Structural Preferences and ADMET Regulation of Fluorinated Small-Molecule Drugs.

Fluorination is a powerful yet highly context-dependent strategy in medicinal chemistry. The F-DrugDatabase comprising 1,789 fluorinated small molecules was constructed, including 430 FDA-approved or marketed drugs. Using Bemis-Murcko scaffold analysis and a matched molecular pair (MMP) "defluorination-hydrogenation" approach, we reveal that 87.7% of fluorinated drugs contain only 1-3 fluorine atoms, with benzene scaffolds dominating (63.5%). Quantitative MMP analysis indicated that the -CF3 group was associated with an increase in predicted elimination half-life (average Δt1/2 = +12.24 h, P < 0.01), whereas -CH2F showed less favorable and more variable predicted effects. Based on these insights, we propose F_RO5, a fluorination-adapted drug-likeness guideline (MW ≤ 800 Da, cLogP ≤ 7, and TPSA ≤ 200 Å2) that covers 94.2% of marketed fluorinated drugs, providing broader coverage than the conventional Lipinski Rule-of-Five (68.2%). This data-driven framework provides quantitative guidelines for rational fluorination and early stage candidate prioritization.