1 Why This List, and Why Now
As we enter 2026, it’s worth reflecting on the past year that delivered transformative advances across the breadth of otolaryngology. To help busy clinicians navigate an increasingly dense literature landscape, we have curated this series to highlight ten peer-reviewed papers with the potential to shift the needle in ear, nose, and throat (ENT) science and practice.
In this installment centered on head and neck surgery, laryngology (including communication), and sleep, we provide jargon-light, high-level summaries that capture the current state of the art in these subspecialties. Our goal is to move beyond the “what” of the research to explain the “why” and, crucially, how these findings tangibly alter clinical care.
From thousands of peer-reviewed studies published or posted online in 2025 across leading general and specialty journals (e.g.,
The Lancet,
Nature,
Science,
Journal of Clinical Oncology), we picked ten that exemplify practice relevance, methodological rigor, and translational potential across clinical trials, new guidelines, engineering, and basic science. These works advance the discipline on multiple fronts: some challenge long-held assumptions, others redefine clinical standards, and several open credible paths to restoring functions once thought irretrievable. While oncology features prominently, reflecting the rapid progress made amid a substantial global head and neck cancer burden
[1], we also highlight pivotal shifts in sleep medicine, communication, and the burgeoning field of musicality.
This is neither an exhaustive “best of” list nor a formal ranking. Many outstanding papers are not included simply due to space constraints, and we acknowledge that other experts may weigh the literature differently. We offer these ten papers as a curated starting point and welcome your comments, as continuing this dialogue is how the field moves forward.
2 Nivolumab After Surgery: Toward a New Standard for High-Risk Head and Neck Squamous Cell Carcinoma (HNSCC)
Why this matters: For resected, high-risk locally advanced HNSCC, the standard of care has been surgery followed by postoperative cisplatin-based chemoradiotherapy (CRT) for features such as positive margins or extranodal extension, yet recurrence remains common. The randomized phase 3 NIVOPOST-OP trial published in
The Lancet tested whether adding the programmed cell death protein 1 (PD-1) inhibitor nivolumab to standard CRT could improve outcomes in this long-stagnant adjuvant setting
[2].
What the study did: This open-label, international trial enrolled 680 patients with resected, high-risk HNSCC and randomized them to standard CRT or CRT plus nivolumab with continued adjuvant nivolumab. At a median follow-up of 30.3 months, 3-year disease-free survival (DFS) improved from 52.5% to 63.1% with nivolumab (hazard ratio 0.76), with benefit across programmed death-ligand 1 (PD-L1) subgroups. Safety showed more grade 3–4 events with nivolumab but no excess treatment-related deaths, and CRT compliance was similar between arms.
How it changes the field: By demonstrating a significant 3-year DFS advantage over adjuvant cisplatin-based CRT with consistency across PD-L1 subgroups, NIVOPOST-OP supports nivolumab-enhanced CRT as a credible candidate for future standards of care in resected, high-risk HNSCC.
3 Ultra-High Dose-Rate Radiotherapy (UHDR-RT): Overcoming Radioresistance
Why this matters: Radiotherapy is central to HNSCC, yet radioresistance remains a major cause of local failure and poor outcomes. A 2025 study in
Signal Transduction and Targeted Therapy reports that UHDR-RT can overcome this resistance, offering a potential technological leap beyond conventional dosing
[3].
What the study did: Using a high-power accelerator to deliver high-energy X-rays at ultra-high dose rates, investigators tested UHDR-RT in radioresistant HNSCC cell lines and mouse models. UHDR-RT increased direct DNA damage and re-engaged antitumor immunity, boosting CD8+ T-cell activity and shifting tumor-associated macrophages toward a pro-inflammatory M1 phenotype. Mechanistically, the authors identified an IFN-γ/CXCL9 positive feedback loop between T cells and macrophages as a key mediator in remodeling the tumor microenvironment and reversing radioresistance.
How it changes the field: Although preclinical, this dual mechanism—enhanced tumor kill plus immune re-engagement, provides a strong rationale for prospective clinical trials investigating UHDR-RT as a strategy to overcome radioresistance, either alone or in combination with immunotherapy.
4 Targeting Immunosenescence to Restore Immunotherapy Response
Why this matters: Immune checkpoint blockade has reshaped oncology, yet durable responses occur in only a minority of patients. Immunosenescence, the age-associated dysfunction of immune cells within the tumor microenvironment, appears to limit benefit in solid tumors such as HNSCC. A 2025 study in
Nature Medicine links poor response to immunosenescence-related shifts in the T- and B-cell pools and proposes senolytics to restore immune competence
[4].
What the study did: The phase 2 COIS-01 trial tested whether a senolytic cocktail, dasatinib plus quercetin (D+Q), could rejuvenate antitumor immunity when added to anti-PD-1 therapy. Twenty-four patients with resectable HNSCC received neoadjuvant tislelizumab plus D+Q, followed by surgery. The regimen achieved a 33.3% major pathologic response rate with remarkably low (4.2%) high-grade toxicity. Detailed immune profiling revealed that the treatment effectively reversed features of “immune aging” by expanding populations of CCR7+ CD4+ naïve T cells and CD27+ memory B cells, supporting that senescent-cell clearance can effectively “prime” the immune system for checkpoint efficacy.
How it changes the field: This study provides an early clinical signal that targeting immunosenescence is a tolerable, mechanism-driven strategy to augment PD-1 blockade. If validated in larger controlled trials, senolytic adjuncts could offer a practical way to improve neoadjuvant response rates in HNSCC without the significant toxicity often associated with traditional chemoimmunotherapy.
5 When to Save the Larynx: A New International Roadmap for Laryngeal Preservation
Why this matters: Total laryngectomy remains the most reliable oncologic control for advanced laryngeal cancer but results in the loss of natural voice and a permanent stoma. For years, multidisciplinary teams have lacked uniform criteria to balance organ preservation against oncologic safety. A 2025
Lancet Oncology Delphi consensus addresses this gap, providing expert multidisciplinary guidance on patient selection, baseline function, and follow-up for laryngeal and hypopharyngeal cancers
[5].
What the consensus says: For T2–T3 tumors, treatment selection follows a dual-track evaluation of anatomy and physiology. The panel recommends open partial laryngectomy primarily for fit patients under 70 with strong cardiopulmonary reserve, given its demanding recovery. In contrast, transoral laser microsurgery is prioritized for its rapid recovery but is strictly limited by anatomical constraints; it is contraindicated in cases of thyroid cartilage invasion, posterior paraglottic space infiltration, or inadequate surgical exposure. For non-surgical preservation via concurrent chemoradiotherapy, the panel highlights the necessity of preserved baseline function, requiring patients to have an intact swallow and no prior need for a tracheostomy. For T4a disease, total laryngectomy remains standard. Exceptions are limited to highly selected patients with minimal extralaryngeal extension who prioritize organ preservation despite inferior oncologic outcomes. The document also codifies salvage surgery principles, endorses frailty- instead of age-based selection in older adults, and formalizes baseline and post-treatment voice and swallow evaluations using validated tools, underscoring that successful preservation requires functional reserve and structured surveillance.
How it changes the field: By translating heterogeneous practices into actionable, consensus-backed criteria, this policy review provides tumor boards with a shared decision framework. It clarifies when organ-saving approaches are appropriate and when total laryngectomy is the safer choice. Clinicians now have a long-overdue roadmap that aligns oncologic discipline with the preservation of voice and swallow functions in daily care.
6 Transoral Robotic Surgery (TORS) in Oropharyngeal Cancer: The American Society of Clinical Oncology (ASCO) Guideline
Why this matters: Clinicians have lacked a unified, evidence-based framework to determine the optimal role of TORS in modern care for oropharyngeal cancer (OPC). In 2025, ASCO published a landmark guideline in the
Journal of Clinical Oncology to bridge these gaps and standardize surgical integration
[6].
What the guideline says: The guideline mandates a comprehensive pretreatment workup, including multidisciplinary evaluation, high-risk HPV (p16) testing, cross-sectional imaging, and baseline functional speech and swallow assessments. In HPV-positive OPC, TORS should be considered for well-lateralized T1–T2 tumors where R0 resection is highly likely and surgical exposure is adequate. Adjuvant therapy should then be calibrated to pathology: observation for low-risk features, radiotherapy for intermediate-risk, and chemoradiotherapy for high-risk findings (e.g., positive margins or substantial extranodal extension). In HPV-negative OPC, TORS is reasonable for early T-stage tumors with standard adjuvant criteria. Contraindications include poor exposure or anatomy predicting compromised margins, and salvage TORS is appropriate only in selected recurrences within previously irradiated fields.
How it changes the field: By synthesizing diverse evidence into clear selection and postoperative pathways, the ASCO document provides multidisciplinary teams with a common playbook. It legitimizes TORS as a potential de-escalation gateway for HPV-positive patients while ensuring broader oncologic discipline is maintained. Crucially, it directs clinicians toward non-surgical or intensified approaches when anatomy, stage, or functional reserve are not optimal for a robotic approach.
7 From Text to Voice: A Brain Implant Restoring Real-Time Expressive Speech
Why this matters: Most current speech neuroprostheses translate brain activity into text. While functional, this method strips away prosody, emotion, and the immediacy of a person’s unique voice, all of which are the core ingredients of natural conversation. A 2025 study in
Nature reported a brain-to-voice interface that addresses this gap, generating audible, expressive speech in real time directly from intracortical activity
[7].
What the study did: The study involved a participant with amyotrophic lateral sclerosis (ALS) and severe dysarthria who underwent the surgical implantation of 256 microelectrodes into the speech motor cortex. To enable communication, the researchers utilized a Transformer-based decoder that converted neural signals into acoustic features every 10 milliseconds, which a vocoder then synthesized into instantaneous speech to provide the user with closed-loop audio feedback. Because the participant’s natural speech was no longer available, the team developed an innovative training method using time-aligned synthetic speech targets and neural activity, proving that the system could learn to produce voice without prior recordings. Listeners found the resulting speech to be highly intelligible; crucially, the interface successfully conveyed paralinguistic features, allowing the user to vary his intonation, emphasize specific words, adjust his pace, and even sing short melodies.
How it changes the field: This work moves speech neuroprosthetics beyond static text toward a fully expressive, natural-sounding voice by demonstrating that intracortical signals can drive real-time vocal synthesis carrying both content and prosody. While currently invasive and limited to a single participant, it establishes a compelling path for restoring voice, not just words, to patients who have lost the ability to speak.
8 Beyond Attempted Speech: Decoding the Voice Inside the Head
Why this matters: State-of-the-art speech brain-computer interfaces (BCIs) typically require users to attempt physical speech. This process can be physically exhausting for patients with advanced neurodegenerative diseases and raises significant privacy concerns regarding whether a device might inadvertently broadcast thoughts not intended for sharing. A study in
Cell addresses both usability and privacy by demonstrating that “inner speech”, the silent internal monologue used for thinking and reading, has a robust representation in the motor cortex that can be decoded in real time
[8].
What the study did: The researchers recorded intracortical activity from four participants with ALS or stroke, including one with complete anarthria, while they performed tasks involving attempted speech, inner speech, silent reading, and listening. They discovered that the same words evoked highly similar neural patterns across both attempted and inner speech, suggesting that inner speech acts as a lower-amplitude variant of the same motor-cortical code. Utilizing a recurrent neural network, the team developed a real-time BCI that decoded imagined sentences with word-error rates as low as 26% from a massive 125,000-word vocabulary. Participants noted that using inner speech felt more natural and significantly less effortful than attempting physical articulation.
How it changes the field: This work establishes the feasibility of lower-effort inner-speech decoding while introducing essential built-in privacy safeguards. By identifying a specific “motor-intent” neural dimension that distinguishes attempted speech from internal thought, the team implemented “intent-gating”. This ensures that the system remains silent unless the user deliberately activates it, pointing toward a future of communication neuroprostheses that are practical, expressive, and private by design.
9 A Wearable “Intelligent Throat” That Lets Stroke Patients Speak Again
Why this matters: For many stroke survivors with severe dysarthria who retain some laryngeal or facial control, the communication options are often limited to slow, fatiguing eye-tracking systems or invasive brain implants. There has been little middle ground to enable fluent, natural-sounding conversation in daily life. A study in
Nature Communications introduces a non-invasive, neck-worn device designed to fill this gap
[9].
What the study did: Researchers developed an “intelligent throat” consisting of a soft, wearable choker equipped with textile strain sensors. These sensors capture subtle vibrations from throat muscles during silent mouthing, while a parallel channel monitors carotid pulse waves to infer the user’s emotional state. To enable continuous, real-time speech without forced pauses, an artificial intelligence (AI) pipeline tokenizes movement signals into 144-millisecond segments and decodes them in context. In five stroke patients, brief system fine-tuning achieved remarkably low error rates: 4.2% for words and 2.9% for sentences. When articulating full sentences became too taxing, patients could mouth only key words; a large-language-model (LLM) agent then expanded these fragments into complete, emotion-congruent sentences by integrating pulse data and ambient context. This assisted communication feature increased user satisfaction by 55%.
How it changes the field: By providing a portable, non-invasive method to generate real-time voice from residual throat movements, this system offers a practical alternative to more invasive or cumbersome technologies. The integration of LLM-assisted sentence completion allows for fluid communication even when physical effort is limited. This technology provides stroke survivors with a natural-sounding voice that can be used comfortably in everyday settings.
10 Breath Shape Predicts Hypoglossal Nerve Stimulation Response in Obstructive Sleep Apnea (OSA)
Why this matters: For patients with OSA who cannot tolerate positive airway pressure therapy, hypoglossal nerve stimulation (HNS) is an effective alternative. However, current patient selection relies heavily on drug-induced sleep endoscopy (DISE), an invasive and resource-intensive procedure that can delay treatment. Clinicians have long sought a reliable, non-invasive method to identify who is most likely to benefit from the implant.
What the study found: In a 2025 study published in the
European Respiratory Journal[10], researchers confirmed that oropharyngeal lateral wall (OLW) collapse during DISE is a primary driver of poor surgical outcomes. From a prospective cohort of 369 HNS recipients, the team found this specific collapse pattern was associated with an 18% smaller reduction in the apnea-hypopnea index (AHI) and nearly a fourfold decrease in treatment success. Importantly, the study demonstrated that these outcomes could be anticipated using airflow-shape features captured during sleep testing. Patients identified by the model as having a high probability of OLW collapse experienced 22%–26% lower HNS efficacy compared to those with a low predicted probability.
How it changes the field: Airflow-shape analysis offers a practical path to non-invasive triage. Patients identified with a low probability of OLW collapse via polysomnography or home sleep testing may be considered for HNS without the need for routine DISE. Conversely, those with a high probability can be prioritized for endoscopy or alternative therapies. This approach streamlines the diagnostic workflow, bringing faster and more personalized care to the sleep clinic.
11 Monkeys Have Rhythm
Why this matters: Rhythmic perception is a foundational component of human communication and auditory rehabilitation. A long-standing hypothesis in evolutionary biology suggests that predictive beat synchronization is unique to species capable of complex vocal learning, such as humans and songbirds. This implied that non-vocal-learning primates, including macaques, lacked the capacity for true beat-keeping. However, a 2025 study in
Science overturns this assumption by demonstrating that macaques can indeed synchronize to the beat of real music, suggesting that the evolutionary roots of rhythmic ability are deeper and more widely shared than previously thought
[11].
What the study did: Researchers tested two metronome-trained macaques using human songs in a tapping task. The animals were rewarded only for maintaining interval consistency (five taps within a set tempo window), while phase alignment to the actual beat was never reinforced. Remarkably, both monkeys spontaneously locked to the musical beat and adjusted their tapping phase when the timing of the music was shifted. This behavior indicated an alignment with the musical structure itself rather than a reliance on visual cues or reward timing. When the music was temporally scrambled, the synchronization vanished, confirming the monkeys’ reliance on beat structure. Even in a “free-tapping” condition where any consistent interval earned a reward, the monkeys still converged on the song’s true tempo, showing the strongest phase-locking at that specific rate.
How it changes the field: By demonstrating predictive beat synchronization in macaques, this study challenges the binary “vocal-learner vs. non-learner” framework. Instead, it supports a continuum model where beat perception emerges from the integrated coordination of auditory pattern detection, temporal prediction, and audiomotor coupling. This discovery establishes macaques as a viable primate model for exploring the neurobiology and evolution of human musicality, potentially opening new avenues for understanding rhythmic processing in the brain.
12 Closing Note
From the real-world benefit of immunotherapy to the evolutionary roots of rhythm, these studies reflect the diverse and expanding scope of modern otolaryngology. We present these ten papers as a strategic vantage point rather than a final verdict. They challenge us to rethink how we evaluate impact—shifting toward a holistic view that prioritizes not only survival, but also the understanding and preservation of the qualities that make life better: voice, music, sleep, and the ability to swallow and communicate.
As these technologies and therapies move toward the mainstream, they prompt essential questions: Can these results be replicated across diverse clinical settings? Do the gains persist beyond early endpoints and justify the associated costs, workflows, and training demands? Furthermore, as technology often outpaces policy, how can we ensure that care remains equitable and accessible? Addressing these questions will require a continuous, concerted effort from the entire clinical and research community.
We offer this curation as a starting point for discussion in tumor boards, clinics, and labs. Progress in our field is iterative and depends on our collective ability to pressure-test new data against local realities. Ultimately, the goal is to translate these scientific signals into treatments that are more effective, cost-efficient, and capable of improving patient health outcomes worldwide.
The Author(s). This article is published by Higher Education Press at journal.hep.com.cn.
This is an open access article distributed under the Creative Commons Attribution License 4.0 (CCBY), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
PDF
(212KB)
