Background: Positron emission tomography (PET) scanning
is used increasingly to detect and stage lung cancer, but the test
performance characteristics and relationship of PET to patient outcomes
Objective: To determine the test
performance characteristics and relationship of PET scanning stage to
patient outcomes relative to the 1997 International System for the
Staging of Lung Cancer.
Design: Survival analysis
using pathologic staging as the criterion standard for comparison of
survival as predicted by staging by PET and CT.
Setting: University-based hospital.
Patients: All consecutive patients undergoing PET scanning
for the evaluation of possible non-small cell lung cancer (NSCLC)
during a 5-year period.
Main outcome measures:
Long-term survival of patients with NSCLC after staging by PET.
Results: One hundred fifty-two thoracic PET scans were
obtained for the staging of possible NSCLC during a 5-year period. One
hundred twenty-three patients (81%) demonstrated increased
18F-fluorodeoxyglucose uptake. The overall sensitivity and
specificity of PET for detecting malignancy were 95% and 67%,
respectively, compared with 100% and 27% for chest CT. PET and CT had
similar accuracy for staging the overall extent of disease (91% and
89%, respectively). PET stage correlated highly with survival using
either nodal location or overall stage (p = 0.003, p = 0.002), as
did pathologic staging (p = 0.0001, p = 0.0001). CT scan results
did not accurately predict survival (p = 0.608, p = 0.338).
Conclusion: PET scanning is a highly sensitive technologic
advance in detecting and staging of thoracic malignancy and may more
accurately predict the likelihood of long-term survival in patients
with NSCLC than chest CT does.