Virtual bioequivalence: simulating the study before anyone is dosed
A pivotal bioequivalence study costs serious money and, if it fails, sends the whole program back to formulation. Virtual bioequivalence - running the trial as a simulation first - exists to move that failure to the cheapest possible stage: before a single human is dosed.
How it works
A physiologically based pharmacokinetic (PBPK) model describes the body as compartments - gut, liver, plasma, tissues - with the drug's known properties as inputs: solubility, permeability, clearance pathways, protein binding, the physicochemical constants measured or reconstructed for the molecule. The formulation contributes its release behavior, either from the candidate's predicted dissolution profile or the reference's measured one.
Run the model on both the reference product and the candidate and you get predicted concentration-time curves for each. The same statistics a clinical study would compute - AUC and Cmax ratios with their 90% confidence intervals - can then be estimated in silico, across the fed and fasted states and across the population variability a real trial would sample.
This is the virtual bioequivalence trial: an in silico estimate of whether the candidate's exposure should land inside the 80-125% equivalence window, before the batch that would feed a real study is even made.
What it can and cannot do
What it de-risks: the catastrophic surprise. A formula that matches dissolution on paper but diverts in vivo - because absorption is permeability-limited, because a food effect the dissolution never showed, because a metabolite pathway shifts the ratio. Seeing that risk in simulation costs a compute run; seeing it in a failed pivotal study costs the program. Simulation also ranks candidates: when several formulas all pass f2, the virtual trial suggests which deserves the first lab batch.
What it cannot do: it cannot replace the clinical study. Regulators accept model-informed evidence in defined contexts, but a generic filing still rests on measured bioequivalence in people. A virtual BE is a decision tool - it answers "is this formula worth developing?" with evidence instead of intuition. It de-risks the batch, it does not certify it.
Why it matters in a dossier: a candidate formula backed by dissolution similarity AND a virtual trial arrives with the two cheapest failure modes already checked. The licensee's first real money goes toward confirming a candidate that earned its place, not auditioning one that merely looks plausible.
| Stage | Question answered | Cost of failing here |
|---|---|---|
| Deformulation + specs | Is the candidate plausibly the same product? | Research time |
| f2 dissolution similarity | Does it release like the reference, all media? | A simulation run or a lab day |
| Virtual bioequivalence | Should its exposure match in vivo? | A compute run |
| Pilot / pivotal BE study | Does it match, measured, in people? | The study - the expensive one |