The central postulation of hydrops in MD suggests a link between fluid homeostasis of the vestibular system and the clinical manifestations of vertigo and aural fullness [7]. While the cause of this abnormal fluid homeostasis is unknown, it has been variously postulated to be linked to abnormal expression of aquaporins or claudins, degradation of ‘adhesive’ vascular endothelial cells, thickening of the basement membrane or abnormal deposits (such as cochlin) within the supporting stroma [5]. This study utilised specimens collected from surgical patients, to compare the histology of the membranous labyrinth between MD and VS (control) patients. It is one of the largest comparative studies of MD histology using non-cadaveric material and builds on the evidence behind the aetiology of MD.

Histology

Our morphological investigations have not demonstrated any unique abnormality in the sensory epithelium or underlying stroma of patients undergoing surgery for intractable MD. There was no evidence of acute or chronic inflammation, fibrosis or denervation. No abnormal deposits were seen within the stroma. Perhaps these findings are not unexpected, as for patients to experience acute symptoms, they need to have an intact sensory apparatus and be able to transmit neural signals. In this study, all MD tissue was collected during labyrinthectomy in the setting of unremitting symptoms of MD. Intact receptor morphology supports the theory that the epithelium is suffering episodic overstimulation due to some mechanical event within the labyrinth, such as a sudden fluid shift between compartments, rather than intrinsic damage to the neuroepithelium.

Basement membrane and neuroepithelia

McCall (2009) demonstrated non-specific basement membrane thickening, organelle changes, and an absence of hair cells when assessing post-mortem temporal bones [9], which was in keeping with the postulation of cellular stress. It is unclear whether this basement membrane thickening extended into the semicircular canals which are postulated to be involved in the ionic regulation of the endolymph [10, 11] or if this thickening was exclusive to the sensory epithelium. Calzada (2012) further demonstrated a disorganised supportive matrix with neuroepithelial degeneration and a thickened basement membrane in MD temporal bones when compared with control specimens with normal audiovestibular functions [12].

On the contrary, our study did not consistently find basement membrane thickening and neuroepithelial monolayering in MD specimens unlike McCall and Calzada [9, 12]. We found thickening of the basement membrane below the lateral canal sensory epithelium in only a single MD patient who had been previously treated with gentamicin. Our finding of considerable variation in the thickness of the sub-epithelial basement membrane according to the site (as well as orientation) of the sample (in both MD and VS patients) indicates that care should be taken when interpreting isolated measurements of basement membranes thickness. Additionally, orientation of these tiny fragile samples is difficult, and “tilt” due to plane of embedding may also affect measurement of membrane thickness.

Tsuji identified a selective loss of Type II hair cells and Scarpa’s ganglion cells in the neuroepithelium of MD samples [13]. This was later challenged by McCall [9] and Calzada [12] who noted no preferential loss of Type II vs. Type I hair cells, but rather a vacuolisation of hair cells and supporting cells with a disorganisation of the basement membrane in MD samples. We identified considerable variation in the cellular constituents and staining profile of the sensory epithelium depending on whether the sample was from the centre or the edge of the sensory apparatus, suggesting similar caution must be applied when assessing relative numbers of certain cell types in small (particularly ultrastructural) samples, as the exact location influences epithelial cell composition and innervation.

Immunohistochemical evaluation of structural elements

Evaluation of the immunohistochemical profile of various structures within the vestibular system identified several novel findings;

i)

No lymphatic vessel of “usual” type (i.e. lined by endothelial cells staining for D2-40) were identified within the stroma beneath the vestibular epithelium. This is similar to the central nervous system where there is an absence of true lymphatic vessels, and has implications for fluid handling within the vestibular system.

ii)

The subepithelial plump stromal cells showed uniformly strong positivity for S100. This is not a characteristic of lamina propria/fibroblast-containing stroma of other organs and raises the possibility that the stroma is of perineurial origin.

iii)

The entire vestibular epithelium stained positively for pan-keratin and (except for type 1 cells) was also strongly positive for S100; supporting cells also labelled for GFAP. The presence of strong epithelial staining for S100 is unusual as it is absent in most normal epithelia.

Immune/inflammatory features

The absence of significant inflammatory cell infiltrate within the stroma is similar to that reported by McCall [9]. We also demonstrated no expression of CD35 (which labels dendritic cells involved in antigen presentation), no perivascular inflammation, no vessel wall thickening and no eosinophilic deposits in the subepithelial stroma suggesting that local FDC-mediated secretion of cochlin is not involved in the pathogenesis of MD. This is in contrary to Calzada who demonstrated increased cochlin immunoreactivity in the vestibular stroma of patients with MD [12]. The interest in cochlin deposition was based on the work of Robertson (2001, 2006) who demonstrated increased eosinophilic deposits of cochlin in the ampullary stroma of patients with hereditary DFNA9-associated hearing loss and vestibular dysfunction [14, 15]. Furthermore, Py (2013) demonstrated that cochlin plays a role in innate immunity and is secreted by follicular dendritic cells [16].

Fluid handling

Monsanto (2017) proposed an anatomical predisposition to hydrops and symptom development due to a decreased endolymphatic volume following volumetric analysis of 16 MD, 16 endolymphatic hydrops and 16 non-diseased cadaveric temporal bones [17]. The 3D reconstructions of the vestibular aqueduct and the endolymphatic sac within the temporal bones demonstrated lower volumes in the MD group with a comparatively smaller vestibular aqueduct, suggestive of an anatomical predisposition to developing MD and the associated symptoms.

Ishiyama (2017, 2018) demonstrated tight junction integrity with stromal and perivascular basement membrane disorganisation, oedema of the pericytes and swelling of the blood-labyrinthine barrier (BLB) in vestibular end organs of MD patients [18, 19]. It is unclear if these changes were the cause or effect of the ionic and volume disturbances leading to endolymphatic hydrops (and the resulting symptomatology). Dixon Johns (2023) demonstrated a reduction of KCNJ10 expression, which plays a role in cellular membrane potential, in temporal bone specimens of those with MD [20]. Lopez (2007) demonstrated the presence of aquaporin (AQP) 1, AQP 4 and AQP 6 by immunofluorescence of inner ear samples from non-diseased temporal bones [21]. AQP 1 was noted to be distributed to fibrocytes and blood vessels of the underlying stroma within the cochlear, AQP 4 was localised to the basal pole of vestibular supporting cells in the macula utricle and cristae ampullaris, and AQP 6 was localised to the apical portion of the vestibular supporting cells, the hair cells themselves being non-immunoreactive.

By contrast, in our study staining for AQP 1 did not show specific labelling in any of the vestibular sections, AQP 4 demonstrated non-specific background staining in apical portions of the sensory epithelia and AQP 6 demonstrated strong apical granular cytoplasmic staining in supporting cells of sensory epithelia, but not in adjacent attenuated lining epithelium. No obvious difference was present between MD and VS samples. While the apical localization of AQP 6 was similar to that reported by Lopez [21], the localization of AQP 1 and AQP 4 appears discordant, suggesting that either post-mortem autolysis affected the staining in the cadaveric samples [20], or that antibodies need further optimization for use in paraffin sections.

We identified no staining for Claudin 3 in our examinations of both MD and VS samples. Moderately strong staining for Claudin 4 was seen in attenuated canal epithelium, with the sensory epithelium staining weakly in most samples, indicating the presence of preserved tight junctions in both the MD and VS samples. We noted the apparent increased epithelial membranous staining for Claudin 4 in the MD patient with past history of gentamicin therapy; the reason for this is unclear, but it would not appear to support a thesis of “leaky” epithelium. (Fig. 3).

Fig. 3

Fluid-handling antibodies of lateral canal of VS (a) Antibody to Aquaporin 1; (b) Antibody to Aquaporin 6; (c) Antibody to Claudin 3 was negative; (d) Antibody to Claudin 4

Study limitations

A limitation of the study was the inability to examine tissues from endolymphatic sac in patients undergoing either labyrinthectomy or vestibular schwannoma resection. This has been a limitation of all reported studies in non-cadaveric samples, as it is not a routine part of transmastoid labyrinthectomy due to the risk of CSF leak. Further analysis of endolymphatic sac tissue in both MD patients and controls would therefore appear to be a priority for future work, to better understand the blood-labyrinth barrier in the sac and its role in MD.