A Canadian perspective on the revised 2020 ASHP–IDSA–PIDS–SIDP guidelines for vancomycin AUC-based therapeutic drug monitoring for serious MRSA infections

Abstract

Background

A revised consensus guideline on therapeutic drug monitoring (TDM) of vancomycin for serious methicillin-resistant Staphylococcus aureus (MRSA) infections was recently published with endorsement of numerous American pharmacy and medical societies. Changing practice from trough TDM to area-under-the-curve-(AUC)-guided dosing was suggested.

Methods

Recent literature was critically appraised to determine whether AUC TDM is appropriate for Canadian hospital practice.

Results

Previous 2009 vancomycin consensus guidelines recommended trough levels of 15–20 mg/L for serious MRSA infections, based on relatively poor evidence for efficacy or safety. In the past decade, aggressive trough targets have led to unnecessary toxicity. Adoption of a TDM strategy using an alternative parameter (AUC) has been suggested, although the evidence for any outcome benefits is low quality. In addition, implementation would require greater resources at health care institutions in the forms of more frequent serum levels or acquisition of costly Bayesian software programs. Most studies on this subject have been observational and retrospective; therefore, relationships between TDM parameters and outcomes have not been convincingly and consistently demonstrated to be causal in nature. Despite claims to the contrary, based on few in silico experiments, available clinical data suggest correlation of trough levels and AUC is high. TDM with lower target trough levels is a simpler solution to reduce risk of toxicity.

Conclusions

There are serious concerns with adoption of AUC TDM of vancomycin into routine practice in Canada. Trough-based monitoring with modest reduction in target levels remains the most evidence-informed practice at this time.

Background

Vancomycin (VAN) is a glycopeptide antibiotic commonly used for the treatment of gram-positive infections, including those caused by methicillin-resistant Staphylococcus aureus (MRSA). A revised consensus guideline on therapeutic monitoring of VAN for serious MRSA infections was released in March 2020 and endorsed by the American Society of Health-System Pharmacists, Infectious Diseases Society of America, Pediatric Infectious Diseases Society, and Society of Infectious Diseases Pharmacists (¹). In this commentary, we discuss the evidence behind one of the primary recommendations, which is to use area under the concentration-time curve (AUC) as the preferred method to dose and monitor VAN instead of the previously recommended monitoring of target trough levels of 15–20 mg/L.

Previous 2009 VAN Therapeutic Drug-Monitoring Consensus Recommendations

Therapeutic drug monitoring (TDM) of VAN has continued to be a greatly debated topic despite more than half a century of use, with controversies regarding the optimal target exposure and approach to monitoring. In an effort to standardize practice, a consensus review of VAN TDM for serious MRSA infections was published in 2009 and has guided clinical practice (in both the United States and Canada) for the past decade (²). This guideline recommended dosing VAN to aim for target trough levels of 15–20 mg/L for serious MRSA infections because these levels were considered to be an adequate surrogate marker for achieving an AUC over the minimum inhibitory concentration (MIC) of more than 400 (provided the MIC of the pathogen was ≤1 mg/L) (²). This AUC:MIC target was based primarily on a small, single-centre retrospective study by Moise et al that found an association between AUC:MIC of more than 345 and clinical success (³) while providing optimal in vitro killing activity with serum levels exceeding MICs of typical pathogens by fourfold (⁴). Moreover, concerns at the time regarding therapeutic failure from underdosing, the MIC creep, and heteroresistant VAN intermediate Staphylococcus aureus (hVISA) were often discussed in the literature, which contributed to recommendations for more aggressive dosing (⁵,⁶). Since the time of these guidelines, systematic study has reported the clinical importance of MIC creep and hVISA to be generally minor (⁷).

Updated 2020 VAN TDM Consensus Guidelines

Previously, despite the demonstration of a relationship between AUC:MIC and activity in animal models of infection, the necessity for multiple serum levels and the resources and time required for AUC:MIC estimation were identified as barriers to TDM use of AUC:MIC (²). However, these limitations are now felt to be surmountable, and the revised consensus statement recommends targeting an AUC of 400–600 mg/L for serious MRSA infections (based on an assumed MIC of 1 mg/L for the pathogen) and retracts the recommendation to target trough levels of 15–20 mg/L (¹). In addition, use of software programs based on Bayesian methods embedded with population pharmacokinetic models to predict steady-state VAN parameters are recommended (¹). The improved ability of such programs to determine AUC with a single VAN level taken in either steady-state or non-steady-state conditions is cited as an advantage in support of this new strategy (¹). Nonetheless, a minimum of two levels is recommended for certain patient populations when determining AUC (eg, obesity, critical illness, unstable renal function); thus, Bayesian AUC TDM may still require an increased number of levels in these groups (¹). In lieu of Bayesian-derived AUC, one-compartment trapezoidal rule analysis has been suggested, but this method also requires the collection of two levels (¹). Potential barriers to use of Bayesian software programs include accuracy, precision, evidence for clinical use, training requirements, integration capability with electronic charting systems, user friendliness, specific pharmacokinetic population modelling, and pricing of the numerous TDM platforms available (⁸). The demonstrated variability of Bayesian software and pharmacokinetic population models (31 models in a recent systematic review) (⁹) raises consistency and quality control issues in this unregulated sub-industry.

Evidence for VAN TDM

The rationale for VAN TDM is based on the intra- and inter-patient variability in pharmacokinetics, physiological changes in those who are acutely ill, and the narrow therapeutic range of VAN, which are all reasons that support individualized dosing and monitoring. A relationship between VAN serum concentrations and clinical response is necessary for TDM to be an effective strategy. However, response to drug therapy is inherently unpredictable and multifactorial (¹⁰). Therefore, although ‘defined’ optimal values for VAN pharmacokinetic–pharmacodynamic (PK–PD) parameters have been promoted, a great deal of uncertainty still exists, and breakpoints for efficacious AUC:MIC values in retrospective studies range widely from 211 to 660 (¹¹). Consideration of patient goals, condition, and response has a major role when adapting the general TDM target for specific patient needs (¹²). In the case of AUC or AUC:MIC monitoring of VAN, the data are primarily drawn retrospectively from patients managed through usual practice (when explicitly reported in studies), which can generally be assumed to be through trough-based adjustments. Therefore, although statistical associations between AUC and outcome are demonstrated through retrospective analysis, clinical management of VAN dosing by trough assessment precludes a conclusion that such associations are evidence of a causal relationship. In addition, many of these studies have a high risk of bias because threshold values for AUC were not predefined (¹¹).

We are not aware of any randomized controlled trials evaluating the efficacy of the VAN AUC targeted approach. There is one completely prospective pre–post study that assessed efficacy and toxicity outcomes of 252 patients who received VAN trough TDM (year 1) or AUC TDM (year 2 and 3) after implementation of AUC monitoring (¹³). Patient characteristics were similar between the two groups, but the duration of VAN therapy was longer in the first year compared with the subsequent 2 years. No treatment failures or deaths occurred during all 3 years, which precluded the ability to make any efficacy assessments; however, nephrotoxicity was statistically lower during the years of AUC monitoring, although it was relatively infrequent throughout the entire study (6 [8%], 0 [0%], and 2 [2%] cases in years 1, 2, and 3, respectively). The low event rate and possible confounders of this study suggest that future large prospective studies are required to support the assertion that AUC-based monitoring is safer than trough-based monitoring. Given the equipoise of the evidence regarding trough versus AUC TDM, what would compel us to proceed with the change to AUC TDM?

Nephrotoxicity and VAN TDM

Since the 2009 guidelines, clinical literature has emerged suggesting that nephrotoxicity is more commonly associated with higher VAN exposure, including among patients with troughs of 15–20 mg/L (¹⁴,¹⁵). A quasi-experimental study found a lower rate of nephrotoxicity when AUC monitoring was implemented with an overall reduction in VAN dosing (median first 24 h 3,000 mg versus 3,250 mg) and trough levels (12.0 mg/L versus 15.0 mg/L) compared with previous years (¹⁶).

Given the lack of clear clinical outcome benefit using a VAN trough-level target of 15–20 mg/L (¹⁷,¹⁸) and a more certain signal of increased nephrotoxicity with trough levels of more than 15 mg/L (¹⁴,¹⁵), a revised target of 10–15 mg/L for serious MRSA infections would be appropriate. This lower range is supported by the consensus endocarditis guidelines, which recommend a target trough of 10–20 mg/L for staphylococcal endocarditis with the caveat of greater nephrotoxicity risk at the higher end of the range (¹⁹).

Correlation Between VAN trough and AUC

According to the 2009 guidelines, trough monitoring was thought to be a useful surrogate for AUC, but the 2020 guidelines cite several Monte Carlo computer simulations (MCS) that suggest a poor correlation between AUC and trough concentrations (²⁰–²²). In contrast, we performed a PubMed search for clinical studies using the terms ‘vancomycin’ [all fields], AND (‘area under the curve’ OR ‘AUC’ [all fields]) AND ‘correlat*’ [all fields] OR ‘regress*’ [all fields] AND pharmacokinetics [all fields] on September 2, 2020. We filtered to English language and human studies. We excluded studies that reported only AUC:MIC to trough correlation (n = 3) because the MCS to which we are comparing did not use MIC in the test variable, and we excluded studies of VAN given by continuous infusion because a trough value cannot be determined. We retrieved 112 citations, from which we were able to determine by means of abstract, full-text review, or both that 17 reported correlation coefficients or coefficients of determination of AUC to trough in their results. Two additional studies obtained from searching bibliographies were also found. We rated the reported correlation using the rule of thumb for interpreting correlation coefficients (²³), recognizing that correlation coefficient is the square root of the coefficient of determination (R²) (²⁴)_. Of the 19 clinical studies, 7 reported very high correlation coefficients (r ≥ 0.9) (²⁵–³¹), 10 studies had high correlations (0.7 > r ≥ 0.89) (³²–⁴¹), none reported moderate correlation (0.5 > r ≥ 0.69), and 2 reported low correlations (r < 0.5) (⁴²,⁴³). Hence the MCS data referred to in the guidelines is not representative of the available human data, which largely demonstrate substantially better correlation. We note that a more recent MCS with full description of methods reported very high correlation of AUC to trough in a morbidly obese population model (⁴⁴). One would not expect trough and AUC to correlate perfectly in the clinical setting because records of timing of blood draws and doses, use of pre-steady-state levels, and calculation errors can contribute to variability. However, we believe that the sum of available data support the opinion from the 2009 guidelines that trough levels are a reasonable surrogate for AUC (²).

Contrast between Evidence for VAN and 5-Fluorouracil TDM

In general, prospective randomized controlled trials are uncommon in TDM. However, high-quality research, including randomized controlled trials for the AUC monitoring and dosing of 5-fluorouracil (a common chemotherapeutic agent with a narrow therapeutic range), has adequately demonstrated improved patient tumour response and decreased toxicity (⁴⁵,⁴⁶). The practice is associated with cost-effectiveness (⁴⁷) and has been endorsed by the International Association of Therapeutic Drug Monitoring and Clinical Toxicology through a thorough and complete evaluative process (⁴⁸). These reports demonstrate that randomized controlled trials in TDM are possible and would be of great benefit in resolving the controversies that persist with VAN TDM.

Final Thoughts on VAN TDM

At this time, VAN AUC-based monitoring has not been shown to improve patient outcomes compared with trough-based monitoring or to reduce nephrotoxicity compared with lower target trough concentrations of less than 15 mg/L. On the basis of the current evidence, it is unclear whether the adoption of AUC-based VAN monitoring for serious MRSA infections provides any clinical benefit. Given the increased nephrotoxicity associated with higher VAN trough concentrations, we recommend adoption of the range of 10–15 mg/L for efficacy and safety (or 10–20 mg/L for MRSA endocarditis infections, with the recognition that nephrotoxicity risk is increased with the higher therapeutic range) (¹⁴,¹⁵). Moreover, because of the dearth of evidence regarding VAN TDM for non-MRSA infections and non-life-threatening MRSA infections, avoidance of toxicity risk should be considered, and therefore a trough range of 10–15 mg/L would be reasonable. Dose titration when levels are within 20% of the targeted range is unnecessary and should be avoided, given the weak evidence for target ranges, provided patient status is improving (¹²). Although the study of precision dosing of VAN may advance understanding of PK–PD parameters and demonstrate the role of laboratory assessment, the time required for interpretation of results is also a potential threat to the duties of careful bedside assessment of clinical parameters, if the priority for the latter is not maintained. At this time, we believe that the most evidence-informed option for patient-focused TDM of VAN is to maintain the practice of trough-based monitoring with a modest reduction in target levels. Further higher-quality research in this field may eventually lead to different strategies of VAN TDM, and we compliment past and future investigators on their contributions to this important area of study.

Ethics Approval:

N/A

Informed Consent:

N/A

Registration and the Registration No. of the Study/Trial:

N/A

Funding:

No funding was received for this work.

Disclosures:

Dr Thirion reports personal fees from Sunovion outside the submitted work.

Peer Review:

This manuscript has been peer reviewed.

Animal Studies:

N/A