Objective radiographic density measurements of sinus opacities are not strong predictors of noninvasive fungal disease.

BACKGROUND: High-density paranasal sinus opacities are often deemed consistent with fungal elements. No studies of objective quantitative radiographic density measures have been performed to support this assertion.
METHODS: A consecutive series of 120 patients with chronic rhinosinusitis who underwent maxillary antrostomy with microbiological evaluation of contents within 60 days of sinus computed tomography scanning was investigated. Radiographic density characteristics of opacities in cultured maxillary sinuses (minimum, maximum, average, and standard deviation of Hounsfield units [HUstd]) were recorded. Receiver operator characteristic (ROC) curves were used to analyze the accuracy of radiographic characteristics in predicting fungal opacities.
RESULTS: Of 133 maxillary sinus opacities, 22 were ultimately consistent with noninvasive fungal disease: 11 allergic fungal rhinosinusitis and 11 fungal balls. Fungal balls had higher-density components and were more heterogeneous and allergic fungal mucin was generally more radiodense. These findings were reflected by statistically significant ROC curves for maximum HU (p = 0.019) and HUstd (p = 0.023) for fungal balls and for average HU (p = 0.002) for allergic fungal mucin. A maximum HU cutoff of 334.0 detected fungal balls with 90.9% sensitivity and 72.7% specificity. An average HU cutoff of 42.9 HU detected allergic fungal mucin with 100% sensitivity and 46.3% specificity, although specificity improved to 73.2% with inclusion of nasal polyposis as a second requirement.
CONCLUSION: Higher average HU more accurately predicts allergic fungal mucin whereas heterogeneity/high-density components more accurately predict fungal balls. No objective radiographic density measure, in isolation, is both sensitive and specific in predicting noninvasive fungal sinusitis.