KEY POINTS:Anosmia/hyposmia affected 14.8%-19.6% of participants in the UK and the USA in the post-pandemic era. Since symptom onset, 37.1%-48.8% of affected participants reported no improvement, while 15.3%-17.6% reported worsening symptoms. Anosmia/hyposmia remains strongly associated with reduced quality of life across both countries. Among respondents reporting anosmia or hyposmia, a high proportion (88.8%-93.5%) also experienced concurrent parosmia.

Supplemental Figure 3. Boxplot of NF- κB values across the three cluster subtypes. By unsupervised consensus clustering and self-organizing map (SOM) approaches, we identified three distinct biological subtypes within HPV-positive HNSCC: Cl1 (immune-related), Cl2 (epithelial-mesenchymal transition [EMT]-related), and Cl3 (proliferation-related). This classification holds prognostic significance, with Cl1 exhibiting the most favorable survival outcomes, Cl2 associated with the poorest prognosis, and Cl3 displaying an intermediate survival rate [Locati, 2019]. In contrast to recent studies that classified HPV-related HNSCC into immune and keratinocyte subtypes, our analysis confirmed the immune-associated subtype while further resolving the keratinocyte group into two distinct clusters with differing biological profiles and prognostic implications. NF-κB signature is significantly activated in Cl1-immune related cluster compared to the other 2 clusters: A) Metanalyisis-HPV267 cohort (1); B) BD2-HPV286 cohort (2).

Supplemental Figure 8A. UMAP dimensional reduction plots with marker gene validating cell type assignments. Marker genes related to epithelial cells, endothelial cells, fibroblasts, mast cells, dendritic cells, and myeloid cells. Supplemental Figure 8B. UMAP dimensional reduction plots with marker gene validating cell type assignments. Marker genes related to natural killer, B, T and cycling cells.

Supplemental Figure 9. Marker genes by cluster using Leiden method.

Supplemental Table 2. Association of NF-κB high and low groups with clinical parameters in both Metanalysis-HPV267 (1) and BD2-HPV286 (2). P-value assessed by χ2 test, with the exception of age (t-test). Abbreviations: CHT, chemotherapy; RT, radiotherapy.

Importance:Standard treatment for human papillomavirus (HPV)-associated oropharyngeal carcinoma (OPC) often includes 7 weeks of intensive chemoradiotherapy with substantial toxic effects. Lowering the radiation dose to gross disease may maintain cancer control with fewer toxic effects. Objective:To evaluate long-term oncologic outcomes among patients with HPV-associated OPC treated with definitive deintensified (chemo)radiotherapy. Design, Setting, and Participants:This cohort study evaluated consecutive patients treated with (chemo)radiotherapy between September 2014 and August 2022 at multiple institutions under one academic hospital system. Eligible patients had p16-positive OPC with a clinical stage from T0 to T3 and N0 to N2c (according to the American Joint Committee on Cancer [AJCC] Staging Manual, seventh edition) and favorable smoking history. Data were analyzed between August 2023 and October 2025. Interventions:Treatment was 60-Gy intensity-modulated radiotherapy with first-choice concurrent chemotherapy of weekly cisplatin, 30 mg/m2, or weekly cisplatin, 40 mg/m2. Patients with OPC with a clinical stage from T0 to T2 and N0 to N1 (according to AJCC, seventh edition) were recommended 60-Gy radiotherapy alone. Main Outcomes and Measures:Evaluation of locoregional recurrence was the primary outcome. Distant recurrence, progression-free survival (PFS), and overall survival (OS) were secondary outcomes. Results:A total of 240 patients received definitive deintensified (chemo)radiotherapy (207 receiving prospective protocol-adherent treatment and 33 receiving non-protocol-adherent treatment). Mean (SD) patient age was 60.3 (9.9) years (range, 33.0-84.0 years); 206 (85.8%) were male, and 34 (14.2%) were female. The cohort included 139 never smokers (57.9%) and 101 former smokers (42.1%) (median smoking history, 9.0 pack-years [range, 0.25-50.0 pack-years]). Concurrent chemotherapy was prescribed for 205 patients (85.4%), with 185 (77.1%) receiving cisplatin, and 165 (68.8%) receiving weekly cisplatin, 30 mg/m2. Median follow-up for surviving patients was 6.5 years (range, 0.44-11.0 years; 208 of 210 [99.0%] with minimum 2-year follow-up). Outcomes were as follows: 2-year OS was 97.9% (95% CI, 96.1%-99.7%), and 5-year OS, 92.4% (95% CI, 89.0%-96.0%); 2-year PFS, 94.1% (95% CI, 91.2%-97.2%), and 5-year PFS, 86.5% (95% CI, 82.1%-91.0%); 2-year locoregional recurrence, 1.3% (95% CI, -0.2% to 2.7%), and 5-year locoregional recurrence, 3.4% (95% CI, 1.1%-5.8%); 2-year distant recurrence, 4.6% (95% CI, 1.9%-7.3%), and 5-year distant recurrence, 7.3% (95% CI, 3.9%-10.7%). The median time to progression events was 1.9 years (range, 0.3-5.1 years); 12 of 26 recurrences (46.2%) occurred after 2 years. Thirty-three patients with low-risk disease (clinical stage T0-T2, N0-N1, according to AJCC, seventh edition) received radiotherapy alone, which resulted in a 5-year PFS of 93.8%, with no patients experiencing a locoregional recurrence. Conclusions and Relevance:In this cohort study, modest dose deintensification of 60-Gy radiotherapy in patients with HPV-associated OPC shows favorable rates of locoregional control. These findings provide support for ongoing investigations of deintensified radiotherapy.

Supplemental Figure 2. UMAP dimensional reduction plot with malignancy probabilities from scMalignantFinder. Low probability predictions (<0.5) were imputed with to a value of 0.5 for visual clarity.

Supplemental Figure 6. Survival Plot. Prognostic Value of NanoString NAC Classifier Measurements. In Figure 4, PFS was shown for (E) all patients and (F) de-escalation candidates in UNC and E1308 cohorts (non-smoker,

Supplemental Table 4. Association of NF-κB high and low groups with clinical parameters in combined UNC &E1308 cohorts (1) as pictured in Figure 4D-E. P-value assessed by χ2 test, except for age (assessed with t-test).

Supplemental Figure 4. Comprehensive comparison of all gene sets and potential summary metrics for the NF-κB signature using RNAseq from the UNC cohort. UNC Cohort originally presented in Schrank and colleagues PNAS 2023 (1). Correlation plot with scatter and Pearson correlation values. Nano str – the gene signature utilized in the nanostring assay including 232 NF-κB related genes, which are compared to 27 control genes, these genes were identified in our prior publications as outlined below and represent the combination of the PNAS and Onco large gene sets as defined below. Onco small – Top 50 NF-κB related genes identified in prior work using machine learning improved differential expression, prioritized by p-value in Schrank et at, Oncotarget 2022 (2), 49 of which are included (one excluded by probe design) in the NanoString panel. 29 of these genes are also identified as NF-κB related genes with an orthogonal approach in our prior work using WGCNA (1). Onco large – Top 75 NF-κB related genes identified in prior work using machine learning improved differential expression, prioritized by p-value in Schrank et at, Oncotarget 2022 (2), 74 of which are included (one excluded by probe design) in the NanoString panel 45 of these genes are also identified as NF-κB related genes with an orthogonal approach in our prior work using WGCNA (1). Pnas – 203 NF-κB related genes identified in our prior work using WGCNA, a network-based approach in Schrank and colleagues PNAS 2023 (3). Onco only – NF-κB genes identified by differential expression but not by WGCNA. Pnas only – NF-κB genes identified by WGCNA but not by differential expression. Mean – mean expression of signature genes. Median – median expression of signature genes. PCA – sample value of PC1, for PCA performed on signature genes only after expression normalization. ssGSEA – Signature sized normalized enrichment score (NES) calculated by the single sample gene set enrichment method.

Supplemental Figure 1. Correlation plots for the: A) Metanalyisis-HPV267 (1) cohort; B) BD2-HPV286 (2) cohorts, comparing various summary metrics (mean, ssGSEA enrichment score and for the NF-κB RNA signature in microarray gene expression data. Values represent Pearson correlation coefficients. *** p-value of Pearson correlation <0.0005.

Supplemental Table 1. Association of NF-κB pathway variants with patient clinical features. P-value assessed by χ2 test, except for age (assessed with t-test).

Supplemental Figure 5. NF-κB pathway mutations are found in cases with high NF-κB scores. A) NanoString NF-κB expression scores and NF-κB pathway mutational data from UNC cohort where both metrics were available. B) Non-linear least squares fit of mono-exponential function relating NanoString and RNAseq data. This fit was used to transpose the threshold defined by NanoString to the RNAseq data, where 3 more cases were available.

Supplemental Figure 7. IF staining was performed at UNC Pathology Services Core. Patients in the UNC RNAseq cohort were stratified by RNA NF-κB Signature score as previously published. (1) The following antibodies were used: NIK antibody: Cat#SC-8417 (Santa Cruz) and Pan CK: Leica NCL-L-AE1/AE3-601 (Leica). High resolution images of IF slides were obtained performed with the Aperio Versa 200 scanner (Leica Biosystems Inc.) at an apparent magnification of 10X. Images were uploaded to the eSlideManager database (Aperio) at the Pathology Services Core at UNC. ** p-value <0.00005, Wilcox test.

Tobacco, alcohol, and human papillomavirus (HPV) are major risk factors for oral cavity (OC) and oropharyngeal (OPC) cancers, yet their specific contributions to the tumor molecular landscape remain incompletely understood. We analyzed somatic alterations in 1,086 tumors (606 OC, 480 OPC, including 287 HPV(+)OPC) profiled with a 1,109-gene panel and evaluated their associations with exposures. PIK3CA alterations were widespread but context-specific: HPV(+)OPC featured APOBEC-driven mutations and co-occurrence of FGFR3 mutations in a subset of cases, while HPV(-)OPC and OC were marked by 3q26/28 (PIK3CA/TP63/SOX2) amplifications. TP53 was the most frequently mutated gene in HPV(-) tumors, with strong enrichment among smokers in HPV(-)OPC, whereas TP53 mutation frequencies in OC were consistently high regardless of smoking status. CCND1 amplifications were largely restricted to HPV(-) tumors and were most frequent in individuals reporting combined tobacco/alcohol use. This combined effect was replicated in the TCGA OC-OPC cohort and mirrored at the transcriptomic level, where CCND1 mRNA expression was highest in dual users. Within OC, characterization of one of the largest cohorts of tumors from never smoker-never drinkers to date (N = 100) showed they were more likely to be from female patients and to exhibit enrichment for FAT1 and CASP8 mutations compared to tumors from ever smoker-ever drinkers (N = 353). Together, these findings define distinct, exposure-driven patterns of somatic alterations by subsite and HPV status, and provide a molecular framework that may inform etiologic-based risk stratification in head and neck cancer.

Supplemental Table 3. NF-κB high and low patient samples with clinical parameters in combined scRNAseq data from Kurten and colleagues (1) No recurrence or progression events were noted in these 6 patients.

PURPOSE:Patients with human papillomavirus-associated head and neck squamous cell carcinoma (HPV+ HNSCC) have relatively favorable outcomes, but standard treatments like radiation or chemoradiation frequently result in long-term toxic side effects. Appropriate patient selection has been a barrier to effectively de-escalate therapy for HPV+ HNSCC, and the absence of accurate biomarkers likely contributed to failure of recent promising de-escalation trials that rely on histologic tumor characteristics, history of tobacco use, or tumor response to chemotherapy; This deficiency underlines the need to develop and validate an assay to accurately detect two subtypes of HPV+ HNSCC-one with good prognosis and one with poor prognosis. EXPERIMENTAL DESIGN:These two subtypes are distinguished by the activity of NF-κB in tumors. We first developed a DNA-based marker panel consisting of genes that when mutated would lead to NF-κB activation. Additionally, we developed a custom NanoString assay to determine the expression of NF-κB target genes. These assays were tested to determine their accuracy in detecting tumor subtype. RESULTS:We demonstrate that the NF-κB gene signature score, as determined using the NanoString assay, could more accurately classify HPV+ HNSCC as compared with the DNA-based marker panel. Patients with a high NF-κB gene signature score demonstrated significantly increased overall survival, indicating more sensitivity to (chemo)radiation treatment. CONCLUSIONS:The NF-κB gene signature score can accurately predict response to standard (chemo)radiation in HPV+ HNSCC. This molecular biomarker holds promise for clinical use in identifying patients who are likely to benefit from treatment de-escalation strategies, potentially reducing long-term side effects without compromising therapeutic efficacy.

Abstract Background: Improving understanding of head and neck squamous cell carcinoma (HNSCC) has led to the discovery of novel molecular markers to stratify risk and guide treatment decisions. Recently, we have identified two subtypes of HPV+ OPSCC – one with good prognosis and one with poor prognosis. The subtypes are distinct in many ways – somatic genes mutated, HPV viral oncogenes expression, the physical state of the viral genome (integrated vs episomal), as well as global mutation and methylation profiles. Intriguingly, all differences converge on intrinsic tumor NF-κB activity with constitutively active NF-κB (usually arising from genetic defects in NF-κB regulators, including TRAF3 and CYLD) driving significantly elevated radiation sensitivity and improved patient survival. NF-κB-based groups define an important clinically actionable prognostic biomarker in HPV+ HNSCC: one important goal is to de-escalate therapy for patients with molecularly low-risk tumors to improve quality of life amongst survivors, while another goal is to improve the cure for of patients with a high risk of treatment failure. Material and methods: We utilized HPV+ head and neck cancer cells, HPV+ HNSCC xenografted mouse model, patient-derived xenografts (PDX), as well as our novel fully immunocompetent mouse model of high risk HPV+ HNSCC; we applied targeted fractionated radiation. To activate NF-κB, we used two clinically applicable drugs with different mechanisms of action: Toll-like receptor 5 (TLR5) agonist and mitochondria-derived activator of caspases (SMAC) mimetic. Custom NF-κB NanoString panel contains our recently developed NF-κB gene signature along with HPV genes and housekeeping genes that are used to normalize the gene expression data. Results: Both drugs activated NF-κB in HPV+ HNSCC and significantly improved response of HPV+ head and neck cancer cells and tumors to radiation. Interestingly, TLR5 agonist alone had a strong tumor suppressing effect in low NF-κB HPV+ HNSCC; this effect was significantly better than 32 Gy focused radiation given in fractions of 4 Gy. Custom NF-κB NanoString panel is robustly prognostic with NF-κB active tumors having improved outcomes after chemoradiation. Conclusions: Because of the average relatively favorable prognosis of HPV+ HNSCC, urgent need for improved therapy for patients with aggressive high risk tumors is greatly undervalued. Here, we present a clinical assay to risk stratify patients with HPV+ HNSCC and begin to explore pharmacological activation of NF-κB as a strategy to sensitize high risk HPV+ HNSCC to radiation therapy. TLR5 agonist is well known for the remarkable normal tissue protective effects against radiation and other damaging conditions; the dual effect of entolimod – sensitization of HPV+ HNSCC to radiation, while simultaneously protecting normal tissues from severe side effects - may be useful for patients with high-risk HPV+ head and neck cancer in combination with radiotherapy. Citation Format: Sri Vemulamanda, Natalia Issaeva1, Aditi Kothari, Travis Schrank, Wendell G. Yarbrough. Novel strategy to improve response of high risk HPV+ HNSCC to radiation therapy. [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Translating Targeted Therapies in Combination with Radiotherapy; 2025 Jan 26-29; San Diego, CA. Philadelphia (PA): AACR; Clin Cancer Res 2025;31(2_Suppl):Abstract nr B030.

Abstract Introduction: Patients with locally advanced head and neck squamous cell carcinoma (LA-HNSCC) have high risk of disease progression and death after standard concurrent chemoradiotherapy (CCRT). We are conducting a prospective trial to investigate the efficacy of combining immune checkpoint blockade (ICB) with PARP inhibition prior to and following CCRT. Here we report the feasibility of utilizing pre- and post-induction CT simulations to assess early imaging response in patients treated on trial. Methods: NCT05366166 is an ongoing multi-institution, phase II trial. Patients with HPV negative or high-risk HPV positive LA-HNSCC were eligible. All received induction pembrolizumab 400mg IV and olaparib 150mg BID for 3 weeks followed by CCRT to 70Gy with weekly cisplatin (40mg/m2) followed by one year of maintenance pembrolizumab 400mg every 6 weeks and olaparib 150mg BID. Per protocol, all patients underwent CT simulation in the radiation treatment position with intravenous contrast pre- and post-induction. Dedicated post-induction imaging by radiology was not obtained. Gross tumor volumes of the primary (GTVp) and nodal (GTVn) disease were contoured on the pre- and post-induction simulation CTs by the treating physician in RayStation (RaySearch Laboratories). Pre- and post-induction simulation CTs were evaluated by two neuroradiologists independently using the RECIST criteria (v1.1). Statistical analyses were conducted in Microsoft Excel. Results: We evaluated the first 9 patients treated on-study at our center only. Median age was 67 years (range 59-76) and 78% were male. Location of primary site was oropharynx (N=6; 4 were HPV-positive), oral cavity (N=2), and larynx (N=1). 78% were T3-T4 and 89% had N2-N3 nodal disease. The median duration from the start of induction therapy to post-induction simulation was 22 days (range 7-51). Following induction, all patients experienced symptomatic benefit and had stable disease without progression per RECIST. Comparing pre- and post-induction GTVp and GTVn, overall mean volumes were not statistically different (mean±standard deviation -- 34.0±17.9 vs 35.6±19.5 cm3 for GTVp and 38.7±51.8 vs 39.5±54.2 cm3 for GTVn, respectively). At the individual patient level, the percent volume change after induction ranged from -21.9 to +16.5% (GTVp) and -48.7 to +53.2% (GTVn). Two patients (22%) had a decrease in GTVp, while five (56%) had a decrease in GTVn. There was a positive correlation (Pearson coefficient 0.77) between the percent volume changes in GTVp and GTVn (p=0.016). No differences in response by TNM stage or HPV status were observed. Conclusion: Favorable clinical and radiographic responses to three weeks of induction pembrolizumab and olaparib suggest anti-tumor activity. Performing pre- and post-induction radiation CT simulations is necessary to ensure adequate and accurate assessment of tumor extent for radiation planning. Therefore, omitting conventional post-induction scans by radiology to assess tumor response may be feasible. Formal cost-effective analysis will be conducted. Citation Format: Xuguang S. Chen, Colette J. Shen, Benjamin Y. Huang, Valerie L. Jewells, Michael C. Repka, Travis P. Schrank, Christopher B. Sullivan, Wendell G. Yarbrough, Trevor G. Hackman, Catherine Lumley, Jeffrey M. Blumberg, Shetal A. Patel, Sumita Trivedi, Jared Weiss, Bhisham Chera, Siddharth Sheth. Imaging response to induction PD-1 and PARP inhibition in patients with locally advanced head and neck squamous cell carcinoma. [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Translating Targeted Therapies in Combination with Radiotherapy; 2025 Jan 26-29; San Diego, CA. Philadelphia (PA): AACR; Clin Cancer Res 2025;31(2_Suppl):Abstract nr A008.