PET/CT imaging of tuberculosis lung lesions in marmosets treated with different drug regimens aligns with human clinical outcomes

Overview

A study tested whether lesion-level radiographic metrics from 2-deoxy-2-[18F]fluoro-d-glucose PET/CT in Mycobacterium tuberculosis–infected marmosets provide predictive information about drug regimen efficacy that aligns with human clinical outcomes. Twenty-two treatment arms, including monotherapies and combinations, were applied for two months. Multivariate lesion response profiles were derived to compare radiographic dynamics with terminal bacterial burden per lesion and to assess correspondence with necropsy histopathology.

Methods and approach

A cohort of infected marmosets was allocated across 22 drug regimens encompassing single agents and combination therapies for a fixed two-month treatment interval. Longitudinal PET/CT imaging quantified lesion-level radiographic attributes (including standardized uptake values and volumetric measures) at baseline and during treatment. Terminal bacterial burden was measured per lesion at necropsy and lesions were classified histopathologically. Unsupervised clustering of combined imaging-derived quantitative changes and lesion bacterial counts generated multivariate treatment response profiles at the lesion level.

Results

Unsupervised clustering of PET/CT-derived quantitative measures together with terminal lesion bacterial burden produced multivariate response profiles that corresponded with known clinical regimen outcomes. The method correctly indicated the clinical inferiority of the four-month moxifloxacin-rifampicin-pyrazinamide-ethambutol regimen relative to the six-month standard regimen in cavitary disease. Lesion-level profiles differentiated cavitary granulomas that exhibited radiographic improvement from those that failed to improve or progressed after one month of therapy. Overall, combined quantitative PET/CT metrics yielded greater discriminative information about treatment response than lesion bacterial burden alone.

Implications

Integration of lesion-resolved PET/CT metrics with bacterial burden in a multivariate framework enhances the translational interpretation of preclinical TB treatment studies and can anticipate regimen performance observed clinically. The approach provides a quantitative means to resolve intra-host lesion heterogeneity and to identify lesion subtypes (for example, cavitary granulomas) with differential responsiveness, which may inform regimen selection and trial design decisions. Further validation across additional models, extended treatment durations, and prospective correlation with clinical endpoints will be necessary to establish generalizability and to operationalize these imaging-derived endpoints for regulatory and drug-development use.