RADCURE: An open-source head and neck cancer CT dataset for clinical radiation therapy insights.

PURPOSE: This manuscript presents RADCURE, one of the most extensive head and neck cancer (HNC) imaging datasets accessible to the public. Initially collected for clinical radiation therapy (RT) treatment planning, this dataset has been retrospectively reconstructed for use in imaging research. ACQUISITION AND VALIDATION METHODS: RADCURE encompasses data from 3346 patients, featuring computed tomography (CT) RT simulation images with corresponding target and organ-at-risk contours. These CT scans were collected using systems from three different manufacturers. Standard clinical imaging protocols were followed, and contours were manually generated and reviewed at weekly RT quality assurance rounds. RADCURE imaging and structure set data was extracted from our institution's radiation treatment planning and oncology information systems using a custom-built data mining and processing system. Furthermore, images were linked to our clinical anthology of outcomes data for each patient and includes demographic, clinical and treatment information based on the 7th edition TNM staging system (Tumor-Node-Metastasis Classification System of Malignant Tumors). The median patient age is 63, with the final dataset including 80% males. Half of the cohort is diagnosed with oropharyngeal cancer, while laryngeal, nasopharyngeal, and hypopharyngeal cancers account for 25%, 12%, and 5% of cases, respectively. The median duration of follow-up is five years, with 60% of the cohort surviving until the last follow-up point. DATA FORMAT AND USAGE NOTES: The dataset provides images and contours in DICOM CT and RT-STRUCT formats, respectively. We have standardized the nomenclature for individual contours-such as the gross primary tumor, gross nodal volumes, and 19 organs-at-risk-to enhance the RT-STRUCT files' utility. Accompanying demographic, clinical, and treatment data are supplied in a comma-separated values (CSV) file format. This comprehensive dataset is publicly accessible via The Cancer Imaging Archive. POTENTIAL APPLICATIONS: RADCURE's amalgamation of imaging, clinical, demographic, and treatment data renders it an invaluable resource for a broad spectrum of radiomics image analysis research endeavors. Researchers can utilize this dataset to advance routine clinical procedures using machine learning or artificial intelligence, to identify new non-invasive biomarkers, or to forge prognostic models.

Development and Validation of a Genomic Profile for the Omission of Local Adjuvant Radiation in Breast Cancer.

PURPOSE: Adjuvant radiotherapy (RT) is used for women with early-stage invasive breast cancer treated with breast-conserving surgery. However, some women with low risk of recurrence may safely be spared RT. This study aimed to identify these women using a molecular-based approach. METHODS: We analyzed two randomized trials of women with node-negative invasive breast cancer to ± RT following breast-conserving surgery: SweBCG91-RT (stage I-II, no adjuvant systemic therapy) and Princess Margaret (age 50 years or older, T1-T2, adjuvant tamoxifen). Transcriptome-wide profiling was performed (Affymetrix Human Exon 1.0 ST microarray). Patients with estrogen receptor-positive/human epidermal growth factor receptor 2-negative tumors and with gene expression data were included. The SweBCG91-RT cohort was divided into training (N = 243) and validation (N = 354) cohorts. A 16-gene signature named Profile for the Omission of Local Adjuvant Radiation (POLAR) was trained to predict locoregional recurrence (LRR) using elastic net regression. POLAR was then validated in the SweBCG91-RT validation cohort and the Princess Margaret cohort (N = 132). RESULTS: Patients categorized as POLAR low-risk without RT had a 10-year LRR of 6% (95% CI, 2 to 16) and 7% (0 to 27) in SweBCG91-RT and Princess Margaret cohorts, respectively. There was no significant benefit from RT in POLAR low-risk patients (hazard ratio [HR], 1.1 [0.39 to 3.4], P = .81, and HR, 1.5 [0.14 to 16], P = .74, respectively). Patients categorized as POLAR high-risk had a significant decreased risk of LRR with RT (HR, 0.43 [0.24 to 0.78], P = .0055, and HR, 0.25 [0.07 to 0.92], P = .038, respectively). An exploratory analysis testing for interaction between RT and POLAR in the combined validation cohort was performed (P = .066). CONCLUSION: The novel POLAR genomic signature on the basis of LRR biology may identify patients with a low risk of LRR despite not receiving RT, and thus may be candidates for RT omission.

Multi-institutional Prognostic Modeling in Head and Neck Cancer: Evaluating Impact and Generalizability of Deep Learning and Radiomics.

Artificial intelligence (AI) and machine learning (ML) are becoming critical in developing and deploying personalized medicine and targeted clinical trials. Recent advances in ML have enabled the integration of wider ranges of data including both medical records and imaging (radiomics). However, the development of prognostic models is complex as no modeling strategy is universally superior to others and validation of developed models requires large and diverse datasets to demonstrate that prognostic models developed (regardless of method) from one dataset are applicable to other datasets both internally and externally. Using a retrospective dataset of 2,552 patients from a single institution and a strict evaluation framework that included external validation on three external patient cohorts (873 patients), we crowdsourced the development of ML models to predict overall survival in head and neck cancer (HNC) using electronic medical records (EMR) and pretreatment radiological images. To assess the relative contributions of radiomics in predicting HNC prognosis, we compared 12 different models using imaging and/or EMR data. The model with the highest accuracy used multitask learning on clinical data and tumor volume, achieving high prognostic accuracy for 2-year and lifetime survival prediction, outperforming models relying on clinical data only, engineered radiomics, or complex deep neural network architecture. However, when we attempted to extend the best performing models from this large training dataset to other institutions, we observed significant reductions in the performance of the model in those datasets, highlighting the importance of detailed population-based reporting for AI/ML model utility and stronger validation frameworks. We have developed highly prognostic models for overall survival in HNC using EMRs and pretreatment radiological images based on a large, retrospective dataset of 2,552 patients from our institution.Diverse ML approaches were used by independent investigators. The model with the highest accuracy used multitask learning on clinical data and tumor volume.External validation of the top three performing models on three datasets (873 patients) with significant differences in the distributions of clinical and demographic variables demonstrated significant decreases in model performance. Significance: ML combined with simple prognostic factors outperformed multiple advanced CT radiomics and deep learning methods. ML models provided diverse solutions for prognosis of patients with HNC but their prognostic value is affected by differences in patient populations and require extensive validation.

Antitumor impact of interferon-γ producing CD1d-restricted NKT cells in murine malignant mesothelioma.

CD1d-restricted natural killer T (iNKT) cells have been shown to provide adjuvant activity against cancer by producing interferon (IFN)-γ. However, the role of invariant NKT (iNKT) cells in the tumor microenvironment has not yet been fully addressed. Our aim is to elucidate the antitumor effect of iNKT cells in the tumor microenvironment by using an intrathoracic murine malignant pleural mesothelioma model that we had previously developed and to provide pleural effusion as a good surrogate of the tumor microenvironment. We found that the number of iNKT cells increased dramatically in the pleural effusion after intrathoracic tumor cell injection at an earlier phase compared with accumulation of CD8 T cells. These iNKT cells showed increased expression of CD25 and increased ratio of cells positive for IFN-γ intracellular staining. iNKT cells sorted from pleural effusion of tumor burden mice produced larger amount of IFN-γ compared with naive mice. Mice pretreated in vivo with anti-CD1d-blocking Ab showed increased amount of pleural effusion and decreased ratio of total and effector-type CD8 T cells as well as decreased intracellular IFN-γ expression of CD8T-cell in the pleural effusion. In vivo administration of α-galactosylceramide (α-GalCer) showed prolonged survival associated with less pleural effusion and increased ratio of IFN-γ-positive iNKT cells and CD8 T cells in the pleural effusion. Therefore, this study suggests that iNKT cells accumulating in the tumor microenvironment play an antitumor effect by producing IFN-γ and enhancing subsequent CD8 T-cell response. Furthermore, in vivo administration of α-GalCer could suppress mesothelioma growth by activating iNKT cells.

CTLA-4 blockade expands infiltrating T cells and inhibits cancer cell repopulation during the intervals of chemotherapy in murine mesothelioma.

Cancer immunotherapy has shown promising results when combined with chemotherapy. Blocking CTLA-4 signaling by monoclonal antibody between cycles of chemotherapy may inhibit cancer cell repopulation and enhance the antitumoral immune reaction, thus improve the efficacy of chemotherapy in mesothelioma. The impact of CTLA-4 blockade on the early stage of tumor development was evaluated in a subcutaneous murine mesothelioma model. CTLA-4 blocking antibody was administered following each cycle of chemotherapy, and monotherapy was included as controls. Antitumor effect was evaluated by tumor growth delay and survival of the animals. Tumor cell repopulation was quantified by bromodeoxyuridine incorporation and Ki67 by immunohistochemistry and/or flow cytometry. In vitro cell killing was determined by classic chromium-released assay, and reverse transcription PCR (RT-PCR) was carried out to determine the gene expression of associated cytokines. Anti-CTLA-4 monoclonal antibody was able to inhibit tumor growth at early stage of tumor development. Antitumor effect was achieved by administration of CTLA-4 blockade between cycles of chemotherapy. Tumor cell repopulation during the intervals of cisplatin was inhibited by CTLA-4 blockade. Anti-CTLA-4 therapy gave rise to an increased number of CD4 and CD8 T cells infiltrating the tumor. RT-PCR showed that the gene expression of interleukin IL-2, IFN-γ, granzyme B, and perforin increased in the tumor milieu. Blockade of CTLA-4 signaling showed effective anticancer effect, correlating with inhibiting cancer cell repopulation between cycles of chemotherapy and upregulating tumor-infiltrating T lymphocytes, cytokines, and cytolytic enzymes in a murine mesothelioma model.

Tumor cell repopulation between cycles of chemotherapy is inhibited by regulatory T-cell depletion in a murine mesothelioma model.

INTRODUCTION: Malignant pleural mesothelioma is a highly aggressive cancer with poor prognosis. We have previously demonstrated that regulatory T cells (Treg) depletion can impact tumor microenvironment when combined with chemotherapy. The aim of this study is to analyze the impact of Treg depletion on tumor cell repopulation during cycles of chemotherapy in a murine mesothelioma model. METHODS: Tumor-bearing mice were treated with chemotherapy once weekly to mimic clinical settings and with PC61 to cause Treg depletion after each cycle of chemotherapy. Tumor cell repopulation was evaluated by BrdU labeling index with immunohistochemistry and flow cytometry, and Ki67 gene expression was determined by real-time reverse-transcribed polymerase chain reaction. The proportion of CD4+ CD25+Foxp3+ Tregs, CD4+, and CD8+ T cells in the tumor, spleen, draining lymph node, and peripheral blood from tumor-bearing mice was determined by using flow cytometry, and gene expression of activated T-cell-related cytokines was quantified by enzyme-linked immunosorbent assay and reverse-transcribed polymerase chain reaction. RESULTS: Tumor growth delay was achieved by cisplatin followed by PC61 or cyclophosphamide. The BrdU labeling index indicated that tumor cell repopulation between cycles of cisplatin was significantly inhibited by PC61. The CD4+CD25+Foxp3+ Tregs in tumor and lymphoid organs were almost completely depleted, whereas the CD4+ or CD8+ T cells did not change. PC61 after chemotherapy resulted in an increase of gene expression of interferon-γ, granzyme B, perforin, and IP-10, thus leading to tumor cell lysis in cytotoxic lymphocyte assay. Nevertheless, cell killing induced by cyclophosphamide combined with cisplatin was due to cytotoxicity rather than specific immune response. CONCLUSION: Treg depletion between cycles of chemotherapy could improve the outcome of mesothelioma. Nevertheless, this effect seems limited, and more effective approaches need to be developed.