The study, published in the Archives of Dermatological Research by a team led by Ashley N. Houff and Monique Kumar of the University of Central Florida College of Medicine, together with colleagues at Mass Eye and Ear and Harvard Medical School, set out to answer a deceptively simple question: do individuals with port-wine stains or Sturge-Weber syndrome experience hearing loss more often than the rest of the population? The answer, based on both a systematic review of the medical literature and a survey of patient support communities, is a cautious but intriguing maybe, one that could reshape how dermatologists, neurologists, and audiologists think about these conditions.

The biological rationale for the investigation rests on a single gene. Most cases of both port-wine stains and Sturge-Weber syndrome are driven by somatic mutations in GNAQ, a gene encoding the alpha q subunit of a heterotrimeric G protein, a molecular switch that relays signals from cell-surface receptors into the interior of the cell. The landmark 2013 discovery in the New England Journal of Medicine showed that a specific activating mutation in GNAQ, arising after fertilization and therefore present in only a mosaic subset of the body’s cells, underlies both the isolated birthmarks and the full syndrome. Because the mutation occurs during embryonic development, its distribution across tissues is unpredictable, which helps explain why the same genetic lesion can produce anything from a small skin patch to a life-altering neurological disorder.

What makes the hearing question scientifically plausible is that G proteins are not strangers to the inner ear. Earlier work on the mammalian cochlea identified G protein isoforms localized presynaptically in the efferent nerve terminals, the fibers that modulate how hair cells transmit sound information to the brain. If GNAQ signaling participates in the maintenance or function of cochlear structures or their vascular supply, a mosaic mutation affecting those tissues could, in principle, compromise hearing. The literature also contains tantalizing case reports, including a description of port-wine stains associated with large vestibular aqueduct syndrome in a patient carrying mutations in both GNAQ and SLC26A4, a gene classically linked to hereditary hearing loss.

To gather the available evidence, the researchers first combed PubMed for published reports of hearing impairment in patients with either condition. The search identified hearing loss documented in three individuals with port-wine stains alone and in 77 individuals with Sturge-Weber syndrome. Prior reports have described several possible mechanisms, including vascular malformations of the external auditory canal, as in a documented case of multiple angiomas within the ear canal, and broader neuro-otological manifestations in patients with hemifacial port-wine stains. Yet the total number of cases remained small, and no systematic estimate of how common hearing problems might be had ever been attempted.

To fill that gap, the team turned to an unconventional but increasingly popular research tool: electronic surveys distributed through Facebook support groups for people with port-wine stains and Sturge-Weber syndrome. The survey asked members about their diagnosis, whether they had experienced hearing impairment or loss, and, for those who had, the characteristics of that loss, including its severity, duration, and treatment. Eighty-four responses were included in the analysis. Of these, 50 respondents, or 59.5 percent, had port-wine stains alone, 31, or 36.9 percent, had both port-wine stains and Sturge-Weber syndrome, and 3, or 3.6 percent, had Sturge-Weber syndrome alone.

The headline finding was striking: 53 of the 84 respondents, a sample proportion of 63.1 percent with a 95 percent confidence interval spanning 52.4 to 72.6 percent, reported hearing loss. The researchers are careful, and rightly so, to emphasize what that number does and does not mean. Because the sample was self-selected, drawn from people motivated enough to join support groups and respond to a survey, it cannot be taken as a measure of prevalence in the general population of people with these conditions. The confidence interval, calculated using standard methods for binomial proportions, reflects the precision of the sample estimate rather than a true population rate. Still, the sheer proportion of respondents reporting hearing difficulties is difficult to dismiss, particularly when compared with background rates of hearing loss in the United States, which rise steeply with age but remain comparatively low in children and young adults.

The characteristics of the reported hearing loss add further texture. Among the 53 respondents with hearing loss, the overwhelming majority, 36 people or 67.9 percent, did not know what type of hearing loss they had, a gap that itself underscores how rarely these patients undergo formal audiological evaluation. Those who did know reported sensorineural loss, arising from the inner ear or auditory nerve, in 7 cases, mixed loss in 6, and conductive loss, involving the outer or middle ear, in 4. Severity spanned the full clinical spectrum: 21 respondents described their loss as mild, 18 as moderate, and 13 as severe. Most strikingly, the largest group, 21 respondents or 39.6 percent, reported that their hearing symptoms had persisted for more than ten years, suggesting a chronic process rather than a transient one.

Treatment patterns revealed both the burden of the problem and the gaps in care. Thirty-four respondents with hearing loss, or 64.2 percent, reported receiving at least one form of treatment. Hearing aids were by far the most common intervention, used by 23 respondents, while 10 reported other surgeries, 5 reported other treatments, 2 reported medication, and one reported a cochlear implant. Because some respondents selected multiple treatment types, the individual percentages sum to more than the overall treatment rate. The predominance of hearing aids is consistent with the profile of chronic, largely irreversible hearing loss, and the presence of surgical interventions hints at the structural ear abnormalities that case reports have long suggested in this population.

The authors are explicit about the limits of their work. Self-perceived symptoms were not confirmed by audiometry, the gold-standard objective measurement of hearing thresholds, and the support-group sample cannot establish prevalence or prove an association between GNAQ-driven vascular malformations and hearing impairment. What the study does accomplish is to elevate a scattered handful of case reports into a coherent, testable hypothesis: that somatic GNAQ mutations, already known to disrupt blood vessels in the skin, brain, and eye, may also affect the delicate vascular and neural architecture of the auditory system. The authors call for future studies with objective audiological assessments to determine whether hearing loss is genuinely associated with port-wine stains and Sturge-Weber syndrome, and if so, to characterize its prevalence and mechanism well enough to inform screening recommendations. If those studies confirm the signal, the implications could be significant, since early identification of hearing loss in children is critical for language development, and a simple audiological screening protocol could prevent years of undiagnosed impairment in a vulnerable population. For now, the message for clinicians and families is one of heightened awareness: a birthmark on the face may be more than skin deep, and the possibility that it reaches the inner ear deserves a careful listen.

Subject of Research: Hearing impairment in patients with port-wine stains and Sturge-Weber syndrome linked to GNAQ mutations

Article Title: Hearing impairment in individuals with Port-wine stains and Sturge-Weber syndrome

Article References: Houff, A. N., Velez, D. E., Guerin, P., Gupta, S., Gupta, I., & Kumar, M. (2026). Hearing impairment in individuals with Port-wine stains and Sturge-Weber syndrome. Archives of Dermatological Research, 318(1), Article 514. https://doi.org/10.1007/s00403-026-04914-y

Image Credits: AI Generated

DOI: 10.1007/s00403-026-04914-y

Keywords: port-wine stains, Sturge-Weber syndrome, GNAQ, hearing loss, vascular malformations, neurocutaneous syndromes, cochlea, sensorineural hearing loss, audiology, somatic mutation, dermatology, patient survey

The research team, led by investigators at the University of Hong Kong and Ningbo Hospital of Integrated Traditional Chinese and Western Medicine in China, took a deliberately modular approach. Rather than asking an AI to produce an end-to-end diagnosis from an image, they split the problem into two separate modules. The first tested whether a large multimodal model could transcribe and interpret pure-tone audiometry and tympanometry images. The second tested whether AI workflows could calculate twenty-five prespecified clinical fields and draft professional and patient-facing reports in Chinese, given already-verified structured data. This separation matters because a single end-to-end score can hide whether errors come from reading the image, applying specialty rules, or communicating the result.

The study drew on 158 outpatient audiology encounters collected between December 2024 and March 2025, of which 155 records representing 151 unique patients and 302 ears were eligible. Reference standards were built painstakingly: two trained transcribers independently entered every threshold while masked to each other, and two audiologists with 13 and 16 years of clinical experience classified tympanogram curves, agreeing on 296 of 300 dually classified ears, a Cohen’s kappa of 0.979. Air-bone gaps, hearing-loss degrees, and loss types were derived under prespecified rules, including a local convention that a meaningful air-bone gap required at least two comparable frequencies with gaps of 15 dB or more.

The heart of the first module was the audiologist skill: a carefully engineered prompt, not a fine-tuned model, that encoded audiological practice. Early development errors were revealing. The model initially produced thresholds not on the standard 5-dB grid, misapplied degree boundaries, overcalled conductive components, treated insufficient bone-conduction evidence as a negative air-bone gap, and read cancelled 95-dB acoustic-reflex marks as present responses. The refined skill imposed a fixed sequence: verify the image and ear, inspect axes and legends, assign symbols, transcribe before calculating, preserve no-response entries, then derive and cross-check. When masking could not be assigned unambiguously, the prompt instructed the model to abstain rather than guess.

The results of the locked test were striking. In 50 development records, the skill raised hearing-loss-type agreement from 85.0 to 94.0 percent and acoustic-reflex agreement from 70.4 to 98.4 percent compared with a schema-only prompt. In the held-out evaluation of 101 independent patients using a Codex GPT5.5 agentic workflow, the system achieved 1809 of 1820 exact numeric thresholds, 99.4 percent, and 91 of 101 patients met every numeric and no-response criterion. Degree was correct in all 101 patients, hearing-loss type in 100, and tympanometry measurements in all 578 entries. Yet the Achilles’ heel persisted: only 9 of 15 no-response entries were correct, and 18 of 19 air-bone-gap mismatches occurred because the model forced a negative label when the evidence supported an indeterminate one.

A post hoc robustness analysis using DeepSeek-V4-Flash-Vision-Exp on the same patients showed how much performance depends on the implementation. With the same locked skill, hearing-loss-type agreement rose from 37.6 to 70.3 percent, a dramatic improvement, but exact numeric-threshold agreement reached only 72.1 percent, reflex agreement hovered near 69 percent, and no-response agreement was zero. The authors are careful to note this was a descriptive cross-model comparison, not a matched foundation-model experiment, since the execution environments differed. The lesson, however, is clear: specialty guidance can substantially improve rule-dependent interpretation, but raw visual accuracy remains tied to the underlying model and workflow.

The second module addressed reporting. Here the AI received adjudicated structured values rather than its own image predictions, calculated twenty-five prespecified fields, and drafted separate Chinese professional and patient-facing reports. After two audiologists reviewed 60 initial cases and identified overreliance on the speech-frequency average, omission of high-frequency losses, and weak integration of history with results, the prompts were refined and locked. In the formal evaluation of 91 independent patients, all twenty-five rule-derived fields were correct in 91 of 91 Codex-workflow cases and 87 of 91 DeepSeek-workflow cases. Both audiologists rated every single report from both workflows as accurate or basically accurate; no report received a rating of clear error or potentially misleading.

An exploratory lay evaluation added a human dimension. Five lay raters compared pre-refinement patient-facing reports from the two workflows in 48 patients. DeepSeek reports were preferred for explanations in 44 of 48 patients and for next steps in 30, but they were also significantly longer, with a median of 392 versus 235 Chinese characters. Overall preference and perceived ease did not differ significantly. The authors emphasize that preference and readability proxies do not establish comprehension, and that no lay evaluation of the refined reports was conducted. This matters because hearing-health materials often exceed recommended reading levels, and limited health literacy can coexist with hearing loss in older adults.

The study’s limitations are candidly enumerated. It came from a single hospital with two devices over four months. Only 15 no-response entries existed, from just three validation patients. The two audiologists who rated the final reports were the same ones who had refined the prompts, raising the possibility of incorporation bias. Even with zero unfavorable ratings among 91 patients, the statistical upper bound on the unfavorable-report rate remains roughly 4.1 percent. Reproducibility was constrained by reliance on proprietary services: the exact Codex snapshot and sampling settings were unavailable, and provider data retention could not be excluded. No end-to-end test connected the two modules, so error propagation from image to report was never measured.

What the study ultimately offers is an architecture rather than a product. The authors envision a safety-conscious pipeline in which image transcription, deterministic validation and calculation, report drafting, uncertainty flags, and clinician approval remain visible, auditable handoff points. This aligns with what Chinese audiologists themselves have said in qualitative work: AI may absorb repetitive technical work, but communication, judgment, and responsibility should remain clinician-led. The findings support prospective evaluation of modular, clinician-supervised assistance, not autonomous diagnosis. The next step, the authors argue, is a silent prospective deployment that connects the modules while preserving intermediate outputs, measuring abstention, correction burden, review time, patient comprehension, and downstream clinical decisions. Until then, the audiogram-reading AI remains a promising apprentice, one that can transcribe nearly every threshold perfectly yet still needs its audiologist to teach it what silence means.

Subject of Research: Audiologist-guided multimodal AI for interpreting pure-tone audiometry and tympanometry and generating clinical reports

Article Title: Audiologist-Guided Multimodal AI for Pure-Tone Audiometry and Tympanometry Interpretation and Reporting

Article References: Wu, X., Shen, X., Mo, C., Shao, S., Wang, J., & Wang, S. (2026). Audiologist-Guided Multimodal AI for Pure-Tone Audiometry and Tympanometry Interpretation and Reporting. Journal of Medical Systems, 50(1), Article 140. https://doi.org/10.1007/s10916-026-02463-5

Image Credits: AI Generated

DOI: 10.1007/s10916-026-02463-5

Keywords: artificial intelligence, audiology, pure-tone audiometry, tympanometry, large language models, multimodal AI, clinical decision support, hearing loss, medical imaging AI, patient communication, prompt engineering, Journal of Medical Systems

The research team analyzed medical records from 240 adults with documented type 2 diabetes who had undergone comprehensive audiological evaluation, alongside 105 normoglycemic control individuals drawn from the same clinical archive. Strict exclusion criteria were applied to sharpen diagnostic specificity: patients with type 1 diabetes, occupational noise exposure, acoustic trauma, conductive middle ear pathology, previous otological surgery, or current use of ototoxic medications were systematically removed from the analysis. Severe renal impairment, hereditary hearing loss, and syndromic deafness were also grounds for exclusion, minimizing confounding from uremia-induced auditory dysfunction and genetic forms of impairment. All participants provided written informed consent in accordance with Declaration of Helsinki principles.

Hearing function was assessed using standardized pure-tone audiometry performed with a clinical audiometer in a soundproof room. Micro-otoscopy ruled out middle ear pathology and active infection, while tympanometry confirmed stable middle ear mechanics. The primary outcome was the pure-tone average at speech frequencies, known as PTA-mean, calculated as the arithmetic mean of air-conduction thresholds at 0.5, 1, 2, and 4 kilohertz, the frequencies most critical for speech intelligibility. Lower PTA-mean values indicate better-preserved hearing, while higher values reflect greater impairment. For participants with asymmetric loss, thresholds from each ear were averaged to provide a representative global estimate, reducing ear-specific variability.

When the researchers compared diabetic and non-diabetic participants using analysis of covariance, adjusting for age, biological sex, body mass index, and smoking status, the presence of type 2 diabetes emerged as a statistically significant predictor of worse hearing thresholds. On average, diabetic patients showed PTA-mean values approximately 8 decibels higher than controls, a difference the authors describe as clinically meaningful, since a 10 decibel rise in auditory threshold is perceptually equivalent to halving perceived loudness. Notably, none of the demographic covariates, including age, reached statistical significance in the adjusted model, a result the authors attribute to the fact that the entire cohort consisted of individuals already diagnosed with sensorineural hearing loss, which compressed the natural variation in auditory function across ages.

The most provocative portion of the study examined whether specific antidiabetic treatments were associated with different auditory outcomes within the diabetic subgroup. Using multivariable linear regression adjusted for age and BMI, the team found that several agents stood out. Semaglutide and dulaglutide, two widely prescribed GLP-1 receptor agonists, were associated with substantially lower PTA-mean values, with large regression coefficients reflecting more favorable hearing thresholds. Metformin, the first-line biguanide used by 55 percent of the cohort, was also linked to better thresholds, and pioglitazone, a thiazolidinedione, showed a similar individual-level association. The remaining drug classes, including sulfonylureas, DPP-4 inhibitors, SGLT2 inhibitors, and glinides, showed no statistically significant connections to hearing outcomes.

When medications were grouped into pharmacological classes, GLP-1RAs again showed the strongest association, with roughly a 9.95 decibel reduction in PTA-mean compared with the reference category, and biguanides were associated with a smaller but significant reduction of about 4.05 decibels. Glinides showed a quantitatively relevant trend that narrowly missed statistical significance, while thiazolidinediones did not reach significance at the class level, likely because only 11 patients in the cohort were taking them. The overall drug-class model explained just over 5 percent of the variance in hearing thresholds, a modest figure consistent with the multifactorial nature of hearing loss, but the medication-related associations persisted even after adjustment for age and BMI, neither of which remained a significant predictor once drug class entered the model.

Why might GLP-1 receptor agonists and metformin be linked to healthier cochleae? The authors point to a convergence of plausible biological mechanisms. Diabetes-related hearing loss is thought to arise from cochlear microangiopathy that compromises blood flow to the stria vascularis, the metabolically demanding tissue that maintains the ionic gradients needed for sound transduction. Chronic hyperglycemia also floods the cochlea with reactive oxygen species, damaging the vulnerable outer hair cells, while advanced glycation end products, mitochondrial dysfunction, low-grade inflammation, and disruption of potassium and calcium homeostasis further erode cochlear and neural function. GLP-1RAs, in this framework, are appealing candidates for auditory protection because they enhance mitochondrial function, suppress inflammasome activation and pro-inflammatory cytokine production, reduce oxidative stress, and improve endothelial function, all of which may preserve the delicate microvascular perfusion the cochlea depends upon.

Metformin’s potential benefit rests on a different but complementary foundation. The drug is a pleiotropic agent that activates AMP-activated protein kinase, a cellular metabolic sensor governing mitochondrial biogenesis, oxidative stress defenses, and inflammatory responses. Through AMPK signaling, metformin may counteract the oxidative damage and microvascular dysfunction that characterize diabetic cochlear pathology, and its systemic metabolic and cardiovascular benefits could secondarily nourish ischemia-sensitive auditory tissue. Intriguingly, the authors note that metformin’s association with better hearing persisted after statistical adjustment, hinting that its potential auditory effects may not be mediated by glycemic control alone. The same logic applies across the medication spectrum: because every drug class studied effectively lowers blood glucose, the differential associations observed suggest that auditory preservation, if real, may stem from antioxidant, anti-inflammatory, and microvascular mechanisms rather than glucose lowering per se. Sulfonylureas, which stimulate insulin secretion without meaningful anti-inflammatory or antioxidant effects, fit neatly into this interpretation by showing no auditory association.

The authors are careful to frame these results as hypothesis-generating rather than practice-changing. The retrospective, non-interventional design precludes causal conclusions, and residual confounding from unmeasured variables, including diabetes duration, HbA1c levels, hypertension, renal function, cumulative drug exposure, treatment duration, and adherence, cannot be excluded. Because longitudinal audiometric follow-up was unavailable, hearing preservation was defined operationally as lower thresholds at a single assessment rather than demonstrated slowing of decline. Some treatment subgroups were small, making the large coefficients for individual agents particularly sensitive to sparse-data effects, outliers, and unmeasured clinical differences. The study also relied on pure-tone audiometry, which does not capture retrocochlear or central auditory processing, and included only hearing-impaired patients, which may explain why age failed to predict thresholds. The authors explicitly caution that the evidence is insufficient to support any change in antidiabetic prescribing based on hearing outcomes.

Even with these caveats, the implications are substantial. Diabetes already affects an estimated 589 million adults worldwide, and hearing loss, a historically underappreciated complication that erodes communication, social integration, and psychological well-being, may deserve a firmer place in routine diabetes care. The study’s authors suggest that audiological monitoring could become a useful adjunctive screening component of comprehensive diabetes management, particularly for older patients with long-standing disease or additional microvascular risk factors. If prospective studies confirm that GLP-1RAs and metformin are genuinely associated with slower auditory decline, hearing outcomes could eventually join cardiovascular and renal endpoints as extra-glycemic measures of treatment success, a notable prospect given the enormous clinical attention these drug classes already command. The mechanistic pathways outlined, from oxidative stress and microvascular inflammation to endothelial dysfunction, now await testing in experimental and translational studies that examine cochlear tissue directly, potentially yielding insights relevant to both diabetes-related and age-related hearing loss.

Subject of Research: Associations between antidiabetic pharmacotherapy and hearing thresholds in adults with type 2 diabetes and sensorineural hearing loss

Article Title: Associations Between Antidiabetic Pharmacotherapy and Hearing Thresholds in Adult Hearing‐Impaired Patients With Type 2 Diabetes

Article References: Martines, F., Salvago, P., Vaccaro, F., Vaccaro, D., Malta, G., Plescia, F., Cascioferro, S., Lavanco, G., & Plescia, F. (2026). Associations Between Antidiabetic Pharmacotherapy and Hearing Thresholds in Adult Hearing‐Impaired Patients With Type 2 Diabetes. Pharmacology Research & Perspectives, 14(5), Article e70316. https://doi.org/10.1002/prp2.70316

Image Credits: AI Generated

DOI: 10.1002/prp2.70316

Keywords: type 2 diabetes, hearing loss, GLP-1 receptor agonists, metformin, pure-tone audiometry, sensorineural hearing impairment, cochlear microangiopathy, oxidative stress, semaglutide, pioglitazone, diabetes complications, audiology