Table 6. Open in new tab Suggested...

Suggested guidelines for conducting studies

1) Consider sample-size calculations to determine whether an exposure–response relationship has a reasonable likelihood of being identified. Conducting analyses on too few patients is unlikely to yield meaningful results.

2) Consider reporting CIs or credible intervals measuring the level of uncertainty in the results to avoid overinterpretation when PK–PD index–response relationships are not identified.

3) Ensure the population is as homogeneous as possible with respect to infection and infecting pathogen, or control for these factors. This will facilitate the detection of PK–PD index–response relationships. Similarly, it may be desirable to standardize concurrent treatments.

4) Derive the PK parameters of the antimicrobial in a robust manner from sufficient samples (the sampling framework may be derived using Stochastic Optimal Design or similar methodology) and using an appropriate population PK model. If free concentrations of antimicrobials are important, then these should be measured rather than adjusting for protein binding using a fixed rate to allow for the fact that protein binding may vary. For some infections, concentration at the site of infection may be important and this should be measured if possible.

5) Perform antimicrobial susceptibility testing of infecting pathogens before the start of therapy and determine results using standardized methodology. If possible, it may be preferable to store strains to test concurrently in one laboratory.

6) Consider having standardized outcomes and timing of outcomes. The outcomes should be relevant to the patient. PK–PD index–response relationships may be more likely if continuous rather than dichotomous outcomes are used.

7) Follow a pre-specified analysis plan. The most appropriate way of statistically analysing the relationship between PK–PD indices and outcomes needs to be further investigated. Explicit modelling of the PK–PD index, for example using fractional polynomials, ⁸⁸ may be preferable to recursive partitioning if the sample size is sufficient. It may be advisable to produce a standardized list of covariates that should be assessed to see if they are associated with outcome, for example severity of illness and presence of comorbidities. This list may vary by indication. The investigation of the influence of covariates on the PK–PD relationship for efficacy using multivariable analyses and through the findings of interactions with the PK–PD index may help further the understanding of which subsets of patients are at increased risk of suboptimal drug exposure.

8) The evaluation of PK–PD indices achieved among patients relative to non-clinical PK–PD targets for efficacy is useful. Such information provides dose-selection support, especially if PK–PD indices achieved are on the upper plateau of the non-clinical PK–PD relationship for efficacy.

Table 6.

Open in new tab

Suggested guidelines for conducting clinical PK–PD studies

Suggested guidelines for conducting studies
1) Consider sample-size calculations to determine whether an exposure–response relationship has a reasonable likelihood of being identified. Conducting analyses on too few patients is unlikely to yield meaningful results.
2) Consider reporting CIs or credible intervals measuring the level of uncertainty in the results to avoid overinterpretation when PK–PD index–response relationships are not identified.
3) Ensure the population is as homogeneous as possible with respect to infection and infecting pathogen, or control for these factors. This will facilitate the detection of PK–PD index–response relationships. Similarly, it may be desirable to standardize concurrent treatments.
4) Derive the PK parameters of the antimicrobial in a robust manner from sufficient samples (the sampling framework may be derived using Stochastic Optimal Design or similar methodology) and using an appropriate population PK model. If free concentrations of antimicrobials are important, then these should be measured rather than adjusting for protein binding using a fixed rate to allow for the fact that protein binding may vary. For some infections, concentration at the site of infection may be important and this should be measured if possible.
5) Perform antimicrobial susceptibility testing of infecting pathogens before the start of therapy and determine results using standardized methodology. If possible, it may be preferable to store strains to test concurrently in one laboratory.
6) Consider having standardized outcomes and timing of outcomes. The outcomes should be relevant to the patient. PK–PD index–response relationships may be more likely if continuous rather than dichotomous outcomes are used.
7) Follow a pre-specified analysis plan. The most appropriate way of statistically analysing the relationship between PK–PD indices and outcomes needs to be further investigated. Explicit modelling of the PK–PD index, for example using fractional polynomials, ⁸⁸ may be preferable to recursive partitioning if the sample size is sufficient. It may be advisable to produce a standardized list of covariates that should be assessed to see if they are associated with outcome, for example severity of illness and presence of comorbidities. This list may vary by indication. The investigation of the influence of covariates on the PK–PD relationship for efficacy using multivariable analyses and through the findings of interactions with the PK–PD index may help further the understanding of which subsets of patients are at increased risk of suboptimal drug exposure.
8) The evaluation of PK–PD indices achieved among patients relative to non-clinical PK–PD targets for efficacy is useful. Such information provides dose-selection support, especially if PK–PD indices achieved are on the upper plateau of the non-clinical PK–PD relationship for efficacy.

Suggested guidelines for conducting studies
1) Consider sample-size calculations to determine whether an exposure–response relationship has a reasonable likelihood of being identified. Conducting analyses on too few patients is unlikely to yield meaningful results.
2) Consider reporting CIs or credible intervals measuring the level of uncertainty in the results to avoid overinterpretation when PK–PD index–response relationships are not identified.
3) Ensure the population is as homogeneous as possible with respect to infection and infecting pathogen, or control for these factors. This will facilitate the detection of PK–PD index–response relationships. Similarly, it may be desirable to standardize concurrent treatments.
4) Derive the PK parameters of the antimicrobial in a robust manner from sufficient samples (the sampling framework may be derived using Stochastic Optimal Design or similar methodology) and using an appropriate population PK model. If free concentrations of antimicrobials are important, then these should be measured rather than adjusting for protein binding using a fixed rate to allow for the fact that protein binding may vary. For some infections, concentration at the site of infection may be important and this should be measured if possible.
5) Perform antimicrobial susceptibility testing of infecting pathogens before the start of therapy and determine results using standardized methodology. If possible, it may be preferable to store strains to test concurrently in one laboratory.
6) Consider having standardized outcomes and timing of outcomes. The outcomes should be relevant to the patient. PK–PD index–response relationships may be more likely if continuous rather than dichotomous outcomes are used.
7) Follow a pre-specified analysis plan. The most appropriate way of statistically analysing the relationship between PK–PD indices and outcomes needs to be further investigated. Explicit modelling of the PK–PD index, for example using fractional polynomials, ⁸⁸ may be preferable to recursive partitioning if the sample size is sufficient. It may be advisable to produce a standardized list of covariates that should be assessed to see if they are associated with outcome, for example severity of illness and presence of comorbidities. This list may vary by indication. The investigation of the influence of covariates on the PK–PD relationship for efficacy using multivariable analyses and through the findings of interactions with the PK–PD index may help further the understanding of which subsets of patients are at increased risk of suboptimal drug exposure.
8) The evaluation of PK–PD indices achieved among patients relative to non-clinical PK–PD targets for efficacy is useful. Such information provides dose-selection support, especially if PK–PD indices achieved are on the upper plateau of the non-clinical PK–PD relationship for efficacy.

This Feature Is Available To Subscribers Only