Asymptomatic tuberculosis can remain difficult to identify when symptoms are absent and routine sampling does not confirm infection. A longitudinal prospective cohort published in The Lancet Respiratory Medicine assessed whether PET–CT could detect tuberculosis-consistent lung changes in adults before diagnosis and whether chest x-ray computer-aided detection could perform against that benchmark. The work focused on HIV-uninfected adult household contacts of patients with rifampicin-resistant tuberculosis in Khayelitsha, Cape Town. These contacts belonged to a high-risk group and were not eligible for preventive therapy under guidance in place at the time. Participants underwent baseline PET–CT, chest x-ray, blood testing and intensive sputum collection, followed by active screening and health register review. The findings show that many later tuberculosis diagnoses were linked to lung abnormalities already visible at baseline, even while participants remained asymptomatic.
Baseline PET–CT Reveals Lung Abnormalities
The cohort included 250 asymptomatic adults aged 18–65 years. All participants had recent household contact with patients who had at least rifampicin-resistant tuberculosis. Follow-up extended over a median of almost five years, combining active visits, sputum testing when indicated and review of provincial health data for diagnosed or treated tuberculosis.
Baseline PET–CT separated participants into four lung categories. Some had normal lungs, while others had abnormalities considered consistent with tuberculosis, consistent with inactive tuberculosis or classified as other lung lesions. More than half of participants had lung parenchymal abnormalities on PET–CT. A smaller group had scans consistent with tuberculosis or inactive tuberculosis. Lymph node abnormalities were also common, especially among participants whose scans were consistent with tuberculosis or inactive tuberculosis.
Participants with PET–CT findings consistent with tuberculosis or inactive tuberculosis differed from those with normal lungs or other lesions. They were generally older and more often had positive immune testing for tuberculosis infection. Participants with PET–CT scans consistent with tuberculosis also showed differences in BMI and selected blood markers. These baseline patterns helped distinguish a group with more visible lung disease despite the absence of symptoms.
Visible Disease Often Comes Before Diagnosis
Tuberculosis was diagnosed and treated in 18 participants during baseline assessment or follow-up. Most had bacteriological confirmation, while a small number were treated after clinical diagnosis. Some cases were detected immediately at baseline, but most were diagnosed later, after a median of more than two years. Several baseline cases required induced sputum culture for confirmation, showing that spontaneous sputum testing alone did not identify all cases.
Baseline PET–CT category was strongly linked to later tuberculosis diagnosis. Participants whose scans were consistent with tuberculosis accounted for most treated cases. Smaller numbers of diagnoses occurred among those with scans consistent with inactive tuberculosis, other lung lesions or normal lungs. Overall, most treated participants had baseline PET–CT scans consistent with tuberculosis or inactive tuberculosis.
Must read: Optimising CT for Lung Screening
Among participants with bacteriologically confirmed tuberculosis, many were still asymptomatic at the time of confirmation. This was especially evident among those whose baseline PET–CT scans were consistent with tuberculosis. The risk of diagnosis over five years was highest in that group, compared with participants whose baseline scans showed normal lung parenchyma. Baseline immune sensitisation, previous tuberculosis and duration of household exposure did not show the same association with subsequent diagnosis in the reported analyses.
Computer-Aided X-Ray Offers a Practical Comparison
PET–CT provides sensitive imaging but is not feasible for routine tuberculosis screening. Chest x-ray computer-aided detection offers a more practical option for intensified screening in high-burden settings. Three systems were assessed: CAD4TB, qXR and Lunit. The software generated scores from chest x-rays, with higher scores indicating findings suggestive of tuberculosis and warranting further sputum investigation.
Computer-aided chest x-ray scores were higher among participants whose PET–CT scans were consistent with tuberculosis than among those with other lesions or normal lungs. Scores were also higher among participants with scans consistent with inactive tuberculosis. When benchmarked against PET–CT findings, chest x-ray computer-aided detection performed comparatively well. Performance was around 0.8 when PET–CT-consistent tuberculosis served as the reference and slightly higher when the benchmark focused on participants with PET–CT-consistent tuberculosis or inactive tuberculosis who were bacteriologically confirmed.
Repeat PET–CT in a subgroup showed that lung lesions could change over time without treatment. Changes were most frequent among participants whose baseline scans were consistent with tuberculosis. Some lesions improved, some worsened and some showed mixed patterns. Complete resolution was rare. Short-term imaging changes over several months did not clearly determine long-term tuberculosis outcomes, but the repeated imaging showed that asymptomatic lung abnormalities could remain dynamic.
The findings indicate that many asymptomatic adults later diagnosed with tuberculosis already had visible lung disease at baseline when assessed with PET–CT. The strongest risk signal came from PET–CT scans consistent with tuberculosis, while routine symptom assessment and spontaneous sputum testing missed some cases. PET–CT is not suitable for routine screening, but it can provide a sensitive benchmark for diagnostic development. Chest x-ray computer-aided detection performed comparatively well against that benchmark and may support intensified screening where practical tools are needed.
Source: The Lancet Respiratory Medicine
Image Credit: iStock
References:
Esmail H, Thienemann F, Sossen B et al. (2026) PET–CT benchmarked detection and 5-year progression of asymptomatic tuberculosis: a longitudinal, prospective cohort study. The Lancet Respiratory Medicine: In Press.
