Widespread ground-glass opacity of the lung in consecutive patients undergoing CT: Does lobular distribution assist diagnosis?

OBJECTIVE: Our purpose was to establish the most frequent diagnoses associated with widespread ground-glass opacity on CT and to determine if the distribution of ground-glass opacity at a lobular level affects the likelihood of a given diagnosis.

MATERIALS AND METHODS: A retrospective search of our department's CT database from January 1998 through December 2000 revealed 153 cases of ground-glass opacity. Patients evaluated using bronchoscopy or open biopsy within 30 days (mean, 6.9 days) of imaging with ground-glass opacity as the predominant high-resolution CT finding were selected. There were 21 men and 16 women with a mean age of 51.4 years. Two chest radiologists, unaware of the clinical diagnoses, independently recorded lobular distributions. Consensus interpretation was used for discrepancies. Primary distributions were recorded as lobular (geographic ground-glass opacity marginated by septal anatomy), centrilobular (ground-glass opacity related to bronchovascular anatomy), or random.

RESULTS: Infectious and other histologic diagnoses fell into four diagnostic groups: atypical pneumonia, chronic infiltrative interstitial disease, acute air-space filling, and drug toxicity. Ground-glass opacity was most frequently associated with acute atypical pneumonia (n = 12, 32%), chronic infiltrative disease (n = 10, 27%), acute air-space filling (n = 6, 16%), and drug toxicity (n = 4, 11%). In five patients, a definitive diagnosis was not established. Ground-glass opacity was most commonly randomly distributed (n = 16, 43%), followed by lobular (n = 15, 41%) and centrilobular (n = 6, 16%) distributions. Distribution did not correlate with diagnostic group.

CONCLUSION: In unselected cases of ground-glass opacity evaluated at a tertiary institution, atypical infection and chronic infiltrative interstitial disease accounted for 59% of diagnoses. Distribution at a lobular level did not differentiate underlying causes.