Hepatosplenic fungal disease: diagnostic accuracy and spectrum of appearances on MR imaging.

OBJECTIVE

We describe our 6-year experience in the prospective examination of patients with suspected hepatosplenic fungal disease to show the diagnostic accuracy of MR imaging and the spectrum of appearances on MR images.

METHODS

All patients who underwent MR examination for suspected hepatosplenic fungal disease from January 1990 to January 1997 in three university institutions were included in the study. Patients presented with persistent fever or no response to antibacterial antibiotics. Patients were grouped as acute, subacute treated, and chronic treated, according to the duration of their symptoms. Patients with 2 weeks or fewer of possible infection were acute presentation, patients on antifungal therapy longer than 2 weeks but shorter than 3 months were subacute treated presentation, and patients on antifungal therapy for 3 months or longer or who had completed antifungal therapy and had a history of hepatosplenic fungal disease were chronic treated presentation. MR studies were prospectively interpreted for the presence of hepatosplenic fungal lesions. The appearances of fungal lesions in patients in each category were determined. The sensitivity, specificity, positive predictive value, negative predictive value, and accuracy for detecting lesions in patients with acute presentation were also determined.

RESULTS

Sixty-nine patients were included in the study. MR imaging revealed signs consistent with hepatosplenic fungal lesions in 22 (32%) of 69 patients and no demonstration of hepatosplenic lesions in 47 (68%) of 69 patients. In the 60 patients with acute presentation, MR imaging in 13 patients revealed hepatosplenic lesions that were interpreted as fungal disease. True-positive lesions were present in 11 of these 13 patients. These lesions measured less than 1 cm in diameter and were best shown as well-defined high-signal-intensity foci on T2-weighted images. The remaining two of the 13 patients had false-positive lesions; one was shown to have tuberculosis, and the other had graft-versus-host disease. For acute presentation, MR sensitivity was 100%, specificity was 96%, positive predictive value was 85%, negative predictive value was 100%, and accuracy was 97%. In the five patients with subacute presentation, lesions were present that measured less than 1 cm in diameter and were best shown as mildly hyperintense on T1-weighted images. A perilesional ring nearly void of signal intensity was seen on unenhanced and gadolinium-enhanced T1-weighted images in all five patients. The four patients with chronic healed lesions all had lesions that were 1-3 cm in diameter with irregular, angular polygonal margins. These lesions, which were best shown on images obtained immediately after gadolinium administration, appeared as regions of diminished enhancement with no perilesional changes.

CONCLUSIONS

MR imaging has high diagnostic accuracy for the diagnosis of acute hepatosplenic fungal disease. Patients with acute, subacute treated, and chronic healed presentations may have lesions that can be distinguished by their MR appearances.