Researchers led by Radojka M. Savic and Véronique Dartois built a validated translational tool connecting how well anti-tuberculosis drugs penetrate lung lesions to the chance of success in shortening therapy. The approach, reported in Nature Communications, uses lesion-compartment drug distribution data rather than relying on plasma concentration surrogates. The team generated site-of-disease distribution data for 24 drugs and candidates in a rabbit model of cavitary tuberculosis, enabling more granular modeling of what bacteria experience inside cavities. Their framework is designed to reduce the recurring mismatch between earlier promising signals and disappointing Phase 3 outcomes where the pharmacology at the actual site of infection was unclear. If adopted broadly, the method could change how shortened regimen candidates are selected and how dosing strategies are modeled before large trials—particularly for lesion-rich disease forms.