Quantitative polarized light microscopy of human cochlear sections

Dysfunction of the inner ear is the most common cause of sensorineural hearing loss, which is the most common sensory deficit worldwide. Conventional imaging modalities are unable to depict the microanatomy of the human inner ear, hence the need to explore novel imaging modalities. We provide the first characterization of the optical properties of human cochlear sections using quantitative polarized light microscopy (qPLM). Eight pediatric cadaveric cochlear sections, aged 0 (term) to 24 months, were selected from the US National Temporal Bone Registry, imaged with qPLM and measured using Image J. Retardance of the bony otic capsule and basilar membrane were substantially higher than that of the stria vascularis, spiral ganglion neurons, organ of Corti and spiral ligament across the half turns of the spiraling cochlea. qPLM provides quantitative information about the human inner ear, and awaits future exploration in vivo.


Introduction
Hearing loss affects over 36 million Americans (National Institute on Deafness and Other Communication Disorders 2010) and 600 million people worldwide (Traynor 2014). Although majority of sensorineural hearing loss originates from the inner ear, individual cells and structure within the human inner ear cannot yet be imaged using current state-of-the-art clinical tools, including computed tomography (CT) and magnetic resonance imaging (MRI) scans. Today, the only source of information about the cellular basis of human deafness is cadaveric human temporal bones that house the inner ear. To enable future cellular-level intracochlear imaging in alive humans, we have been exploring optical imaging tools (Yang et al. 2013, Kalwani et al 2013 because optics provides high spatial resolution at a lower cost than techniques based on ionizing radiation or magnetic resonance. We have recently demonstrated that quantitative polarized light microscopy (qPLM) detects differences in optical properties of key intracochlear structures, as evaluated using unstained mouse cochlear sections (Kalwani et al. 2013), while being faster and cheaper than immunohistochemistry. The current study explores the utility of qPLM in imaging human cochlear sections. Polarized light microscopy (PLM) is based on the principle that biological specimens alter the polarization state of the polarized light that passes through them. When the refractive index of the specimen is dependent on the polarization and direction of incoming light (a property known as birefringence), nonhomogeneous alterations in wave propagation velocity can be detected, which is referred to as sample retardance. Specimens that exhibit birefringence have directional structure, such as seen in collagen and myelin. Within hearing research, PLM has been used to examine the microcirculation within the guinea pig cochlea (Ren et al. 1993), and to study the organization of collagen within the guinea pig (Wenzel et al. 2004) and mouse (Wenzel et al. 2007) basilar membrane after laser irradiation.
Although PLM has proven useful for qualitative comparisons, qPLM has the major advantage of being quantitative. There are two key parameters in qPLM: retardance (expressed in nanometers) and the average orientation of the polarization axis with the greater index of refraction (Oldenbourg and Mei 1995). qPLM has been applied in many clinical specialties, including reproductive biology to quantify spindle aberrations in oocytes (Molinari et al. 2012), dermatology to improve detection of melanoma (Herman 2012), and orthopedics to quantify microstructural remodeling of articular cartilage following defect repair with osteochondral autograft (Raub et al. 2013). We were the first to apply qPLM to the inner ear (Kalwani et al. 2013).
Specifically, we used qPLM to quantify the optical properties of sectioned cochlear tissues in two different, commonly used mouse models, C57BL/6J and CBA/CaJ. We showed that the networks of collagen fibers known to exist in mammalian cochlea are better-identified using qPLM than using conventional differential interference contrast microscopy.
Encouraged by these results in mice, which are commonly used animal models to investigate hearing and deafness, we undertook the current study to characterize the optical properties of the cadaveric human cochlear sections using qPLM, with an eye to ultimate clinical applications in vivo. We focused on pediatric specimens with no known pathology of the inner ear so to establish a reference for future explorations of pathologic specimens. Many diseases that cause hearing loss are characterized by defects in the quantity, quality and organization of collagen -a major birefringent molecule in the inner ear (Merchant et al. 2004, Merchant andNadol 2010). We show that cochlear structures with the greatest retardance are the bony otic capsule and basilar membrane, suggesting that qPLM may be most useful for detecting pathology in these structures.

Human cochlear sections
All neonatal human cochleas (n=79), aged between 0 (term) and 24 months, were reviewed from the Massachusetts Eye and Ear collection of the US National Temporal Bone Registry. Twenty one temporal bones reported to have normal labyrinths were selected; the remaining bones were excluded because they had signs of autolysis or otitis media. Of the 21 temporal bones with no documented cochlear histopathology, only 8 had midmodiolar sections with excellent preservation of all cells. These 8 temporal bones were therefore selected for imaging with qPLM. The average time between death and tissue fixation was 8 hours, ranging from 2 hours to 17 hours; 2 cases out of the 8 selected did not have a reported port-mortem time.

Quantitative polarized microscopy
Archival cochlear specimens that we studied had been embedded in celloidin, sectioned at 20 µm, and stained with hematoxylin and eosin as per a routine protocol (Merchant and Nadol 2010) in the US temporal bone registry. The sections were mounted onto an inverted microscope (Nikon Eclipse TE 2000-S) and imaged using a commercial birefringence microscopy system (ABRIO TM ; Hinds Instruments) to characterize optical anisotropy. This instrument acquires pixwelwise-resolved measurements of retardance and the orientation of the slow optical axis in the specimens, as previously described (Kalwani et al. 2013). In short, monochromatic (wavelength 546 nm) circularly polarized light travels through the specimen and a computer-controlled compensator optic. The change in state of polarization of the light wave was recorded at five different compensator settings, and sample birefringence was calculated based on a polarmetric algorithm. This system resolves retardance values to within 0.02 nm, and it accounts for light absorbance in the sample and a background correction to correct for intrinsic birefringence in the microscope optical elements. The birefringence of the following inner ear structures were measured in each of the 4 cochlear half turns: organ of Corti, stria vascularis, spiral ligament, basilar membrane, otic capsule and cell bodies of spiral ganglion neurons. The images were analyzed in ImageJ (NIH). Retardance of each cochlear structure was measured 3 times for each sample to arrive at the mean and standard error of the mean for each structure across different samples. For each cochlear structure with measured retardance, a single factor analysis of variance (ANOVA) was used (in Excel) to test the null hypothesis that the means of retardance for different cochlear turns were all equal; p < 0.05 was considered significant.

Results
Quantitative PLM of the human cochlear sections revealed differences in birefringence across cochlear structures (Figure 1, 3), and typically more structural detail than seen under simple light microscopy ( Figure 2). Of the six structures measured, the otic capsule and basilar membrane were the most birefringent ( Figure   4). For all structures, retardance is expressed as mean +/-standard error of the mean.

Discussion
To the best of our knowledge, this is the first study to characterize the optical properties of human cochlear structures using qPLM. Our results reveal that the otic capsule and basilar membrane are the most birefringent structures whereas the spiral ligament is the least birefringent cochlear structure. While the basilar membrane and the otic capsule tended to have decreasing retardance with distance away from the cochlear base -unlike the more uniform retardance of the organ of Corti, stria vascularis, spiral ligament and spiral ganglion neurons along the cochlear length-this trend did not meet our criterion for significance. The results of this study are comparable with our findings in mice (Kalwani et al. 2013). The level of retardance measured in the human otic capsule was similar to, albeit larger than that of the mouse otic capsule of different strains. However, retardance of spiral ganglion cell bodies was notably higher in humans than mice. The greatest difference between human and mouse cochlear section was in retardance of the spiral ligament, which was substantially smaller in humans than in mice. There are several possible reasons for this difference. First, mice are higher-frequency hearing animals -they can hear up to 70 kHz and have maximal sensitivity at 16 kHz whereas humans hear up to 20 Hz with the maximal sensitivity at 2 kHz. Studies in bats, which can hear up to 200 kHz, have suggested that the intricately organized collagen fibers within the spiral ligament that attach to the basilar membrane contribute to the tension in the mechanically-tuned basilar membrane, hence enabling high frequency hearing (Henson et al. 1984).
Second, the time between death and cochlear extraction is substantially longer for human specimens, averaging 8 hours in our series, compared to just minutes in mice.
The prolonged post-mortem time in human specimens may contribute to collagen breakdown and tissue disorganization. In addition, due to the archival nature of human specimens, they were imaged years and decades after cochlear fixation whereas mouse specimens were imaged within days of cochlear extraction; continued tissue breakdown is possible even for fixed specimens. Although our specimens were from young pediatric patients, it is relevant that the human spiral ligament begins to lose collagen-producing fibrocytes early, within the first decade of life (Wright and Schuknecht 1972), and that degeneration of the spiral ligament occurs at a faster rate than that of other cochlear structures (Kusunoki et al. 2004). In addition, fibrocytes of the spiral ligament are particularly sensitive to noise trauma and mild exposure to noise, such as seen in neonatal intensive care units, can cause hearing damage undetectable by threshold based audiogram (Wang et al. 2002).
Establishing the optical properties of normal healthy cochlear tissue is the first step in considering qPLM as a diagnostic tool in vivo. Although the histological sections used in this study were fixed, they are likely representative of unfixed tissue, based on our comparison of optical properties of fixed and unfixed cochlear tissues using two photon microscopy (Yang et al. 2013). By establishing a normal baseline, out study enables future interpretation of pathologic human specimens, and classification of pathologies that are reflected in abnormal qPLM signatures. Many human diseases are characterized by disorganization or malfunction of collagen in the inner ear (Merchant and Nadol 2010), including Alport's disease (Merchant et al. 2004), otosclerosis, osteogenesis imperfect and Paget's disease.
Our study indicates that qPLM is capable of providing quantitative details of human cochlear structures, in particular the organization of collagen networks. Future studies are needed to explore whether application of qPLM in vivo could detect fine pathologic changes early, thus enabling early treatment and possible prevention of hearing loss. Combining qPLM with optical coherence tomography (OCT) may be particularly promising because OCT can generate cross-sectional images with a high spatial resolution based on back-scattered light from a focused beam of infrared light directed at the tissue (Huang et al. 1991). The resolution of OCT can range from 2-15 µm (Drexler et al. 2001), which has been sufficient to visualize the microanatomy of the rat cochlea ex vivo (Wong et al. 2000;Wong et al. 2004) and the mouse cochlea in vivo (Subhash et al, 2010), revealing the three cochlear fluid-filled spaces, modiolus, spiral ligament, the organ of Corti and spiral limbus.
Polarization-sensitive OCT (PS-OCT) combines qPLM and OCT, thus allowing measurement of the birefringence and the orientation of the fast axis of a specimen.
This generates a quantitative 3D cross sectional image of a specific region of interest.
PS-OCT has allowed identification of pathological microstructural changes in the concentration and organization of various tissue types such as tendons and ligaments in the knee (Martin et al. 2003;Bagnaninchi et al. 2010), and vocal folds of the larynx (Burns et al. 2009). PS-OCT has been used during laryngeal surgery to provide immediate feedback (Burns et al. 2009), and identify scar tissue within vocal folds to enable precise treatment of the scar tissue that contributes to dysphonia (Kim et al. 2010). It would be important to explore whether PS-OCT could similarly generate 3D cross sectional images of the inner ear in vivo, hence providing diagnostic information and invaluable real-time feedback during inner-ear surgery. Such feedback is currently not available but may be crucial in preventing microstructural damage and the associated hearing loss during procedures such as cochlear implantation, stapedectomy or labyrinthectomy. PS-OCT may also enable precise delivery of emerging cell-based and genetic therapies to the diseased regions of the inner ear, and quantitative monitoring of the response to treatment.
A limitation of our study is that it is based on healthy neonatal cochlear sections so that further research is required to explore birefringent patterns in other age groups and in pathological states. Nonetheless, our study highlights the utility of qPLM in visualizing human cochlear structures, and suggests that PS-OCT may be a useful tool to investigate the inner ear and guide precise therapy.

Conclusions
This study provides the first characterization of the optical properties of the microstructures within the human cochlea using qPLM. We note many similarities between the retardance of the human and mouse cochlear tissues, with the notable exception of the spiral ligament, which is much less birefringent in humans than in mice. Taken together, this suggests that techniques developed in mice will have quantitative utility in humans. Our measurements of normal human tissue motivate future exploration of qPLM in diseased tissue associated with hearing loss so to determine whether birefringence can serve as an imaging biomarker for inner ear diseases that are currently lumped under the umbrella term "sensorineural hearing loss". Our results, combined with others' reports of intracochlear imaging using OCT in rodent models, suggest that PS-OCT may have multiple applications in microstructural diagnosis and targeted therapy of disorders of the human inner ear, while providing real-time feedback during surgical intervention.