Benjamin Huang
Cholesteatomas of the middle ear and mastoid are a major cause of morbidity among patients with chronic otitis media and eustachian tube dysfunction, due primarily to their destructive potential. While surgical resection successfully eradicates disease in the majority of patients, recurrences after tympanomastoidectomy are still quite common, occurring in anywhere from 10–40% of cases.1,2 For this reason, many surgeons commonly perform second-look procedures to diagnose and address recurrent or residual disease, particularly when clinical examination and postoperative CT findings are equivocal.
Over the last decade, developments in magnetic resonance diffusion-weighted imaging pulse sequences have pointed to a shift in the paradigm for postoperative surveillance away from second-look surgery to noninvasive MR imaging. Similar to intracranial epidermoids, most middle ear cholesteatomas show characteristically high signal intensity on DWI, allowing differentiation from other processes such as granulation tissue, fibrosis, and mucoid tissue, which also frequently inhabit the postoperative middle ear. The articles featured in this month’s AJNR Digest focus on the evolution of DWI in the middle ear for detection of recurrent middle ear cholesteatoma.
The first article, by Mukherji, a case report from the May 2002 issue, is included largely for historical interest, as it was among the earliest published reports on the potential utility of DWI (in this case, echo-planar DWI) for detecting cholesteatoma, and the first to demonstrate the ability of DWI to distinguish recurrent cholesteatoma from granulation tissue after mastoidectomy. Only a month later, Fitzek et al3 reported in the Journal of Magnetic Resonance Imaging that DWI was able to prospectively identify 13 of 15 surgically proven cholesteatomas, while excluding cholesteatoma correctly in 10 of 12 patients with acute otitis media and in 100% of healthy volunteers. Over the next several years, a number of studies published in both the radiology and ENT literature showed similar results, with DWI demonstrating sensitivities of 77–86% and a specificity of 100% for diagnosing residual or recurrent cholesteatoma postoperatively.4-6
The second article included in this month’s Digest was among the first to compare DWI with delayed postcontrast imaging (DPI), an alternative MR method for cholesteatoma evaluation first described by Williams et al in 2003.7 In their report, Venail et al found that the two techniques performed comparably for cholesteatomas over 5 mm in size, but that the DWI sequence they utilized (again EPI DWI) was much less sensitive when smaller cholesteatomas were included.
Common to the early studies thus far mentioned was their use of a single-shot EPI DWI sequence, which is used primarily in neuroradiology for stroke imaging. In the middle ear, however, EPI sequences suffer from poor intrinsic spatial resolution and significant susceptibility artifacts at air-bone interfaces, limiting sensitivity for detecting small mural cholesteatomas. Because of these shortcomings, alternative non-EPI DWI pulse sequences soon entered the picture. The next three articles in this month’s Digest nicely illustrated the advantages of these techniques over EPI DWI. Although these sequences vary slightly from manufacturer to manufacturer, non-EPI spin-echo DWI sequences all utilize a 180º refocusing pulse for each measured echo, which significantly reduces susceptibility artifacts at the skull base.
De Foer and colleagues were among the earliest to utilize single-shot turbo spin-echo DWI, which is the technique we use at our own institution (ie, HASTE DWI), while the research by Dremmen et al and Lehmann et al utilized multishot acquisitions—the latter study utilizing a rotating k-space acquisition scheme (PROPELLER). The study by Lehmann and colleagues was also the first, and thus far only, to have directly compared the diagnostic accuracy of a non-EPI DWI sequence with EPI DWI (in this