Purpose: To compare the accuracy of magnetic resonance (MR) imaging with that of the current clinical standard of endoscopy and endoscopic biopsy, to determine whether MR imaging depicts subclinical cancers missed at endoscopy and endoscopic biopsy, and to determine whether MR imaging can identify patients without nasopharyngeal carcinoma (NPC) who do not need to undergo invasive sampling biopsy.
Materials and methods: The study protocol was approved by the institutional review board; written informed consent was obtained from all patients. Patients suspected of having NPC underwent MR imaging, endoscopy, and endoscopic biopsy. Endoscopic biopsy targeted the suspected tumor or sampled the endoscopically normal nasopharynx. The final diagnosis was based on results of the endoscopic biopsy or on results of a repeat biopsy directed at the lesion detected at MR imaging. The sensitivity and specificity of the three investigations were compared by using the Fisher exact test.
Results: NPC was present in 77 (31%) of 246 patients and absent in 169 (69%) patients. The combined sensitivity, specificity, and accuracy, respectively, were 100%, 93%, and 95% for MR imaging, 90%, 93%, and 92% for endoscopy, and 95%, 100%, and 98% for endoscopic biopsy. Benign disease was mistaken for NPC in 12 (7%) of 169 patients at MR imaging and in 11 (6%) patients at endoscopy. The sensitivity of MR imaging was significantly higher than that of endoscopy (P = .006) and was similar to that of endoscopic biopsy (P = .120). The specificity of MR imaging was similar to that of endoscopy (P = .120) and was significantly lower than that of endoscopic biopsy (P < .001).
Conclusion: MR imaging is an accurate test for the diagnosis of NPC. MR imaging depicts subclinical cancers missed at endoscopy and endoscopic biopsy and helps identify the majority of patients who do not have NPC and who therefore do not need to undergo invasive sampling biopsies.
© RSNA, 2010.