The case for ‘conservative pharmacotherapy’—authors’ response

Dear Editor,

We thank Alsultan et al.¹ for their interest in our article.² We reiterate the fundamental premise of our article was that the value of precision dosing for any drug, using therapeutic drug monitoring (TDM) or other new technologies, should be convincingly proven and shown to outweigh competing priorities before these new methods are adopted into routine clinical practice. We are concerned that over reliance on numerical values can lead to unnecessary cascades of care and distract from careful clinical assessment, particularly if achieving a pharmacokinetic/pharmacodynamic (PK/PD) target has not been consistently shown to improve outcomes. We were disappointed that Alsultan et al.¹ did not directly address our premise, but rather focused on the value of preclinical data in regulatory approval. Further, they make assertions that are not sufficiently supported, for example the unreferenced statement ‘The targets generated by preclinical studies are reproducible and they can be highly predictive of what happens in vivo.’¹

While we agree that in vitro hollow fibre models are important for understanding the action of antimicrobials under controlled conditions, we do not agree that we can assume these data are directly applicable to patient care. Proof from high quality clinical studies is essential. Furthermore, although the preclinical data generated from hollow fibre models might be important to the EMA for regulatory purposes, regulatory activities and bedside patient assessment are very different functions, with the former being irrelevant to our commentary.

There is ample literature that directly contradicts Alsultan et al.’s statement that ‘we have a good understanding of the relationships between drug exposure and efficacy, and, in some cases, toxicity…[including] vancomycin, the β-lactams, linezolid and the fluoroquinolones’.¹ With regards to β-lactams, a recent review of TDM stated β-lactam ‘PD targets still remain to be elucidated’ and noted that ‘Current β-lactam TDM practices are aimed at achieving 100% fT > MIC or 100% fT > 4 × MIC, but perhaps lower percentage targets (40% to 70%)…are adequate.’³ So in fact, at this point, the target is unclear. Likewise, the evidence supporting the proposed vancomycin AUC-MIC therapeutic range is weak and inconsistent.^4–8 Lower limits of the therapeutic range have varied from 211 to 667 in observational clinical studies and preclinical data have been equally, if not more, inconsistent (AUC-MIC lower limits from 86 to >700).^4–8 In fact, in one of the highest quality clinical studies on this topic, the PROVIDE study, the authors were unable to show an association between AUC-MIC and clinical failure.⁹ These examples demonstrate that we do not have a good understanding of relationships between drug exposure and response.

Alsultan et al.¹ cite a 2003 review of basic PD and clinical applications of antimicrobials,¹⁰ to support their statement that ‘Recent clinical trial designs have effectively incorporated preclinical PK/PD data.’, referring to vancomycin, β-lactams, fluoroquinolones and linezolid. This citation does not focus on clinical trials designed to establish causality as implied by its use here. Additionally, several statements in the review underscore the variability in preclinical PK/PD. First, when discussing the relationship between vancomycin PK/PD parameters in preclinical models, the review authors note ‘the 24-hour AUC-MIC required for 50% of maximum killing over 24 hours ranged from 86 to 460’, but, for other strains, ‘peak-MIC ratio was the best parameter with values of 4.4 to 9.5’ and later ‘increasing concentrations of vancomycin…did not affect the rate or extent of bacterial killing of S. aureus’.¹⁰ With regards to PD of linezolid, the review authors note that no clinical trials ‘have yet associated success or failure with any PD target’.¹⁰ So, to summarize, critical review of Alsultan et al.’s supporting reference reveals that it, in fact, does not support what they assert in their letter.

We disagree that the use of tools with substantial limitations is ‘better than no information at all’.¹ More information is not always useful and tests with suboptimal performance characteristics can trigger inappropriate or even harmful cascades in care.¹¹ A PK/PD result must have established value to patients and be able to inform clinical decision making.¹² Without this, it is just distraction from other relevant monitoring parameters and wastes resources. Use of unnecessary medical services is increasingly recognized worldwide and often coexists with unmet healthcare needs, suggesting the former divert resources from the latter.¹²

Finally, we object to the claim that if a patient ‘die[s] subsequent to an infection, but without live pathogens within them…the antimicrobial did its job’. This is certainly not the standard that the EMA, the FDA or Health Canada set for approval of new antimicrobials. We care most about patients, not dead pathogens. Patient care outcomes should be the ultimate focus.

Transparency declarations

S.C.J.J. has received speaker’s honoraria from Sunovion and Melinta. J.J.S. and B.R.D.: none to declare.

References