The value of four stage vestibular hydrops grading and asymmetric perilymphatic enhancement in the diagnosis of Menière’s disease on MRI

More than 150 years ago, Prosper Menière described for the first time that vertigo can be elicited from the inner ear. Until then, it was believed vertigo was a disorder of the brain. The pathology named after him is characterized by a clinical triad of fluctuating low-frequency sensorineural hearing loss, tinnitus and/or aural fullness, and vertigo attacks lasting for at least 20 min. In 1995, the American Academy of Otolaryngology - Head and Neck Surgery (AAO HNS) established a specific set of criteria for the diagnosis of Menière’s disease (MD). In this classification, the disease is divided into certain (with postmortem histologic confirmation), definite, probable, and possible categories [1]. In 2015, the Bárány society formulated simplified diagnostic criteria for MD, including only two categories: definite MD and probable MD [2]. Unfortunately, in clinical practice, patients rarely present as textbook cases. Especially in the first years, the symptoms may be transient and incomplete. In about 9% of cases, the time period between the onset of vertigo and hearing loss can last more than 10 years [3]. In the advanced stage of Menière’s disease (MD), the symptoms are permanent. The clinical diagnosis of MD can be complemented with a battery of audiological, vestibular, and electrophysiological tests; however, the lack of a definitive gold standard diagnostic test still complicates the process of diagnosis [3].

Until recently, the role of imaging has been to exclude other pathologies mimicking MD, such as schwannomas, labyrinthitis, intracranial liquor hypotension, and so on. However, at present, the development of delayed gadolinium (Gd)-enhanced high-resolution magnetic resonance imaging (MRI) of the inner ear has enabled us to visualize the histopathological basis of MD, namely the endolymphatic hydrops (EH) as described by Hallpike in 1938 on temporal bone histologic studies [4]. Four-hours delayed intravenous contrast-enhanced three-dimensional fluid attenuated inversion recovery (3D FLAIR) MRI has become a validated technique for demonstration of this EH by allowing dilute contrast to accumulate within the perilymphatic compartment, where the blood-labyrinth barrier is permeable, outlining the impermeable endolymphatic compartment [5].

Nevertheless, it is reported that 10–33% of patients with MD do not have MRI-demonstrable changes of hydrops [6‐8]. This clinical-radiologic discrepancy still reflects an incomplete understanding of the disease process and the need for additional imaging biomarkers of disease activity in MD beyond EH. The purpose of this study is to investigate the reliability and validity of the current MRI diagnostic criteria and grading system for EH [8] on delayed intravenous contrast-enhanced 3D FLAIR MRI in a larger study population and the clinical utility of possible new imaging signs such as a supplementary fourth low-grade vestibular EH and the degree of perilymphatic enhancement (PE).

Of the cohort of 148 patients, 105 patients had unilateral disease resulting in 105 normal ears, 26 undefined (not enough data) ears, 1 probable MD ear, and 78 definite MD ears. Thirteen patients had bilateral disease resulting in 5 undefined (not enough data) ears, 5 probable MD ears, and 16 definite MD ears. Thirty patients were classified with undefined problems. The average age (standard deviation) of the cohort was 56 (12) years (range, 25–84 years), with a female-to-male ratio of 81:67.

For the assessment of the intra- and inter-reader reliability of the readings by the radiologists by means of Cohen’s kappa analysis, all 148 cases were used. The intra-reader agreement was assessed by comparing per ear the first reading with the second reading for each individual reader in a test-retest design. This yields per parameter six kappa values. Table 1 reports the averaged kappa. In general, the kappa is excellent for all the measures (0.87 < kappa < 0.92). The inter-reader agreement was based on the agreement per ear for all three readers, calculated for the first reading and averaged. The agreement between the three readers was also high for every measurement (kappa 0.73–0.83) (Table 2). In parallel, the inter-reader agreement of the clinicians for assessing the diagnosis of MD (no, probable, definite) based on the clinical data yields a kappa (± se) = 0.66 ± 0.05.

For the logistic regression, to assess the ability of the imaging characteristics to predict MD, we used only patients with unilateral definite MD (78 patients). Their contralateral ears were normal and served as controls in this analysis. Patients with bilateral MD were not included, since the notion that the concerned diseased ear has more or less enhancement than the control ear becomes irrelevant. We obtained thereby 78 definite MD ears and 78 control ears. 17 (21%) of these definite MD ears received previous intratympanic therapy (ITT). Analysis was done both for the entire definite MD ears (78) and for the definite MD ears (61) without ITT.

Logistic regression was performed both for the three-stage hydrops grading system according to Baráth and the new four-stage hydrops grading system. To go from the four-stage to the three-stage grading system, grade 0 and grade I are combined to grade 0, grade II becomes grade I, and grade III becomes grade II.

Multivariate logistic regression analysis yielded two characteristics, out of the four investigated, which are needed to identify the definite MD ears versus the control ears, being cochlear PE and vestibular EH. Table 3 demonstrates our data broken down by these two criteria for the entire definite MD ears (78). Using solely the cochlear PE, a very high specificity was obtained of 97.4% (76/78), but the sensitivity was too low, i.e., 65.4% (51/78). This is because the grading “equal” yields a 50/50 MD versus control. So, this variable by itself is insufficient. However, by adding—as calculated by the logistic regression—the vestibular EH to an equal (in comparison with the normal ear) cochlear PE, the sensitivity increased significantly to 79.5% in case of the three-stage grading system. The specificity slightly decreased to 93.63% which is still very satisfactory. The introduction of the additional low-grade vestibular hydrops further increased the sensitivity from 79.5 to 84.6% without losing specificity (92.3%). The other variables (cochlear EH, vestibular PE) did not contribute significantly to this classification.

For MD ears with exclusion of ITT (61 patients), the same tendency in sensitivity and specificity can be found. Cochlear PE remains the most specific parameter with a maximum increase in sensitivity by adding vestibular EH in a four-stage grading system (Table 4).

In a following analysis, we correlated the duration of the MD illness with the different MRI-based variables. Cochlear and vestibular EH correlated significantly with the duration of the MD disease (Spearman rho = 0.23, p = 0.047 and rho = 0.32, p = 0.004, respectively). Also, cochlear PE correlated significantly with the duration of MD (rho = 0.28, p = 0.014).

Our study shows that delayed gadolinium-enhanced 3D-FLAIR is a reliable and accurate diagnostic technique with a high intra- and inter-reader agreement of, respectively, 0.88 and 0.79. These results are in line with previous studies with smaller patient populations [8]. In addition, this is the first study to report on the high intra-reader agreement of the technique.

We investigated the value of four variables in delayed gadolinium-enhanced 3D FLAIR MRI in patients with definite MD: that is cochlear EH, vestibular EH, cochlear PE, and vestibular PE. Multivariate logistic regression analysis yields only two discriminative parameters which are necessary to identify definite MD ears versus control ears: vestibular EH and cochlear PE.