Summary

A preclinical mouse study found that the investigational tuberculosis regimens PaBQU and DBQU acted more slowly in complex lung lesions than in mice without such lesions. The findings identify a possible factor behind their failure to meet a clinical treatment-shortening target.

A preclinical study in mice has identified slower drug activity inside complex tuberculosis lung lesions as a possible weakness of the investigational PaBQU and DBQU regimens.

The study, posted as a bioRxiv preprint on September 17, 2026, compared the regimens in two mouse models: BALB/c mice, which lack complex lung lesions, and C3HeB/FeJ mice, which develop complex lesions resembling important features of human TB disease. The researchers assessed bacterial killing, relapse prevention and drug exposure in caseum, the necrotic material found within some TB lesions.

The findings are relevant to the development of shorter TB treatments, but the evidence is preclinical and comes from animal models rather than a new human trial.

Contents

Why the regimens were reevaluated

A recent human Phase 2 trial of four-month PaBQU or DBQU regimens was terminated early after the investigational treatments failed to meet the treatment-shortening criterion in the Target Regimen Profile: a regimen duration of three months or less.

The researchers examined whether a factor called lesional liability could have contributed. In this context, lesional liability means that a regimen begins killing bacteria slowly, or has a reduced effect, within the caseum of complex lung lesions. Drug behaviour in that compartment can differ from behaviour measured in simpler experimental infections.

The study used a translational design: activity was compared between an easier-to-treat mouse model and a model that develops more complex lesions. This allowed the researchers to examine whether regimen performance changed when the disease environment became more difficult to treat.

How the mouse comparison exposed the difference

The team measured conventional colony-forming units, a direct count of viable bacteria, alongside the RS ratio, a newer pharmacodynamic marker. It also tracked relapse after treatment and measured drug concentrations in caseum outside the animals.

PaBQU and DBQU were slower in the C3HeB/FeJ mice to produce bactericidal activity, changes in the RS ratio and protection against relapse than they were in BALB/c mice. In other words, the regimens performed less rapidly when tested in mice with complex, human-like lung lesions.

The reference regimen BPaMZ showed less lesional liability than PaBQU and DBQU in the comparison. The pharmacokinetic analysis also projected that fewer drugs in PaBQU would reach their target exposure in caseum than in BPaMZ, particularly early in treatment. The researchers linked this early difference to the slow accumulation of bedaquiline in caseum.

These measurements connect two parts of treatment performance: whether a drug reaches the relevant lesion compartment and whether the regimen produces bacterial killing and prevents relapse there.

What the findings mean for regimen development

The study provides a way to test candidate TB regimens against a more demanding disease environment before they advance to human trials. Comparing an easy-to-treat model with a complex-lesion model may reveal weaknesses that are less visible when efficacy is assessed only through bacterial counts in simpler infections.

For PaBQU and DBQU, the results identify delayed activity in complex lesions as a potential explanation for why four-month regimens did not achieve the required shortening target in the human Phase 2 programme. They also show why total drug exposure is not the only consideration: the timing and location of exposure within a lesion may influence how quickly treatment works.

The authors conclude that additional, diverse regimens need to be examined to determine how reliably preclinical lesional liability predicts clinical treatment-shortening activity. The current evidence therefore supports lesional liability as a useful research question and screening concept, while its value as a general predictor of human outcomes remains to be established.

Sources