Artificial intelligence-based epigenomic, transcriptomic and histologic signatures of tobacco use in oral squamous cell carcinoma.

Oral squamous cell carcinoma (OSCC) biomarker studies rarely employ multi-omic biomarker strategies and pertinent clinicopathologic characteristics to predict mortality. In this study we determine for the first time a combined epigenetic, gene expression, and histology signature that differentiates between patients with different tobacco use history (heavy tobacco use with ≥10 pack years vs. no tobacco use). Using The Cancer Genome Atlas (TCGA) cohort (n = 257) and an internal cohort (n = 40), we identify 3 epigenetic markers (GPR15, GNG12, GDNF) and 13 expression markers (IGHA2, SCG5, RPL3L, NTRK1, CD96, BMP6, TFPI2, EFEMP2, RYR3, DMTN, GPD2, BAALC, and FMO3), which are dysregulated in OSCC patients who were never smokers vs. those who have a ≥ 10 pack year history. While mortality risk prediction based on smoking status and clinicopathologic covariates alone is inaccurate (c-statistic = 0.57), the combined epigenetic/expression and histologic signature has a c-statistic = 0.9409 in predicting 5-year mortality in OSCC patients.

Artificial intelligence-based morphologic classification and molecular characterization of neuroblastic tumors from digital histopathology.

A deep learning model using attention-based multiple instance learning (aMIL) and self-supervised learning (SSL) was developed to perform pathologic classification of neuroblastic tumors and assess MYCN-amplification status using H&E-stained whole slide digital images. The model demonstrated strong performance in identifying diagnostic category, grade, mitosis-karyorrhexis index (MKI), and MYCN-amplification on an external test dataset. This AI-based approach establishes a valuable tool for automating diagnosis and precise classification of neuroblastoma tumors.

Acquired resistance to immunotherapy and chemoradiation in MYC amplified head and neck cancer.

The proto-oncogene MYC encodes a nuclear transcription factor that has an important role in a variety of cellular processes, such as cell cycle progression, proliferation, metabolism, adhesion, apoptosis, and therapeutic resistance. MYC amplification is consistently observed in aggressive forms of several solid malignancies and correlates with poor prognosis and distant metastases. While the tumorigenic effects of MYC in patients with head and neck squamous cell carcinoma (HNSCC) are well known, the molecular mechanisms by which the amplification of this gene may confer treatment resistance, especially to immune checkpoint inhibitors, remains under-investigated. Here we present a unique case of a patient with recurrent/metastatic (R/M) HNSCC who, despite initial response to nivolumab-based treatment, developed rapidly progressive metastatic disease after the acquisition of MYC amplification. We conducted comparative transcriptomic analysis of this patient's tumor at baseline and upon progression to interrogate potential molecular processes through which MYC may confer resistance to immunotherapy and/or chemoradiation and used TCGA-HNSC dataset and an institutional cohort to further explore clinicopathologic features and key molecular networks associated with MYC amplification in HNSCC. This study highlights MYC amplification as a potential mechanism of immune checkpoint inhibitor resistance and suggest its use as a predictive biomarker and potential therapeutic target in R/M HNSCC.

Generative Adversarial Networks Accurately Reconstruct Pan-Cancer Histology from Pathologic, Genomic, and Radiographic Latent Features.

Artificial intelligence models have been increasingly used in the analysis of tumor histology to perform tasks ranging from routine classification to identification of novel molecular features. These approaches distill cancer histologic images into high-level features which are used in predictions, but understanding the biologic meaning of such features remains challenging. We present and validate a custom generative adversarial network - HistoXGAN - capable of reconstructing representative histology using feature vectors produced by common feature extractors. We evaluate HistoXGAN across 29 cancer subtypes and demonstrate that reconstructed images retain information regarding tumor grade, histologic subtype, and gene expression patterns. We leverage HistoXGAN to illustrate the underlying histologic features for deep learning models for actionable mutations, identify model reliance on histologic batch effect in predictions, and demonstrate accurate reconstruction of tumor histology from radiographic imaging for a 'virtual biopsy'.

Slideflow: deep learning for digital histopathology with real-time whole-slide visualization.

Deep learning methods have emerged as powerful tools for analyzing histopathological images, but current methods are often specialized for specific domains and software environments, and few open-source options exist for deploying models in an interactive interface. Experimenting with different deep learning approaches typically requires switching software libraries and reprocessing data, reducing the feasibility and practicality of experimenting with new architectures. We developed a flexible deep learning library for histopathology called Slideflow, a package which supports a broad array of deep learning methods for digital pathology and includes a fast whole-slide interface for deploying trained models. Slideflow includes unique tools for whole-slide image data processing, efficient stain normalization and augmentation, weakly-supervised whole-slide classification, uncertainty quantification, feature generation, feature space analysis, and explainability. Whole-slide image processing is highly optimized, enabling whole-slide tile extraction at 40x magnification in 2.5 s per slide. The framework-agnostic data processing pipeline enables rapid experimentation with new methods built with either Tensorflow or PyTorch, and the graphical user interface supports real-time visualization of slides, predictions, heatmaps, and feature space characteristics on a variety of hardware devices, including ARM-based devices such as the Raspberry Pi.

Deep learning generates synthetic cancer histology for explainability and education.

Artificial intelligence methods including deep neural networks (DNN) can provide rapid molecular classification of tumors from routine histology with accuracy that matches or exceeds human pathologists. Discerning how neural networks make their predictions remains a significant challenge, but explainability tools help provide insights into what models have learned when corresponding histologic features are poorly defined. Here, we present a method for improving explainability of DNN models using synthetic histology generated by a conditional generative adversarial network (cGAN). We show that cGANs generate high-quality synthetic histology images that can be leveraged for explaining DNN models trained to classify molecularly-subtyped tumors, exposing histologic features associated with molecular state. Fine-tuning synthetic histology through class and layer blending illustrates nuanced morphologic differences between tumor subtypes. Finally, we demonstrate the use of synthetic histology for augmenting pathologist-in-training education, showing that these intuitive visualizations can reinforce and improve understanding of histologic manifestations of tumor biology.

Integration of clinical features and deep learning on pathology for the prediction of breast cancer recurrence assays and risk of recurrence.

Gene expression-based recurrence assays are strongly recommended to guide the use of chemotherapy in hormone receptor-positive, HER2-negative breast cancer, but such testing is expensive, can contribute to delays in care, and may not be available in low-resource settings. Here, we describe the training and independent validation of a deep learning model that predicts recurrence assay result and risk of recurrence using both digital histology and clinical risk factors. We demonstrate that this approach outperforms an established clinical nomogram (area under the receiver operating characteristic curve of 0.83 versus 0.76 in an external validation cohort, p = 0.0005) and can identify a subset of patients with excellent prognoses who may not need further genomic testing.

Machine learning for the prediction of toxicities from head and neck cancer treatment: A systematic review with meta-analysis.

INTRODUCTION: The aim of the present systematic review (SR) is to summarize Machine Learning (ML) models currently used to predict head and neck cancer (HNC) treatment-related toxicities, and to understand the impact of image biomarkers (IBMs) in prediction models (PMs). The present SR was conducted following the guidelines of the PRISMA 2022 and registered in PROSPERO database (CRD42020219304). METHODS: The acronym PICOS was used to develop the focused review question (Can PMs accurately predict HNC treatment toxicities?) and the eligibility criteria. The inclusion criteria enrolled Prediction Model Studies (PMSs) with patient cohorts that were treated for HNC and developed toxicities. Electronic database search encompassed PubMed, EMBASE, Scopus, Cochrane Library, Web of Science, LILACS, and Gray Literature (Google Scholar and ProQuest). Risk of Bias (RoB) was assessed through PROBAST and the results were synthesized based on the data format (with and without IBMs) to allow comparison. RESULTS: A total of 28 studies and 4,713 patients were included. Xerostomia was the most frequently investigated toxicity (17; 60.71 %). Sixteen (57.14 %) studies reported using radiomics features in combination with clinical or dosimetrics/dosiomics for modelling. High RoB was identified in 23 studies. Meta-analysis (MA) showed an area under the receiver operating characteristics curve (AUROC) of 0.82 for models with IBMs and 0.81 for models without IBMs (p value < 0.001), demonstrating no difference among IBM- and non-IBM-based models. DISCUSSION: The development of a PM based on sample-specific features represents patient selection bias and may affect a model's performance. Heterogeneity of the studies as well as non-standardized metrics prevent proper comparison of studies, and the absence of an independent/external test does not allow the evaluation of the model's generalization ability. CONCLUSION: IBM-featured PMs are not superior to PMs based on non-IBM predictors. The evidence was appraised as of low certainty.

Data augmentation and multimodal learning for predicting drug response in patient-derived xenografts from gene expressions and histology images.

Patient-derived xenografts (PDXs) are an appealing platform for preclinical drug studies. A primary challenge in modeling drug response prediction (DRP) with PDXs and neural networks (NNs) is the limited number of drug response samples. We investigate multimodal neural network (MM-Net) and data augmentation for DRP in PDXs. The MM-Net learns to predict response using drug descriptors, gene expressions (GE), and histology whole-slide images (WSIs). We explore whether combining WSIs with GE improves predictions as compared with models that use GE alone. We propose two data augmentation methods which allow us training multimodal and unimodal NNs without changing architectures with a single larger dataset: 1) combine single-drug and drug-pair treatments by homogenizing drug representations, and 2) augment drug-pairs which doubles the sample size of all drug-pair samples. Unimodal NNs which use GE are compared to assess the contribution of data augmentation. The NN that uses the original and the augmented drug-pair treatments as well as single-drug treatments outperforms NNs that ignore either the augmented drug-pairs or the single-drug treatments. In assessing the multimodal learning based on the MCC metric, MM-Net outperforms all the baselines. Our results show that data augmentation and integration of histology images with GE can improve prediction performance of drug response in PDXs.

Uncertainty-informed deep learning models enable high-confidence predictions for digital histopathology.

A model's ability to express its own predictive uncertainty is an essential attribute for maintaining clinical user confidence as computational biomarkers are deployed into real-world medical settings. In the domain of cancer digital histopathology, we describe a clinically-oriented approach to uncertainty quantification for whole-slide images, estimating uncertainty using dropout and calculating thresholds on training data to establish cutoffs for low- and high-confidence predictions. We train models to identify lung adenocarcinoma vs. squamous cell carcinoma and show that high-confidence predictions outperform predictions without uncertainty, in both cross-validation and testing on two large external datasets spanning multiple institutions. Our testing strategy closely approximates real-world application, with predictions generated on unsupervised, unannotated slides using predetermined thresholds. Furthermore, we show that uncertainty thresholding remains reliable in the setting of domain shift, with accurate high-confidence predictions of adenocarcinoma vs. squamous cell carcinoma for out-of-distribution, non-lung cancer cohorts.

Development and Validation of a Decision Analytical Model for Posttreatment Surveillance for Patients With Oropharyngeal Carcinoma.

Importance: Clinical practice regarding posttreatment radiologic surveillance for patients with oropharyngeal carcinoma (OPC) is neither adapted to individual patient risk nor fully evidence based. Objectives: To construct a microsimulation model for posttreatment OPC progression and use it to optimize surveillance strategies while accounting for both tumor stage and human papillomavirus (HPV) status. Design, Setting, and Participants: In this decision analytical modeling study, a Markov model of 3-year posttreatment patient trajectories was created. The training data source was the American College of Surgeon's National Cancer Database from 2010 to 2015. The external validation data set was the 2016 International Collaboration on Oropharyngeal Cancer Network for Staging (ICON-S) study. Training data comprised 2159 patients with OPC treated with primary radiotherapy who had known HPV status and disease staging information. Patients with American Joint Committee on Cancer, 7th edition stage III to IVB disease and those with clinical metastases during the time of primary treatment were included. Data were analyzed from August 1 to October 31, 2020. Main Outcomes and Measures: Main outcomes included disease stage and HPV status, specific disease transition probabilities, and latency of surveillance regimens, defined as time between recurrence incidence and disease discovery. Results: Training data consisted of 2159 total patients (1708 men [79.1%]; median age, 59.6 years [range, 40-90 years]; 401 with stage III disease, 1415 with stage IVA disease, and 343 with stage IVB disease). Cohorts predominantly had HPV-negative disease (1606 [74.4%]). With model-optimized regimens, recurrent disease was discovered a mean of 0.6 months (95% CI, 0.5-0.8 months) earlier than with a standard surveillance regimen based on current clinical guidelines. Recurrent disease was discovered using the optimized regimens without significant reduction in sensitivity. Compared with strategies based on reimbursement guidelines, the model-optimized regimens found disease a mean of 1.8 months (95% CI, 1.3-2.3 months) earlier. Conclusions and Relevance: Optimized, risk-stratified surveillance regimens consistently outperformed nonoptimized strategies. These gains were obtained without requiring any additional imaging studies. This approach to risk-stratified surveillance optimization is generalizable to a broad range of tumor types and risk factors.

Risk and response adapted de-intensified treatment for HPV-associated oropharyngeal cancer: Optima paradigm expanded experience.

BACKGROUND: Favorable prognosis for Human papillomavirus-associated (HPV+) oropharyngeal cancer (OPC) led to investigation of response-adaptive de-escalation, yet long-term outcomes are unknown. We present expanded experience and follow-up of risk/response adaptive treatment de-intensification in HPV+ OPC. METHODS: A phase 2 trial (OPTIMA) and subsequent cohort of sequential off-protocol patients treated from September 2014 to November 2018 at the University of Chicago were reviewed. Eligible patients had T3-T4 or N2-3 (AJCC 7th edition) HPV+ OPC. Patients were stratified by risk: High-risk (HR) (T4, ≥N2c, or >10PYH), all others low-risk (LR). Induction chemotherapy (IC) included 3 cycles of carboplatin and nab-paclitaxel (OPTIMA) or paclitaxel (off-protocol). LR with ≥50% response received low-dose radiotherapy (RT) alone to 50 Gy (RT50). LR with 30-50% response and HR with ≥50% response received intermediate-dose chemoradiotherapy (CRT) to 45 Gy (CRT45). All others received full-dose CRT to 75 Gy (CRT75). RESULTS: 91 patients consented and 90 patients were treated, of which 31% had >10PYH, 34% had T3/4 disease, and 94% had N2b/N2c/N3 disease. 49% were LR and 51% were HR. Overall response rate to induction was 88%. De-escalated treatment was administered to 83%. Median follow-up was 4.2 years. Five-year OS, PFS, LRC, and DC were 90% (95% CI 81,95), 90% (95% CI 80,95), 96% (95% CI 90,99), and 96% (88,99) respectively. G-tube placement rates in RT50, CRT45, and CRT75 were 3%, 33%, and 80% respectively (p < 0.05). CONCLUSION: Risk/response adaptive de-escalated treatment for an inclusive cohort of HPV+ OPC demonstrates excellent survival with reduced toxicity with long-term follow-up.

Immune-related adverse events are associated with improved response, progression-free survival, and overall survival for patients with head and neck cancer receiving immune checkpoint inhibitors.

BACKGROUND: The authors hypothesized that patients developing immune-related adverse events (irAEs) while receiving immune checkpoint inhibition (ICI) for recurrent/metastatic head and neck cancer (HNC) would have improved oncologic outcomes. METHODS: Patients with recurrent/metastatic HNC received ICI at 2 centers. Univariate and multivariate logistic regression, Kaplan-Meier methods, and Cox proportional hazards regression were used to associate the irAE status with the overall response rate (ORR), progression-free survival (PFS), and overall survival (OS) in cohort 1 (n = 108). These outcomes were also analyzed in an independent cohort of patients receiving ICI (cohort 2; 47 evaluable for irAEs). RESULTS: The median follow-up was 8.4 months for patients treated in cohort 1. Sixty irAEs occurred in 49 of 108 patients with 5 grade 3 or higher irAEs (10.2%). ORR was higher for irAE+ patients (30.6%) in comparison with irAE- patients (12.3%; P = .02). The median PFS was 6.9 months for irAE+ patients and 2.1 months for irAE- patients (P = .0004), and the median OS was 12.5 and 6.8 months, respectively (P = .007). Experiencing 1 or more irAEs remained associated with ORR (P = .03), PFS (P = .003), and OS (P = .004) in multivariate analyses. The association between development of irAEs and prolonged OS persisted in a 22-week landmark analysis (P = .049). The association between development of irAEs and favorable outcomes was verified in cohort 2. CONCLUSIONS: The development of irAEs was strongly associated with an ICI benefit, including overall response, PFS, and OS, in 2 separate cohorts of patients with recurrent/metastatic HNC.

The impact of site-specific digital histology signatures on deep learning model accuracy and bias.

The Cancer Genome Atlas (TCGA) is one of the largest biorepositories of digital histology. Deep learning (DL) models have been trained on TCGA to predict numerous features directly from histology, including survival, gene expression patterns, and driver mutations. However, we demonstrate that these features vary substantially across tissue submitting sites in TCGA for over 3,000 patients with six cancer subtypes. Additionally, we show that histologic image differences between submitting sites can easily be identified with DL. Site detection remains possible despite commonly used color normalization and augmentation methods, and we quantify the image characteristics constituting this site-specific digital histology signature. We demonstrate that these site-specific signatures lead to biased accuracy for prediction of features including survival, genomic mutations, and tumor stage. Furthermore, ethnicity can also be inferred from site-specific signatures, which must be accounted for to ensure equitable application of DL. These site-specific signatures can lead to overoptimistic estimates of model performance, and we propose a quadratic programming method that abrogates this bias by ensuring models are not trained and validated on samples from the same site.

Deep learning prediction of BRAF-RAS gene expression signature identifies noninvasive follicular thyroid neoplasms with papillary-like nuclear features.

Noninvasive follicular thyroid neoplasms with papillary-like nuclear features (NIFTP) are follicular-patterned thyroid neoplasms defined by nuclear atypia and indolent behavior. They harbor RAS mutations, rather than BRAFV600E mutations as is observed in papillary thyroid carcinomas with extensive follicular growth. Reliably identifying NIFTPs aids in safe therapy de-escalation, but has proven to be challenging due to interobserver variability and morphologic heterogeneity. The genomic scoring system BRS (BRAF-RAS score) was developed to quantify the extent to which a tumor's expression profile resembles a BRAFV600E or RAS-mutant neoplasm. We proposed that deep learning prediction of BRS could differentiate NIFTP from other follicular-patterned neoplasms. A deep learning model was trained on slides from a dataset of 115 thyroid neoplasms to predict tumor subtype (NIFTP, PTC-EFG, or classic PTC), and was used to generate predictions for 497 thyroid neoplasms within The Cancer Genome Atlas (TCGA). Within follicular-patterned neoplasms, tumors with positive BRS (RAS-like) were 8.5 times as likely to carry an NIFTP prediction than tumors with negative BRS (89.7% vs 10.5%, P < 0.0001). To test the hypothesis that BRS may serve as a surrogate for biological processes that determine tumor subtype, a separate model was trained on TCGA slides to predict BRS as a linear outcome. This model performed well in cross-validation on the training set (R2 = 0.67, dichotomized AUC = 0.94). In our internal cohort, NIFTPs were near universally predicted to have RAS-like BRS; as a sole discriminator of NIFTP status, predicted BRS performed with an AUC of 0.99 globally and 0.97 when restricted to follicular-patterned neoplasms. BRAFV600E-mutant PTC-EFG had BRAFV600E-like predicted BRS (mean -0.49), nonmutant PTC-EFG had more intermediate predicted BRS (mean -0.17), and NIFTP had RAS-like BRS (mean 0.35; P < 0.0001). In summary, histologic features associated with the BRAF-RAS gene expression spectrum are detectable by deep learning and can aid in distinguishing indolent NIFTP from PTCs.

Academics as leaders in the cancer artificial intelligence revolution.

The successful translation of artificial intelligence (AI) applications into clinical cancer care practice requires guidance by academic cancer researchers and providers who are well poised to step into leadership roles. In this commentary, the authors describe the landscape of the deep learning-based AI innovation boom in cancer research. For progress in applied AI research to continue, 4 essential components must be present: algorithms, data, computational resources, and domain-specific expertise. Each of these components is available to researchers and providers in academic settings; in particular, cancer care domain-specific expertise in academia is superb. Three common pitfalls for deep learning research also are detailed along with a discussion of how the academic oncology research environment is well suited to guard against these challenges. In this rapidly developing field, there are few established standards, and oncology researchers and providers must educate themselves about emerging AI technology to avoid common pitfalls and ensure responsible use.

Machine Learning-Guided Adjuvant Treatment of Head and Neck Cancer.

Importance: Postoperative chemoradiation is the standard of care for cancers with positive margins or extracapsular extension, but the benefit of chemotherapy is unclear for patients with other intermediate risk features. Objective: To evaluate whether machine learning models could identify patients with intermediate-risk head and neck squamous cell carcinoma who would benefit from chemoradiation. Design, Setting, and Participants: This cohort study included patients diagnosed with squamous cell carcinoma of the oral cavity, oropharynx, hypopharynx, or larynx from January 1, 2004, through December 31, 2016. Patients had resected disease and underwent adjuvant radiotherapy. Analysis was performed from October 1, 2019, through September 1, 2020. Patients were selected from the National Cancer Database, a hospital-based registry that captures data from more than 70% of newly diagnosed cancers in the United States. Three machine learning survival models were trained using 80% of the cohort, with the remaining 20% used to assess model performance. Exposures: Receipt of adjuvant chemoradiation or radiation alone. Main Outcomes and Measures: Patients who received treatment recommended by machine learning models were compared with those who did not. Overall survival for treatment according to model recommendations was the primary outcome. Secondary outcomes included frequency of recommendation for chemotherapy and chemotherapy benefit in patients recommended for chemoradiation vs radiation alone. Results: A total of 33 527 patients (24 189 [72%] men; 28 036 [84%] aged ≤70 years) met the inclusion criteria. Median follow-up in the validation data set was 43.2 (interquartile range, 19.8-65.5) months. DeepSurv, neural multitask logistic regression, and survival forest models recommended chemoradiation for 17 589 (52%), 15 917 (47%), and 14 912 patients (44%), respectively. Treatment according to model recommendations was associated with a survival benefit, with a hazard ratio of 0.79 (95% CI, 0.72-0.85; P < .001) for DeepSurv, 0.83 (95% CI, 0.77-0.90; P < .001) for neural multitask logistic regression, and 0.90 (95% CI, 0.83-0.98; P = .01) for random survival forest models. No survival benefit for chemotherapy was seen for patients recommended to receive radiotherapy alone. Conclusions and Relevance: These findings suggest that machine learning models may identify patients with intermediate risk who could benefit from chemoradiation. These models predicted that approximately half of such patients have no added benefit from chemotherapy.

A randomized phase 2 study of temsirolimus and cetuximab versus temsirolimus alone in recurrent/metastatic, cetuximab-resistant head and neck cancer: The MAESTRO study.

BACKGROUND: Patients with cetuximab-resistant, recurrent/metastatic head and neck squamous cell carcinoma (HNSCC) have poor outcomes. This study hypothesized that dual blockade of mammalian target of rapamycin and epidermal growth factor receptor (EGFR) would overcome cetuximab resistance on the basis of the role of phosphoinositide 3-kinase signaling in preclinical models of EGFR resistance. METHODS: In this multicenter, randomized clinical study, patients with recurrent/metastatic HNSCC with documented progression on cetuximab (in any line in the recurrent/metastatic setting) received 25 mg of temsirolimus weekly plus cetuximab at 400/250 mg/m2 weekly (TC) or single-agent temsirolimus (T). The primary outcome was progression-free survival (PFS) in the TC arm versus the T arm. Response rates, overall survival, and toxicity were secondary outcomes. RESULTS: Eighty patients were randomized to therapy with TC or T alone. There was no difference for the primary outcome of median PFS (TC arm, 3.5 months; T arm, 3.5 months). The response rate was 12.5% in the TC arm (5 responses, including 1 complete response [2.5%]) and 2.5% in the T arm (1 partial response; P = .10). Responses were clinically meaningful in the TC arm (range, 3.6-9.1 months) but not in the T-alone arm (1.9 months). Fatigue, electrolyte abnormalities, and leukopenia were the most common grade 3 or higher adverse events and occurred in less than 20% of patients in both arms. CONCLUSIONS: The study did not meet its primary endpoint of improvement in PFS. However, TC induced responses in cetuximab-refractory patients with good tolerability. The post hoc observation of activity in patients with acquired resistance (after prior benefit from cetuximab monotherapy) may warrant further investigation.

A randomized phase 2 network trial of tivantinib plus cetuximab versus cetuximab in patients with recurrent/metastatic head and neck squamous cell carcinoma.

BACKGROUND: MET signaling is a well described mechanism of resistance to anti-EGFR therapy, and MET overexpression is common in head and neck squamous cell carcinomas (HNSCCs). In the current trial, the authors compared the oral MET inhibitor tivantinib (ARQ197) in combination with cetuximab (the TC arm) versus a control arm that received cetuximab monotherapy (C) in patients with recurrent/metastatic HNSCC. METHODS: In total, 78 evaluable patients with cetuximab-naive, platinum-refractory HNSCC were enrolled, including 40 on the TC arm and 38 on the C arm (stratified by human papillomavirus [HPV] status). Patients received oral tivantinib 360 mg twice daily and intravenous cetuximab 500 mg/m2 once every 2 weeks. The primary outcome was the response rate (according to Response Evaluation Criteria in Solid Tumors, version 1.1), and secondary outcomes included progression-free and overall survival. After patients progressed on the C arm, tivantinib monotherapy was optional. RESULTS: The response rate was 7.5% in the TC arm (N = 3; 1 complete response) and 7.9% in the C arm (N = 3; not significantly different [NS]). The median progression-free survival in both arms was 4 months (NS), and the median overall survival was 8 months (NS). Both treatments were well tolerated, with a trend toward increased hematologic toxicities in the TC arm (12.5% had grade 3 leukopenia). The response rate in 31 HPV-positive/p16-positive patients was 0% in both arms, whereas the response rate in HPV-negative patients was 12.7% (12.5% in the TC arm and 13% in the C arm). Fifteen patients received tivantinib monotherapy, and no responses were observed. CONCLUSIONS: Combined tivantinib plus cetuximab does not significantly improve the response rate or survival compared with cetuximab alone but does increase toxicity in an unselected HNSCC population. Cetuximab responses appear to be limited to patients who have HPV-negative HNSCC. MET-aberration-focused trials for HNSCC and the use of higher potency, selective MET inhibitors remain of interest.