From the discovery of penicillin to PK-PD principles: A brief historical overview and what we have learned from recent clinical trials.
Almost a century ago, Alexander Fleming discovered by coincidence in his laboratory the antimicrobial effects of penicillin, a finding that has dramatically changed the landscape of medicine.1 More than 20 years later, Eagle and his colleagues were among the first to establish that β-lactam antibiotics act in a time-dependent manner and wrote, “There is a considerable body of experimental evidence that the therapeutic action of penicillin rests in large part on its direct bactericidal action, and that the factor that primarily determines its therapeutic efficacy is the total time for which the drug remains at effective levels at the focus of infection.”2 It wasn’t until 1985, however, that Williamson et al gave a molecular explanation for the time dependency of β-lactams that appeared to be related to the acylation reaction that occurs when β-lactams bind to the serine active site of penicillin-binding proteins (PBPs). Based on experiments with penicillin in Streptococcus pneumoniae, Williamson et al showed that bacterial cell wall synthesis appears to be inhibited only when sufficient acylation of different PBPs has occurred.3 In 1998, Craig published on the pharmacokinetic (PK) and pharmacodynamic (PD) indices based on preclinical studies, which now serves as a guide to rational dosing of antibiotics.4 It became clear that various classes of β-lactams require different time periods during the dosing interval where free (unbound) drug concentrations remain above the minimum inhibitory concentration (MIC) (expressed as ƒT > MIC) to achieve maximal bacterial killing. Compared with other β-lactams, meropenem requires the shortest ƒT > MIC (approximately 40% of the dosing interval) to achieve maximal bacterial killing.4 Although the antibiotic concentration needs to be above the MIC to achieve and maintain the acylation of PBPs, it also became clear that further dose increments did not lead to enhanced bacterial killing.5 Importantly, nearly 100 years after the discovery of penicillin, we are still learning how to best achieve dose optimization in different patient populations. On balance, shortly after the discovery of penicillin, cases of resistance were reported and 2 major causes of resistance were identified: structural changes in the PBPs, β-lactamases, or a combination of both. From an evolutionary perspective, it is assumed that PBPs were the precursors of β-lactamases and that both enzymes existed already millions of years before penicillin was introduced. Interestingly, both enzymes (β-lactamases and PBPs) show remarkable molecular similarity and mechanism of action,6,7 making them partners in crime, threatening our present-day arsenal of β-lactam antibiotics.
Bearing in mind the time-dependent action of β-lactams and the PK-PD targets defined in preclinical studies, multiple experiments have been performed with prolonged (infusion time up to 4 hours) or continuous (infusion time 24 hours) treatment with β-lactam antibiotics. Lodise et al published on the clinical implications of extended infusion of piperacillin-tazobactam therapy for critically ill patients infected with Pseudomonas aeruginosa. Retrospectively, the investigators compared patients who received extended infusions of piperacillin- tazobactam (3.375 g intravenously for 4 hours every 8 hours) with those who received intermittent infusions of piperacillin-tazobactam (3.375 g intravenously for 30 minutes every 4 or 6 hours).
The authors reported a statistically significant decrease in mortality and improved clinical outcomes in critically ill patients treated with extended infusion.8 Since then, multiple similar studies have been performed in patients with severe sepsis, with a tendency toward improved clinical outcomes and reduced mortality.9 However, the 2 most recent notable randomized controlled trials, MERCY (NCT03452839) and BLING III (NCT03213990), failed to show a difference in mortality, although the BLING III study did show improved clinical outcomes.10,11 Possible explanations for these inconsistent results should be sought in the heterogeneity of the included patients, MIC values against infecting pathogens, dosing strategies, and the absence of therapeutic drug monitoring (TDM).12 The most widely used PK-PD target in these clinical trials is ƒT > MIC, with variations in the percentage of the dosing interval that free drug concentrations remain above the MIC (up to 100%; referred to as the ceiling effect) and the height of the concentrations (as expressed by multiples of the MIC, up to 4xMIC). It should be noted that PK-PD indices are derived from preclinical studies that mainly focused on the PD target bacterial killing, and numerous limitations are increasingly being recognized. On the one hand, the PK component of the model is subject to multiple variables such as renal function, volume of distribution (Vd), and infusion time. On the other hand, the limitations of the PD component (represented by the static MIC value–determined in vitro conditions) are also significant, including the inoculum effect, bacterial growth dynamics and resistance development.13
Although the β-lactams act in a time dependent manner, evidence suggests that resistance suppression of β-lactams appears to be more concentration related, for which minimum blood plasma concentration reached by a drug during a dosing interval Cmin /MIC and the area under the curve (AUC)/MIC appear to be more predictive in PK-PD models.14-17 Subtherapeutic levels of β-lactams may be associated with genetic mutations and resistance development.18-20 In addition, duration of treatment also appears to be a major risk factor for resistance development.21,22 However, a clear correlation between subtherapeutic concentrations of β-lactams that trigger genetic mutations resulting in the expression of altered PBPs and or transcription of β-lactamases remains to be proven. Furthermore, the bacterial inoculum size appears to be important, as rapid target attainment with inoculum reduction at the site of infection in the early phase of infection minimizes the risk of resistance.23
Simply prolonging the infusion time of β-lactams does not necessarily translate to resistance suppression, as demonstrated by the experiment done by Felton et al, who showed in a hollow-fiber infection model that the Cmin /MIC values required to achieve resistance suppression were significantly higher with prolonged infusion than with intermittent infusion.24 Al-Shaer et al performed a retrospective study to assess whether achieving early free β-lactam concentrations above the MIC during 100% of the dosing interval is associated with improved outcomes in the intensive care unit. They reported that a conservative PK-PD target of 100% ƒT > MIC and a stricter target of 100% ƒT > 4xMIC were both significantly associated with clinical cure, microbial eradication, and suppression of resistance. A delay in measuring β-lactam concentrations was associated with clinical failure and higher mortality, indicating the importance of TDM.25 Scharf et al tried to define the optimal β-lactam target for critically ill patients to reach infection resolution and improve outcomes in critically ill patients. The authors reported that achieving the target of 100% ƒT > MIC leads to faster infection resolution in critically ill patients, but there was no benefit for patients who reached the highest target of 100% ƒT > 4xMIC.26 Maranchick et al performed a retrospective study to identify associations between β-lactam PK-PD targets and the emergence of resistance in patients with gram-negative infections. The authors concluded that a ƒAUC/MIC ≥ 494 may be associated with decreased gram-negative resistance emergence. Of note, no associations between resistance emergence and ƒT > MIC and ƒT > 4xMIC were found.17 In a very recent meta-analysis (based on observational studies), Gatti et al showed that attaining an aggressive PK-PD target (defined as 100% ƒT > 4xMIC) was associated with significantly higher odds of clinical cure (OR, 1.69; 95% CI, 1.15-2.49) and lower odds of β-lactam resistance development (OR, 0.06; 95% CI, 0.01-0.29) in critically ill patients.27
Despite a considerable degree of inconsistencies and heterogeneity, it can be carefully concluded from the available evidence that increasing the exposure of β-lactams by prolonged infusion combined with β-lactam concentrations that equal multiple times the MIC during 100% of the dosing interval is associated with improved clinical outcomes and resistance suppression. These dosing strategies could be further optimized by the use of a loading dose, especially in critically ill patients with high inoculum infections and increased Vd and/or those with slightly elevated MICs against the infecting organisms.28 However, as stated earlier, it should be noted that these concepts are based on the assumption that PK-PD indices are static, which is not realistic. Both the PK and PD components of the model are subject to multiple variables, which make them highly dynamic and unpredictable, especially in critically ill patients with septic shock.12
In addition, increasing the dose with prolonged or continuous infusion in an attempt to improve target attainment also carries an increased risk for toxicity. As we move forward with clinical trials that study the relationship between β-lactam exposure and outcome, we should be more aware of these limitations and apply several PK-PD indices for different outcome parameters in more homogenous patient populations with severe sepsis or septic shock. Clinical outcome appears to be best predicted with 100% ƒT > 1-4xMIC and resistance suppression with concentration-related PK-PD indices such as Cmin /MIC and/or fAUC/MIC. One size doesn’t fit all, and this also holds true for the application of PK-PD indices in clinical trials.