Radiomics is a rapidly growing area of research within radiology that involves the extraction and modeling of high-dimensional quantitative imaging features using machine learning/artificial intelligence (ML/AI) methods. In this review, we describe the published clinical evidence on the application of ML methods to improve the performance of ultrasound (US) in head and neck oncology. A systematic search of electronic databases (MEDLINE, PubMed, clinicaltrials.gov) was conducted according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. Of 15,080 initial articles identified, 34 studies were selected for in-depth analysis. Twenty-five out of 34 studies (74%) focused on the diagnostic application of US radiomics while 6 (18%) studies focused on response assessment and 3 (8%) studies utilized US radiomics for modeling normal tissue toxicity. Support vector machine (SVM) was the most commonly employed ML method (47%) followed by multivariate logistic regression (24%) and k-nearest neighbor analysis (21%). Only 11/34 (~32%) of the studies included an independent validation set. A majority of studies were retrospective in nature (76%) and based on single-center evaluation (85%) with variable numbers of patients (12–1609) and imaging datasets (32–1624). Despite these limitations, the application of ML methods resulted in improved diagnostic and prognostic performance of US highlighting the potential clinical utility of this approach.

Xerostomia is a common side effect of radiation therapy (RT) in patients with head and neck cancer. However, limited information is available on the temporal dynamics of parenchymal and vascular changes in salivary glands following RT. To address this gap in knowledge, we conducted experimental studies in mice employing ultrasound (US) with coregistered photoacoustic imaging (PAI) to noninvasively assess the early and late changes in salivary gland size, structure, vascularity, and oxygenation dynamics following RT. Multiparametric US-PAI of salivary glands was performed in immune-deficient and immune-competent mice before and after RT along with correlative sialometry and ex vivo histologic-immunohistochemical validation. US revealed reduction in gland volume and an early increase in vascular resistance postradiation. This was accompanied by a reduction in glandular oxygen consumption on PAI. Imaging data correlated strongly with salivary secretion and histologic evidence of acinar damage. The magnitude and kinetics of radiation response were impacted by host immune status, with immunodeficient mice showing early and more pronounced vascular injury and DNA damage response compared to immunocompetent animals. Our findings demonstrate the ability of noninvasive US-PAI to monitor dynamic changes in salivary gland hemodynamics following radiation and highlight the impact of the host immune status on salivary gland radiation injury.

Figure 1 Bar graph shows average gray value distributions of the various sub-sites in the oral cavity in 20 healthy subjects as reported by Izzetti et al. 2020. Errors bars represent the standard deviation (*p<0.05 -one-way analysis of variance)

Tumor hypoxia has long been recognized as a negative prognosticator in patients with head and neck squamous cell carcinomas (HNSCC). Recent studies have also demonstrated that hypoxia promotes an immunosuppressive tumor microenvironment. However, the potential of non-invasive imaging of tumor hypoxia as a biomarker of immunotherapeutic efficacy has not been previously studied in HNSCC. To address this gap in knowledge, in the present study, we employed photoacoustic imaging (PAI) with co-registered ultrasound (US) for spatiotemporal profiling of tumor hypoxia in orthotopic models of HNSCC. Orthotopic RP-MOC1 and SCCVII tumors were established in female albino C57Bl/6 and C3H/HeJ mice, respectively. Longitudinal measurements of PAI-based biomarkers of tumor vascularity (hemoglobin concentration; HbT) and hypoxia (oxygen saturation; sO2) were obtained during initial tumor growth and following treatment with anti-programmed death-1 antibody treatment (αPD1; 200 µg, 3 doses administered once every 3 days). PAI-based measures of HbT and sO2 were validated using immunohistochemical markers of proliferation (Ki67), vascularity (CD31), hypoxia (CAIX), CD8+ T cell levels and treatment outcomes (US-based tumor volumes). Prior to treatment, SCCVII tumors exhibited rapid growth kinetics on US that was associated with decreased HbT and sO2 levels on PAI . Consistent with these findings, immunohistochemical analysis of SCCVII tumors revealed higher numbers of Ki67+ cells, higher levels of CAIX and low CD8+ T cell levels. In comparison, RP-MOC1 tumors showed a slower tumor growth rate on US and higher levels of HbT and sO2 on PAI prior to treatment. Ex-vivo analysis of RP-MOC1 tumors showed lower CAIX staining and higher CD8+ T cells validating PAI observations. Treatment with αPD1 yielded distinct therapeutic response profiles between the two tumor models. While SCCVII tumors were consistently resistant to αPD1, RP-MOC1 tumors exhibited a differential response to immunotherapy with 47% of tumors showing regression (responders) and 53% of tumors displaying continued increase in growth (non-responders). Notably, αPD1 responsive tumors showed an early (24h post 1st dose) increase in HbT and sO2, likely reflective of increased vessel perfusion and alleviation of hypoxia mediated by CD8+ T cell accumulation and IFNγ production, as previously reported in colon cancer models. Collectively, our observations highlight the impact of hypoxia on the tumor immune microenvironment and demonstrate, for the first time, the potential of PAI with co-registered US as an early non-invasive biomarker of immunotherapeutic efficacy in head and neck cancer. Citation Format: Celia DeJohn, Aparajita Verma, Vui King Vincent-Chong, Mukund Seshadri. Non-invasive imaging of hypoxia as a biomarker of immunotherapeutic efficacy in head and neck cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 2388.