JCM, Vol. 12, Pages 202: Reliability of Endolymphatic Hydrops Qualitative Assessment in Magnetic Resonance Imaging

1. IntroductionMénière’s disease (MD) is a chronic disorder of the inner ear characterized by spontaneous attacks of vertigo, fluctuating sensorineural hearing loss, tinnitus, and aural fullness [1,2].

The MD diagnosis depends on the patient’s symptoms, medical history, and functional inner ear test results. Since the symptoms of MD are heterogenous and it takes time to develop from monosymptomatic to fully symptomatic disease, diagnosing early stages of MD or atypical forms of the disease is often complicated.

Therefore, scientists have been searching for a distinct characteristic of this disease. Approximately 80 years ago, a hydropic enlargement of the endolymphatic structures (cochlear duct within cochlea, utricle, and saccule of the vestibule) in temporal bone specimens of patients with MD was observed [3]. Even though it was an important discovery, its usefulness was limited, as it was impossible to visualize inner ear structures in vivo. Consequently, the diagnosis of MD has been based on the patient’s symptoms, clinical findings, and functional inner ear test results. The only role of radiology was to exclude other pathologies with symptoms similar to MD [1,2]. With the breakthrough discovery of Nakshima and Naganawa that, in magnetic resonance imaging (MRI) using the delayed post-gadolinium contrast sequences, perilymphatic structures enhance and endolymphatic structures do not, visualizing endolymphatic hydrops (EH) became possible [4]. Since then, many studies have been performed to evaluate if an endolymphatic hydrops is a biomarker of MD or if there are any other imaging features characteristic of this disease. A few imaging methods for the assessment of EH have been described in the literature, as follows: semiquantitative scale [5], qualitative [6], comparison of the size of saccule and utricle [7], and a modification of the qualitative scale by adding an extra-low vestibular hydrops grade [8].Furthermore, a more robust enhancement of the affected inner ear structures has been reported in MD ears as a sign of blood–labyrinth barrier breakdown [8]. Many studies have described the presence of EH and increased perilymphatic enhancement in MD patients. However, a discrepancy in the reported frequency of EH exists, potentially due to the radiological criteria used.

One of the critical features of imaging in medicine is its reliability concerning image reading. It should be reliable regardless of the observers involved. In clinical settings, usually, it is one expert observer who studies a large number of images (cases), rather than multiple observers studying each case and comparing their results to determine the final image result. In the outpatient department setting, an otolaryngologist takes patients’ history and examines and reads the MRI scans bought to the visit by the patient (sometimes without a result written by a radiologist).

The described frequency of endolymphatic hydrops differs among studies, as mentioned above. Some authors described the presence of endolymphatic hydrops also in healthy patients. It is not clear if endolymphatic hydrops is present only in Ménière’s disease ears or might be present also in other inner ear diseases. This is similar to the asymmetry of perilymphatic enhancement. The debate is still going on, as to whether endolymphatic hydrops and increased perilymphatic enhancement can be biomarkers of Ménière’s disease. From a clinical point of view, it is interesting and important to investigate if the differences between studies presented in the literature might result from MRI scan interpretation by different observers.

This study aimed to evaluate the qualitative assessment of inner ear hydrops to verify if it is consistent between observers and easy to learn from a practical clinical point of view for the needs of the daily work of a radiologist and otolaryngologist in clinical and outpatient department settings. In addition, all the MRI-assessed endolymphatic hydrops features were analyzed to calculate the sensitivity and specificity of the method.

In this study, we performed an MRI interpretation of endolymphatic hydrops qualitative assessment of inner ear structures by independent observers and compared the consistency of their MRI interpretation. We wanted to perform this investigation mimicking real clinical and outpatient department settings; that is, each of the observers interpreted the MRI scan once, and their results were compared. For that reason, three observers were engaged in this study, two radiologists and one otorhinolaryngologist. In addition, engaging an otorhinolaryngologist aimed to check if the evaluated method of endolymphatic hydrops assessment was easy to learn after reasonably short training, so that an ENT specialist (otorhinolaryngologist) in the outpatient department could read inner ear MRI scans by themselves during a patient’s visit.

4. DiscussionIn the last few years many MRI studies have been performed on MD but, still, there is a lack of a gold standard for MRI protocols and assessment methods. Different protocols of examinations (both methods of contrast administration and sequences employed on MRI), various scales, and biomarkers are used in radiological diagnosis. Some researchers have used intratympanic [4,11,12,13,14,15,16,17], others intravenous administration of gadolinium-based contrast-agent [6,8,18,19,20,21]. Yamazaki et al. [22] compared two methods of contrast injection and suggested that the intratympanic contrast administration provides better perilymphatic enhancement than intravenous and probably should be used in unilateral MD. However, the intratympanic method is invasive, allows for unilateral examination only, is off-label GBCA use, and requires 24-h waiting time [18,23], so the intravenous method is widely used as more feasible and comfortable for the patient. Furthermore, this method of contrast administration allows the assessment of EH and blood–perilymph barrier permeability.The two principal sequences applied to EH imaging are 3D-FLAIR and 3D-REAL IR, compared visually with the heavily T2 cisternography sequence for anatomical reference. Furthermore, Naganawa et al. developed a series of subtraction sequences such as HYDROPS [23], HYDROPS2 [24], and Hydrops-Mi2 [25], but those techniques are still used as clinical research methods and are not commonly available. Moreover, there are two main EH grading methods, and each evaluates cochlear and vestibular EH separately. Some researchers have used the semiquantitative scale proposed by Nakashima [5], while others have used the qualitative scale described by Barath et al. [6]. In addition, the latter has been recently modified by adding a grade to the VEH evaluation [8]. Each of the factors mentioned above can affect image quality and final assessment.Furthermore, in some studies, the diagnosis of EH was made by one author [26], or if more observers were involved, the diagnosis of EH was made by consensus [4,23,27,28] or there is a lack of information about the observers’ agreement [29,30,31,32]. Consequently, the reported frequency of EH in MD and asymptomatic ears varies in the literature, and debate still exists if the EH, especially CoEH, is a valuable biomarker of MD [7].

Our study aimed to evaluate if the discrepancy in the prevalence of EH in MD might result from the radiological criteria used. We wanted to investigate if the criteria, such as assessment of the presence of EH in the Barath scale and Bernaerts’ modification of this scale and increased perilymphatic enhancement, are repeatable and easy to evaluate. For this purpose, the assessment of MRI examinations of all the ears in our group was made by two experienced radiologists and one otorhinolaryngologist in training.

Our study showed that the frequency of cochlear endolymphatic hydrops in definite MD ears was 76–80%. This type of endolymphatic hydrops was more often detected by experienced observers (Rad 1 and 2) than by the observer in training (Oto). Our results can be compared with other studies based on the same MRI sequence—delayed post-contrast 3D-FLAIR. In Barath et al.’s [6] study, this type of hydrops was described in 86.9% of MD ears, van Steekelenburg et al. [21] reported the presence of CoEH in 85% of definite MD ears, and Pai et al. [30] recognized CoEH in 100% of MD ears. In authors that used other MRI sequences (most often three dimension inversion recovery sequence—3D-IR), the frequency of CoEH in MD patients was 90% in Shi et al. [33], while Ito et al. [28] reported this biomarker in 62% of MD ears, and Yoshida et al. [34] observed CoEH in 87% of MD ears. Suárez Vega et al. [27] compared 3D-FLAIR and 3D-IR and found CoEH in 75% of definite MD ears. In our study, in a group of ears without a diagnosis of MD, CoEH was rare, ranging from 4.5 to 6% for ears with Menieriform symptoms and 2.7–8.2% for asymptomatic contralateral ears in MD patients. Similar results were described by van Steekelenburg et al. [21]. They observed CoEH in 3.1% of ears with Menieriform symptoms and 2% of asymptomatic ears MD patients. Ito et al. [28] observed this in 6.3% of asymptomatic ears.In our study, all the observers detected vestibular endolymphatic hydrops using the Barath scale in 74.7% of definite MD ears. Referring to other studies that used the 3D-FLAIR sequence, our results are similar to those reported by Paskoniene et al. [29], who revealed VEH in 76.4% of MD ears, and lower than results obtained in studies by Barath et al. [6] (92%), van Steekelenburg [21] (89%) and by Pai et al. [30] (86%). Similar results were obtained with other MRI sequences, as follows: Shi et al. [33] described VEH in 88% of MD ears, Ito et al. [28] in 66%, and Yoshida et al. [34] in 94%. Suárez Vega et al. [27] described VEH in 92% of MD ears. In our study, for ears with other symptoms that do not fulfil the criteria for MD but presented some symptoms, VEH was assessed only in 4.5%, and similarly in asymptomatic contralateral ears in 2.7%.In our study, adding one grade to Barath’s classification of vestibular endolymphatic hydrops, as was described by Bernaerts, changed the frequency of VEH identification in definite MD ears to 81–83%, depending on the observer. However, this criterion changed the frequency of recognizing VEH in ears with Menieriform symptoms to 7.5% and 11% in asymptomatic ears. Consequently, the sensitivity increased, but the specificity decreased. Similarly, in Bernaerts et al.’s study [8], this feature increased the sensitivity of VEH from 79.5% to 85%, and in the study of Jasinska et al. [10], from 82% to 92%. It is worth emphasizing the extra low VEH in asymptomatic ears of six patients diagnosed with unilateral definite MD and two with sudden deafness. This finding might be explained by the fact that MD is often a bilateral disease with different onset in each ear, and dilation of the saccule might be an early sign of MD before symptoms appear [10,35,36].In the literature, increased perilymphatic enhancement of the inner ear structures as a sign of blood–perilymph barrier impairment was described as the next probable biomarker of MD [6,8,21,22,37,38,39]. Our study observed it in 59–63% of definite MD patients, in the group of ears with other symptoms in 13%, and only 2.7% in asymptomatic ears. In Bernaerts et al.’s [8] study, 67.9% of MD patients had increased PE on the affected side. Van Steekelenburg et al. [21] observed increased PE in 82.6% of MD, 9.4% of Menieriform symptoms, and 3.4% of asymptomatic ears. However, this parameter should probably be combined with the EH because it may also be present in other ear diseases, such as sudden sensorineural hearing loss or vestibular neuritis [21,40].Ménière’s disease is a complex disease with heterogeneous symptoms. Moreover, it is chronic and gradually progresses from monosymptomatic to fully symptomatic. The actual diagnostic criteria (AAO–HNS) for diagnosing MD include patients with advanced-stage disease. Therefore, the diagnosis of this disease is often difficult. According to the literature, in 20% of patients, it takes more than five years to diagnose MD [41]. Additionally, it was discovered that the presence of EH may precede symptoms in MD patients [8,41], can progress with the disease duration, and is correlated with clinical symptoms [10,42,43,44,45,46,47,48]. Therefore, it can serve as a method of early detection and support the diagnosis in clinically atypical cases, help to choose proper treatment, and potentially monitor therapeutic effects [39,47,49,50].In our study, when assessing the radiological features, the highest differences between the observers occurred for the evaluation of cochlear endolymphatic hydrops. A significant difference was found between all pairs of observers, with a more considerable difference between the pairs of radiologists and otorhinolaryngologist. Earlier studies suggest that the assessment of CoEH on delayed post-contrast 3D-FLAIR sequence might be interpreted variably [27,51,52]. Our study confirmed this finding; however, it also showed 96% interobserver agreement for the differentiation of hydropic and normal ears. Consequently, it showed that it is much easier to classify ears as normal or hydropic than to choose the proper grade of endolymphatic hydrops.The assessment of vestibular endolymphatic hydrops in both Barath and Bernaerts scales, comparing both radiologists, was precisely the same. The only differences were present between radiologists and otorhinolaryngologist when assessing the VEH on the Barath scale, not on the Bernaerts scale. However, like with CoEH, the main non-accordance was in the proper grading of hydropic ears, not in assessing the presence of any hydrops. All observers differentiated normal and hydropic ears almost the same way. When comparing interobserver agreement, our results slightly differ from the literature. Most researchers showed a higher concordance coefficient for CoEH than VEH recognition, as follows: Barath et al. [6], 0.97 for CoEH and 0.94 for VEH grading, van Steekelenburg et al. [21], 0.93 for CoEH and 0.92 for VEH, Bernaerts et al. [8], 0.83 for CoEH and 0.81 for VEH. However, Suárez Vega et al. [27] reported that degree of concordance was higher for VEH (0.66) than for CoEH (0.39) using the 3D-FLAIR sequence. Moreover, they also found that the concordance coefficient was higher (0.82) when diagnosing any EH than scoring it. The latter is similar to our findings.As for the perilymphatic enhancement of the inner ear structures, there were no significant differences between the three observers and the pairs of observers for this parameter, which is in line with the literature [8,21].

In our study, three observers evaluated MRI scans independently to evaluate if the assessment method was repeatable. By engaging a less experienced MRI observer (otorhinolaryngologist) who was trained with a few cases of MRI scans of inner ears, the study aimed to evaluate if the method was easy to learn. It showed that the assessment of vestibular endolymphatic hydrops is easy to learn and repeatable. For experienced radiologists, the readings for 220 ears were the same, whereas, for the otolaryngologist, the number of different observations was very low. Typically, the saccule and utricle are easily identified in the vestibule’s inferior part as two “black dots” surrounded by a bright (enhancing) rim of perilymph. Furthermore, the saccule is smaller than the utricle. Even for an untrained observer, it is not difficult to find that this configuration is changed. That is probably the reason for such high agreement in assessing this parameter.

Similarly, with the perilymphatic enhancement qualitative assessment, it is not difficult to visually compare two sides and find an asymmetry.

When it comes to cochlear endolymphatic hydrops assessment using a 3D-FLAIR MRI sequence, this is more complicated. In this sequence, differentiating endolymphatic structures from surrounding bone is impossible because both have low signals [27]. First, changes in the contours of the cochlea should be found and compared with the FIESTA sequence as an anatomic reference. Second, how much outlines are changed should be evaluated, and then CoEH should be graded. Evaluation of this parameter is more complicated, which is probably why it is so difficult to properly stage EH, even for experienced users. According to existing literature, the solution to this problem might be using a post-contrast 3D-IR inversion recovery sequence [23,24,27,53]; however, this sequence is not widely available, takes longer, and is more prone to motion artifacts. The repeatability of CoEH assessment using different sequences needs to be confirmed in further studies.

留言 (0)

沒有登入
gif