Main Article Content

Integrated positron emission tomography/computed tomography for evaluation of mediastinal lymph node staging of non-small-cell lung cancer in a tuberculosisendemic area: A 5-year prospective observational study

JA Shaw
EM Irusen
F von Groote-Bidlingmaier
JM Warwick
B Jeremic
R du Toit
CFN Koegelenberg


Background. Integrated positron emission tomography/computed tomography (PET-CT) is a well-validated modality for assessing mediastinal lymph node metastasis in non-small-cell lung cancer (NSCLC), which determines management and predicts survival. Tuberculosis (TB) is known to lead to false-positive PET-CT findings.
Objectives. To assess the diagnostic accuracy of PET-CT in identifying mediastinal lymph node involvement of NSCLC in a high TB-endemic area.
Methods. Patients who underwent both PET-CT and lymph node tissue sampling for the investigation of suspected NSCLC were prospectively included in this observational study. Results were analysed per patient and per lymph node stage. A post-hoc analysis was performed to test the validity of a maximum standardised uptake value (SUVmax)
cut-off for lymph node positivity.
Results. PET-CT had a sensitivity of 92.6%, specificity of 48.6%, positive predictive value of 56.8% and negative predictive value (NPV) of 90.0% in the per-patient analysis. Diagnostic accuracy was 67.2%. Similar values were obtained in the per-lymph node stage analysis. TB was responsible for 21.1% of false-positive results. A SUVmax cut-off of 4.5 yielded an improvement in diagnostic accuracy from 64.0% to 84.7% compared with a cut-off of 2.5, but at the cost of decreasing the NPV from 90.6% to 83.5%.
Conclusion. In a high TB-endemic area, PET-CT remains a valuable method for excluding mediastinal lymph node involvement in NSCLC. Patients with a negative PET-CT may proceed to definitive management without further invasive procedures. However, PET-CT-positive lymph nodes require pathological confirmation, and the possibility of TB must be considered.

Journal Identifiers

eISSN: 2078-5135
print ISSN: 0256-9574