--- module: 097-02 language: en chapter: 97 title: "Receptor Theory, Experimental Pharmacology, Drug Discovery, and Pharmacogenomics" module_title: "Drug discovery, preclinical development, and clinical trials" source_sha256: 87e531aebe333aaa10983015e9601227c067370e2a9d7ddc966c41bcf86952d8 --- # Drug discovery, preclinical development, trials ## Development as iterative translation ### Connects a modifiable mechanism with benefit ### Rarely a simple target-to-cure sequence #### Disease may not depend on the target #### Models capture only one feature #### Active compound may lack exposure or safety ### Failure at any interface invalidates results ## Target identification and validation ### Genetics, pathology, omics, phenotypic screens ### Association is not validation ### Stronger target evidence #### Manipulation changes disease-relevant biology #### Direction and timing understood #### Human variation predicts compatible effects ### Lifelong genetic loss differs from late inhibition ### Valid targets may be inaccessible or unsafe ## Screening and hits ### Phenotypic: integrated biology, harder mechanism ### Target-based: tractable, may oversimplify ### Orthogonal assays confirm activity ### Counter-screens catch artefacts and off-targets ### A hit is a starting compound, not a drug #### Confirm identity, purity, concentration-response ### Structure-activity relationships guide chemistry #### Lipophilicity aids passage, adds binding ## Lead optimisation and biomarkers ### Pharmacokinetics and pharmacodynamics together ### Absorption through active metabolites set exposure ### Target engagement biomarkers ### Downstream biomarkers show biological consequence ### Clinical endpoints: feel, function, survive ### A biomarker is not automatically a surrogate #### Non-causal marker or off-target harm ## Preclinical models ### In vitro: purified protein to co-culture #### Complexity costs control and throughput #### Cell lines mutate and lose phenotypes #### Organoids lack circulation and immunity ### Compare with achievable unbound human exposure ### Animals integrate kinetics, physiology, toxicity #### Species differ in targets, metabolism, immunity #### Homologous, phenotypic, or empirically predictive ### Rigorous design reduces exaggerated effects ## Safety, toxicology, and manufacturing ### Safety pharmacology: heart, lungs, CNS ### Toxicology: repeat dosing, genotoxicity, reproduction ### NOAEL is a boundary, not zero risk ### Starting dose integrates exposure and uncertainty ### Manufacturing determines delivered exposure #### Impurities may have distinct toxicity #### Biologics: folding, glycosylation, immunogenicity #### Process changes need comparability evidence ## Early clinical development ### First-in-human: safety, tolerability, kinetics ### Single and multiple ascending dose cohorts #### Predefined stopping rules ### Healthy volunteers or patients by risk ### Escalate on cumulative exposure, not dose alone ### Phase 2: dose, regimen, preliminary efficacy #### Dose choice is an estimation problem #### Several doses and exposure-response modelling ### Enrichment raises signal, narrows applicability ## Confirmatory trials and endpoints ### Phase 3 tests benefit and harm ### Randomisation balances prognostic factors ### Allocation concealment protects enrolment ### Blinding reduces differential care and assessment ### Intention-to-treat preserves randomisation #### Per-protocol is vulnerable to selection ### Composite endpoints driven by minor components ### Surrogates need an established link to benefit ### Superiority, non-inferiority, equivalence #### Non-inferiority needs a working active control #### Poor adherence, crossover mimic similarity ### Significance is not clinical importance ## Adaptive designs and pharmacovigilance ### Prespecified adaptation preserves error control ### Platform trials add or remove arms ### Interim looks need statistical adjustment ### Independent data-monitoring committees ### Post hoc adaptation is ordinary bias ### Rare, latent, pregnancy effects after approval ### Spontaneous reports lack denominators ### Investigate signals proportionately ## Development decisions ### Probability, consequence, opportunity cost ### Coherence outweighs a single small p value ### Update target, model, molecule, dose separately ### Answers must stay coherent over time