Background
Organ impairment pharmacokinetic (PK) studies typically require healthy control participants, raising ethical concerns due to their exposure to potential risks without therapeutic benefit [1,2]. It remains unclear at which stage of drug development a model can reliably predict the PK of healthy controls. This study investigates best practices for generating virtual control populations using population PK (PopPK) modelling at various drug development milestones.
Methods
Five drugs were selected to evaluate methodological designs for virtual control populations. For each drug, data were grouped into five scenarios reflecting increasing stages of development:
- Single ascending dose (SAD) and multiple ascending dose (MAD) studies
- SAD, MAD, and other Phase 1 studies
- SAD, MAD, and Phase 2 proof-of-concept studies
- All Phase 1 and Phase 2 studies
- All Phase 1–3 studies
PopPK models were developed and validated for each drug and scenario. Simulations were conducted in two populations: (1) healthy participants from organ impairment studies, and (2) 500 matched individuals from the NHANES database. Geometric mean ratios (GMR) of the simulated and the observed (referenced organ impaired study) were obtained.
Results
Preliminary results from two drugs show GMR between observed and simulated PK parameters were generally close to 1 across scenarios, with the best predictions seen in Scenario 4 as shown in table 1. Simulated data showed greater variability than observed data, especially in Scenarios 3–5. GMRs and confidence intervals were consistent across both NHANES and organ-impaired reference populations.
Conclusion
This study supports the use of PopPK modelling to generate virtual healthy controls in organ impairment studies. Including Phase 2 data improves prediction accuracy, offering an ethical and practical alternative to enrolling healthy volunteers.
Table 1: Geometric mean ratio of the AUC and Cmax
| Drug 1 | Drug 2 | |||
| Scenarios | AUC | Cmax | AUC | Cmax |
| 1 | 0.871 (0.697–1.12) | 1.11 (0.77–1.55) | 1.14 (0.961–1.42) | 1.45 (1.28–1.84) |
| 2 | 0.997 (0.787–1.30) | 1.02 (0.702–1.42) | 1.13 (0.927–1.44) | 1.42 (1.24–1.79) |
| 3 | 1.07 (0.476–2.45) | 0.99 (0.420–1.960) | 1.01 (0.43-2.37) | 0.934 (0.34-2.0) |
| 4 | 1.04 (0.484–2.26) | 1.00 (0.443–1.89) | 1.02 (0.450–2.30) | 0.959 (0.367–1.99) |
| 5 | 1.10 (0.448–2.74) | 1.01 (0.439–2.73) | 1.06 (0.434–2.55) | 0.963 (0.352–2.08) |
| RI study | 1.01 (0.819–1.23) | 0.973 (0.769–1.36) | 0.987 (0.788–1.23) | 0.973 (0.803–1.38) |
References
- Moore KT, Fossler MJ Jr, Younis I. The importance of participant tracking when conducting clinical pharmacology drug trials. J Clin Pharmacol. 2022;62:1465-1467.
- Walker RL, Cottingham MD, Fisher JA. Serial participation and the ethics of phase 1 healthy volunteer research. J Med Philos. 2018;43:83-114
