Laryngology Brief, September 2026, Issue 4

Welcome to the fourth issue of Laryngology Brief,

Every month I scan the laryngology literature across my 11 core journals, this time 138 articles published between 1 August and 30 August 2026, and select the handful that I believe deserve a clinician's attention. Each featured paper comes with a short summary of its design and numbers, followed by my own reading: why it matters at the bedside or in the office, where the data is solid, and where I would hold back. This issue leans toward one uncomfortable question that turned up in paper after paper, namely what our outcome measures are actually measuring: who with exercise-induced laryngeal obstruction should ever reach the operating room, whether a consensus document in which all sixty statements came out GRADE A is as solid as that looks, what frailty really predicts before a type 1 thyroplasty, whether the complication side of the superior laryngeal nerve block changes the consent conversation we started last issue with the neurogenic cough meta-analysis, and how much of what we call regeneration is simply bulk.

My aim is not to replace your own reading but to save you time and give you a starting argument. Comments and disagreements are welcome; they make the next issue better and increase my motivation to continue!

By the way, this issue comes a little later than it should be, accept my apologies! I was focused on IFOS 2026 which was in Istanbul and kept me away from brief. But also I have the excitement to be on Fall Voice with my team will present 3 oral presentations and 4 posters!

Featured Papers in This Issue

1. Sandvik L, et al. Exercise-Induced Laryngeal Obstruction: A Randomized Controlled Trial of Surgical Treatment. Laryngoscope.

2. Lechien JR, et al. European Practice Consensus on Office-Based Laryngology. Laryngoscope.

3. Hoffmeister JD, et al. Pitch Strength and Percent Vocal Creak to Assess Treatment Response in Adductor Laryngeal Dystonia. Laryngoscope.

4. Fletcher BN, et al. Linguistically Informed Automated Estimates of Creak in Adductor Laryngeal Dystonia. J Speech Lang Hear Res.

5. Schlegel LE, et al. Variable Effects of Vagus Nerve Stimulator Cycling on Vocal Fold Motion: A Videographic Case Series. Laryngoscope.

6. Schuman AD, et al. Frailty Severity and Voice Outcomes After Type 1 Thyroplasty. Otolaryngol Head Neck Surg.

7. Hernández-García E, et al. Office-Based Serial Platelet-Rich Plasma Injection for Structural Vocal Fold Disorders: A Prospective Observational Study. J Voice.

8. Miller SM, et al. Rare Complications of a Popular Procedure: A Multi-institutional Case Series of Superior Laryngeal Nerve Block Complications. J Voice.

9. McClelland TQ, et al. Intraoperative Bradycardia: Evaluating Vasovagal Events in Adult Suspension Microlaryngeal Surgery. Laryngoscope.

10. Ma Y, et al. Evaluation of the Effects of Testosterone on Voice in Transmasculine Adolescents and Young Adults. Laryngoscope.

11. Shah P, et al. Association of Glucagon-Like Peptide-1 Receptor Analogues and Laryngeal Symptoms in Obese Adults. Laryngoscope.

12. Massonet H, et al. Improving Functional Oral Intake in Head and Neck Cancer Survivors with Chronic Radiation-Associated Dysphagia: The Multicenter Randomized Controlled HIT-CRAD Trial. Dysphagia.

9 short items follow at the end, under In Brief.

---

⭐ Featured

Sandvik L, et al. Exercise-Induced Laryngeal Obstruction: A Randomized Controlled Trial of Surgical Treatment. Laryngoscope. 2026.

​10.1002/lary.70820 · Level of evidence 2 · NCT04620343

This is phase 3 of the HelpILO programme from Bergen, and the first randomised surgical trial in EILO. 303 patients entered phase 1, none of whom had previously undergone continuous laryngoscopy during exercise (CLE) testing or any EILO treatment. All received structured conservative treatment in phases 1 and 2. Only those with persisting severe symptoms despite that, a supraglottic CLE score of 2 or more, and high motivation for surgery were eligible for phase 3. Forty-eight qualified, and 41 were randomised 1:1:1 to full supraglottoplasty (FSP, n = 13), minimally invasive supraglottoplasty (MISP, n = 15) or watchful waiting (n = 13). FSP means bilateral lengthening of the aryepiglottic folds with releasing incisions anterior to the cuneiform tubercles plus circular excision of the cranial part of the tubercles, preserving the posterior commissure and pyriform sinuses. MISP uses the same laser at the same settings but non-ablatively: the fibre is used like a needle to deliver submucosal thermal energy at six bilateral points less than 5 mm deep, removing no tissue, the aim being localised fibrosis and stiffening. One surgeon performed all operations.

Against control, FSP improved the supraglottic score at maximum exercise intensity dramatically(OR 0.002, 95% CI 0.000 to 0.108, p < 0.001) and also improved the glottic score at maximum intensity and the total score. MISP did not differ from control on any CLE outcome (ORs 0.33 to 0.49, all confidence intervals crossing 1). Yet on patient-reported breathing difficulty during activity, MISP did beat control (OR 0.15, p = 0.022), while FSP did too (OR 0.13, p = 0.039). For general breathlessness in daily life, only FSP separated from control.

It is nice they have shared the surgical videos!! FSP and MISP videos are here!

​https://onlinelibrary.wiley.com/doi/10.1002/lary.70820​

​https://onlinelibrary.wiley.com/doi/10.1002/lary.70820​

Why it matters. The number I keep coming back to in this paper is not a treatment effect at all. Three hundred and three patients went in, forty-eight came out as surgical candidates, and forty-one were randomised. That is sixteen percent, after a full course of conservative treatment delivered by a team that does this properly. Alanazi reported surgery in six of seventy-eight. So the first thing this trial teaches is not how to operate on EILO, it is how rarely/often you should. I see adolescent athletes in the office almost every week who arrive already convinced that something in the larynx needs to be cut, and this paper is now the cleanest thing I can hand them to explain why we are going to spend a few months on breathing technique first. But at the same time in patients who are not getting benefits from conservative therapy surgery might be still an option.

The second thing, and I find it more interesting than the headline, is MISP. It improved how patients felt while doing nothing measurable to the airway. The authors say it themselves, that these effects may reflect non-specific mechanisms rather than a direct physiological effect on obstruction. Surgery almost never gets a placebo arm, and here one appeared by accident. I would read that arm very carefully before adopting any minimally invasive laryngeal procedure whose justification rests on symptom scores alone.

Where I would hold back: thirteen patients in the FSP arm, of whom only six reached the lowest supraglottic score at maximum intensity, six retained moderate residual obstruction and one remained severe. That distribution is in the body of the paper and not in the abstract, and it is the number I would quote to a family, not the odds ratio. There is also no established minimal clinically important difference for CLE scores, the assessors could not be fully blinded because a post-surgical supraglottis is recognisable, and the trial cites Walsted's work showing substantial inter- and intra-rater variability in CLE scoring. Finally, the paper itself cites Carlsen's finding that a third of asymptomatic athletes show supraglottic obstruction at peak exercise. So the score alone does not make the diagnosis, and this trial does not answer the question I most want answered: which of the sixteen percent would have done just as well with another year of conservative treatment.

---

Lechien JR, Al Yaghchi C, Bidaye R, Denizoglu I, Enver N, et al. European Practice Consensus on Office-Based Laryngology. Laryngoscope. 2026.

​10.1002/lary.70869​

Disclosure: I was one of the twenty-five laryngologists in this Delphi panel, so read what follows knowing that.

Twenty-five laryngologists from the British Laryngological Association, the European Laryngological Society and the Union of European Phoniatricians ran a modified Delphi with up to four anonymous voting rounds and an 80 percent agreement threshold. Four systematic reviews were performed before voting to fill the gaps in the literature, a non-voting biostatistician analysed the rounds, and three external experts who did not vote performed the GRADE assessment. Sixty statements were validated over two rounds, 49 in the first and 11 after revision, with one statement rejected outright. Inter-expert agreement was strong (ICC 0.798, 95% CI 0.645 to 0.911) and every statement had an interquartile range of 1 or less. The statements cover pre-procedural assessment and patient information, indications and contraindications, technique, vocal fold augmentation, post-procedural care and success criteria, and environmental, economic and societal consequences.

Why it matters. A consensus document does not produce evidence, it organises practice, and the most useful thing a panel member can do is show you how to read it rather than tell you it is good.

Start with the GRADE column. The methods section states plainly that because comparative studies in office-based laryngology are scarce, several statements were expected to receive low or very low certainty, GRADE C or D, and that in those cases the three external experts could still assign a strong recommendation when they judged that the benefits clearly outweighed the risks, based on panel consensus and clinical experience rather than comparative data. The results then report that after external expert consensus, all of the validated statements range between A and B. That is not an irregularity, it is a path GRADE explicitly permits. But a reader who sees sixty A and B ratings and assumes sixty comparative trials underneath them has misread the document, and I would rather say that here than have someone else say it later.

The second place to look is where the panel failed. Twenty-five experienced laryngologists could not reach consensus on office-based treatment of dysplasia and carcinoma in situ after histological confirmation. Miri's systematic review, cited in the paper, gives a cumulative complication rate of 2.3 percent in leukoplakia and 1 percent in dysplasia, which sounds reassuring, until you see that 39 percent of cases required repeat treatment for complete resolution. That is why the room did not agree, and the honest part of any consensus is the part it could not settle.

The third is the most practical line in the whole document and the one almost nobody will notice: failure of a previous office-based procedure is listed as a contraindication to another one. To be honest I still try again in many patients. What is the right move to do no-one knows

---

Neurolaryngology

Hoffmeister JD, et al. Pitch Strength and Percent Vocal Creak to Assess Treatment Response in Adductor Laryngeal Dystonia. Laryngoscope. 2026.

​10.1002/lary.70817 · open access · Level of evidence 3

Fletcher BN, et al. Linguistically Informed Automated Estimates of Creak in Adductor Laryngeal Dystonia. J Speech Lang Hear Res. 2026.

​10.1044/2026_JSLHR-25-00997​

Two papers a fortnight apart, both trying to turn creak into a number, and they reach opposite-looking conclusions because they ask different questions.

Hoffmeister took 16 patients with adductor laryngeal dystonia, all with documented prior benefit from thyroarytenoid or lateral cricoarytenoid botulinum toxin, and measured them 7 days before injection and 4 to 6 weeks after. Four acoustic measures were used: pitch strength, percent vocal creak from a pretrained neural network with every creak instance hand-verified against spectrogram and waveform, AVQI and CSID. After injection, manual aberration counts improved (p = 0.02), VHI-10 improved (p = 0.012), patient-reported voice quality improved with a large effect (d = 1.1), vocal effort improved and CAPE-V improved.Percent creak did not change (p = 0.355). Pitch strength did not change (p = 0.091). CSID did not change. AVQI moved in the wrong direction, worsening significantly (p = 0.03), although below its established minimal important difference. And yet at the individual level the change scores correlated strongly: change in pitch strength against change in aberration count, rho = -0.83, and against overall dysphonia severity, rho = -0.88.

Fletcher asked whether the problem is that we are counting the wrong creak. Fifty speakers with AdLD and 50 matched controls read a standard passage. Creak occurring outside the phonological contexts where English speakers normally creak, that is phrase-final creak, /t/ glottalisation and word-initial vowel glottalisation, was separated from total creak.

Why it matters. Adductor laryngeal dystonia is the diagnosis where I most wish I had a number, not because I doubt my ears but because I cannot show my ears to the patient or to the next clinician. So my interest in both papers is the same: can this become a tool for judging the treatment we already give?

Hoffmeister's answer is yes, but only in one direction. For following a single patient across injections, the individual correlations are strong enough to be useful, and the authors say so themselves, that this may have clinical value for individual treatment optimisation, dose, timing and laterality. For comparing groups or study arms, no. And the reason is the part of the paper I would make every trainee read. The measure is blind to the form the pathology takes. A patient can go from ten frequency shifts to two aperiodic segments, be obviously better in clinic, and score the same or worse, because aperiodic segments are penalised more heavily. Phonatory breaks are silent, so percent creak, AVQI and CSID do not measure them at all by definition. Pre-treatment breathiness, adopted to compensate for hyperadduction, adds noise that masks post-treatment recovery of pitch strength. And creak is not pathological to begin with; normophonic speakers creak for sociolinguistic reasons, and patients may be using creak deliberately to smooth over aperiodic breaks.

---

Schlegel LE, et al. Variable Effects of Vagus Nerve Stimulator Cycling on Vocal Fold Motion: A Videographic Case Series. Laryngoscope. 2026.

​10.1002/lary.70819 · Level of evidence 4

To be clear at the outset, because the abbreviation is used for two different things in our field: this is the implanted vagus nerve stimulator placed in the chest wall for epilepsy and treatment-resistant depression, not intraoperative nerve monitoring during thyroid surgery. The cuff sits on the cervical vagus, distal to the take-off of the superior laryngeal nerve and proximal to the recurrent laryngeal nerve.

Eight patients over twelve years, all with conventional left-sided devices, presenting to a laryngology clinic with dysphonia and having flexible laryngoscopy video available. Mean presenting VHI-10 was 25.0. In five of the eight, symptoms began after a revision lead placement rather than after the index surgery. Three had complete left vocal fold immobility at rest, which persisted during cycling, but all three showed lengthening and increased tension of the left fold during stimulation. Of the five who had motion at rest, three developed transient immobility during cycling, all in an adducted position. One patient with pre-existing left immobility and right hypomobility lost both adduction and abduction on the right side during active cycling, with glottic narrowing and audible stridor. In one patient, tonic lateral cricoarytenoid contraction during cycling was treated with 1 unit of onabotulinumtoxinA into the left adductor compartment under EMG guidance, after which the static position during cycling shifted from adduction to relative abduction. Device parameters were reduced in all eight as part of management, but only three reported subjective improvement, and two could not tolerate the reduction because seizure control deteriorated.

Why it matters. The reason to read this is that it changes what you think you are looking at. A patient with a vagus nerve stimulator and a motionless left fold is not necessarily a paralysis. It may be a cyclical, stimulation-locked motion disorder, and the difference is everything, because one of those gets a thyroplasty and the other gets a device reprogramming or a unit of toxin.

The authors put the practical conclusion better than I could: laryngeal imaging during active cycling is essential to guide management, and without it the intervention chosen may worsen rather than improve the glottic configuration. That is a concrete change to the examination protocol. If you scope one of these patients, you have to scope them with the device running, and you have to watch through a full cycle.

The physiology underneath is the part I find genuinely useful. Roughly 75 percent of recurrent laryngeal nerve axons serve adduction and 25 percent abduction, and the muscle architecture mirrors that ratio, one abductor against an adductor complex. On population anatomy, non-selective vagal stimulation should reliably produce a medialised configuration. It does not, and the authors argue this is clinical evidence that adductor and abductor fibres are not topographically segregated within the main trunk. For anyone working on reinnervation, that is not a footnote. It is the reason selective reinnervation is hard, stated in a human case series rather than an animal model.

---

Phonosurgery and the Office

Schuman AD, et al. Frailty Severity and Voice Outcomes After Type 1 Thyroplasty. Otolaryngol Head Neck Surg. 2026.

​10.1002/ohn.70395 · open access

One hundred and twenty-seven patients from a single institution between 2017 and 2024, revisions excluded, all operated by two laryngologists. Frailty was scored with the MFI-5, giving 39 robust, 48 pre-frail and 40 frail patients. Seventy-nine had motion impairment and 48 atrophy. Frail patients were older (68.2 versus 48.0 years) and more often female. The only significant pre-operative difference was breathiness, which was worse in frail patients. After surgery, breathiness improved more in the frail group (-1.4 versus -0.7, p = 0.0203), while grade, asthenia and VHI-10 gains did not differ between frailty levels. There were no deaths within 30 days and seven complications in the whole cohort, none of them in the frail group.

Why it matters. We are operating on older and sicker patients than we were ten years ago. People live longer, and a patient with advanced systemic disease and a paralysed fold now gets offered a thyroplasty rather than being told to live with it. So a paper that asks whether frailty should change that offer is worth the twenty minutes.

The answer is reassuring at the level I care about: frailty was not associated with worse voice outcomes and not associated with complications. Zero complications in the frail group is a small number doing a lot of work, but combined with no 30-day mortality it is enough for me to say that frailty alone is not a reason to withhold a type 1 thyroplasty.

The finding that is not in the abstract is the better one. The authors separated out the cardiopulmonary components of frailty and reran the analysis, and found nothing: no difference in VHI-10, grade, breathiness or asthenia. Their inference is that the functional component of frailty matters more to voice outcome than the cardiopulmonary component. That is a real change to how I would read a pre-operative assessment. Anaesthetic risk and voice benefit are not on the same axis, and a patient's COPD tells you about the first and apparently nothing about the second.

Where I would hold back, and this matters for how you quote the paper: VHI-10 data exist for only 48 of the 127 patients. That is enough to support "frailty is not a contraindication" and not enough to support "frail patients gain more". And there is a quieter number in the subgroup table that I would carry into the consent conversation: VHI-10 gain was -6.3 in the atrophy group against -13.1 in the immobility group. Most frail patients are atrophy patients, so if you are counselling an eighty-year-old with bilateral atrophy, the expectation you set should come from the smaller number, not the headline. There are also no acoustic measures at all, and no comparison with injection laryngoplasty, which for a frail patient is the alternative that actually competes.

---

Hernández-García E, et al. Office-Based Serial Platelet-Rich Plasma Injection for Structural Vocal Fold Disorders: A Prospective Observational Study. J Voice. 2026.

​10.1016/j.jvoice.2026.07.041​

Twenty-four patients and 41 vocal folds, treated prospectively between November 2023 and November 2025 at a Spanish university hospital. Inclusion required a structural lamina propria disorder (atrophy, sulcus or scar), VHI-10 of 11 or more and failure or only partial response to voice therapy. Sulcus was diagnosed on high-resolution videostroboscopy and restricted to Ford type II and III. PRP was prepared from peripheral blood immediately before the procedure with a single centrifugation step at 1800 rpm for 8 minutes, no exogenous activation, and 0.3 to 0.5 mL of pure PRP was injected per fold through a transcricothyroid approach under local anaesthesia, subepithelially for sulcus and scar and deeper for atrophy. Three serial injections were given. Diagnoses were atrophy in 10, scar in 9 and sulcus in 5. Forty-two percent were professional voice users.

Across the whole sample, VHI-10 fell from a median of 18.5 to 14.5 (p = 0.001) and GRBAS from 5 to 2 (p < 0.001). Closed phase on stroboscopy rose from 38.5 to 47.5 percent (p = 0.015). CSID fell from 38 ± 14 to 22 ± 14 (p < 0.001) while jitter, shimmer and noise-to-harmonics ratio did not move at all. Two to three patients required reinjection at around one year for recurrent dysphonia. No significant adverse events.

Why it matters. PRP in the office is one of the genuine expansion areas in our field right now, and the thing I value most about this paper is simply that it is prospective, with a defined protocol and a stated injection plan, in a literature that is mostly retrospective series.

But the full text takes the commentary somewhere the abstract does not. The biological rationale for PRP is restoration of lamina propria viscoelasticity, which means sulcus and scar should be the natural targets. In the subgroup table, the only diagnosis reaching significance on VHI-10 is atrophy. Also atrophy of the muscles get more benefit from regeneration treatments than scar and sulcus. The authors attribute this to lack of power in small subgroups and they may well be right.

Two more things worth carrying into practice. The acoustic pattern is the same one we saw in the AdLD papers above: perturbation measures flat, cepstral measure moving sixteen points. If you are still using jitter and shimmer to judge a dysphonic, aperiodic voice, this is your third reminder this month. And the reinjection rate, two or three of twenty-four at about a year, belongs in the consent conversation, because a patient who has sat through three office injections needs to know that this is maintenance and not a cure. What the study does not tell us, and what any of us doing this needs, is what was actually injected: platelet concentration and leukocyte content were not measured, so the biological dose behind the word PRP is unknown.

Also the preparation styles of PRP in the literature are so different, It is not easy to compare results too. I see many sulcus patients whom my preference is 3 PRP session with voice therapy. I double spin the blood with 1500rpm 5 min 3000rpm 10 min and use both buffer zone and the upper part of plasma. Buffer zone goes to muscle and low-platelet part of it goes to lamina propria. Which becomes a game-changer for my practice. We are collecting the data of ours retrospectively, will publish soon.

---

Complications and Perioperative Practice

Miller SM, et al. Rare Complications of a Popular Procedure: A Multi-institutional Case Series of Superior Laryngeal Nerve Block Complications. J Voice. 2026.

​10.1016/j.jvoice.2026.07.054​

Sixteen laryngologists at 15 institutions were surveyed, and 10 reported an unusual complication, defined as an unexpected symptom or finding the physician had not previously seen or heard of and did not include in pre-procedure counselling. All respondents had at least five years in practice and career injection volumes between 85 and 1500. Thirteen patients and seven complication types: marginal mandibular nerve palsy in five, fungal pharyngitis of the pyriform sinus in two, transient ischaemic attack in two, and one each of Horner syndrome, anxiety-related response, adrenal suppression and hypoglossal palsy. Ninety-two percent of the affected patients were being treated for chronic cough and 38 percent had a history of anxiety. Every marginal mandibular palsy resolved spontaneously within one to four hours. On the Clavien-Dindo scale only the fungal pharyngitis and the adrenal suppression, in a patient who had received twenty injections, exceeded grade I.

Why it matters. Last issue carried the meta-analysis of superior laryngeal nerve block for neurogenic cough. A month later, the complication side arrives, and the two belong together.

The practical message is about consent, not about avoidance. I have not had marginal palsy before, but it happens sometimes.

Where I would hold back: this is a survey series with no denominator. Thirteen complications out of an unknown number of injections. You can make a rough estimate from ten physicians at roughly fifty injections a year, but the paper does not, and neither should we. Read it as a list of what can happen, not as a rate.The existing literature quoted in their Table 2 reports adverse event rates anywhere from 1 to 44.2 percent, which tells you how unsettled this is.

I am lucky, I have not seen any of these complications, yet! But I will add these in to my informed consent to be on safe side.

---

McClelland TQ, et al. Intraoperative Bradycardia: Evaluating Vasovagal Events in Adult Suspension Microlaryngeal Surgery. Laryngoscope. 2026.

​10.1002/lary.70855 · open access · Level of evidence 4

One hundred and ninety consecutive adult suspension microlaryngoscopy cases by a single surgeon between September 2023 and June 2025. A vasovagal response was defined as a fall of at least 10 beats per minute within 5 minutes together with a nadir below 60. Events during intubation were excluded by using a standard window from 5 to 30 minutes after intubation. Seventeen patients, 8.9 percent, had a vasovagal response. Heart rate fell by a mean of 16 beats per minute, or 22.5 percent, over two minutes, to a mean nadir of 49, with two patients dropping below 40, and took a mean of 9 minutes to return to baseline. The laryngoscope was removed in all seventeen; eight settled with that alone and nine received intravenous glycopyrrolate. No asystole, no chest compressions, no ACLS, no desaturation. Nothing predicted the event: not demographics, not comorbidity, not laryngoscope size (p = 0.204), not remifentanil.

Why it matters. Any of us who does phonosurgery is in this position several times a week, and 8.9 percent puts vasovagal bradycardia at roughly the same frequency as sore throat and dental injury, the complications everyone counsels about. It is not rare. It is just not discussed.

The negative result is the useful one. There is no patient you can identify in advance and no scope you can choose to avoid it. Which means you do not protect yourself by selection, you protect yourself by preparation: stand where you can see the monitor, say the word vasovagal to your anaesthetist before you suspend rather than after, and have glycopyrrolate drawn. The authors' escalation is sensible and worth committing to memory: remove the laryngoscope immediately, then intravenous glycopyrrolate if needed, then topical 4 percent lidocaine to the glottis, then a smaller or different laryngoscope.

I should also say, one of my laryngologist colleagues had an bradichardia reaction in one of her patients which resulted with cardiac arrest and a temporary optic nerve palsy. So this happens for some of us. Sometimes we start believing it will not happen in our hands, but it does!

---

Gender-Affirming Voice

Ma Y, et al. Evaluation of the Effects of Testosterone on Voice in Transmasculine Adolescents and Young Adults. Laryngoscope. 2026.

​10.1002/lary.70729 · Level of evidence 4

A prospective observational cohort from the UCSF Child and Adolescent Gender Center: adolescents and young adults assigned female at birth, naive to both testosterone and voice therapy, enrolled before starting hormones and followed monthly for twelve months. Fourteen participants were analysed, mean age 18.4 years, range 14 to 31. Dose, formulation and titration were left to the treating clinician. Recordings were made by participants on their own smartphones and analysed in Praat.

Sustained vowel fundamental frequency began to fall significantly at month 2 and continued throughout, from 212.80 Hz to 128.58 Hz. Speaking fundamental frequency fell significantly by month 3 and plateaued at month 9. Semitone pitch range did not change at all over twelve months (p = 0.47). Cepstral peak prominence smoothed fell significantly at month 3 and, although it recovered, remained significantly below baseline at months 9 and 12. VHI-10 worsened transiently at month 3 and returned to baseline by month 9. The Trans Woman Voice Questionnaire score, used here for voice-gender congruence, improved from 51.3 to 26.5, and PROMIS anxiety fell, while depression did not change.

Why it matters. This is the decision point immediately before the part of the pathway I am responsible for, and the single most useful number in the paper is the plateau. Speaking fundamental frequency stops moving at month 9. I

The two findings I would put in front of every patient starting testosterone are the ones the abstract passes over. Pitch range does not widen. Frequency drops, usable range stays where it was. For a singer, an actor or anyone who uses their voice professionally, that is a more consequential piece of information than the drop in F0, and it is not what most patients expect when they start. And CPPs is still below baseline at twelve months. The authors attribute this to rapid structural change temporarily destabilising phonation, which is plausible, but twelve months is a long time to call temporary.

Put those together and the summary is that testosterone masculinises the voice without improving it, may be with some destabilizing the quality. That is precisely the space where voice therapy belongs. I have been doing gender-affirming voice surgery since 2016 and the conversation I regret most is the one where expectations were set by someone who only looked at pitch.

Where I would hold back: fourteen participants, one centre, a cohort that is 92.9 percent white, uncontrolled dose and formulation, blood sampling not timed to the last dose, and recordings made unsupervised on participants' own phones after the first session. None of that invalidates the F0 trajectory, which is large and consistent. It does mean I would treat the CPPs finding as a hypothesis worth a properly instrumented replication rather than as an established fact.

---

Systemic Medicine and the Larynx

Shah P, et al. Association of Glucagon-Like Peptide-1 Receptor Analogues and Laryngeal Symptoms in Obese Adults. Laryngoscope. 2026.

​10.1002/lary.70808 · open access · Level of evidence 3

A retrospective cohort from the TriNetX US Collaborative Network covering January 2016 to December 2025: 617,296 adults with obesity started on a GLP-1 receptor analogue or GIP agent, against 3,503,102 controls with a similarly coded obesity visit who were not, propensity matched on age, sex, race and ethnicity. Head and neck and pulmonary neoplasia, head and neck radiotherapy, neurodegenerative and autoimmune disease with neuromuscular involvement, tobacco use and chronic airway disease were excluded. At six months, the odds of any laryngeal finding were higher in the treated group (OR 1.19, 95% CI 1.16 to 1.21, p < 0.0001), driven by cough (OR 1.46), globus (OR 1.41) and voice and resonance disorders (OR 1.19). By agent, exenatide, liraglutide, semaglutide, dulaglutide and lixisenatide all showed significantly elevated six-month risk. Albiglutide and tirzepatide did not.

Why it matters. Last issue's In Brief carried a single case of revision thyroplasty after semaglutide. One case is a hypothesis. Six hundred thousand patients is a signal, and the speed of this is worth stating: GLP-1 receptor analogue use in the United States rose elevenfold between 2014 and 2022, and monthly semaglutide prescriptions went from 0.7 million to 2.6 million by February 2024. Whatever the laryngeal effect turns out to be, the denominator is now enormous.

The detail that is easy to miss is the agent breakdown. Tirzepatide, one of the most prescribed agents in the class, shows no significant association. Neither does albiglutide. So this may not be a class effect, and "GLP-1 users have more laryngeal symptoms" is already too coarse a statement to be useful. The clinically actionable version is that when a patient on a weight-loss agent presents with cough, globus or a changed voice, the name of the drug is worth asking for, not just the category. Of course reflux should be always the part of conversation too. But the effect of GLP1's are not limited with reflux on larynx.

Where I would hold back, and here I would hold back quite hard: this is a coded-claims study. The outcomes are ICD-10 and CPT codes, not clinical assessments, so what is being measured is coding behaviour as much as symptoms. A patient on a GLP-1 agent sees clinicians more often and gets coded more often. Detection bias is the central threat to this design and the paper does not address it.

---

▪️ In Brief

Synthetic laryngeal images are now good enough to fool us, and the training does not need to be long. Morrison and colleagues generated videostroboscopic images with StyleGAN3 from age-stratified datasets and had 114 clinicians classify a balanced 36-image survey. Clinicians identified true clinical frames far better than synthetic ones (70.9 versus 57.8 percent, p < 0.001), and realism peaked at 20,120 kimg, where accuracy fell to 44.8 percent, worse than chance. Perceptual realism plateaued early, which matters for anyone thinking about synthetic training sets. 10.1002/lary.70870​

By the way, I can not keep my self feel prouder for the studies coming from my dear colleagues, Anais Ramaeu’s team. She was inspired me for innovation in the field during my fellowship. I was lucky to work with her.

How loudly you speak depends on how far away the listener looks, not on how large the room sounds. Daşdöğen and colleagues put fifteen vocally healthy participants in immersive virtual reality across nine conditions varying room size and speaker-to-listener distance, holding auditory feedback constant so that only visual input changed. Sound pressure level rose with listener distance, room size had a significant but secondary effect, and the two interacted. A small study, but it makes a case for building visual context into therapy environments rather than expecting carryover from a quiet clinic room.10.1044/2026_JSLHR-25-00798​

Awake blue laser excision plus balloon dilation for oropharyngeal stenosis in airways you would not want to suspend. Landini and colleagues describe two head and neck cancer survivors with severe trismus and altered laryngeal landmarks, managed under monitored anaesthesia care with a flexible channelled nasolaryngoscope. Both reported symptomatic improvement with objective gains in airway patency at follow-up. Two patients is two patients, but the technique is worth knowing about before you meet the airway that needs it. 10.1002/ohn.70363​

Three idiopathic subglottic stenosis papers landed in the same month

1-Dronkers and colleagues surveyed 193 clinicians in 42 countries: endoscopic dilation was near universal at 95 percent but heterogeneous in execution, in-office serial intralesional steroid was used by 57 percent, the Maddern procedure by 17 percent, and there was significant variation even in how treatment failure is defined. 2-Larkin and colleagues report three patients who started JAK inhibitors for rheumatoid arthritis after their iSGS diagnosis, in whom the mean surgical-free interval went from 317 to 631 days and the annualised dilation rate from 1.24 to 0.38, with single-cell data showing JAK-STAT activation highest in iSGS CD4+ T cells and neutrophils. And 3-Weiss and colleagues surveyed 682 women through an international support group, finding Dyspnea Index inversely correlated with sexual quality of life (rho = -0.305) and each one-point rise in DI predicting a 0.61-point fall in SQOL-F. A disease with no agreed outcome definition, a three-patient hint at a medical adjuvant, and a reminder of what we are not asking about in clinic. 10.1002/lary.70830 · 10.1002/lary.70872 · 10.1002/ohn.70405​

Photoangiolytic laser in professional voice users: twenty-two studies, no comparative data. Wu and colleagues conducted a PRISMA-ScR scoping review of benign vocal fold lesions in professional voice users. KTP was the most studied laser, polyps the most studied lesion, subjective and acoustic outcomes improved consistently, and complications ran at 5 to 10 percent, all minor and temporary: transient hoarseness, oedema, self-limited bleeding. What is missing is any head-to-head comparison against cold steel microflap in the population where the margin for error is smallest. 10.1016/j.jvoice.2026.08.016​

If you are interested in using laser for your singer patients read this paper too. 10.1002/lary.31484

---

Next issue: October 2026. Reply with corrections, disagreements, or a paper I missed.

---

Necati Enver, MD

Istanbul Voice Center

​Youtube, Instagram​