Module 15: Pharmacokinetics & DMPK
Connect molecular properties to absorption, distribution, metabolism, excretion, exposure, and dosing so that computational design optimizes the full pharmacokinetic profile rather than drug-likeness alone.
Learning outcomes
- Distinguish PK, PD, exposure, unbound concentration, and ADME drivers.
- Reason about solubility-permeability, distribution, metabolism, and clearance tradeoffs.
- Interpret common PK quantities and their relationship to dose and target coverage.
- Use in silico DMPK models with assay context, uncertainty, and modality awareness.
Interactive PK exposure simulator
Explore how oral dose, bioavailability, absorption, clearance, and distribution shape a one-compartment concentration-time profile.
- Cmax
- 1.35 mg/L
- Tmax
- 2.1 h
- AUC₀–∞
- 14.0 mg·h/L
- Half-life
- 5.5 h
Model boundary: This is an educational, linear one-compartment oral model with first-order absorption and elimination. It omits distribution phases, saturation, repeated dosing, variability, and target-site exposure, so it must not be used to choose a clinical dose.
1. PK, PD, and the ADME system
Pharmacokinetics describes what the body does to a drug: absorption, distribution, metabolism, and excretion determine exposure over time. Pharmacodynamics describes what the drug does to the biological system. Potency matters only if adequate unbound concentration reaches the target for long enough.
Module 12 uses drug-likeness rules and toxicity models for early triage. This module begins at the next decision layer: translating compound and modality properties into exposure, clearance, dose, and pharmacological effect.
Exposure
The concentration-time profile is summarized with quantities such as Cmax, Tmax, AUC, trough concentration, bioavailability, and half-life.
Unbound drug
Only the unbound fraction is generally available for diffusion, clearance, and target engagement. Total plasma concentration can be misleading when protein binding differs.
Dose and route
Oral, intravenous, inhaled, topical, and other routes impose different absorption and first-pass constraints. A PK objective must specify route and dosing schedule.
PK/PD linkage
Relate exposure to an effect metric such as receptor occupancy, biomarker modulation, growth inhibition, or microbial killing instead of optimizing PK in isolation.
2. Absorption: dissolving and crossing membranes
| Driver | Useful in silico view | Common design tension |
|---|---|---|
| Solubility | pKa, logP/logD, crystal packing, melting point proxies, and aqueous solubility models | Adding polarity may improve solubility but reduce passive permeability. |
| Permeability | Polar surface area, hydrogen-bonding, size, ionization, conformational flexibility, and transporter models | Masking polarity can improve permeability but increase lipophilicity and metabolic risk. |
| Dissolution | Particle and solid-state properties plus dose/volume context | A potent compound can still be dose limited if the required mass cannot dissolve. |
| First-pass loss | Gut-wall and hepatic metabolism, efflux, and extraction models | Oral exposure may remain low even when permeability is high. |
3. Distribution: where the compound goes
Volume of distribution
Vd relates the amount of drug in the body to measured plasma concentration. A large apparent Vd often indicates extensive tissue partitioning, not a literal anatomical volume.
Plasma and tissue binding
Albumin, alpha-1-acid glycoprotein, membranes, and tissue components influence free concentration and can produce species- or concentration-dependent behavior.
Blood-brain barrier
CNS exposure reflects passive permeability, ionization, efflux transporters, plasma binding, brain-tissue binding, and local clearance. No single descriptor is decisive.
Target-site exposure
Plasma PK is a proxy. Distribution into tumor, lung, skin, intracellular compartments, or infection sites can be the true determinant of efficacy.
4. Metabolism and drug-drug interaction risk
Phase I reactions commonly introduce or expose functional groups through oxidation, reduction, or hydrolysis. Phase II enzymes conjugate molecules with groups such as glucuronide, sulfate, or glutathione. Metabolites may be inactive, active, reactive, or toxic.
| Question | Computational approach | Experimental partner |
|---|---|---|
| Where will metabolism occur? | Site-of-metabolism and CYP-reactivity models; docking as a supporting hypothesis | Metabolite identification and soft-spot assays |
| How fast is intrinsic clearance? | QSAR/ML models stratified by assay and species | Microsomes, hepatocytes, or recombinant enzymes |
| Could the compound inhibit CYPs? | Classification/regression and structure alerts | Reversible and time-dependent inhibition assays |
| Are reactive metabolites plausible? | Bioactivation rules, structural alerts, and metabolite enumeration | Glutathione trapping and covalent-binding studies |
5. Clearance, half-life, and exposure
For a simple one-compartment intravenous model, total clearance is the dose divided by AUC. The terminal half-life depends on both distribution and clearance: t1/2 = 0.693 × Vd / CL. Improving stability can increase exposure, but excessive persistence may complicate safety or dosing control.
Hepatic clearance
Depends on hepatic blood flow, unbound fraction, and intrinsic metabolic capacity. The limiting factor changes between low- and high-extraction compounds.
Renal clearance
Combines filtration of unbound drug, active secretion, and reabsorption. Molecular size, charge, transporters, and urine pH can matter.
Bioavailability
Oral bioavailability is the product of absorbed fraction and survival through intestinal and hepatic first pass. Diagnose which component limits exposure.
Dose projection
Combine target potency, required unbound coverage, clearance, bioavailability, safety margin, and variability. Potency alone cannot determine a human dose.
6. In silico DMPK is a decision system, not one score
- 1
Define the product profile
State route, dosing frequency, target tissue, onset, duration, acceptable interaction risk, and species before selecting endpoints.
- 2
Use assay-matched models
A model trained on one solubility protocol, pH, species, or matrix should not be silently applied to another. Preserve assay metadata and units.
- 3
Interpret within an applicability domain
Flag unusual chemistry, ionization, molecular size, modality, or predicted uncertainty. Do not force a precise value for out-of-domain molecules.
- 4
Triangulate predictions
Combine physicochemical calculations, empirical models, mechanistic reasoning, and targeted experiments. Resolve disagreements instead of averaging them away.
- 5
Optimize a profile
Track potency, selectivity, solubility, permeability, clearance, safety, and synthetic feasibility together. Pareto thinking is more honest than a hidden weighted sum.
7. Biologics and multispecific antibodies
Large biologics are not governed by small-molecule rules. Their PK can be driven by target-mediated drug disposition, FcRn recycling, proteolysis, immunogenicity, tissue convection, nonspecific binding, charge patches, and aggregation. Multispecific formats add valency, geometry, and multiple target sinks.