Imaging of the saccule for the diagnosis of endolymphatic hydrops in Meniere disease, using a three-dimensional T2-weighted steady state free precession sequence: accurate, fast, and without contrast material intravenous injection

Meniere disease is a chronic disease of unknown aetiology, causing vertigo, hearing loss, sensation of clogged ear, and tinnitus. The diagnosis is based on a combination of the patient’s symptoms, the results of the clinical examination, and functional tests [1]. Knowledge of normal anatomy of the saccule is important in order to identify size and shape modifications in pathological conditions. Meniere disease symptoms are known to be correlated to the degree of endolymphatic hydrops, which is the distension of endolymph-filled structures (Fig. 1a) [2, 3]. The endolymphatic hydrops can be easily studied with 3 T magnetic resonance imaging (MRI). However, all published studies used delayed images acquired 4 h after intravenous injection of a Gd-chelate or 24 h after intratympanic injection of a Gd-chelate [4–10].

Our aim was to compare high-resolution T2-weighted images of the saccule in normal subjects with histological sections from cadavers and to identify its changes in Meniere disease compared to normal subjects.

The Ethics Committee of our institution approved the study (Internal Research Project, authorisation number HUS-PRI 5012). Patients and healthy volunteers gave their written informed consent to participate in the study. The temporal bones of premature foetuses born after 7 months of gestation were provided by the Anatomical Institute of the Strasbourg University.

On the histological sections, the saccule occupied the inferior, medial, and anterior part of the vestibule, facing the oval window, located under the utricle and medially to the perilymph. It had the oval shape of a rugby ball, lying in a coronal plane through the anterior vestibule at the level of the lateral and anterior ampullas, as shown in Fig. 2. It followed the craniocaudal axis of the vestibule and its orientation was close to that of the anterior semicircular canal. Its medial wall was made of the saccular macula, which was not visible, as it blended in the osseous adjacent vestibule. On the histological sections in a coronal plane, the measurements were 1.6 mm for the saccular height and 0.9 mm for the saccular width.

The images of 60 healthy volunteers were analysed as four of them were excluded due to motion artefacts. In these healthy subjects, the saccule was always visible (100% of the cases) on the reference coronal section with excellent interobserver correlation (κ = 1.0). It always presented with the ovoid shape of a rugby ball, as observed on the equivalent histological sections, occupying the anterior and medial quadrant of the vestibule (under the linear hypointense signal of the utricular macula), as shown in Fig. 3. Its lateral wall was distant from the footplate, separated by the perilymphatic cistern and its craniocaudal axis appeared to be nearly in the same direction as the anterior semicircular canal. The vertical saccular macula, located close to the medial wall of the bony vestibule, was not visible on MRI, as it blended with the hypointense signal of the surrounding bony vestibule on the T2-weighted images. In volunteers, the mean height of the saccule was 1.40 ± 0.10 mm (mean ± SD) and the maximal height was 1.6 mm, with a range from 1.10 to 1.60 mm (Fig. 5); the mean normal width of the saccule was 1.20 ± 0.10 mm and the maximal width was 1.40 mm. The height measurements in volunteers were similar for both observers, with an ICC of 0.89 on the right side and 0.68 on the left side; so were the width measurements, with an ICC of 0.86 on the right side and 0.72 on the left side. The intraobserver agreement on volunteers was also good with an ICC of 0.87 for the height measurement and 0.77 for the width.

Sixty-two patients with Meniere disease were analysed as 2 patients were excluded because of motion artefacts (mostly related to swallowing). Fifty-two of the 62 patients with Meniere disease (84%) presented with signs of saccular hydrops. The measurement on the coronal image was carried out easily, with an average of 2 min and 30 s for each patient. The saccular height was increased (greater than 1.6 mm) in 45 out of 52 cases (86%) and gave the saccule an elongated oval shape (Fig. 4). The mean saccular height in Meniere disease was 1.69 mm ± 0.24 (mean ± SD) with a range from 1.20 to 2.30 mm, which was significantly different from the control group (1.40 ± 0.10 mm, range from 1.10 to 1.60 mm, p < 0.001) or the measures on the histological sections (1.60 mm) (Figs. 2, 3, 4 and 5). The interobserver agreement for the evaluation of the hydrops in Meniere disease patients was good (ICC 0.69 on the right side and 0.84 on the left) and the intraobserver agreement was good with an ICC of 0.87.

The saccular width was also increased (greater than 1.40 mm) in 23 of 52 cases (44%) (Fig. 5a, b). The width of the saccule in the Meniere group was 1.30 ± 0.21 mm (range 1.10–1.70 mm), which was significantly different from the control group (1.20 ± 0.13 mm, range 0.90–1.40 mm, p < 0.001) or the histological sections (0.9 mm). A round shape appearance of the saccule with an isolated increase of the saccular width (4% of the cases, n = 2) was also considered as pathological (Fig. 5b), as the normal saccule always presented with an oval shape in the control group. The interobserver agreement was very good (ICC 0.72 for the right side and 0.94 for the left side). The intraobserver agreement was good, with an ICC of 0.86.

Bilateral saccular involvement was found in 19 of 52 cases (35%) and the vast majority had no clinical signs on the contralateral side. Furthermore, in cases of bilateral involvement; the clinically affected side showed no greater distension compared to the contralateral side.

In 8% of the cases (n = 5), the saccule on the side of the clinical symptoms was poorly defined (Fig. 6), while the contralateral saccule was clearly visible, without any artefact interfering with its interpretation. As the saccule was visible on both sides in the control group, a non-visualisation of the saccule on the clinical pathological side, observed only in Meniere disease patients, was considered as pathological.

The Fisher exact test did not find a correlation between the degree of saccular dilation and the hearing loss, the vestibular dysfunction, duration of the disease, and the delay of completion of the MRI.

The essential point of our method is the fast measurement of the saccule, enabling us to visualise a saccular hydrops in 84% of patients with Meniere disease, in agreement with previously published data [4–13]. In Meniere disease, the saccule appeared elongated with an increased average height, round shape, or in few cases not visualised on the side of the clinical symptoms, compared to normally shaped and sized saccules in the control group and on histological sections, where it was always visible and oval-shaped. The non-visualisation of the saccule in Meniere disease has a possible histopathological explanation. In fact, Kimura and Schucknecht [14] and Schucknecht and Rüther [15] described ruptures and fistulas of the membranous labyrinth on temporal bone sessions, which could cause this non-visualisation.

The endolymphatic hydrops associated with Meniere disease was firstly described by Hallpike and Cairns in 1938 [16] and confirmed by Altman and Fowler in 1965 [17]. In a more recent study on temporal bones, Rauch et al. [18] described the involvement of the saccule in almost all cases of endolymphatic hydrops. According to Horner [19], the degree of saccular dilation appeared correlated to the degree of membranous distortion towards the footplate. This was confirmed by Morita et al. [20], who described a saccular hydrops on the pathological side in Meniere disease patients, while the cochlear and utricular hydrops were less frequent.

To our knowledge, our study is the first one to analyse both size and morphology of the saccule in patients with Meniere disease without intravenous or intratympanic contrast material injection. Most of the authors used fluid attenuated inversion recovery (FLAIR) or three-dimensional inversion-recovery turbo spin-echo sequences, acquired 4 h after intravenous Gd-based contrast material injection [5, 6, 20–24]. Naganawa et al. [8] have been able to individualise the endolymph from the perilymph using a three-dimensional inversion-recovery turbo spin-echo sequence after intravenous injection of a Gd-based contrast material, but they were unable to separate the utricle from the saccule, contrary to Attye et al. [5]. One of the downsides of these methods using contrast injection is the time needed for the penetration of the Gd-based contrast material in the perilymph (approximately 4 h) while the required injection also exposes the patient to potential Gd-related risks. The main advantage of the method proposed by Naganawa et al. [8] is given by the morphological visualisation of the anterior membranous labyrinth, although Attye et al. [5] gave examples of dilation of the cochlear duct in their normal population; thus, the significance of a dilated cochlear duct is uncertain. On the contrary, in our study the saccule only appeared dilated in Meniere disease and no pathological dilation of the saccule was observed in the control group.

Other authors, such as Seo et al. [25], Sun et al. [9], or Fiorino et al. [10], used FLAIR images acquired 24 h after intratympanic gadolinium injection. Seo et al. [25], using MRI images acquired 24 h after intratympanic injection, found a hydrops in 81% of their patients (21/26) with saccular involvement in the majority of the cases (19 patients, 69%).

Our method enables a fast diagnosis of Meniere disease in less than 8 minutes. Its other major strength is that it does not require intravenous or intratympanic injection of a Gd-based contrast material. The major downside of the T2-weighted gradient-echo SSFP sequence (FIESTA-C) may be its sensitivity to motion artefacts. In our study, it involved 6/128 subjects (5%), mostly related to swallowing. When compared to the literature, all sequences used for the imaging of the membranous labyrinth are quite long: 9 minutes in Attye et al. [5] and 14 min in Sun et al. [9], compared with 7 min and 49 s in our study.

Notably, three-dimensional T2-weighted turbo-spin-echo sequences (such as SPACE, CUBE, VISTA) would also work for the measurement of the saccule as long as the special resolution is sub-millimetric, but unlike gradient-echo sequences, spin-echo sequences are not sensitive to fluid composition [26, 27]. Thus, using gradient-echo SSFP sequences, such as FIESTA-C in our study, true fast imaging with steady state precession (trueFISP) sequences, or constructive interference steady state (CISS), sequences, additional information can be obtained, such as signal changes of the liquids of the inner ear that might be seen in case of schwannomas or meningiomas of the internal auditory canal, or labyrinthitis [26, 27].

To conclude, a simple 8-minute T2-weighted gradient-echo SSFP sequence (FIESTA-C), performed at 3 T, is a fast and anatomically accurate method to diagnose Meniere disease without injection of Gd-based contrast material. The measures are sub-millimetric, but the interobserver and intraobserver agreements were good, with κ ranging from 0.68 to 0.94. In Meniere disease, an endolymphatic hydrops with saccular abnormality was present in 84% of the cases, an elongated saccule (height > 1.60 mm) in 86% of the cases, an increased saccular width (>1.40 mm) in 44% of the cases (isolated in 4%), a missing saccule in 8% of the cases.