Medicines discovery for auditory disorders: Challenges for industry

Inner ear disorders have few treatment options, and no medicines are currently approved for the prevention or treatment of hearing loss, despite the current estimate that approximately 500 × 10⁶ people worldwide suffer from hearing loss (Wilson , 2017). The available drug treatment options for inner ear disorders, such as for patients with sensorineural hearing loss, are limited to the off-label use of medicines where there is a lack of good-quality evidence of their effectiveness in these indications (Crane , 2015; Wei , 2013). In the absence of effective drug treatments, devices such as hearing aids and cochlear implants remain the mainstay for those with mild to profound hearing loss, respectively (Sprinzl and Riechelmann, 2010). Cochlear implants are transformational for patients such as deaf pre-lingual children, but these devices are unable to reproduce natural hearing and require significant postoperative training or rehabilitation of the users to ensure adequate hearing performance (Macherey and Carlyon, 2014; Sprinzl and Riechelmann, 2010). Patients with successful cochlear implantation can readily understand speech in quiet situations, but most struggle in noisy situations, while poor pitch perception impairs the enjoyment of music and can prevent discrimination of pitch differences in separate competing voices (Sprinzl and Riechelmann, 2010; Zeng, 2017). Hearing aids have a low level of uptake in eligible patients and then poor subsequent continued use due in part to the non-natural hearing these devices provide, such as poor performance in noisy environments (Lesica, 2018; McCormack and Fortnum, 2013). There remains an unmet medical need for medicines to treat hearing loss and other inner ear disorders, as highlighted by analyses of global health burdens (Wilson , 2017).

The societal burden of untreated hearing loss is receiving greater recognition, as hearing loss has been associated with increased risk of anxiety, social isolation and cognitive decline (Lin and Albert, 2014; Lin , 2011; Nirmalasari , 2017; Thomson , 2017). A commission has identified hearing loss as the most important modifiable risk factor in middle age for developing dementia, accounting for nearly 10% of the overall risk, although, at present, causality is unproven (Livingston , 2017). Beyond hearing loss, other disorders of the inner ear, including Meniere's disease, vestibular disorders, and tinnitus, also lack effective treatments (Baguley , 2013; Maldonado Fernández , 2015; Phillips and Westerberg, 2011). The unmet medical need of inner ear disorders combined with advances in the understanding of the underlying biology has fuelled the creation of new life science companies focused on the hearing loss therapeutic area, with novel investigational drugs being advanced toward and into the clinic (Li, 2017). In a review of the commercial landscape, a total of 43 biotechnology and pharmaceutical companies were identified as developing therapeutics for inner ear and central hearing disorders with a range of drug modalities (Schilder , 2019). This non-exhaustive list of companies and potential therapeutics will change over time, so updates of this list will be made freely available through the website of the UK based charity Action on Hearing Loss.

This article will provide an overview of the challenges in discovering medicines of value and describe the critical decision points in the drug discovery process. The review will provide insight to how lessons learned from past poor productivity in drug discovery across all disease areas are now being applied to enable data-informed decisions to discover high-quality clinical candidate molecules and the implementation of best practice.

The discovery of a transformative medicine involves thousands of individuals in a truly multidisciplinary enterprise from the initial idea to final delivery of a medicine to the patient, and, often, collaborations involving industry and academia remain a key to future innovations, particularly in uncharted therapeutic areas such as hearing loss (Vallance, 2016). Despite examples of industry successes in medicines discovery (Bolger , 2017; Jarvis, 2019), there remains significant challenges and a high failure rate associated with the medicine discovery process across all therapeutic areas. An analysis published by Biotechnology Innovation Organisation (BIO) of the commercial research and development (R&D) activity for the period from 2005 to 2015 found the average Likelihood of Approval (LOA) for all therapeutic indications upon an experimental medicine entering Phase I was only 9.6%, and across different diseases ranged from 26.1% for haematological diseases to just 5.1% for oncology, whilst neurology indications had a below-average LOA of 8.4% (Thomas , 2016). These findings are consistent with other published reviews of industrial R&D productivity across all therapeutic indications (DiMasi , 2010; DiMasi , 2013; Harrison, 2016; Hay , 2014; Smietana , 2016). The combination of poor LOA in neuroscience indications with the often frequently lengthy clinical studies and subjective endpoints, in particular for slowly progressing neurodegenerative diseases, has contributed to the decision for some large pharmaceutical companies to abandon the neuroscience therapeutic area in order to focus on other therapy areas with a perceived higher probability of success (PoS) (Mullard, 2018).

Pharmaceutical R&D requires substantial financial and time investments to deliver a medicine approval, and decreasing industrial productivity as measured by regulatory approvals per dollars invested have drawn attention to the rising costs of pharmaceutical R&D. Recent estimates on the cost of innovation in the pharmaceutical industry indicate a total pre-approval cost estimate of $2558 × 10⁶ per medicine (DiMasi , 2016). The current estimated cost compares to historical values of $1044 × 10⁶ per medicine from the 1990s to the mid-2000s, and of $413M per medicine from the 1980s to mid-1990s (DiMasi , 2016). The ever-rising cost of drug development has led to much analysis within and across the industry to understand why the PoS remains low despite the increased investment, and how the LOA can be increased to improve pharmaceutical R&D productivity. A review by the Tufts Center for the Study of Drug Development of clinical studies across all therapeutic indications from 2000 to 2009 determined that the three most common reasons that Phase III trials failed to progress to approval were lack of efficacy (∼50%), an unacceptable safety profile or unexpected adverse events (∼30%), or a failure to demonstrate differentiation over the existing standard of care (∼15%) (CSDD, 2013) and is consistent with another recent published analysis (Harrison, 2016).

From published and internal industry reviews of the drug discovery processes, several strategies for good practice were developed and implemented to improve productivity and to eliminate, as quickly as possible, targets and projects which cannot deliver the desired medicine product profile. A team at Pfizer undertook a detailed analysis of the 44 Phase II clinical programmes from the internal Pfizer portfolio that had reached a decision point from 2005 to 2009 (Morgan , 2012). Their published analysis revealed that in 43% of the failed Phase II studies, it was not possible to make an informed decision on the role of the molecular target in the disease indication clinically evaluated. Importantly, it was therefore impossible to make an informed decision as to whether to terminate commercial interest, or to repeat the clinical study with a different protocol, or to evaluate the target/compound in a different disease. The conclusions of this analysis made clear the importance of acquiring data and information on the fundamental pharmacokinetic and pharmacodynamic properties of each investigational drug to ensure the biological mechanism of action is evaluated thoroughly to enable an informed progression beyond Phase II development. In this landmark review, the authors proposed what has become known as the “Three Pillars of Survival” framework, which provided a basis from which to increase the likelihood of an investigational drug surviving Phase II testing and moving on to Phase III testing. These principles determined that for an investigational drug to deliver the desired effect over the necessary period, three fundamental elements need to be demonstrated: exposure at the target site of action over the desired period, binding to the pharmacological target as expected for its mode of action, and expression of pharmacological activity commensurate with the demonstrated target exposure and target binding. Their analysis exposed the high risk of progressing the clinical development of an investigational medicine to Phase III studies without an understanding of these three critical components.

Before the “Three Pillars of Survival” strategy, across the pharmaceutical industry, data supporting these three critical components had not been rigorously collected. For some therapeutic areas such as neuroscience, accessing all these types of data is challenging and may at present be impossible. As such, for neuroscience, clinical progression will require innovative solutions to be developed to mitigate the higher risk of proceeding without these attendant data. The challenge of accessing all these data in neuroscience indications may have contributed to the decision of some pharmaceutical companies to deprioritise neuroscience indications and focus resource on therapeutic areas where these data are more readily measured (Mullard, 2018). The principles set out in the “Three Pillars of Survival” have become embedded across the industry and drive a range of activities and studies in modern industrial drug discovery to enable data-informed decisions to be made in clinical development to assess the likelihood of success and whether investment in a project or investigational drug should continue or terminate.

In response to the challenges set by the “Three Pillars of Survival,” the pharmaceutical industry has focused on improving the quality of drug discovery activities across the three cornerstones of medicine discovery: target selection, clinical candidate selection, and clinical studies.

Biological target selection is the lynchpin of medicine discovery, as once chosen, all the risks associated with the selected target are carried the entire way through the discovery and development process. The primary importance of target selection has led to significant investment in the pharmaceutical industry with a broadening range of activities directed at the critical decision of biological target selection to ensure that only the best and most appropriate biological targets are taken forward into drug discovery programmes. The Human Genome Project provided an opportunity for identifying potential genes that play a role in human disease, and with the increased amount of genetic data now available, an analysis was undertaken by a team to determine if there was a correlation between clinically successful drug efforts and known genetic target associations with disease (Nelson , 2015; Plenge , 2013). For diseases where multiple clinical studies had completed, the proportion of drug targets with direct genetic support increased significantly through drug development, from 2.0% of targets at the preclinical stage to 8.2% among targets for approved drugs. Selecting molecular targets with evidence of human genetic association meant the success rate in clinical development was doubled, and so has led to increased prioritisation of targets with human genetic association within pharmaceutical company portfolios (Hurle , 2016). However, genetic association with human disease does not necessarily mean causality, and significant investment is required in target validation studies before committing to drug discovery to increase confidence in causality and non-redundancy of the biological target of interest (Page , 2003).

Studies examining the genetics of hearing loss have found more than 100 genetic loci associated with non-syndromic deafness (Muller and Barr-Gillespie, 2015; Vona , 2015). Furthermore, mutations in many hundreds of genes are believed to cause or predispose individuals to congenital, progressive, noise-induced, and age-related forms of hearing loss (Dror and Avraham, 2010; Raviv , 2010). The challenges for identifying genes linked to hearing loss in highly heterogeneous disorders such as age-related hearing loss (ARHL) have been described (Bowl and Dawson, 2019). Strategies to enhance the likelihood of finding a gene target include evaluating as a control, against a population of individuals with good hearing for their age rather than an unselected population (Lewis , 2018). Nonetheless, genome-wide association studies (GWAS), whole-genome, and whole-exome sequencing studies have been undertaken for ARHL. In an example of how such an approach may reveal biological targets suitable for drug discovery, a GWAS identified the glutamate metabotropic receptor 7 gene (GRM7) from several single nucleotide polymorphisms (SNPs) as potentially playing a role in susceptibility to age-related hearing impairment and, in particular, in speech recognition (Friedman , 2009; Luo , 2013; Newman , 2012). Additional genetics studies have supported the identification of GRM7 variants conferring susceptibility to sensorineural hearing impairment (Matyas , 2018) and found evidence of GRM7 variants may reduce susceptibility to noise-induced hearing loss (NIHL) (Yu , 2018). These genetic findings combined with the knowledge that the protein mGluR7 is expressed in the human cochlea and has a central role in glutamate synaptic transmission and homeostasis in the cochlea encouraged Pragma Therapeutics¹ to include hearing disorders as an indication for their drug discovery activities around the GRM7 molecular target (Lu, 2014). Although the biological and genetic data for the GRM7 target are encouraging, another critical factor in the decision to take GRM7 target into drug discovery is the knowledge that the mGluR7 receptor is “druggable,” and that small molecule modulators of mGluR7 receptor are known (Kalinichev , 2013; Suzuki , 2007). Going forward, the prioritisation of biological targets for progression to drug discovery activities will demand an understanding of the genetic human disease association and insights concerning disease causation, and so further studies directed to the genetics and the genomics of human hearing loss are required.

Not all biological targets are equal with respect to suitability for drug discovery; in short, the probability of discovering small molecule or antibody therapeutics, which remain the engines of drug discovery, varies significantly across the different classes of biological targets and is a crucial factor in the decision as to whether to take a target into drug discovery or not. The term “druggability” was developed in the early 2000s from the analysis of the human genome for druggable protein targets that would be suitable for small molecule or therapeutic antibody modulation and estimated that perhaps only 10%–14% of human gene targets are amenable to modulation by these modalities (Hopkins and Groom, 2002; Russ and Lampel, 2005). Initially, the druggability analysis of a new biological target was binary and depended on whether the protein target belonged to a target family that had at least one approved drug, which was an unsatisfactory approach, as a novel biological target or novel biological target class was immediately classified as non-druggable, which may be inappropriate and inhibit progression. Therefore, biological target evaluation was modified to include all publicly available data of small molecule binders for the biological targets of interest or related proteins, and so started to provide a scale of biological target tractability (Imming , 2006; Overington , 2006) and eventually providing a comprehensive map of drug targets (Santos , 2017). A range of bio- and chemo-informatic methods have been developed to enable a systematic analysis of new targets and to determine the innate potential suitability of the biological target of interest for drug discovery activities using known available modalities (Abi Hussein , 2017). However, a number of these analyses only determine the ability to identify small molecule ligands or binders for the protein of interest and do not regard the suitability of these binders to generate drug-like molecules or to induce a functional effect. It has been proposed to extend the concept of “druggability” to “inhibitability” to allow the prioritisation of targets with low affinity for their cognate binding partner (Vukovic and Huggins, 2018). The concepts of druggability, target tractability, ligandability, and target quality are now better defined and have become established strategies in the pharmaceutical industry for determining if a biological target is suitable for drug discovery and which drug modality might be best used to clinically modulate that biological target (Brown , 2018).

Initial biological target validation studies are often undertaken in the absence of suitable tool compounds and so will use genetic target validation strategies, such as gene knock out/knock-in and gene editing methodologies. Chemical tool compounds are applied in target validation studies to increase confidence in the biological target and that the selected biological target is amenable to modulation by the chosen modality before being taken forward into full drug discovery activities. Scientists from Pfizer published a framework setting out what they believe is needed to be achieved preclinically to increase confidence of the relevance of a biological target using chemical tools: (1) knowledge of exposure at site of action; (2) demonstration of biological target occupancy; (3) proof of mechanistic pharmacology; and (4) the use of disease-relevant phenotypic assays (Bunnage , 2013). The framework developed parallels the requirements set out in the “Three Pillars of Survival” charter and with these target validation data would provide a platform to enable the translation of preclinical biology to potentially effective drugs that are most likely to deliver clinical success, with a series of questions to be addressed. The proposed framework aimed to promote a deep understanding of a biological target, including identifying where gaps in knowledge existed (Bunnage , 2015; Gashaw , 2012). In addition to these target validation activities, a safety assessment of the chosen biological target must be undertaken to understand the potential adverse effects and risks of modulating the protein target of interest and how these could be monitored and mitigated for (Roberts, 2018). All the biological target insights obtained will inform the decision for which drug modalities can or cannot be used to modulate the protein target of interest, and which modality is most likely to address the unmet medical need of the patient and to deliver commercial success.

Marketed medicines are mostly small molecule entities or biologics such as therapeutic antibodies, and in the short term, these established modalities are expected to continue to be the mainstay of drug discovery projects, as demonstrated in the published list of 59 U.S. Food and Drug Administration (FDA) drug approvals for 2018 where there were 38 small molecule drugs and 11 antibody therapeutics (Jarvis, 2019). However, there are significant limitations with these two modalities that limit their utility against the full range of biological targets of potential therapeutic interest, and it has been estimated that perhaps only 10%–14% of human gene targets are amenable to modulation by these modalities (Hopkins and Groom, 2002). In simple terms, small molecule entities are restricted to protein targets with defined small binding pockets that can modulate function, such as the catalytic site of an enzyme, but small molecules have the advantage that they can access intracellular located biological targets. In comparison, biologics such as therapeutic antibodies are limited to extracellular or cell surface targets but can disrupt large and often featureless protein surfaces that lack discrete binding pockets, such as often found in protein-protein interactions. It is most likely that biological targets for hearing loss will be located within the inner ear or the central nervous system (CNS), therefore biological targets that are amenable to small molecule modulation should perhaps be prioritised, since small molecules can access the CNS and the likely target tissues associated with hearing disorders more easily than therapeutic antibodies or other biologic modalities.

Innovations in medicinal chemistry have led to the creation of novel chemical technologies to extend the range of protein targets that can be modulated, and include DNA encoded chemical libraries, cyclic and stapled natural and non-natural amino acid containing peptides, nucleic acid based therapeutics, and bifunctional molecules such as proteolysis targeting chimeras (PROTACs) (Arico-Muendel, 2016; Cary , 2017; Churcher, 2018; Machutta , 2017; Sawyer , 2018; Valeur , 2017). These technologies have allowed the evaluation of more challenging biological target classes, often labelled as “undruggable” target classes such as protein-protein interactions, and in the short term will aid the prioritisation of targets and pathways in the target validation phase (Valeur and Jimonet, 2018; Verdine and Walensky, 2007). Some of these new technologies, such as PROTACs, are unproven modalities for a marketed medicine, but current research and clinical development activity will provide insights to their suitability as medicine modalities. Of these novel modalities, PROTACs with a catalytic mode of action for target protein degradation and an expected low dose might prove applicable to inner ear disorders for a suitable biological target, where the PROTAC drug is administered using direct or local topical delivery.

A crucial cornerstone of drug discovery is the selection of the molecular entity for progression to clinical evaluation, often referred to as the clinical candidate. Once chosen, all the hope and promise, along with all the yet unknown weaknesses of a molecule, are locked into the physical structure of that entity going forward. For small molecule drug discovery, an essential publication in 1997 from Pfizer scientists proposed that physicochemical properties of previously successful Phase II small molecule drugs may be able to define a favourable drug-like chemical space and became known within the industry as the “Lipinski's Rule of 5” (Ro5) (Lipinski , 1997). The concept of drug-like chemical properties was attractive and became a guiding principle that sparked initiatives within the industry to reduce compound attrition, improve pharmacokinetics, and increase drug discovery PoS (Lipinski, 2016; Young and Leeson, 2018). Multiple physicochemical criteria were henceforth applied and used by medicinal chemists in small molecule tool and drug design. The physicochemical criteria included easily measured or calculated properties such as molecular weight, or the calculated partition coefficient (clog P), to drive medicinal chemistry to deliver clinical candidate molecules that were within the perceived drug-like chemical space (Leeson, 2016; Leeson and Young, 2015). The primary focus of small molecule medicinal chemistry has shifted away from the discovery of molecules with the highest biological potency for the target of interest to a multiparameter optimisation that balances the biological target potency with desired specificity and selectivity and the perceived need to stay within the proposed drug-like chemical space (Holenz and Stoy, 2019). This strategy is believed to increase the likelihood of delivering the desired absorption, distribution, metabolism, elimination (ADME) properties, and to afford the necessary unbound drug concentrations at the desired target site of action with acceptable toxicity characteristics (Campbell , 2018). The concept of drug-like chemical space has become embedded within the industry and has led to the development and application of a range of novel calculated metrics, such as ligand efficiency and lipophilic ligand efficiency. These compound metrics allow compound selection and enable comparison across different chemical structure series from screening activities, or to evaluate how modifications to the chemical structure and concomitant changes in molecular size and lipophilicity within a chemical series have impacted on biological potency (Hopkins , 2004; Hopkins , 2014; Scott and Waring, 2018; Young and Leeson, 2018). These practical techniques to aid and guide medicinal chemistry in the multiparametric optimisation of small molecules have found wide acceptance but are not without critics and so should be best considered as tools to support decision making rather than decision making tools in and of themselves (Cavalluzzi , 2017; Kenny, 2019; Murray , 2014). As contrary to the idea of a drug-like chemical space it had been noted that 42% of oral drugs with molecular weight >500 Daltons have doses of <50 mg/day (Doak , 2014), challenging the perceived risk of small molecules with a molecular weight greater than 500 Daltons not being suitable for oral drug delivery (Lipinski, 2016; Lipinski , 1997). Another published analysis indicated that the favourable drug-like chemical space had changed in the 20 years since the Lipinski Ro5 publication and noted that some parameters, such as molecular weight had increased substantially (Shultz, 2018). Scientists at GSK published a guideline for the selection oral clinical candidates using the predicted dose, solubility and the property forecast index of each candidate molecule, which is a composite measure of lipophilicity using chromatographically determined logD and aromaticity (Bayliss , 2016). Application of this guideline enabled molecules outside of the original perceived drug-like space to progress due to their low predicted dose. Many important and biologically attractive targets are perceived to be difficult to drug or to be undruggable with respect to finding small molecule compounds compliant with the Lipinski Ro5 (Doak and Kihlberg, 2017). New U.S. FDA drug approvals in the 3 year period (2013–2017) included 12 new oral beyond Ro5 drugs, accounting for 21% of new oral drug approvals in that time period, providing encouragement, that difficult to drug biological targets requiring beyond Ro5 drug-like chemical space can be successfully prosecuted to deliver orally active molecules (DeGoey , 2018). Therefore, the use of metrics and the concept of drug-like chemical space should always be used instead as a guide, rather than as a rigid, unchanging rule set.

A biological target of relevance to hearing loss or impairment will most likely require biological target engagement within the inner ear or in the CNS. In each case, accessing the biological target within these tissues presents significant additional challenges to drug distribution as molecules will need to cross the blood labyrinthine barrier (BLB) to access the inner ear (Glueckert , 2018) or the blood-brain barrier (BBB) to access the CNS (Daneman and Prat, 2015). General principles for designing drugs with physicochemical properties suitable for crossing the BBB to treat diseases of the CNS have been outlined and require small molecule drugs to meet the more restrictive physicochemical properties observed for putative CNS active drugs (Di , 2013; Geldenhuys , 2015). It is unclear at present whether the rules that apply to small molecules crossing the BBB are the same as could be applied to small molecules to cross the BLB. The more restrictive physicochemical demands of CNS active compounds in conjunction with modern medicinal chemistry design concepts has led to the development of specific multiparameter optimisation frameworks for CNS drugs (Ghose , 2017; Wager , 2016).

Biological targets that only require delivery of drug substance into the inner ear rather than the CNS may offer an opportunity for local topical delivery via permeation through the membranes covering the round or oval windows following injection into the middle ear void, or direct delivery via intracochlear delivery. It is common in inner ear disorders in the clinic to use local delivery to the ear via intratympanic injection. Although advances have been made in delivery techniques and analytics to support topical administration to the inner ear, the physicochemical properties of a drug can nonetheless impede access to the final site of action. The appropriate physicochemical properties must be designed into drug molecules during medicinal chemistry optimisation to leverage the intratympanic topical route of administration and to the strategy of using intratympanic delivery as a rescue approach to drug administration of a non-optimal molecule (Salt and Hirose 2018; Salt and Plontke, 2018). Nonetheless, the intratympanic route of administration may broaden the range of biological targets for drug discovery as the acceptable physicochemical properties for an intratympanic route of administration may lie outside the restrictive physiochemical property range of orally delivered CNS active drugs.

The clinical evaluation of an investigational therapy to demonstrate clinical efficacy and safety is the third cornerstone of medicine discovery. The initial clinical trials of a novel investigational medicine are used to collect data that will enable an informed decision as to whether to continue the clinical development of the clinical candidate or to terminate the development activities of the clinical candidate as quickly and as early as possible. Such a clinical development approach then allows interest in a clinical candidate to be terminated with a relatively low cost and impact and is often referred to as a fail-fast strategy (DiPiro and Chisholm-Burns, 2013). The fail-fast strategy in the development of new CNS drugs has been driven by the recognition that that although CNS therapeutic drugs at Phase I and at Phase II were no less likely to progress than non-CNS drugs, at Phase III CNS drugs were 45% more likely to fail than non-CNS drugs (Kesselheim , 2015). Phase III is the most expensive point in drug discovery and failure at this point at reinforced the need for rigour in CNS drug discovery to ensure that there is evidence-based support for progression. Such data includes appropriate pharmacokinetics, target engagement, drug exposure dependent effects, and validation of biomarkers of drug action and disease status with measured objective data, as set out in the “Three Pillars of Survival” (Morgan , 2012).

Fast to fail clinical development strategies include experimental medicine studies at the earliest stage of clinical evaluation, which use small numbers of subjects, where the patients are stratified to focus initial clinical evaluation on the sub-populations believed to be most likely to respond to the treatment (Grabb , 2016; Roses, 2004). Other trial methods such non-superiority designs, adaptive randomisation designs, and integrated Phase II and Phase III studies are also used to focus the clinical study on fundamental properties or to enhance efficiency and the speed of the clinical evaluation of a potential medicine (Dorsey , 2015). In parallel, methodologies and tools have been developed to improve human dose prediction (Nair , 2018), which along with in silico quantitative systems pharmacology modelling (Knight-Schrijver , 2016), and physiologically based pharmacokinetic modelling and simulation (Sager , 2015) are enabling prospective criteria to be put in place to enable fast to fail decisions. These simulations, when coupled with human clinical drug exposure data, are crucial for understanding if the proposed doses can provide the necessary drug concentration at the target tissue with a duration of exposure expected to deliver the desired biological effect. These requirements within pharmaceutical drug discovery and development will require innovative solutions to obtain these data for the inner ear tissues.

For disorders of the inner ear and hearing, more in-depth knowledge of disease natural history and pathophysiology is required for these new approaches to be truly impactful. In patients presenting with hearing loss and other inner ear disorders, the underlying pathology and site of lesion driving their disease may be unknown or uncertain. Therefore, the selection of the most appropriate patients for initial evaluation of new targeted treatments may not be straightforward and is likely to increase risk of failure due to inappropriate subject inclusion (Schilder , 2018). Methods to enable the detailed assessment of the cochlea in both patients and healthy individuals are vital, and without their development, the advancement of new treatment options will be delayed or will proceed with a significant risk of inappropriate patient selection. For example, regenerative therapeutic approaches that aim to (re)-create new sensory hair cells will need to understand that sensory hair cells are absent in the cochlea of the patient. However, the modern diagnostics such as otoacoustic emissions cannot differentiate between the absence of outer hair cells from outer hair cells that are present, but dysfunctional. The standard audiology diagnostic methodologies can only provide limited insight into the condition of the auditory nerve, while the health and the competency of the stria vascularis cannot be determined in patients. The limited diagnostics presently available for determining the presence, absence, damage or the health status of neurons, synapses, hair cells, supporting cells, stria and other cells and structures in the inner ear, coupled with the lack of inner ear pharmacokinetic tools and visualisation methods in patients will mean that the development of new medicines for hearing loss will continue to be high risk. Potentially, when clinical efficacy is not achieved in a clinical study in the absence of a full understanding of the pharmacokinetic profile, it will not be genuinely understood whether a novel biological target hypothesis has been thoroughly tested in patients. It is also possible that where there is a functional benefit to the patient, we may not be able to demonstrate the true causality. In other therapeutic indications, increasing precision is being applied to patient selection for the initial clinical proof of concept studies, typically Phase IIa, as well as for proven treatments, through examining and including or excluding patients based on biomarkers, as well as their genetic and genomic profiles (Dijkstra , 2016; Sloan-Heggen and Smith, 2016). As patient selection technologies develop, and the knowledge of human inner ear pharmacokinetics improves, greater confidence can be applied to small, focused clinical experimental medicine studies to provide a reason to believe and continue or to terminate the development of the investigational drug and or the biological target at the earliest stage possible.

The patient populations as defined by the origin of their hearing loss demand different considerations during the clinical evaluation of new therapeutics but these must match of the demands of convincing the regulatory authorities that relevant clinical efficacy has been measured and achieved along with an appropriate safety profile, whilst still enabling clinical evaluation to be completed in a suitably rapid and reasonable resource-demanding fashion (Le Prell and Lobarinas, 2015; Lynch , 2016; Mukherjea , 2015).

Across the pharmaceutical industry, a range of experiments and strategies have been initiated with the goal of improving current levels of R&D productivity. Different frameworks have been developed applying the learnings and insights gained through analyses of the history of successful and unsuccessful drug discovery activities to enable informed decision making and increase PoS, with AstraZeneca publishing their “5Rs Framework” (Cook , 2014), and Merck sharing their Translational Medicine Guide decision framework (Dolgos , 2016). A recent update from AstraZeneca published an analysis of their portfolio changes since the application of the “5Rs Framework,” and the authors claim to have realised nearly a fourfold increase in PoS from 5% to 19% in progressing from Phase II to Phase III (Morgan , 2018). The team from AstraZeneca also highlighted a sixth factor—the right culture, where “truth seeking” is encouraged by rigorous and quantitative decision-making.

Hearing loss therapeutic drug discovery is focused around sensorineural dysfunction, but different underlying pathologies may be driving the hearing loss. It is essential that the correct underlying pathology is diagnosed when a patient presents with hearing loss or dysfunction to ensure that the most appropriate intervention and medicine can be used and that the patient is not put at risk of harm (Primum non nocere—“first, do no harm.”). The population of hearing loss patients may be broken down into three groups, with patients suffering noise-induced hearing loss (NIHL), hearing loss due to ototoxicity, and ARHL (also termed as presbyacusis) (Lustig, 2018). With respect to the last patient group, it is reasonable to assume that a significant proportion of individual patients presenting with suspected ARHL may have more than one underlying pathology due to various life events leading to a complex array of structural changes in the inner ear and so acquired hearing loss may be a better non-specific descriptor (Schuknecht and Gacek, 1993). Nonetheless, after considering all these patient groups and the putative pathologies driving auditory dysfunction, there are three therapeutic strategies available: otoprotection, restoration of hearing, and reduction of tinnitus.

Preventing hearing loss in patients where exposure to known ototoxic agents is anticipated in advance represents an attractive drug discovery opportunity, and it is probable that an otoprotective medicine in many circumstances may only require a limited dosing regimen. Drug ototoxicity induced hearing loss presents an opportunity for developing and establishing translational understanding in the uncharted hearing loss therapeutic area, and could potentially provide a platform from which to find medicines for the likely more heterogeneous pathology in other hearing loss patient groups with acquired hearing loss, such as ARHL.

The therapeutic strategy of protection against aminoglycoside or cisplatin ototoxicity is the most common approach taken in hearing loss drug discovery (Li, 2017; Schilder , 2019). A key consideration is that the otoprotective prophylaxis treatment does not reduce or change the efficacy of the antibacterial or anticancer ototoxic medicines respectively, and significant progress has been made in the discovery and advancement of prospective otoprotective therapies (Schilder , 2019). For example, a biological target agnostic phenotypic screen was undertaken using Zebrafish to identify compounds that could protect against aminoglycoside-induced hair cell loss using the lateral line neuromast hair cells of the Zebrafish as surrogates of human sensory hair cells (Owens , 2008). Oricula Therapeutics² undertook the medicinal chemistry lead optimisation of the screening hits, which led to the nomination of the molecule, ORC-13661, as the clinical candidate, and the physicochemical properties of this molecule meet the desired perceived oral drug-like chemical space described previously (Chowdhury , 2018). The new chemical entity ORC-13661 has recently been exclusively licensed by Decibel Therapeutics,³ and the renamed molecule, DB 041, is reported as being progressed to clinical trials for the prevention of hearing loss and balance disorders that can occur following treatment of severe infections with aminoglycoside antibiotics. In a phenotypic screening strategy as used to discover DB 041 (ORC-13661), the identity of the biological mechanism of action or the biological target of active molecules identified remains unknown and could be different for the different compounds that are initially identified as hits in the screen. Therefore, target deconvolution studies are required to enable biological target identification and confirmation of the mechanism of action. In an alternative drug discovery strategy, where the biological target is known, Sound Pharmaceuticals Inc.⁴ has advanced their investigational compound SPI-1005 (ebselen) into clinical evaluation for the prevention of aminoglycoside and cisplatin-induced hearing loss (ClinicalTrials.gov Identifier: NCT02819856 and NCT01451853, respectively). It is proposed that SPI-1005, a novel heterocyclic organoselenium molecule, mediates catalytic anti-oxidant activity and is a potent glutathione peroxidase mimetic (Kil , 2007; Lynch , 2005). The outcome of clinical studies with these and other molecules will provide data and information on the translational value of the ototoxicity preclinical models and drug discovery screening strategies (Schilder , 2019).

Prolonged or brief high-intensity episodes of noise can lead to NIHL, and, in certain situations such as the military and industrial processes, the timing of the noise exposure may be predicted. Although a range of cellular mechanisms that might have a role in NIHL has been described, this continues to be an emerging field (Kurabi , 2017; Le Prell , 2007). Potential therapeutic interventions against NIHL have been described (Lynch , 2016; Sha and Schacht, 2017) and include the investigational compound SPI-1005. In animal studies, it was found that glutathione peroxidase 1 (GPx1) is expressed in the cochlea in cell types that are involved in NIHL and that post noise exposure, GPx1 expression is reduced in these cochlear structures, resulting in cell injury and death (Kil , 2007). In animal models of NIHL, treatment with SPI-1005 reduced the extent of noise-induced cell injury and death, and acute and chronic NIHL was prevented (Kil , 2007). A human genetics study of a Chinese Han population identified that GPx1 SNP rs1987628 might be associated with the susceptibility to NIHL, so encouraging a possible role in human NIHL (Wen , 2014). Noise exposures can result in temporary threshold shifts (TTS) that occur and resolve over time and in a double-blind, placebo-controlled Phase 2 trial, healthy adults received SPI-1005 (200, 400, or 600 mg), or placebo orally twice daily for four days, beginning two days prior to a sound challenge (ClinicalTrials.gov Identifier: NCT01444846), the primary outcome measure was mean TTS at 4 kHz (Kil , 2017). Only healthy volunteers receiving the twice daily 400 mg SPI-1005 treatment evidenced a clinically relevant reduction in the noise-induced TTS at 4 kHz and neither the 200 mg nor 600 mg doses provided clinically significant benefit in preventing TTS, although significant effects were seen at these doses at other frequencies. Nonetheless, as SPI-1005 appears from human studies to be safe and tolerable after oral administration, and although the relevance of TTS to human auditory disorders is not yet fully understood, then SPI-1005 should be considered for evaluation in other types of acquired hearing loss or inner ear disorders where uncontrolled oxidation may be an underlying factor driving damage and functional impairment.

N-Acetyl cysteine (NAC) is another antioxidant molecule, which has been extensively studied, where the proposed mechanism of action for NAC is through the replacement of depleted intracellular glutathione (GSH) such as might occur in the cochlea as a consequence of noise exposure and efficacy with NAC has been observed in preclinical animal studies (Bielefeld , 2007; Kopke , 2015). However, translation to clear clinical benefit with NAC has not been observed in human subjects (Kopke , 2015; Sha and Schacht, 2017). In the absence of pharmacokinetic data, it is not possible to make an informed decision as to whether the mechanism of action of NAC has been tested, but it is reported from other human clinical studies that the oral bioavailability of NAC is low (∼4–10%) (Coles , 2018). A clinical pharmacokinetic repeat dose study in healthy volunteers and Parkinson's patients found similar oral bioavailability in human subjects and that 28 days of high dose NAC (6 g/day) administration did not lead to significant increases in brain GSH, which may be consequence of the low levels of NAC that would have been expected to have entered the systemic circulation following oral dosing (Coles , 2018). It is expected that there are similar challenges for a compound to access the inner ear tissue as the CNS because of the need to be able to cross the BLB. The likely low concentration levels of NAC drug in the blood and poor CNS permeation of NAC limit its ability to deliver efficacy in the clinic for prevention of hearing loss. The clinical example of NAC exemplifies the importance of not only understanding the pharmacokinetic profile of a proposed investigational drug but also the ability to measure or estimate the target tissue drug exposure.

The repurposing of known drugs for new indications offers the opportunity to leverage the available human safety and pharmacokinetic data of the known molecule for the new clinical application and potentially a shortened discovery and development path to a new medicine (Pushpakom , 2019). It may be that repurposing applies the same known reported mechanism of action of a medicine to a new indication or that a new previously unknown biological target activity has been revealed and this new biological mechanism of action is applied to the new indication of interest. An example of the latter is the small molecule investigational drug, SENS-401, being advanced by Sensorion Pharmaceuticals⁵ into Phase II clinical studies (AUDIBLE-S) in patients with severe or profound sudden sensorineural hearing loss (SSNHL) (ClinicalTrials.gov Identifier: NCT03603314 and EudraCT Number: 2018-000812-47). The small molecule compound SENS-401 is R-azasetron besylate, an enantiomer of Serotone^® which is marketed in Asia for chemotherapy-induced nausea and vomiting, where the primary pharmacology of compound SENS-401 is antagonism of the 5-HT3 receptor. However, SENS-401 has also been found to be an inhibitor of calcineurin, a calcium/calmodulin-dependent protein phosphatase, which is implicated in noise-induced trauma in the inner ear and subsequent NIHL (Dyhrfjeld-Johnsen, 2017). SENS-401 has a human pharmacokinetic profile suitable for oral dosing and recent in vivo pre-clinical data provided evidence of the distribution of SENS-401 into the inner ear tissues and perilymph of animals (Minami , 2004; Uemaetomari , 2005). The combined efficacy and pharmacokinetic data from in vivo NIHL animal studies provided insight that efficacy observed was most likely due to total drug exposure rather than maximal drug concentration and so informed the decision on the likely doses and dosing regimen to be used in the future clinical studies (Petremann , 2017; Petremann , 2019). With these extensive preclinical data and subsequent clinical data, the translational value of the preclinical models can be examined and insights into human inner ear biology will be established.

Specific biological targets or pathways that can prevent or ameliorate the impact of increased oxidative stress can provide opportunities for drug discovery. The detrimental effects of oxidative pathway activation and reactive oxygen species (ROS) observed in preclinical in vivo studies of hearing loss have been found to be ameliorated through elevation of the nucleotide nicotinamide adenine dinucleotide (+) (NAD+) levels (Brown , 2014; Kim , 2014a; Kim , 2014b; Kim , 2015). The intracellular levels of NAD+, an essential metabolic co-factor, decline both with age and with increased oxidative damage, so mechanisms that can elevate NAD+ levels may be protective against hearing loss due to oxidative stress (Chini , 2017). It is known that noise exposure causes hearing loss by inducing degeneration of spiral ganglia neurites and that administering nicotinamide riboside, an NAD+ precursor, in mice activates an NAD+-Sirtuin3 pathway reducing neurite degeneration caused by noise exposure (Brown , 2014). The cochlea has a high metabolic demand sustained by mitochondrial activity, but elevated ROS levels caused by noise trauma can lead to deleterious impact on multiple structures, and cell types within the cochlea, whilst even low-level ROS levels can induce inhibition of mitochondrial function and lead to significantly enhanced hair cell death (Baker and Staecker, 2012; Gonzalez-Gonzalez, 2017). The homeostasis of NAD+ involves a range of proteins, including CD38 (cyclic ADP ribose hydrolase) and NAMPT (nicotinamide phosphoribosyltransferase). Target tractability analysis would indicate that these proteins and others involved in the homeostasis of NAD+ levels are druggable to small molecules and so could make for an attractive drug discovery approach. Appropriate animal translational studies are required to determine the reason to believe in the science and to understand the pharmacokinetic-pharmacodynamic relationship prior to advancing to the clinic. Thereafter, an appropriate patient stratification strategy is required to ensure appropriate patient selection for clinical trial studies (Camacho-Pereira , 2016; Chini , 2017; Elhassan , 2017; Hickox , 2017; Kujawa and Liberman, 2019).

Restoration of hearing through a regenerative strategy represents an attractive proposition to patients, physicians, and payers, particularly if regenerative treatment strategies only require short, perhaps one-time, dosing regimens, and has attracted a variety of therapeutic approaches (Mittal , 2017; Simoni , 2017; Valente , 2017). Where clear durable hearing restoration and safety are achieved, the potential for local delivery into the inner ear to reduce systemic exposure may be a viable route of administration for the clinic and commercialisation. The direct topical route of administration has encouraged biological target strategies using gene vector platforms to advance. GenVec Inc.⁶ discovered CGF166, a recombinant adenovirus 5 vector containing a cDNA encoding the human atonal homologue 1 (Atoh1), and Novartis Pharmaceuticals are sponsoring the gene therapy clinical trial with CGF166 (ClinicalTrials.gov Identifier: NCT02132130) evaluating the direct induction of Atoh1 to effect sensory hair cell regeneration and hearing restoration (Baker , 2009; Staecker , 2016). The isolated but accessible nature of the inner ear will encourage further drug discovery and development activity using gene therapy to treat hearing dysfunction in the future (Ahmed , 2017; Landegger , 2017; Richardson and Atkinson, 2015).

Sensory hair cell regeneration has attracted interest as a potential therapeutic intervention strategy for hearing loss through the induction of the critical transcription factor Atoh1, which is necessary and sufficient for sensory hair cell generation in the developing embryo (Bermingham , 1999; Costa , 2017). A modest recovery of hearing was afforded following in vivo topical treatment of mice with LY411575, a small molecule inhibitor of gamma-secretase, which can induce Atoh1 induction through inhibition of the Notch pathway (Mizutari , 2013). Gamma-secretase is a tractable biological target for which small molecule drugs can be found, enabling potentially a drug repurposing opportunity or a shortened drug discovery activity. Indeed, LY3056480, a novel small molecule gamma-secretase inhibitor, was rapidly optimised for intratympanic delivery from a known chemical class of gamma-secretase inhibitors and is being advanced by Audion Therapeutics⁷ for the reversal of hearing loss in an ongoing Phase I/II multiple ascending dose open-label safety and efficacy study (REGAIN) in patients with mild to moderate sensorineural hearing loss (EudraCT Number: 2016-004544-10). In another regenerative approach, Frequency Therapeutics⁸ have selected FX-322 as a clinical candidate, to be delivered via intratympanic administration for the restoration of hearing using their proprietary progenitor cell activation science (McLean , 2017). A first in human safety clinical study of FX-322 (ClinicalTrials.gov Identifier: NCT03300687) in adults undergoing cochlear implantation has completed, where in addition to the safety evaluation, pharmacokinetic measurements of systemic drug exposure and cochlear perilymph pharmacokinetics in each patient were recorded. In light of these data, FX-322 was taken forward into a recently completed Phase I/II randomised, double-blind, placebo-controlled, single-dose study with FX-322 administered by intratympanic injection using a hydrogel formulation in adults with stable sensorineural hearing loss (ClinicalTrials.gov Identifier: NCT03616223). Frequency Therapeutics announced that data from this Phase I/II clinical trial indicated FX-322 was safe and well-tolerated following a single intratympanic injection with no serious adverse events. Additionally, improvements in hearing function were claimed for audiometry and word scores in multiple FX-322 treated patients; however, at present, no publication of these clinical data is available.

Recent advances in the hearing science field have led to the concept of “hidden” hearing loss, wherein the quality of hearing is affected in individual patients despite presenting with healthy audiograms, and has been hypothesised to be caused by cochlear synaptopathy (Kujawa and Liberman, 2015; Liberman and Kujawa, 2017). The role of cochlear synaptopathy in hearing loss and tinnitus is not without some controversy, and its significance in “hidden” hearing loss is yet to been established conclusively (Guest , 2017; Guest , 2018) with further investigations proposed (Dewey , 2018). There are concerns about the translation of synaptopathy with uncertainty as to whether the identified electrophysiological characteristics of synaptopathy in humans are the same as recorded in the animal models, since hearing dysfunction in patients may involve a mixture of pathologies that includes other structures in the cochlea and auditory machinery (Hickox , 2017; Kobel , 2017). Whilst the role of synaptopathy in human hearing impairment remains inconclusive with conflicting conclusions from published studies, there remains the need to further investigate this possible mechanism of hearing loss and will require improved methodologies for diagnosing synaptopathy and agreement within the auditory field to standardise experimental approaches (Bramhall , 2019). Two neurotrophins, brain-derived neurotrophic factor (BDNF) and neurotrophin 3 (NT3), have been implicated to play essential roles in the development of the neuronal connections to the sensory cells of the cochlea (Ramekers , 2015; Suzuki , 2016). Treatment with these growth factors or their mimetics may prevent auditory nerve degeneration or may even enable the reversal of synaptic loss through regeneration. Such a drug discovery strategy would involve the modulation of neurotrophic receptors; however, it has been very challenging to find small molecule modulators of the neurotrophic receptors (Josephy-Hernandez , 2017; Longo and Massa, 2013). Recently published data suggests that previously identified small molecule TrKB agonists do not activate TrkB signalling (Boltaev , 2017). In the absence of small molecule modulators, natural and modified neurotrophin proteins or antibody agonist and antagonist molecules have been developed as tools and potential therapeutics (Rosentha and Lin, 2014; Sahenk , 2010). Ensuring delivery of large-sized proteins or antibody macromolecules to their target tissue is a continuing challenge; however, the access to the inner ear via topical delivery may yet enable specific and limited drug administration. The key to the effectiveness of these types of agents as medicines will be ensuring appropriate innate stability of the bioactive molecule and pharmacokinetics that afford exposure and duration in the inner ear tissue to deliver the desired efficacy with acceptable safety as well as a patient dosing regimen that is sustainable. A proprietary sustained-exposure formulation of BDNF, OTO-413, has been nominated by Otonomy, Inc.⁹ for clinical evaluation, who plan to initiate a Phase I/II clinical trial in 2019 in patients with speech-in-noise hearing loss, however as yet no further details are available from the US FDA and European clinical trial registries.

The most advanced regenerative therapeutic strategies for hearing loss involve auditory nerve ribbon synapse repair and sensory hair cell regeneration, but restorative strategies should not be restricted to just these cell types, as other structures in the inner ear, such as the stria, are essential and vulnerable to damage that can lead to hearing loss (Ohlemiller, 2009; Schuknecht and Gacek, 1993; Ulehlova, 1983). The high metabolic activity of the stria to maintain the required endocochlear potential to sustain full auditory performance suggests that there may be potential pathways and targets in the stria tissue that may play a role in auditory dysfunction, and recent publications have suggested the involvement of S1P signalling in human hearing loss (Chen , 2014; Ingham , 2016). The combination of the possible human genetic association with hearing loss and an immune system pathway which contains biological targets that are believed to be druggable with small molecule therapeutics makes for a potentially attractive drug discovery strategy.

The early science and hope around regenerative strategies for hearing loss drug discovery are exciting, but the translational understanding of the biology remains to be established. It is essential that the models used in preclinical in vivo studies include animals of an appropriate maturity and reflect in some way the potential patient group to be treated, and do not include animals of an age where the auditory systems have not yet reached full maturity. Recent published studies indicate that the responsiveness of sensory tissue in the organ of Corti to antagonism of Notch pathway changes with tissue maturity and is rapidly lost once inner ear development has completed, so translation of Notch inhibition strategy to drive sensory hair cell regeneration through Atoh1 induction within the mature human damaged inner ear is uncertain (Costa , 2017; Maass , 2015; McGovern , 2018). The epigenetic temporal regulation of Atoh1 sensory hair development in mammalian cochlea has been reported and inner ear specific epigenetic changes and regulation have been described (Avraham , 2018; Doetzlhofer and Avraham, 2017; Stojanova , 2016). The epigenetic changes that lead to non-heritable changes in gene expression have become a significant drug discovery opportunity as many of the epigenetic biological targets have been found to be amenable small molecule drug discovery with compounds advancing in the clinic for other non-auditory indications (Conway , 2016; Knapp and Weinmann, 2013).

The treatment period post-trauma or cellular damage in humans where clinical intervention would be possible is unclear and how available animal models will enable translation is unknown, although it is encouraging that cochlea implantation in humans can be successful for many years following the loss of hearing (Pfingst , 2015). For patients, one key area to be addressed is the development of diagnostic methods for ensuring correct identification of the underlying disease pathology to allow selection of the appropriate regenerative treatment. The specific clinical indication that potential regenerative treatments seek to address will need to be well-defined to enable a medicine label, and reimbursement, so active collaborations between industry, patients, clinicians, payers, and regulators are required.

Tinnitus, characterised as the perception of sound such as ringing or buzzing in the absence of external auditory stimuli, can be hugely burdensome for patients, but any medicine for the treatment of tinnitus must possess an appropriate long-term safety profile for a non-life-threatening disorder (Baguley , 2013). A wide range of non-drug interventions to support patients with tinnitus symptoms have been developed, but a medicine with proven efficacy for the treatment of chronic tinnitus has remained elusive (Langguth , 2019; Zenner , 2017). A significant challenge for the clinical development of an effective drug for tinnitus is that outcome measures need to be objective rather than self-reported, but no reliable or reproducible objective measures have as yet been identified (Jackson , 2019). Measurements must also be able to distinguish between tinnitus efficacy and improved hearing resulting from drug treatment (Maldonado Fernández , 2015). TINNET,¹⁰ an ongoing initiative aims to build consensus and unify the outcome measures for future tinnitus clinical studies (Hall, 2017).

For tinnitus, an agnostic biological target strategy using phenotypic screening is not possible as there are no available in vitro assays and there is a lack of a predictive in vivo animal model (Eggermont, 2016; Langguth , 2019). It has been proposed that the majority of tinnitus symptoms may be as a consequence of CNS plasticity and neuronal activity changes (Schaette, 2014; Schaette and McAlpine, 2011). It is currently unclear whether a chronic treatment regimen is required to modulate aberrant CNS activity, or whether a short acute treatment period can be sufficient. It may be that there are differences in the required treatment duration, depending upon how long the tinnitus symptoms have been present. More importantly, there is a lack of biological targets tractable to small molecule drugs, although such biological targets may be revealed as the genetics of tinnitus is resolved (Lopez-Escamez , 2016).

The human genome contains more than 400 genes that encode for ion channels, and these have provided druggable biological targets, as small molecule ion channel CNS medicines have been discovered (Cox, 2015; Gosling, 2015; Imbrici , 2016). A repurposing therapeutic strategy to reduce increased spontaneous neuronal activity, and so potentially ameliorate tinnitus symptoms, led to the evaluation in an in vivo behavioural model of tinnitus of KCNQ2–5 channel openers, including ezogabine, which was used as an adjunctive treatment for epilepsy, as well as the structural analogue SF0034, which is claimed to have improved channel opener selectivity for the KCNQ2/3 sub-type (Kalappa , 2015; Li , 2013). The study found that SF0034 inhibited the development of tinnitus in mice, but that neither of these agents was effective in reversing established tinnitus in the animal models tested. Autifony Therapeutics¹¹ developed small molecule modulators of Kv3 ion channels, which are expressed in fast-spiking neurons throughout the auditory brainstem and enable neurons to sustain high firing rates (Chambers , 2017). Following noise exposure or with ageing, the Kv3.1 current can be reduced, and so compounds that increase Kv3.1 current may have therapeutic potential. Autifony Therapeutics advanced AUT00063, a small molecule Kv 3.1 channel opener compound, to a 28-day randomised placebo-controlled double-blind Phase IIa study (QUIET-1) to investigate the efficacy and safety in subjective tinnitus (ClinicalTrials.gov Identifier: NCT02315508 and EudraCT number: 2014-002179-27). Analysis of the clinical trial data found no safety issues but confirmed a lack of efficacy following 28 days treatment with AUT00063 (800 mg) in patients with mild-to-moderate tinnitus. A key challenge for evaluating the outcome of clinical trials is accessing data that enables drug engagement with the biological target of interest to be determined and whether appropriate drug concentrations were achieved at the desired site of action for the required duration. Without these data, a fully informed decision about whether a biological target has been tested in a therapeutic intervention cannot be determined.

A recent publication demonstrating translation from a preclinical in vivo guinea pig model to an experimental medicine study in patients with tinnitus symptoms provides encouragement that tinnitus may be reversed in suitable patients (Marks , 2018). The clinical study (ClinicalTrials.gov Identifier: NCT02974543) found that 28 days of bimodal auditory and somatosensory stimulation led to a temporary reduction in patient-reported tinnitus loudness and intrusiveness, and long term depression of the dorsal cochlear nucleus was proposed as a treatment option for tinnitus patients. However, it is not yet known whether the benefits of bimodal auditory and somatosensory stimulation treatment can be made durable or whether drug tractable biological targets can be identified to safely deliver equivalent clinical benefit.

Medicine discovery in auditory disorders is immature, but the field is entering a new phase where novel targets and clinical candidates are in or shortly entering clinical evaluation (Schilder , 2019). Specific niche indications such as drug-induced ototoxicity hearing loss provide opportunities to bring forward treatments where the well-defined patient groups can be identified and the opportunity to build confidence in the translational understanding of these forms of hearing loss is presented. The outcomes from these clinical studies will inform the translational understanding and the therapeutic opportunities and guide the future science of the currently uncharted hearing loss therapeutic field. Nonetheless, to maximise the successful development of novel drug therapies for the inner ear, diagnostics are required to enable the underlying specific pathology of inner ear disorders to be fully understood and the appropriate selection of patients for each advanced treatment. Methodologies for measuring inner ear target engagement and pharmacokinetics will also be required. The clinical development of these novel drugs and the prioritisation and progression of new novel biological targets into drug discovery can take advantage of the many learnings that the pharmaceutical industry has realised from past pharmaceutical R&D activities to deliver future medicines of value to the patients.

The author thanks Andrew Benowitz for the helpful comments on an earlier version of this manuscript. R.P.C.C. is employed by Cinnabar Consulting Ltd., and preparation of this manuscript was supported by the visiting professorship at UCL Ear Institute. The views, opinions, and/or findings contained in this article are those of the author and should not be construed as an official UCL Ear Institute position, policy, or decision unless so designated by other documentation. R.P.C.C. reviewed the literature and wrote the manuscript.