Antibiotics aren’t enough

Treatment success in tuberculosis (TB) has long been defined by microbiologic cure in the absence of relapse. Yet microbiological cure alone is not sufficient to the growing community of TB survivors who increasingly face persistent dyspnea, reduced exercise tolerance, impaired quality of life, and increased risk of long-term, post-TB morbidity and mortality.1 An estimated 70 million live with posttuberculosis lung disease (PTLD), defined by limitations in lung function and exercise tolerance.2 In this issue of American Journal of Respiratory and Critical Care Medicine, Zifodya and colleagues evaluate clinical and molecular determinants of PTLD in a cohort of adults in Nairobi, Kenya.3 The study demonstrates that half of those completing standard antibiotic therapy develop PTLD after treatment completion, consistent with global estimates.4 By evaluating RNA sequencing at baseline and 6 months, the study identified that TB survivors with restrictive and obstructive PTLD have unique transcriptomic signatures. Specifically, early inflammation at the time of diagnosis, including increased MMP9 and ARG1 expression, and pathway level increases in IL-6 and TNF signaling at baseline, was associated with restrictive PTLD. This transcriptomic signature is consistent with extracellular matrix remodeling and fibrotic processes implicated in parenchymal lung damage. In contrast, TB survivors with obstructive PTLD had increased TNF–NF-κB, and type I and II IFN signaling that was persistently elevated at the end of TB therapy. These observations suggest that the timing and nature of immune activation, rather than the magnitude of microbial burden alone, drive posttreatment pulmonary impairment. These results reinforce a fundamental concept that the same immune pathways responsible for controlling Mycobacterium tuberculosis can also inflict collateral lung damage. These findings align with prior work demonstrating that approximately 20% of TB patients placed on appropriate antibiotics have worsening lung function between treatment initiation and treatment completion,5 underscoring that inflammation often outlasts infectiousness unless intentionally addressed. Past attempts to attenuate TB-associated lung damage through host-directed therapies (HDTs) such as everolimus,6 metformin,7 dovramilast, doxycycline, azithromycin, and N-acetylcysteine8 have yielded modest benefits. The present work helps explain why. The immune response that augments mycobacterial killing capacity in vitro (TNF and IFN signaling) is associated with immune-induced lung destruction in ­animal models9-11 and in this work by Zifodya et al. is associated with PTLD. If restrictive PTLD arises from early fibrotic pathways and obstructive PTLD from persistent IFN-driven inflammation, broad anti-inflammatory or antifibrotic interventions may be less effective unless matched to the dominant perturbed biological process. These findings emphasize the need to move away from a one-size-fits-all HDT approach and toward an endotype-specific HDT approach. Endotypes are the distinct molecular perturbations within a single disease phenotype.12,13 These results provide further evidence that there is a molecular trade-off between pathogen elimination and immune-induced collateral lung damage. The immune system must balance killing the intracellular pathogen without inducing collateral damage to the lungs (Figure 1A, 2-dimensional endotype-specific HDT). Some individuals with TB will have an immune response with excellent M tuberculosis killing capacity and minimal immune-induced lung damage (Figure 1A, red group) and only need antibiotics, while other TB patients will have severe immune-induced lung damage (Figure 1A, yellow and orange groups) requiring antibiotics and HDT to mitigate immune-induced lung damage and or to restore immune-induced M tuberculosis killing capacity (Figure 1A, purple and orange groups). Endotype-specific HDT for tuberculosis must address both immune-induced Mtb killing capacity and immune-induced tissue pathology. (A) Hypothetical 2-dimensional endotype-specific HDT addresses both immune-induced pathogen killing capacity and immune-induced lung damage. (B) Hypothetical 3-dimensional endotype-specific HDT model addresses immune-induced Mtb killing capacity and mitigating both immune-induced restrictive and obstructive lung damage. Abbreviations: HDT, host-directed therapy; Mtb, Mycobacterium tuberculosis; PFT, pulmonary function testing; PTLD, posttuberculosis lung disease. This work suggests that further nuance will be needed in that the HDT that may be beneficial to mitigate restrictive PTLD may not mitigate obstructive PTLD (Figure 1B, 3-dimensional endotype-specific approach to HDT). Future studies will need to evaluate organ specificity for HDT, but likely the HDT that mitigates immune-induced lung damage is distinct from mitigating immune neuropathology during TB meningitis. Endotype-specific therapy has transformed other diseases. Stratified treatment in asthma now targets eosinophilic Th2-driven inflammation or neutrophilic phenotypes.14 Only 10 years ago, all patients with acute myelogenous leukemia received the same nontargeted cytotoxic chemotherapy. Now, multiple molecular targets (IDH1, IDH2, FLT3, NPM1) have been identified, resulting in endotype-adapted, small molecule–targeted therapy leading to improved outcomes with less toxicity. A recent successful endotype-specific HDT for sepsis used only 2 biomarkers (ferritin and HLA-DR) to identify macrophage-overactivation syndrome or immune paralysis, with contradictory host-directed therapy with recombinant IFN-γ for the latter, and IL-1β inhibition (anakinra) for the former.15 Tuberculosis, a disease of poverty, should not be exempt from nuanced therapy. Rather, the prevalence of PTLD and the profound personal and societal costs of chronic respiratory impairment demand innovation. Practical implementation of endotype-specific strategies for TB will require biomarkers that can stratify individuals at risk early in treatment or preferably at diagnosis.16 While transcriptomic profiling is not be currently feasible, it provides a roadmap for the identification of simplified markers. Zifodya et al. demonstrate that integrating gene expression signatures with key clinical variables substantially improves predictive performance for future PTLD. Future studies will identify the salient 2 to 5 biomarkers that help clinicians distinguish individuals at greatest risk for restrictive or obstructive trajectories. Restrictive and obstructive patterns of PTLD share some clinical risk factors, including reduced mid-upper arm circumference and increased radiographic involvement at diagnosis. These findings reinforce the importance of early TB detection and treatment initiation to minimize immune-mediated injury. Delayed diagnosis prolongs unchecked inflammation and fibrosis, diminishing the potential for recovery even after microbial cure. No HDT can compensate for the irreparable damage done to the lungs by delayed diagnosis of pulmonary TB. Nutritional status is a modifiable clinical factor associated with PTLD risk, and interventions to address undernutrition may represent an accessible means to reduce long-term pulmonary sequelae. Reframing treatment success in TB requires conceptual shifts. Functional recovery must be recognized as a core therapeutic goal alongside microbiologic cure. In addition, the timing and persistence of immune activation must inform interventions that prevent or limit lung injury. Finally, research efforts must prioritize defining and validating endotypes that correspond to meaningful clinical outcomes and guide HDT. Achieving these aims will align treatment success with patient priorities—the ability to breathe comfortably, work, and live without constant reminders of past illness. The findings by Zifodya et al. challenge the field to redefine what it means to cure TB. Antibiotics aren’t enough. Supplementary material is available at American Journal of Respiratory and Critical Care Medicine online. Please see the ICMJE disclosure forms, which have been provided as supplementary material. No artificial intelligence tools were used in writing this manuscript.