Abstract
RATIONALE: Post-tuberculosis lung disease (PTLD), which includes restrictive and obstructive patterns of impairment, develops in 50% or more of survivors of tuberculosis, yet mechanisms and associated risk factors are poorly understood. OBJECTIVES: We sought to identify clinical and transcriptomic risk factors for PTLD phenotypes. METHODS: In a prospective, observational, cohort study, we enrolled adults (month 0) with newly diagnosed pulmonary tuberculosis in Nairobi, Kenya and evaluated clinical and transcriptomic risk factors for PTLD. Participants completed 6 months of standard anti-tuberculosis therapy. PTLD was defined as abnormal spirometry at month 12 (i.e., 6-months post-treatment) with either a restrictive- or obstructive-pattern. MEASUREMENTS AND MAIN RESULTS: We enrolled 205 participants of whom 103 (50.2%) had PTLD, including 60 with restrictive-PTLD and 43 obstructive-PTLD. Participants with PTLD had lower mid-upper arm circumference and cough peak flow. In multivariable analyses, more lung quadrants involved on radiograph at diagnosis was a risk factor for both restrictive- (aOR 2.1, P < .001) and obstructive-PTLD (aOR 2.2, P < .001). Prior tuberculosis was associated with obstructive-PTLD (aOR 5.4, P < .001). Gene expression improved clinical predictive models. In transcriptomic analyses, restrictive-PTLD was associated with upregulation of IL-6/JAK/STAT3 and TNF-α signaling at diagnosis (FDR < 0.2). In contrast, obstructive-PTLD was associated with transcriptomic upregulation of IFN-α and IFN-γ signaling responses (FDR < 0.2) at month 6. CONCLUSIONS: Despite common clinical risk factors (lower mid-upper arm circumference, radiographic lung involvement), PTLD phenotypes have unique transcriptional signatures. Restrictive-PTLD is marked by early pro-fibrotic inflammation. In contrast, obstructive-PTLD is characterized by persistent inflammation at treatment completion.