Background It is well established that biallelic mutations in transmembrane protease, serine 3 (TMPRSS3) cause hearing loss. Currently, there is controversy regarding the audiological outcomes after cochlear implantation (CI) for TMPRSS3-associated hearing loss. This controversy creates confusion among healthcare providers regarding the best treatment options for individuals with TMPRSS3-related hearing loss.
Methods A literature review was performed to identify all published cases of patients with TMPRSS3-associated hearing loss who received a CI. CI outcomes of this cohort were compared with published adult CI cohorts using postoperative consonant-nucleus-consonant (CNC) word performance. TMPRSS3 expression in mouse cochlea and human auditory nerves (HAN) was determined by using hybridisation chain reaction and single-cell RNA-sequencing analysis.
Results In aggregate, 27 patients (30 total CI ears) with TMPRSS3-associated hearing loss treated with CI, and 85% of patients reported favourable outcomes. Postoperative CNC word scores in patients with TMPRSS3-associated hearing loss were not significantly different than those seen in adult CI cohorts (8 studies). Robust Tmprss3 expression occurs throughout the mouse organ of Corti, the spindle and root cells of the lateral wall and faint staining within <5% of the HAN, representing type II spiral ganglion neurons. Adult HAN express negligible levels of TMPRSS3.
Conclusion The clinical features after CI and physiological expression of TMPRSS3 suggest against a major role of TMPRSS3 in auditory neurons.
- otorhinolaryngologic diseases
- clinical decision-making
- disease management
- gene expression profiling
Data availability statement
Data are available on reasonable request.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
- otorhinolaryngologic diseases
- clinical decision-making
- disease management
- gene expression profiling
WHAT IS ALREADY KNOWN ON THIS TOPIC
Biallelic variants in transmembrane protease, serine 3 (TMPRSS3) result in congenital and postlingual sensorineural hearing loss.
Previous reports suggest TMPRSS3 is expressed in spiral ganglion neurons (SGNs) and portend for poor cochlear implantation (CI) outcomes in patients with TMPRSS3-related hearing loss.
WHAT THIS STUDY ADDS
We find that CI outcomes in patients with TMPRSS3-related hearing loss are similar to other cohorts of CI patients.
Tmprss3 is not expressed in acoustic stimuli-transmitting type I SNGs of mice or humans.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE AND/OR POLICY
This study shows that patients with biallelic TMPRSS3-related hearing loss have excellent CI performance and argues against the physiological role of TMPRSS3 in SGNs.
Transmembrane serine proteases are a large family of proteins and indispensable regulators within many tissues, including the inner ear.1 Broadly, they are classified based on how they are anchored to the cellular membrane. Type II transmembrane serine proteases are anchored by their N-terminus to the cellular membrane and contain a group A scavenger receptor domain and an active serine protease domain.1 2 Transmembrane protease, serine 3 (TMPRSS3) is a type II transmembrane serine protease required in the mammalian auditory system for proper hearing.3 4
TMPRSS3 is essential for cochlear hair cell (HC) survival.5–7 Defects or ablation of TMPRSS3 results in hearing loss in both mice and humans. Mice carrying a protein-truncating TMPRSS3 mutation exhibit normal cochlear and vestibular HC development followed by rapid HC degeneration over 48 hours starting at postnatal day 12 (P12).5 In humans, biallelic pathogenic variants of TMPRSS3 are the most common causative hearing loss gene in adults undergoing traditional and hybrid cochlear implantation (CI).8 Additionally, TMPRSS3 is the fifth most common gene causally associated with deafness in a multiethnic cohort of human congenital deafness.9 Furthermore, select missense mutations (eg, A306T, A138E, A426T) are linked to postlingual onset high-frequency hearing loss.10 Despite this robust and dramatic phenotype, the function of TMPRSS3 and the pathomechanism(s) that underlie TMPRSS3-related deafness remains elusive.
Cochlear implantation is the gold-standard treatment for inherited severe-to-profound sensorineural hearing loss (SNHL).11 12 Implants function by directly stimulating spiral ganglion neurons (SGNs), permitting sound detection and speech recognition while circumventing inner ear organs such as cochlear HCs.6 11–13 Currently, the postoperative CI outcomes in patients with TMPRSS3-associated hearing loss remain controversial in the published literature.6 10 13 TMPRSS3-related damage to the SGN has been proposed as one possible explanation for poor CI outcomes in these patients,13 evidenced by TMPRSS3 immunostaining within SGNs in the murine cochlea.14 However, recent single-cell RNA-sequencing (scRNA-seq) studies have demonstrated that Tmprss3 is not expressed in type I SGNs, but only in type II SGNs—which do not transmit auditory signals and make up only 4% of the SGN population.15 16 This disconnect between biological data and patient outcomes prompted us to critically review published reports regarding TMPRSS3 patient CI outcomes and perform in-depth expression studies of TMPRSS3 in the human and mouse auditory system.
Here, we show that patients with TMPRSS3-related hearing loss have overall good postoperative performance and a TMPRSS3 genetic diagnosis should not be a reason to restrict cochlear implantation. In addition, using two different methods, we show that TMPRSS3 is not expressed in the sound transducing neurons of the cochlea and is primarily localised to the sensory epithelium portion of the cochlea. Finally, we call into question the correlation between negative CI performance and TMPRSS3-related hearing loss.
Materials and methods
TMPRSS3-associated cochlear implant outcomes
Two independent searches of PubMed for articles published on TMPRSS3-associated hearing loss treated with CI were conducted. Search terms included the combination of ‘TMPRSS3’ with the major subheading topics ‘hearing loss’ and ‘cochlear implant’. All articles reporting CI outcomes for TMPRSS3-associated hearing loss were identified. Articles that contained subjective (favourable or unfavourable) or objective (speech scores) outcome measures were reviewed. For the adult CI cohort analysis, articles that contained 10 patients and had postoperative consonant-nucleus-consonant (CNC) word scores were reviewed. Nine articles fit these criteria and were included in this study. Demographics, age of SNHL onset, phenotypic presentation, documented TMPRSS3 mutations, patient age at time of CI and audiologic outcomes were recorded. Individuals with hearing loss prior to 2 years of age were classified as having prelingual hearing loss whereas those with onset 2 years of age or older were designated as having postlingual hearing loss. Objective outcomes included preoperative and postoperative pure tone averages (PTA), CNC word recognition scores (WRS) and sentence scores (SS). Subjective outcomes, when reported in published articles, were classified as either favourable or poor: postoperative implant-aided CNC WRS <30% were considered poor outcomes, and scores 50% or greater were considered favourable. Change in PTA and CNC WRS scores were also calculated.
Objective data were evaluated by measures of central tendency using Microsoft Excel, 2016. The mean age difference between favourable and poor postlingual CI performers was evaluated using a t-test using SPSS Statistics, 2020. CI outcomes in patients with TMPRSS3-associated hearing loss were compared with postoperative outcomes of eight representative CI studies in the general hearing loss population. Logistic regression was performed to determine the effects of possible predictor variables on favourable CI outcomes in the TMPRSS3 population. Linear regression was similarly performed to evaluate possible predictor effects on average change (delta) dCNC and dPTA. Mean postoperative CNC WRS analysis of variance (ANOVA) of the two populations was performed using SPSS.
Tmprss3-mutant mice (Tmprss3Y260X ) have been described previously.5 All mice were housed in the animal facility of Indiana University School of Medicine. Genotyping was performed as previously described.3
Human tissue harvest
Human auditory nerves (HAN) were harvested during vestibular schwannoma operation. During the tumour removal, the auditory nerve was dissected away from the schwannoma and facial nerve within the internal auditory canal (IAC). The auditory nerve between the modiolus of the cochlea and porus of the IAC was harvested. Immediately after harvest, the tissues were flash-frozen in liquid nitrogen and stored at −80°C before RNA extraction for qRT-PCR. For hybridisation chain reaction (HCR) experiments, the auditory nerves were immediately fixed in 10% formalin for 24 hours prior to embedding.
The coding sequence (CDS) of human TMPRSS3 isoform 1 from transcript variant A (NM_024022.4) with added 5’ EcoRI and 3’ KpnI restriction sites was generated as gBlocks Gene Fragments (Integrated DNA Technologies, Corralville, Iowa, USA). The construct p3XFLAG-CMV 7.1_TMPRSS3 was generated from the backbone of p3XFLAG-CMV 7.1_syn6 (Addgene, 50012). The Syntaxin-6 gene was replaced by TMPRSS3 gBlocks with EcoRI and KpnI restriction enzymes.
HEK293 cells (ATCC, Manassas, Virginia, USA; CRL-1573), plated on 1% gelatin-coated cover slide in 6-well plates and were transfected with p3XFLAG-CMV 7.1_TMPRSS3 plasmid using Lipofectamine 3000 Reagent (Thermo Fisher, Waltham, Massachusetts, USA; L3000015) for 24 hours as described by the manufacturer. Cells were washed once with ice-cold phosphate-buffered saline (PBS) and fixed with 4% paraformaldehyde for 20 min at room temperature (RT).
HAN were fixed in 10% formalin for 24 hours. Then, fixed samples were cryoprotected in 30% sucrose in PBS. After embedding in TFM Tissue Freezing Medium (General Data, TFM-5), the auditory nerve was frozen and stored at −30°C. Sections of 6 μm were cut on a cryostat and air-dried for 2 hours.
Mouse cochleae were dissected from P11 mice anaesthetised with isoflurane. The mice were perfused with 1X PBS through left ventricle. The cochleae were dissected and ﬁxed in 4% paraformaldehyde overnight at RT and decalcified overnight at RT in solution containing 120 mM EDTA buffered in 1X PBS. Fixed cochleae were cryoprotected in 30% sucrose in PBS. After embedding in TFM Tissue Freezing Medium (General Data, TFM-5), cochleae were frozen and stored at −30°C. Midmodiolar cross-sections of 6 μm were cut on a cryostat and air-dried for 2 hours.
The HAN sections and TMPRSS3-transfected HEK cells were stained with human TMPRSS3 HCR RNA-FISH probes while the mouse cochleae sections were stained with mouse Tmprss3 and mouse myelin basic protein (MBP) HCR RNA-FISH probes (Molecular Instruments, Los Angeles, California, USA) following the manufacture’s protocol with incubations at 37°C (Corning LSE mini 6800). Twenty probes were designed for human TMPRSS3 transcript variant A (NM_024022.4), mouse Tmprss3 transcript variant 1 (NM_001163776) and mouse MBP transcript variant 1 (NM_001025251), which recognises 5’-untranslated region (UTR), 3’-UTR and CDS. Sections were mounted with ProLong Gold Antifade Mountant with DAPI (4′,6-diamidino-2-phenylindole) (Thermo Fisher, P36931) and visualised on Leica DMi8 microscopy.
Total RNA was extracted from HAN, HEK cells and TMPRSS3-transfected HEK cells with PureLink RNA Mini Kit (Invitrogen) and cDNA was synthesised from 50 ng total RNA with SuperScript IV VILO Master Mix (Invitrogen). qRT-PCR was performed using SYBR-Green and primers: TMPRSS3 sense: CAT CCA GGT GGG TCT AGT TTC; TMPRSS33 antisense: AGC CTC TTT GGC TTG TAC TT; MBP sense: GGC AAG GTA CCC TGG CTA AA; MBP antisense: GGG TGG TGT GAG TCC TTG TA; GAPDH sense: CAT CAC TGC CAC CCA GAA GAC TG; GAPDH antisense: ATG CCA GTG AGC TTC CCG TTC AG. The amplification and detection were performed on QuantStudio 5 Real-Time PCR System (Invitrogen). Relative expression for TMPRSS3 and MBP were adjusted to GAPDH levels.
The cells on the slide or tissue sections were blocked with 10% goat serum with 0.1% Tween-20 in PBS for 30 min at RT. Following this, sections were treated with primary antibodies in 3% goat serum with 0.1% Tween-20 in PBS at RT for 1 hour. The antibody we used were mouse anti-TUJ1 (BioLegend, 801202, 1:100). After washing with PBS, sections were incubated with secondary antibodies or Alexa Fluor 568 Phalloidin (Thermo Fisher, A12380, 1:200) in 3% goat serum with 0.1% Tween-20 in PBS at RT for 1 hour. Finally, sections were washed and mounted with ProLong Gold Antifade Mountant with DAPI (Thermo Fisher, P36931). Staining was visualised on Leica DMi8 microscopy.
TMPRSS3 hearing loss causing variants
Eighty-six TMPRSS3 deafness-causing variants were curated from the published literature and the Deafness Variation Database (https://deafnessvariationdatabase.org/)17 or via a PubMed Search (July 2021). Variants were classified as pathogenic, likely pathogenic or variants of uncertain significance (VUS) according to the hearing loss-specific American College of Medical Genetics (ACMG).18
Variants throughout the TMPRSS3 gene cause hearing loss
Of 86 variants curated in this study, 62 were classified as pathogenic/likely pathogenic according to the ACMG criteria while the remaining 24 are considered VUS. The 62 pathogenic/likely pathogenic TMPRSS3 mutations map across all protein domains which are associated with hearing loss (figure 1A, online supplemental table 1). Of these, 35 (~56.5%) are missense or non-protein truncating (NPT) variants (figure 1A–C). The remaining protein truncating (PT) mutations include 10 (~16%) nonsense variants, 8 (~13%) frameshift indels and 9 (14.5%) splice-altering variants (figure 1A–C).
Patients with TMPRSS3 variants have excellent speech recognition outcomes with cochlear implants
A total of 9 articles describing 27 patients (30 ears) who received CI for TMPRSS3-associated hearing loss were identified (table 1).19–22 After a review of the variants identified in individuals with reported CI outcomes, one individual reported in the study by Miyagawa et al 23 was removed due to the reported variant c.212T>C (p.Phe71Ser), being present in >1% of control alleles in gnomAD.23 Of the 19 patients with gender recorded, 13 (68%) were female and 6 (32%) were male. Of 21 patients with data available, 14 (67%) were white and 7 (33%) were Asian (table 1). Postlingual hearing loss occurred in 80% of patients.
The age of implantation was widely variable with a mean (SD) age of 19.8 (±20.2) years (table 1). Favourable hearing outcome after implantation was reported by 85% of patients. The mean PTA improved from 82.4 (±18.2) dB to 26.7 (±6.4) dB with an average change in PTA (dPTA), defined as postoperative PTA minus preoperative PTA, of −55.7 (±17.9). Multivariable linear regression found no significant difference between dPTA and age (0.954), although higher preoperative PTA (worse hearing) was associated with a greater dPTA (b=−0.284, p<0.001). Average dCNC, defined as postoperative CNC score minus preoperative score, was 48.3% (±14.7%) (table 1), with an average of 17.2% (±15.2%) preoperatively and 65.5% (±26.3%) after implantation. Multivariable linear regression found no significant association between dCNC and age (p=0.973) or preoperative CNC (p=0.124). Poor postlingual performers were significantly older than good performers (t(17.4)=4.7, p=0.018). Multivariable logistic regression found no significant association between age and subjective performance (p=0.413).
TMPRSS3 mutations contributing to hearing loss occurred on 13 different alleles in our cohort (figure 1D). Mutations were recorded for each of the two alleles present in each individual (table 1). The most common allele was the frameshift variant c.208delC (p.His70fs), which accounts for ~1/3 of total alleles (17/52) present in this cohort. Other common variants included c.916G>A (p.Ala306Thr), c.1276G>A (p.Ala426Thr) and c.413C>A (p.Ala138Glu), which comprised 10, 6 and 4 alleles, respectively. The remaining 10 variants accounted for 2 or 1 alleles each (figure 1D). Favourable and poor CI performance related to each variant was determined (figure 1D).
The most common allelic combination was PT by NPT, accounting for 14 cases (13 favourable outcomes vs 1 poor outcome). NPTxNPT accounted for the next highest allelic combination with eight cases (five favourable outcomes vs three poor outcomes), and four cases had PTxPT (four favourable outcomes vs zero poor outcomes). Statistical analysis (χ2 test) could not be performed because several conditions had fewer than five events.
Four patients reported subjectively poor outcomes. Three of the four patients carried the c.413C>A (p.Ala138Glu) allele (table 1 and figure 1D), although the second allele was different in each case (p.Arg216Cys, p.Ala306Thr and p.Ala426Thr). The remaining patient also carried the c.1276G>A (p.Ala426Thr) allele along with the c.208delC (p.His70fs) allele. The c.1276G>A (p.Ala426Thr) allele was observed in five patients with favourable outcomes, while the c.413C>A (p.Ala138Glu) allele was observed in one favourable outcome in combination with the c.595G>A (p.Val199Met) allele. Alleles p.His70fs and p.Ala306Thr were also seen in patients with favourable outcomes (table 1). Of the four poor performers, two (both with p.Ala138Glu allele) had experienced hearing loss for >25 years while the remaining two (one with p.Ala138Glu allele, both with p.Ala426Thr allele) did not have age at onset recorded, but were in their 50s and 60s, respectively, at time of implantation. The p.Ala138Glu and p.Ala426Thr mutations typically results in postlingual progressive high-frequency SNHL prior to age 15 years.10 Therefore, it is likely that both of these patients had a significant duration of hearing loss prior to implantation, which is associated with poor CI performance.24 Both mutations are, however, typically associated with less severe hearing loss than other TMPRSS3 mutations,10 such as p.Val199Met and p.Ala306Thr, which were not present in as high of a proportion of poor CI performers. Additionally, one patient (case 8) who carries both the p.Ala306Thr and the p.Ala138Glu variants, required revision CI surgery due to hardware infection.
CI outcomes in patients with TMPRSS3-mediated hearing loss are similar to postlingual adults
Nine large studies of >10 patients reported postlingual adult CI outcomes for 565 individuals and these patients exhibited a mean postoperative CNC word score of 51.2% (22.2%) (table 2).25–32 Of the seven studies with preoperative and postoperative CNC scores recorded, the average improvement in CNC score (defined as final score minus initial score) was 40.4%. An ANOVA test found no significant difference in postoperative CNC WRS between the TMPRSS3 and general hearing loss cohorts (66.2% (25.8) vs 50.1% (12.5); F(1,6)=1.97, p=0.21).
TMPRSS3 expression is minimal in auditory neurons
To determine the spatial and quantitative expression of Tmprss3, we performed HCR on mouse cochlea at P11. We observe robust expression of Tmprss3 in cells throughout the organ of Corti including Myo7A+ inner and outer hair cells (figure 2A–D). Tmprss3 is also detected in the spindle and root cells of the lateral wall, but no expression is detected in the stria vascularis (figure 2A–C). Very faint expression is detected in <5% of the SGNs, but not in the peripheral neuronal processes near the habenula perforate (figure 2E,F). Based on the location and low quantity of these SGN cells, they most likely represent type II SGNs.
Next, we analysed the expression of Tmprss3 in the publicly available Gene Expression Analysis Resource (http://umgear.org).33 We compared the expression of Tmprss3 across SGN subtypes (type Ia, Ib, Ic and type II) across three scRNA-seq datasets from SGNs (online supplemental figure 1).15 16 34 Consistent with our HCR data, we observed no expression of Tmprss3 in all three subtypes (A, B, C) of type I SGNs across all three datasets (figure 2 and online supplemental figure 1). The neuronal marker TuJ1 (Tubb3 gene) was expressed across all subtypes of SGNs and all datasets (figure 2 and online supplemental figure 1). One dataset demonstrated expression of Tmprss3 in type II SGNs in P25–P27 mice, but expression in type II SGNs was not seen in mice aged 9 weeks (figure 2 and online supplemental figure 1).
Next, we harvested HAN from three patients undergoing translabyrinthine craniotomy for vestibular schwannoma surgery. As expected, these patients had preoperative hearing loss with speech reception thresholds (SRT) between 30 and 45 dB and word discrimination scores of 0% (patients 1 and 2) and 84% (patient 3) (figure 3A). To confirm the specificity of our human HCR probe and for qRT-PCR, we cloned the human TMPRSS3 gene into an expression vector and expressed TMPRSS3 in HEK293 cells. Using qRT-PCR on HAN, control HEK cells and transfected HEK293 cells, we observed robust expression of TMPRSS3 in the transfected cells and only minimal expression of TMPRSS3 in the HAN (figure 3B). As expected, we observed robust expression of MBP in HAN and not in HEK cells (figure 3C). Next, we performed TMPRSS3 HCR and confirmed the specificity of the probe using TMPRSS3 transfected HEK293 cells (figure 3D,E). We did not detect TMPRSS3 expression within the HAN of either patients 2 or 3 (figure 3F,G). We confirmed the presence of auditory neurons using TuJ1 immunostaining (figure 3H,I).
Determining factors that impact CI performance is critical for identifying CI candidates and managing patient expectations. CI performance is linked to both age of implantation and hearing loss aetiology.24 35 36 As genomic sequencing has become more feasible, increasing efforts are being made to determine which genetic causes may predict good or poor outcomes after implantation—and, consequently, how these patients should be counselled preoperatively. Much of this interest has centred on TMPRSS3 and the still poorly understood mechanism by which mutations to this gene adversely affect hearing and, possibly, CI outcomes. In this study, however, we find that patients with TMPRSS3-related hearing loss do not have worse CI outcomes compared with the general population. Furthermore, we demonstrate that Tmprss3 is highly expressed in otic sensory epithelia with limited SGN expression.
Poor CI outcomes in TMPRSS3-associated hearing loss were originally attributed to TMPRSS3 expression in SGNs.13 However, these data were obtained with mouse tissue using an antibody that was not validated for specificity with methods such as Tmprss3-knockout tissue or western blot analysis.14 Our compiled CI outcomes data here indicate that TMPRSS3 does not play a significant physiologic role in the neuronal aspects of hearing. Our data using validated and highly specific probes for Tmprss3 in mouse and TMPRSS3 in human tissues corroborates published scRNA-seq datasets from SGNs (online supplemental figure 1),15 16 34 the sensory epithelia37 38 and the lateral wall.39
Postimplantation outcomes in TMPRSS3 patients compared with the general population.25–28 40 These outcomes compare similarly to a large systematic review of adult CI outcomes of 2798 patients across 46 studies and showed CNC scores improved from 8.3% (12.4) to 54.0% (22.5).41 The absence of significantly different outcomes between the TMPRSS3 cohort and the general population also suggest that TMPRSS3 may have no clinically relevant biological role in the auditory nerve.
Duration of deafness is one of the most important factors to predict speech perception after cochlear implantation in postligually deaf patients.24 This may be due to progressive degeneration of SGNs. Mouse models mimicking human deafness with degeneration of sensory hair cells uniformly exhibit progressive neuronal death, a secondary effect to the primary lesion.42–44 Loss of TMPRSS3 in mouse does not alter the total number of sensory hair cells or total number of SGNs during development,3 yet we and others have shown that loss of TMPRSS3 in mouse results in rapid HC degeneration at P12 leading to profound deafness.3 7 Similar to other genetic causes of hearing loss in which the pathology affects the HC or the supporting cells of the sensory epithilium, Tmprss3-mutant mice also exhibt a delayed-onset progressive SGN degeneration started after P90.3 The selected TMPRSS3 patients who exhibited poor performance were all implanted at a significantly older age—and, in at least two cases, with prolonged duration of hearing loss, which most likely affected the overall number and health of residual SGNs. These factors likely played a substantial role in their poor outcome.45
In order to assess possible genotype-phenotype correlations among various TMPRSS3 allelic combinations, we compared hearing outcomes in patients with PTxPT, NPTxPT or NPTxNPT combinations. No difference in performance between these groups.
Taken together, patients with TMPRSS3-associated SNHL have comparable CI outcomes relative to the general population. Duration of deafness appears to be the key factor predictive of poor performance. While the cause of deafness in patients harbouring TMPRSS3 mutations remains poorly understood, our results suggest that TMPRSS3 mutations largely affect cochlear hair cells and not the SGN, which would explain the positive CI outcomes in these patients.
Data availability statement
Data are available on reasonable request.
Patient consent for publication
This study was approved by institutional review board of Indiana University School of Medicine (approval 200634852). Participants gave informed consent to participate in the study before taking part. All animal procedures were carried out in accordance with the approval of the Indiana University Institutional Animal Care and Use Committee protocol.
The authors are grateful to all patients for consenting to biological sampling and data collection
Contributors Conception and design: KTB and RFN. Data collection and analysis: Y-SC, EC, BJT, TJS, JVM, DJT, KTB and RFN. Performed experiments: Y-SC, EC, BJT, TJS, JVM, DJT and RFN. Drafted original manuscript: Y-SC, BJT, KTB and RFN. All authors have reviewed and approved the final manuscript. Guarantor: RFN.
Funding This work was supported by the National Institutes of Health (K08-DC016034 to RFN), the Triological Society and American College of Surgeons (Clinician Scientist Development Award to RFN) and NIGMS T32 Training in Genetics Fellowship (T32 GM007748 to KTB).
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.