Four‐hour‐delayed 3D‐FLAIR MRIs in patients with acute unilateral peripheral vestibulopathy

Abstract Objective Conventionally, MRI aids in differentiating acute unilateral peripheral vestibulopathy/vestibular neuritis (AUPV/VN) from mimickers. Meanwhile, the diagnostic utility of MRIs dedicated to the inner ear remains to be elucidated for diagnosing AUPV/VN. Methods We prospectively recruited 53 patients with AUPV/VN (mean age ± SD = 60 ± 15 years, 29 men). Initial MRIs were performed with a standard protocol, and an additional axial 3D‐fluid‐attenuated inversion recovery (3D‐FLAIR) sequence was obtained 4 h after intravenous injection of gadoterate meglumine. Abnormal enhancement was defined as a signal intensity that exceeded the mean + 2SD value on the healthy side. The findings of neurotologic evaluation and MRIs were compared. Results Overall, the inter‐rater agreement for gadolinium enhancement was 0.886 (Cohen's kappa coefficient). Enhancement was observed in 26 patients (49%), most frequently in the vestibule (n = 20), followed by the anterior (n = 12), horizontal (HC, n = 8), posterior canal (n = 5), and superior (n = 3) and inferior (n = 1) vestibular nerves. In multivariable logistic regression analysis, the enhancement was associated with decreased HC gain in video head‐impulse tests (p = 0.036), increased interaural difference in ocular vestibular‐evoked myogenic potentials (p = 0.001), and a longer onset‐to‐MRI time span (p = 0.024). The sensitivity and specificity were 92.3% and 81.5%, respectively, with an area under the curve of 0.90 for predicting gadolinium enhancement. Interpretation Robust gadolinium enhancement was observed on 4‐hour‐delayed 3D‐FLAIR images in nearly half of the patients with AUPV/VN, with a good correlation with the results of neurotologic evaluation. The positivity may be determined by the extent of vestibular deficit, timing of imaging acquisition, and possibly by the underlying etiology causing AUPV/VN. MRIs may aid in delineating the involved structures in AUPV/VN.


Introduction
Vestibular neuritis (VN) refers to a syndrome of acute unilateral peripheral vestibulopathy (AUPV) due to inflammation of the peripheral vestibular organ. 1 The diagnosis is centered on clinical features and neurotologic findings when other causes are readily excluded. 2Patients invariably suffer from spontaneous vertigo, nausea/vomiting, and unsteadiness.Along with these hallmark symptoms, video-head impulse tests (video-HITs) and caloric tests can identify peripheral vestibular hypofunction in the presence of contralesionally beating horizontaltorsional nystagmus. 2In this context, imaging is mostly adopted for differentiation of other causes of vestibulopathy, especially when neurotologic findings are atypical or inconspicuous (i.e., pseudo-VN).Most importantly, MRIs can differentiate AUPV/VN from posterior circulation strokes, which are potentially dangerous mimickers.
The diagnostic yield of MRIs in AUPV/VN has been sought through earlier studies, 3 but pathologic finding was not found for the vestibular nerve, ganglion, or labyrinth. 3Meanwhile, recent advances in imaging technology have improved the identification of involved structures in various vestibular and auditory disorders.For instance, MRIs can visualize endolymphatic hydrops (EH), a possible radiologically relevant finding of Meniere disease (MD). 4Likewise, gadolinium enhancement can be observed in AUPV/VN either at the primary vestibular afferents 5,6 or in the inner ear. 7However, the relationship between neurotologic findings and gadolinium enhancement remains to be elucidated in AUPV/VN.Thus, we investigated the diagnostic yield of 4-hour-delayed inner ear imaging in patients with AUPV/VN and determined the association between imaging and neurotological parameters.

Patients
We prospectively recruited 66 patients with AUPV/VN who underwent neurotologic evaluation and inner ear MRI between September 2019 and January 2023 at Korea University Medical Center.Among them, patients with a history of vestibulopathy that may affect the results of HITs (n = 8, three with vestibular migraine, three with chronic otitis media, one with MD, and one with posterior fossa large meningioma) were excluded.Additionally, we excluded patients who showed EH in either the affected or healthy ear (n = 5).Even though they had no clinical symptoms compatible with MD, thereby we could rule out the potential influence on the results of neurotologic evaluations.AUPV/VN was diagnosed based on the diagnostic criteria proposed by the Barany Society. 2
The images were independently reviewed by the neuroradiologists who were blinded to the lesion side and clinical characteristics.The signal intensity of each neural structure was calculated using the average value determined independently by neuroradiologists.Later, the signal intensity values were reviewed by the neurologist (S.U.L).And then, abnormal enhancement was defined as when the normalized signal intensities of the affected side exceeded the mean + 2 standard deviation (SD) of the signal intensity of each neural structure derived from the healthy side (upper normal limit <1.49 and <1.62 for the superior and inferior vestibular nerves, respectively; <0.69 for the vestibule; <0.40, <0.61, and <0.63 for the HC, AC, and PC, respectively).

Statistical analysis
Nominal or independent variables were compared using the chi-squared or Fisher's exact test, and continuous or independent variables were compared using the Student's t-test and Mann-Whitney U-test.For logistic regression, all variables with a p < 0.05 using an age-and sex-matched univariate analysis were included in the multivariable analysis.Variables with p < 0.05 were determined to be significant in the multivariable analysis.
Receiver operating characteristic (ROC) curve analysis was conducted to determine the sensitivity and specificity of continuous variables.Sensitivity and specificity were presented as cut-off values that maximized the Youden Index (sensitivity + [specificity -1]) and the area under the curve (AUC).
Statistical analyses were performed using the R software package (version 3.4.0;http://www.r-project.org), and the significance level was set at two-tailed p < 0.05.

Clinical characteristics and neurotologic findings
The detailed clinical profiles of the patients are shown in Table 1.Fifty-three patients were included in the analyses (mean age AE SD = 60 AE 15 years, 29 men).Intravenous or oral corticosteroid treatment was not attempted in any of the patients.Three patients (3 out of 53, 6%) had inferior AUPV/VN.All patients presented with acute spontaneous dizziness/vertigo (n = 53), associated with nausea or vomiting (n = 47).None of the patients reported new-onset headaches, tinnitus, ear fullness, or hearing loss at presentation.

ROC analysis
The sensitivity was 92.3%, and specificity was 81.5% at cutoff values of 0.95 for HC gain, 100% for IAD of oVEMP, and 4 days for the time span from symptom onset to MRI with an AUC of 0.90 (95% CI = 0.82-0.98;Fig. 4).

Discussion
The main findings of our study can be summarized as follows: 1) Abnormal enhancement was observed in 49% of patients with AUPV/VN.2) Gadolinium enhancement    The interaural difference of cervical and ocular vestibular-evoked myogenic potential and VOR gain for each canal were calculated separately for multivariable logistic regression due to multicollinearity.was observed primarily in the vestibule, followed by the AC and HC. 3) Gadolinium enhancement was associated with the degree of vestibular deficit; a robust enhancement was observed in each corresponding neural structure, showing a profound HC gain decrement and a greater IAD of oVEMP.4) Together, HC gain <0.90, absent oVEMP response, and time span from symptom onset to MRI more than 4 days predicted gadolinium enhancement with a sensitivity of 92.3%, specificity of 81.5%, and AUC of 0.90.

Prior reports on MRIs as a diagnostic test for AUPV/VN
Conventionally, the diagnostic yield of brain MRI is limited to differentiating AUPV/VN from central lesions.
Bell's palsy has a similar mechanism to AUPV/VN, but it exhibits frequent gadolinium enhancement in the facial nerve, 16 whereas enhancement has scarcely been observed in AUPV/VN. 5This may be due to the small diameter of the vasculature distributed along the vestibular nerve required to show remarkable enhancement from disrupted blood-labyrinthine barrier (BLB). 3Additionally, the vestibular damage may be subtle and fail to reach the threshold at which it becomes visible. 38][19] Our result suggests that AUPV/VN cannot visualized when the MRIs were conducted too early.Rather, MRIs can show positivity usually after 4 days after symptom onset when AUPV/ VN is suspected.It is widely accepted that the timing of the imaging acquisition is important.Yet, the dosage of gadolinium is at odds.The standard dose and higher dose (0.1 vs. 0.15 vs. 0.2 mmol/kg body weight) allowed for the same differentiating power for visualization of endolymphatic spaces. 19By contrast, other researchers reported gadolinium might be a factor for positivity, 5,7,20 as a higher dose (≥0.3 mmol/kg of body weight) has a higher chance of detecting enhancement. 5Our study showed that inner ear structures can be readily visualized by using a standard dose of 0.1 mmol/kg of gadolinium.Similarly, gadolinium enhancement can be found in up to 60-69% of AUPV/VN using delayed MR imaging when other factors are adequately controlled. 7,12The discrepant results observed across studies may be attributed to differences in the dosage of gadolinium or MRI protocols.Additionally, the method of signal normalization (such as signalto-noise ratio vs. adopting reference ROI) or the definition of ROIs may be another factor to be considered.

Distribution of the gadolinium-enhancement and anatomical site of damage in AUPV/VN
Questions are increasingly being raised regarding the precise anatomical lesions in AUPV/VN.It is still unclear whether the vestibular nerve (i.e., neuritis) or vestibular organ (intralabyrinthine) causes AUPV/VN. 21Some electrophysiological studies may provide clues.For instance, VEMP responses can be impaired by galvanic and sound stimuli in patients with AUPV/VN.Among these, 27% of patients show abnormal VEMP only on sound stimuli, indicating that a subset of patients may have a selective labyrinthine lesion. 22However, prior MRI studies have shown contradictory results: gadolinium enhancement can be found both at the vestibular nerve and the labyrinth. 7,12Notably, gadolinium enhancement can be confined to the labyrinth in some patients. 23We observed a robust association between neurotological test results and gadolinium enhancement, mostly in the labyrinth.Based on our findings, we assume that both the inner ear and nerves can be involved, but the predilection site for damage may be the labyrinth.Therefore, AUPV/VN may be a limited form of labyrinthine inflammation that spares the cochlea.
Gadolinium enhancement was readily found in AC and HC than in PC.The canal of the SVN is seven times longer and has more spiculae than the canal of the IVN. 24hus, the SVN and its supplying arteriole travel through a relatively narrower passage compared to IVN.In addition, PC may be anatomically resilient to inflammation, as it has double innervation of the two nerves, namely the posterior ampullary nerve and the accessory posterior ampullary nerve, running through a separate bony canal. 25,26The wide range of kappa values on the vestibular nerve (0.542 for SVN and 1.000 for IVN) may result from the rarity of enhancement in these structures.It also implies that the inner ear would be suitable for detecting gadolinium enhancement in AUPV/VN.However, our results should be interpreted with caution, given that AUPV/VN is clinically diagnosed, and various etiologies are inferred as the underlying mechanism.For example, the viral etiology is hypothesized by the reactivation of type 1 herpes simplex virus type I (HSV-1) in the vestibular ganglia. 25,27Temporal bone studies mostly indicate that microvascular occlusion and hypoperfusion of the vestibular nerve or organ is an alternative hypothesis. 28Immune-mediated dysregulation may also play a role in the pathogenesis. 12,29,30The damaged site may vary depending on the etiology of AUPV/VN, and a single unifying theory of nerve versus vestibular organ can be misleading.This various etiology causing AUPV/VN may be a factor for determining positivity on MRI.Additionally, 4-hour-delayed 3D FLAIR may be more suitable for anatomic discrimination of the inner ear than the nerve, precluding a comparative discussion of the nerve and vestibular organ. 31Furthermore, gadolinium enhancement tended to be preferentially found in the HC and AC, omitting the PC.This selective involvement supports the hypothesis that the inflammatory process originates from the vestibular nerve or ganglion. 32

Factors as determinants of gadolinium enhancement
Gadolinium enhancement results from a breakdown of either the blood-nerve barrier (BNB) or BLB, or hyperemia of the perineural arteriovenous structure. 33Prior MRI studies have pointed out that positivity is associated with persistent spontaneous nystagmus 7 or the timing of imaging acquisition (onset-to-MRI). 23This implies that the extent of vestibular damage and the subsequent compensation may be a factor for positivity.However, possible covariates were not statistically controlled for in previous studies. 7,23Our findings suggest that the MRI can yield negative results, if the damage is marginal to the threshold to be visible.In addition, current imaging techniques cannot discriminate ischemic injury confined to the labyrinth from inflammation.Future studies on a larger scale can clarify other factors that may influence enhancement in AUPV/VN, other than the extent of vestibular injury or underlying etiology causing AUPV/ VN. 12,29,30,34 Our results align with those of previous reports, showing that the onset-to-MRI time can be a factor for gadolinium enhancement.In fact, temporal changes in BNB and BLB disruption have not been fully delineated.The bimodal temporal profile of the blood-brain barrier (BBB) may provide some insight.6][37] The endothelial caveolae increase in the acute early opening, whereas the tight junction remodeling begins later at delayed time points of ischemic reperfusion injury, exhibiting biphasic temporal changes. 38Our results imply that the BNB or BLB breakdown may be analogous to the BBB breakdown.

BNB and BLB: Anatomical barriers that demarcate the blood-nerve and bloodlabyrinth barriers
The BNB is a physiological boundary between the peripheral nerve axons and the bloodstream that prevents the transfer of substances from the plasma to the nerve fibers. 39The BNB normally maintains and guarantees axonal function in the peripheral nerves.Similarly, the BLB outlines the impermeable endolymphatic compartment and is selectively permeable to solutes within the endothelial cells of the endoneurial continuous capillaries and in the internal layers of the perineurium. 40Endothelial cells from endoneurial vessels are joined tightly together by specialized junctions, minimizing capillary permeability.Adjacent to the capillaries, the perineurial sheath comprises large numbers of tight junctions between perineurial cells, isolating each fascicle from the interfascicular and epineurial environments.When a nerve or labyrinth is damaged, it swells to disrupt the BNB and BLB, which is documented as gadolinium enhancement on MRIs.

Clinical implication and caveats for future study
Our results should be interpreted with caution, as the diagnostic yield of routine screening of inner ear MRIs in AUPV/VN may be limited.Only nearly half of patients showed positivity, and clinical diagnosis of VN/AUPV is already secured based on clinical and neurotological findings.However, our findings suggest that MRI may aid in AUPV/VN diagnosis, potentially expanding the current clinical spectrum of AUPV/VN.Given the various etiologies that should be considered for causing AUPV/ VN, 12,25,27,28 imaging may aid in stratifying the subtypes according to various etiologies and anatomical sites involved.
Our study has some limitations.Most of all, the lack of a control group limits the validity of our findings.By only comparing the (clinically) affected side with the healthy side, potential physiological side differences in contrast agent uptake pose a relevant limitation.In addition, gadolinium enhancement is not pathognomonic to AUPV/VN and can be found in several other diseases involving the vestibular nerve or labyrinth, such as MD, 41 labyrinthitis, 42 vestibular schwannoma, 43 and labyrinthine ischemia/hemorrhage. 17,44 Similarly, gadolinium enhancement of other cranial nerves (e.g., the oculomotor or abducens nerves) does not correlate with the clinical manifestation or underlying etiology. 45,466][37] Gadolinium enhancement, per se, should not be recognized as a pathognomonic imaging finding of AUPV/VN, and it should be interpreted in consideration of clinical manifestations and neurotologic findings.Furthermore, vestibular impairment can evolve over time.For instance, canal paresis evolves over time, and cases with negative initial results can show canal paresis later. 47Thus, the MRIs in AUPV/VN interaction between the nature of vestibular damage and the timing of image acquisition may be a factor to be considered for positivity.

Figure 1 .
Figure1.Quantitative evaluation of a degree of the perilymphatic enhancement.Six freehand round or polygonal regions of interests (ROIs) were set manually in AC (3.22-3.68mm 2 ), HC (6.90-9.39mm 2 ), PC (6.90-9.39mm 2 ), canalicular segments of the superior and inferior vestibular nerves (4.60 mm 2 ), and vestibule (20.40 mm 2 ) on 3D-FLAIR sequence to include as much of the perilymph as possible.An additional 50 mm 2 circular ROI was drawn in the medulla at the same plane as the vestibule.The mean signal intensity was recorded for each ROI.The signal intensity ratios of the vestibular nerves and inner ear structure to that of the signal intensity of the medulla were calculated to avoid bias from patient-related artifacts.AC, anterior canal; HC, horizontal canal; PC, posterior canal.

Figure 3 .
Figure 3. Representative MRIs in patients with acute unilateral peripheral vestibulopathy/vestibular neuritis.(A) No signal changes are present on baseline 3D-FLAIR image in a patient with right acute unilateral peripheral vestibulopathy/vestibular neuritis (upper row).Contrastingly, a robust gadolinium enhancement is observed in the horizontal canal (arrow) and vestibule on the 4-hour-delayed 3D-FLAIR image (red rectangle; lower row).(B) gadolinium enhancement is found in the vestibule (arrowhead) on the 4-hour-delayed 3D-FLAIR image (lower row) while showing no discernible changes on baseline 3D-FLAIR image (upper row).

Figure 4 .
Figure 4. Receiver operating characteristic analysis for VOR gain, IAD of oVEMP, and onset-to-MRI for separating gadolinium enhancement from no enhancement on MRIs.The sensitivity was 92.3%, and specificity was 81.5% at cutoff values of 0.95 for HC gain, 100% for IAD of oVEMP, and 4 days for a time span from symptom onset to MRI with an AUC of 0.90 [95% CI = 0.82-0.98].AUC, area under the curve; HC, horizontal canal; IAD, interaural difference; oVEMP, ocular vestibular-evoked myogenic potentials; VOR, vestibulo-ocular reflex.

ª
2024 The Author(s).Annals of Clinical and Translational Neurology published by Wiley Periodicals LLC on behalf of American Neurological Association.2037 K.-T.Kim et al.

Table 1 .
Comparison of the neurotologic findings with respect to gadolinium enhancement.

Table 2 .
Prediction of gadolinium enhancement in overall inner ear and vestibular nerves in patients with AUPV/VN a a

Table 3 .
Prediction of gadolinium enhancement in overall inner ear and vestibular nerves in patients with AUPV/VN.