PKPD or Exposure-Response modeling

Under construction

PKPD modeling strategies

  • Assume a PK model
    • Estimate PK and PD parameters in a simultaneous fit (SIM)
    • Estimate PK parameters first and then fit PD
      • Condition on individual PK parameter estimates
        • Assume no error in parameters (IPP = Individual PK Parameters)
        • Account for error in parameters (IPPSE = Individual PK Parameters Standard Error) [1]
      • Fix population PK parameters
        • Include individual PK data (PPP&D = Population PK Parameters & Data)
        • Don’t include individual PK data (PPP = Population PK Parameters)

Abbreviations in Zhang et al 2003 [2]

Binary

Logistic regression

Logistic regression is used to model binary data. It is often used for eg E-R or dropout modeling.

E-R relationships of dichotomous endpoints based on logistic regression can be reliably assessed even in the presence of high shrinkage in the pharmacokinetic exposure metric [1,2].

Which exposure metric should be used in E-R?

Ususally, just try all of them (e.g., Cavg, Ctrough, Cmax), and choose the best one (lowest AIC). They are most of the time highly correlated (>0.95), i.e., any of them can be used.

Ordinal data

  • Composite scores
    • IRT
    • BI
    • PO

References

  1. PAGE 31 (2023) Abstr 10663 [https://www.page-meeting.org/?abstract=10663]
  2. PAGE 32 (2024) Abstr 11183 [https://www.page-meeting.org/?abstract=11183]
[1]
Lacroix BD, Friberg LE, Karlsson MO. Evaluation of IPPSE, an alternative method for sequential population PKPD analysis. J Pharmacokinet Pharmacodyn 2012;39:177–93. https://doi.org/10.1007/s10928-012-9240-x.
[2]
Zhang L, Beal SL, Sheiner LB. Simultaneous vs. Sequential Analysis for Population PK/PD Data I: Best-Case Performance. J Pharmacokinet Pharmacodyn 2003;30:387–404. https://doi.org/10.1023/B:JOPA.0000012998.04442.1f.