operated by specialized personnel, and must remain “on stand-by” if results need to be available in less than several hours (which means that the instrument cannot be used for other analyses). Because of their complexity, the instruments are always located in specialized laboratories, which makes an interactive approach with the health care professionals in charge of the individual patient quite difficult. It also drastically reduces the number of Institutions where such TDM is feasible for financial and practical reasons.
  • b-lactams are known to be intrinsically chemically unstable. As the instrumentation needed for the current assays is far from the ward, sample handling and protection is essential, which is not always easy to obtain in routine clinical conditions. This may lead to underestimation of the true blood levels and inappropriate suggestions of dose adjustment.
  • Detection by UV (which is easy and accessible) is rather unspecific. Even if it allows to partially differentiate penicillins (λ: 210 nm) from cephalosprins (λ: 260 nm) or carbapenems (λ: 310 nm) a lack of selectivity occurs when a combination of b-lactams is used with the same or overlapping detection wavelengths. Interferences with other drugs than b-lactams and other constituents present in abnormal levels in patients’ blood makes often chromatograms difficult/impossible to interpret unless significant changes in the extraction and analysis procedures are introduced, which is unpractical for routine assays.
  • Detection by tandem mass spectrometry (MS-MS) is even less subject to interferences but its implementation is not easy especially if quantitative determinations are necessary.
  • Separation of free from protein-bound b-lactams before analysis by HPLC, while being technically easy (ultracentrifugation), requires careful control of the conditions to yield reliable results. This is often neglected in most investigations reporting on the use of TDM of b-lactams in patients.

Recently, the group of J. Lipman and J.A. Roberts (Queensland, Australia) has described a method to determine the free concentration of ten b-lactams in human plasma using HPLC combined with UV detection. Results are available within 12 hours which is currently considered as a “rapid” method. Although, this is an important step towards the development of b-lactam TDM, the method still suffers from the above-mentioned lack of specificity, potential interference concerns and localization far from the patient bed-side.

Current dosing strategies (PK-PD targets) do not often take into account potential emergence of low-level drug resistance and the promising attempts in the field need still to be validated in clinic

Resistance is rising rapidly in many Gram-negative bacteria, which is believed to be due to overuse and misuse of antibiotics (at least in part). This is concerning for many clinicians as there are very few viable treatment options remaining. In a large bacterial population causing a serious infection, pre-existing low-level resistant mutants could be present in a minute proportion due to natural mutations. However, a sub-optimal antibiotic dosing regimen represents a selective pressure on this heterogeneous population. While it eradicates most susceptible bacteria in the population, it also facilitates proliferation of less susceptible mutants. This sub-population is then selectively amplified and emergence of resistance is observed as a result. While resistance obtained is initially of low level only (few multiples of MIC increases), it may in the long term progress to very high level (even in the absence of horizontal gene transfer) causing real epidemiological concerns.

Another alarming observation today is the clear increase of MICs of successive isolates during the treatment (MIC drift). This has been demonstrated, namely, for patients with nosocomial pneumonia and similar examples are on the rise. Until now, however, no efforts have been made to adjust dosing during treatment since MIC data were obtained retrospectively only.

An urgent effort is needed to curb these rising resistance trends and, until new effective agents can be developed and approved, using the available agents rationally and more optimally is an option.

At the clinical level, commonly employed methods to suppress or delay the development of resistance include de-escalation therapy, truncated courses of antibiotics, antibiotic cycling and surveillance cultures. Combination therapy is also sometimes used for curbing resistance.

Studies performed in vitro and in animal models show that the development of a low-level resistance to several classes of antibiotics, including b-lactams, can be suppressed/delayed by using higher doses of antibiotics than commonly used. Actually, the aim should be to target the low-level resistant subpopulations present in the inoculum. But calculations of optimal dosages often ignore them and simply assume that their elimination will be achieved by the natural host defences once the large population of susceptible organisms has been eliminated. However, this may not be the case in severely-ill patients where host-defences are impaired either globally or at the site of infection. Also, bacterial populations at an infected site (such as the lung) may be very large, which means that even a minor subpopulation may actually represent an important bacterial burden. Facing these uncertainties, using a high dose appears to be an option although toxic effects may appear.

Actually, little definitive knowledge is available on which b-lactam dosages are needed to effectively suppress the emergence of low-level bacterial resistance to avoid selecting less susceptible subpopulations. But PK-PD can also be important in developing a strategy in this context. Thus, Tam et al. (UHOUS) used an in vitro model, consisting of a system where bacteria are maintained in a hollow fibre cartridge but exposed to a simulated fluctuating drug concentration-time profile to determine the dosing regimen of meropenem that suppress the development of resistance by P. aeruginosa. However, while theoretically promising, the specific target exposures to suppress resistance development have not have been established for many Gram-negative bacteria. Furthermore, the clinical relevance of this dosing approach needs to be validated.

Even with the “best PK-PD based” approach, little attention is paid to possible adverse effects

In almost all studies dealing so far with the optimization of b-lactam antibiotics, little or no attention is paid about the potential risks of exposing patients to sustained high blood levels of these antibiotics. This is because b-lactams have been long considered as very safe drugs if disregarding their potential to cause allergy and other reactions of hypersensitivity which are very patient specific and unrelated to dosage. However, three issues need to be carefully considered when increasing b-lactam dosages beyond the provisions of the corresponding labelings (SmPC):

  • b-lactams is potentially toxic for the central nervous system (myoclonic seizures). While rare with penicillins (only if very large doses are administered), the toxic effect is significantly more frequent with some cephalosporins (e.g. cefepime) and increases in case of reduced renal function. All carbapenems, particularly imipenem, have been associated with seizures and this risk is mentioned in their labelling (SmPC). This type of toxicity is commonly encountered in severely-ill patients who combine the risk of renal insufficiency and often need large doses of b-lactams. This also explains why imipenem is not recommended in neonates (still immature renal function),
  • b-lactams are also known to cause nephrotoxicity when tested in animals at large doses. While the approved drugs can be considered as safe, when used at approved dosages, little is known about the risks associated with higher dosages,
  • b-lactam chemical instability may result in product degradation and cause additional toxicity. While little is reported about this problem in the clinical literature, it is well known from animal preclinical studies. It has appeared as a potential risk for cefepime.

Thus the overzealous use of excessively high antibiotic doses may predispose patients to unnecessary adverse effects. An optimal dosing regimen for a specific patient should also take into account the likelihood of adverse effects associated with high drug concentrations.
Consequently, prospective monitoring of b-lactams in patients undergoing therapy represents a major and much awaited advancement in patient care for addressing simultaneously improved efficacy, reduction of EDR and avoidance of adverse effects. Significant benefits in patient outcomes improvement, resistance suppression and minimizing adverse effects can be anticipated with b-lactam TDM provided the main bottleneck of yet unsatisfactory b-lactam blood level determination is resolved.

To this end, the applicants of this Project already own encouraging preliminary results regarding an alternative, cost-effective, rapid and accurate method for determining free b-lactam blood levels.


Preliminary data regarding a novel method for determining b-lactam blood levels are available and will greatly help to solve the problems related to dosage adjustment relying on population-only PK-PD

ULG, in collaboration with UCL, have developed a colorimetric assay to determine the concentration of free b-lactam antibiotics in a complex biological sample such as serum or micro-dialyzed blood. A patent application describing the method has been filed (International patent application n° PCT/EP2012/070434 filed on October 15, 2012 based on European Patent Application n° 11185288.5; publication via eg. esp@cenet is foreseen mid-April 2013).

The method is based on the hydrolysis of a reporter substrate by an enzyme (the biosensor). The b-lactam to be assayed acts as competitive inhibitor and can be quantified by the measurement of the apparent loss of enzyme activity. The selected biosensor is a class C b-lactamase, a highly specific enzyme that allows the specific determination of b-lactam concentrations in complex biologic fluids. The optimized colorimetric assay is conceptually simpler and faster than HPLC methods (HPLC-UV, LC-MS-MS): an analysis is carried out in 5 minutes. In addition, by using the same colorimetric approach and the combination of 2 or more different selected b-lactamases, ULG and UCL have demonstrated the accurate determination of individual concentration in a mixture of 2 or more b-lactams (analysis time: 10 minutes for 2 different b-lactams). In this case, the concentration is obtained by mathematical calculus without physical separation of b-lactams. In the laboratory, the colorimetric assay has been adapted and carried out by using an automatic bench top analyser validated for clinical chemistries (Ellipse instrument – Analyser Medical System (AMS)).
Preliminary studies show that the method is highly selective since the analytes are the natural substrates of the enzyme. The concentrations of b-lactams selected in the MON4STRAT project have already been successfully determined by the method with respect to meropenem and piperacillin-tazobactam). No interference is observed with vancomycin and generally the lower limit of b-lactam quantification is 1 μg/ml (10 fold lower than minimal b-lactam concentration required for clinical purposes). Finally, ampicillin has been specifically and accurately determined in serum samples submitted to micro-dialysis and than spiked with the antibiotic.