Diffusion Analysis of Intracranial Epidermoid, Head and Neck Epidermal Inclusion Cyst, and Temporal Bone Cholesteatoma

DISCUSSION

Major advancements in neuroimaging, particularly with DWI and ADC, have been made since the foundational work of Le Bihan.²³ ADC plays a vital role in measuring water diffusion in the brain, which is critical for evaluating microstructure and identifying characteristics of gray and white matter and CSF.²⁴ ADC values are highest in CSF, reflecting its free-flowing nature; they are higher in the cerebral cortex than in white matter due to higher water and blood flow and intermediate in the basal ganglia and thalamus, indicating their unique microstructures.²⁵ For perspective, free water ADC in an aqueous solution at 37.5°C is approximately 3.0 × 10⁻³ mm²/s but 2–10 times lower in the brain parenchyma.²³ This difference leads to consistently lower diffusivity in normal brain parenchyma compared with CSF and CSF-filled structures like arachnoid cysts.

Epidermoid cysts or epidermoid cysts are ectodermal-derived congenital or acquired encapsulated lesions or nodules lined by stratified squamous epithelium and filled with luminal keratin. Similarly, congenital or acquired cholesteatomas show bland, keratinizing, stratified squamous epithelium and anucleate squares with keratinous debris. The lining and composition of these lesions are used for histopathologic differentiation. The keratin content often defines the imaging characteristics and distinguishes them from the more fluid content of other cysts and cystlike lesions.^{12,18–19,26} In 1990, Tsuruda et al¹² first demonstrated that ADC aids in diagnosing “cystic-appearing” extra-axial brain lesions. IEs, which are cystic-appearing lesions but with more solid content, were found to have restricted diffusion relative to CSF but not when compared with normal parenchyma. Their study revealed that the ADC of arachnoid cysts approximates that of stationary water, whereas the ADC of IEs is similar to brain parenchyma.¹² Contrary to these initial descriptions, a major portion of the imaging literature, including book chapters and radiology reports, continues to describe IEs, TBCs, and ECs as exhibiting diffusion restriction without specifying the comparator tissue, leading to erroneous conclusions and issues with interpretation. Numerous research studies show that IEs demonstrate restricted diffusion.^18,19 However, a few studies have otherwise suggested the absence of such diffusion restriction in IEs.^26,27 For TBCs, most articles report that they exhibit diffusion restriction.^20,21 Multiple studies examining the DWI characteristics of ECs describe these lesions as also exhibiting restricted diffusion.^6,22

IEs, TBCs, and ECs are recognized for their high DWI signal, a key distinguishing characteristic, though the reason behind this in these conditions is still debated in some articles.²⁸ In medical imaging, DWI hyperintensity can result from a combination of restricted water movement and the T2 effects (T2 shinethrough).^8,29 Accordingly, the qualitative evaluation of ADC map images plays a crucial role in differentiating these causes.²⁹ Quantitative ADC values offer a more objective assessment of water diffusion in brain lesions or parenchyma.³⁰ Higher ADC values indicate greater water molecule diffusivity, typical of CSF and vasogenic edema, while lower ADC values suggest restricted water movement, as in ischemic cells or densely cellular tumors.³¹

In our study, all 3 entities showed higher ADC values than the white matter, thalamus, and cerebellum, confirming that their hyperintense DWI signal is not due to restricted diffusion relative to the normal brain. Our findings also demonstrated that the ADC values of all 3 conditions are comparable, underscoring their shared origin and similar characteristics. The term “restricted or reduced diffusion” should be relative and needs comparison with a reference structure. For instance, ADC values of CSF are higher than those of any other brain tissue because all tissue has some degree of diffusion restriction compared with free water and CSF. Consequently, all brain regions will display restricted diffusion if the reference point is water or CSF. Preferably, when one evaluates ADC maps, the focus should be identifying diffusion restriction or reduced diffusivity relative to neighboring tissues. Ideally, the ADC value within the lesion should be statistically significantly lower than that of the surrounding tissues and visually distinguishable.

In this study, because all lesions were confined to the head and neck region, we used the cerebellum, white matter, and thalamus as reference points to compare with ADC values of IE, TBC, and EC. For these lesions to be classified as having restricted diffusion, their ADC values should have been lower than those in other brain parts. Nevertheless, our findings revealed that the ADC values of these lesions were, in fact, statistically higher than those of the brain. We used pMRI and the histogram metrics of IE, TBC, and EC to evaluate the ADC values. To ensure the precision of this software, we also used another software, FireVoxel, to calculate the ADC values of these lesions using the same ROIs as in the initial software calculation. On the basis of our findings, the calculated ADC values from pMRI and FireVoxel are similar. When comparing the means of ADC values, the ICC was 0.997. For the ADC medians, the ICC was even higher at 0.999. The lowest ICC value between these 2 applications, 0.878, was found when we compared the maximum ADC values of these lesions. This result suggests a slightly higher heterogeneity and lower degree of agreement when analyzing lesions with high ADC values.

Moreover, the Bland-Altman plot comparing the ADC values between the 2 software applications demonstrates good agreement and consistency in the measurements. The mean difference of measurements between these 2 applications was only 19.23, indicating that, on average, the ADC values obtained from both software applications are very similar. Additionally, the upper and lower limits of agreement, typically set at an SD of 1.96 of the differences (ranging from −84.83 to 46.37), encompass most the data points (93.75%). This finding further confirms the strong concordance between the 2 methods.

In IE, the hyperintensity observed in DWI was initially considered to result from restricted water movement due to the histologic arrangement of concentric keratin filaments.²⁸ However, some studies have demonstrated that IE may exhibit ADC values similar in appearance to the surrounding brain parenchyma.^12,13 Most of the studies that measured the ADC values of IE had small sample sizes. Chen et al¹⁴ reported 1.197 × 10⁻³ mm²/s in 8 patients with IE. Annet et al¹⁵ reported 1.070 × 10⁻³ mm²/s in 6 patients; and Hakyemez et al¹⁶ reported 1.157 × 10⁻³ mm²/s in 15 patients with IE. Our results align with some of these previous studies, with a mean ADC value of 1.116 × 10⁻³ mm²/s obtained from the DWI of 13 pediatric patients with IEC. In a study involving 15 patients with IEs, Hakyemez et al discovered that the ADC values of ECs were notably lower than those of CSF yet higher than those of deep white matter. They also observed that ECs showed similar intensity in exponential DWI to those of brain parenchyma. This finding suggests that the hyperintensity seen in trace images of ECs is primarily due to an enhanced T2 effect in the tissue rather than a decrease in ADC values.¹⁶

EC are typically found subcutaneously.²² However, studies using MRI to evaluate these lesions are scarce because most are diagnosed clinically.²² Imaging evaluations of these lesions in the head and neck region are even rarer. In a previous study, the ADC value of subcutaneous epidermal cysts has been documented to be 0.81 × 10⁻³ mm²/s from a sample of 14 patients.³² Our mean ADC value of 0.918 × 10⁻³ mm²/s is slightly higher. Suzuki et al³² measured epidermoid cysts located intracranially and subcutaneously and found a statistically significant difference, with intracranial cysts exhibiting a much higher ADC value. In contrast, our results did not show a statistical difference among IE, TBC, and EC. However, the sample size in these studies is limited, and further research is needed to clarify the true cause of this discrepancy.

Like epidermoid cyst, cholesteatoma is a benign lesion caused by the excessive growth of keratinizing squamous epithelium.³³ While high-resolution CT is the primary imaging technique for diagnosing cholesteatoma, the characteristic DWI hyperintensity of TBC has been shown to enhance the accurate diagnosis of these lesions.²⁰ This feature has also been used to rule out recurrence during follow-up.³⁴ The DWI and ADC maps of these lesions are often compared with those of the brain because it is often the main tissue with adequate signal for comparison on trace DWI images. Only 1 previous study by Thiriat et al,³⁵ in 2009, measured ADC values quantitatively in 9 patients with cholesteatoma, with a mean value of 0.903 × 10⁻³ mm²/s, different from the mean ADC value of an abscess 0.415 × 10⁻³ mm²/s. These findings generally align with the mean ADC value of our study of 1.252 × 10⁻³ mm²/s in 21 patients with TBC, suggesting that restricted diffusion (relative to brain) is not the cause of lesion hyperintensity on DWI.

This article is the first to address these 3 types of lesions comprehensively in a pediatric cohort by ADC histogram analysis with rigorous inclusion and exclusion criteria. Its strengths include strict inclusion and exclusion criteria, by including pathology-proved cases only, using 2 distinct software tools to calculate ADC to ensure accuracy, and double-checking ROIs. A potential limitation of our retrospective study is the that the ROIs may overlook microscopic pockets of CSF, calcification, or blood products in the lesions.