Multi-omic machine learning predictor of breast cancer therapy response.

Breast cancers are complex ecosystems of malignant cells and the tumour microenvironment1. The composition of these tumour ecosystems and interactions within them contribute to responses to cytotoxic therapy2. Efforts to build response predictors have not incorporated this knowledge. We collected clinical, digital pathology, genomic and transcriptomic profiles of pre-treatment biopsies of breast tumours from 168 patients treated with chemotherapy with or without HER2 (encoded by ERBB2)-targeted therapy before surgery. Pathology end points (complete response or residual disease) at surgery3 were then correlated with multi-omic features in these diagnostic biopsies. Here we show that response to treatment is modulated by the pre-treated tumour ecosystem, and its multi-omics landscape can be integrated in predictive models using machine learning. The degree of residual disease following therapy is monotonically associated with pre-therapy features, including tumour mutational and copy number landscapes, tumour proliferation, immune infiltration and T cell dysfunction and exclusion. Combining these features into a multi-omic machine learning model predicted a pathological complete response in an external validation cohort (75 patients) with an area under the curve of 0.87. In conclusion, response to therapy is determined by the baseline characteristics of the totality of the tumour ecosystem captured through data integration and machine learning. This approach could be used to develop predictors for other cancers.

Precision radiogenomics: fusion biopsies to target tumour habitats in vivo.

High-grade serous ovarian cancer lesions display a high degree of heterogeneity on CT scans. We have recently shown that regions with distinct imaging profiles can be accurately biopsied in vivo using a technique based on the fusion of CT and ultrasound scans.

Ultrasound-guided targeted biopsies of CT-based radiomic tumour habitats: technical development and initial experience in metastatic ovarian cancer.

PURPOSE: To develop a precision tissue sampling technique that uses computed tomography (CT)-based radiomic tumour habitats for ultrasound (US)-guided targeted biopsies that can be integrated in the clinical workflow of patients with high-grade serous ovarian cancer (HGSOC). METHODS: Six patients with suspected HGSOC scheduled for US-guided biopsy before starting neoadjuvant chemotherapy were included in this prospective study from September 2019 to February 2020. The tumour segmentation was performed manually on the pre-biopsy contrast-enhanced CT scan. Spatial radiomic maps were used to identify tumour areas with similar or distinct radiomic patterns, and tumour habitats were identified using the Gaussian mixture modelling. CT images with superimposed habitat maps were co-registered with US images by means of a landmark-based rigid registration method for US-guided targeted biopsies. The dice similarity coefficient (DSC) was used to assess the tumour-specific CT/US fusion accuracy. RESULTS: We successfully co-registered CT-based radiomic tumour habitats with US images in all patients. The median time between CT scan and biopsy was 21 days (range 7-30 days). The median DSC for tumour-specific CT/US fusion accuracy was 0.53 (range 0.79 to 0.37). The CT/US fusion accuracy was high for the larger pelvic tumours (DSC: 0.76-0.79) while it was lower for the smaller omental metastases (DSC: 0.37-0.53). CONCLUSION: We developed a precision tissue sampling technique that uses radiomic habitats to guide in vivo biopsies using CT/US fusion and that can be seamlessly integrated in the clinical routine for patients with HGSOC. KEY POINTS: • We developed a prevision tissue sampling technique that co-registers CT-based radiomics-based tumour habitats with US images. • The CT/US fusion accuracy was high for the larger pelvic tumours (DSC: 0.76-0.79) while it was lower for the smaller omental metastases (DSC: 0.37-0.53).

Radiology and multi-scale data integration for precision oncology.

In this Perspective paper we explore the potential of integrating radiological imaging with other data types, a critical yet underdeveloped area in comparison to the fusion of other multi-omic data. Radiological images provide a comprehensive, three-dimensional view of cancer, capturing features that would be missed by biopsies or other data modalities. This paper explores the complexities and challenges of incorporating medical imaging into data integration models, in the context of precision oncology. We present the different categories of imaging-omics integration and discuss recent progress, highlighting the opportunities that arise from bringing together spatial data on different scales.

Calibrating ensembles for scalable uncertainty quantification in deep learning-based medical image segmentation.

Uncertainty quantification in automated image analysis is highly desired in many applications. Typically, machine learning models in classification or segmentation are only developed to provide binary answers; however, quantifying the uncertainty of the models can play a critical role for example in active learning or machine human interaction. Uncertainty quantification is especially difficult when using deep learning-based models, which are the state-of-the-art in many imaging applications. The current uncertainty quantification approaches do not scale well in high-dimensional real-world problems. Scalable solutions often rely on classical techniques, such as dropout, during inference or training ensembles of identical models with different random seeds to obtain a posterior distribution. In this paper, we present the following contributions. First, we show that the classical approaches fail to approximate the classification probability. Second, we propose a scalable and intuitive framework for uncertainty quantification in medical image segmentation that yields measurements that approximate the classification probability. Third, we suggest the usage of k-fold cross-validation to overcome the need for held out calibration data. Lastly, we motivate the adoption of our method in active learning, creating pseudo-labels to learn from unlabeled images and human-machine collaboration.

Artificial intelligence for early detection of renal cancer in computed tomography: A review.

Renal cancer is responsible for over 100,000 yearly deaths and is principally discovered in computed tomography (CT) scans of the abdomen. CT screening would likely increase the rate of early renal cancer detection, and improve general survival rates, but it is expected to have a prohibitively high financial cost. Given recent advances in artificial intelligence (AI), it may be possible to reduce the cost of CT analysis and enable CT screening by automating the radiological tasks that constitute the early renal cancer detection pipeline. This review seeks to facilitate further interdisciplinary research in early renal cancer detection by summarising our current knowledge across AI, radiology, and oncology and suggesting useful directions for future novel work. Initially, this review discusses existing approaches in automated renal cancer diagnosis, and methods across broader AI research, to summarise the existing state of AI cancer analysis. Then, this review matches these methods to the unique constraints of early renal cancer detection and proposes promising directions for future research that may enable AI-based early renal cancer detection via CT screening. The primary targets of this review are clinicians with an interest in AI and data scientists with an interest in the early detection of cancer.

Integrating Artificial Intelligence Tools in the Clinical Research Setting: The Ovarian Cancer Use Case.

Artificial intelligence (AI) methods applied to healthcare problems have shown enormous potential to alleviate the burden of health services worldwide and to improve the accuracy and reproducibility of predictions. In particular, developments in computer vision are creating a paradigm shift in the analysis of radiological images, where AI tools are already capable of automatically detecting and precisely delineating tumours. However, such tools are generally developed in technical departments that continue to be siloed from where the real benefit would be achieved with their usage. Significant effort still needs to be made to make these advancements available, first in academic clinical research and ultimately in the clinical setting. In this paper, we demonstrate a prototype pipeline based entirely on open-source software and free of cost to bridge this gap, simplifying the integration of tools and models developed within the AI community into the clinical research setting, ensuring an accessible platform with visualisation applications that allow end-users such as radiologists to view and interact with the outcome of these AI tools.

Lesion-specific 3D-printed moulds for image-guided tissue multi-sampling of ovarian tumours: A prospective pilot study.

Background: High-Grade Serous Ovarian Carcinoma (HGSOC) is the most prevalent and lethal subtype of ovarian cancer, but has a paucity of clinically-actionable biomarkers due to high degrees of multi-level heterogeneity. Radiogenomics markers have the potential to improve prediction of patient outcome and treatment response, but require accurate multimodal spatial registration between radiological imaging and histopathological tissue samples. Previously published co-registration work has not taken into account the anatomical, biological and clinical diversity of ovarian tumours. Methods: In this work, we developed a research pathway and an automated computational pipeline to produce lesion-specific three-dimensional (3D) printed moulds based on preoperative cross-sectional CT or MRI of pelvic lesions. Moulds were designed to allow tumour slicing in the anatomical axial plane to facilitate detailed spatial correlation of imaging and tissue-derived data. Code and design adaptations were made following each pilot case through an iterative refinement process. Results: Five patients with confirmed or suspected HGSOC who underwent debulking surgery between April and December 2021 were included in this prospective study. Tumour moulds were designed and 3D-printed for seven pelvic lesions, covering a range of tumour volumes (7 to 133 cm3) and compositions (cystic and solid proportions). The pilot cases informed innovations to improve specimen and subsequent slice orientation, through the use of 3D-printed tumour replicas and incorporation of a slice orientation slit in the mould design, respectively. The overall research pathway was compatible with implementation within the clinically determined timeframe and treatment pathway for each case, involving multidisciplinary clinical professionals from Radiology, Surgery, Oncology and Histopathology Departments. Conclusions: We developed and refined a computational pipeline that can model lesion-specific 3D-printed moulds from preoperative imaging for a variety of pelvic tumours. This framework can be used to guide comprehensive multi-sampling of tumour resection specimens.

Integrated radiogenomics models predict response to neoadjuvant chemotherapy in high grade serous ovarian cancer.

High grade serous ovarian carcinoma (HGSOC) is a highly heterogeneous disease that typically presents at an advanced, metastatic state. The multi-scale complexity of HGSOC is a major obstacle to predicting response to neoadjuvant chemotherapy (NACT) and understanding critical determinants of response. Here we present a framework to predict the response of HGSOC patients to NACT integrating baseline clinical, blood-based, and radiomic biomarkers extracted from all primary and metastatic lesions. We use an ensemble machine learning model trained to predict the change in total disease volume using data obtained at diagnosis (n = 72). The model is validated in an internal hold-out cohort (n = 20) and an independent external patient cohort (n = 42). In the external cohort the integrated radiomics model reduces the prediction error by 8% with respect to the clinical model, achieving an AUC of 0.78 for RECIST 1.1 classification compared to 0.47 for the clinical model. Our results emphasize the value of including radiomics data in integrative models of treatment response and provide methods for developing new biomarker-based clinical trials of NACT in HGSOC.

Deep learning-based segmentation of multisite disease in ovarian cancer.

PURPOSE: To determine if pelvic/ovarian and omental lesions of ovarian cancer can be reliably segmented on computed tomography (CT) using fully automated deep learning-based methods. METHODS: A deep learning model for the two most common disease sites of high-grade serous ovarian cancer lesions (pelvis/ovaries and omentum) was developed and compared against the well-established "no-new-Net" framework and unrevised trainee radiologist segmentations. A total of 451 CT scans collected from four different institutions were used for training (n = 276), evaluation (n = 104) and testing (n = 71) of the methods. The performance was evaluated using the Dice similarity coefficient (DSC) and compared using a Wilcoxon test. RESULTS: Our model outperformed no-new-Net for the pelvic/ovarian lesions in cross-validation, on the evaluation and test set by a significant margin (p values being 4 × 10-7, 3 × 10-4, 4 × 10-2, respectively), and for the omental lesions on the evaluation set (p = 1 × 10-3). Our model did not perform significantly differently in segmenting pelvic/ovarian lesions (p = 0.371) compared to a trainee radiologist. On an independent test set, the model achieved a DSC performance of 71 ± 20 (mean ± standard deviation) for pelvic/ovarian and 61 ± 24 for omental lesions. CONCLUSION: Automated ovarian cancer segmentation on CT scans using deep neural networks is feasible and achieves performance close to a trainee-level radiologist for pelvic/ovarian lesions. RELEVANCE STATEMENT: Automated segmentation of ovarian cancer may be used by clinicians for CT-based volumetric assessments and researchers for building complex analysis pipelines. KEY POINTS: • The first automated approach for pelvic/ovarian and omental ovarian cancer lesion segmentation on CT images has been presented. • Automated segmentation of ovarian cancer lesions can be comparable with manual segmentation of trainee radiologists. • Careful hyperparameter tuning can provide models significantly outperforming strong state-of-the-art baselines.

Clinically Interpretable Radiomics-Based Prediction of Histopathologic Response to Neoadjuvant Chemotherapy in High-Grade Serous Ovarian Carcinoma.

Background: Pathological response to neoadjuvant treatment for patients with high-grade serous ovarian carcinoma (HGSOC) is assessed using the chemotherapy response score (CRS) for omental tumor deposits. The main limitation of CRS is that it requires surgical sampling after initial neoadjuvant chemotherapy (NACT) treatment. Earlier and non-invasive response predictors could improve patient stratification. We developed computed tomography (CT) radiomic measures to predict neoadjuvant response before NACT using CRS as a gold standard. Methods: Omental CT-based radiomics models, yielding a simplified fully interpretable radiomic signature, were developed using Elastic Net logistic regression and compared to predictions based on omental tumor volume alone. Models were developed on a single institution cohort of neoadjuvant-treated HGSOC (n = 61; 41% complete response to NCT) and tested on an external test cohort (n = 48; 21% complete response). Results: The performance of the comprehensive radiomics models and the fully interpretable radiomics model was significantly higher than volume-based predictions of response in both the discovery and external test sets when assessed using G-mean (geometric mean of sensitivity and specificity) and NPV, indicating high generalizability and reliability in identifying non-responders when using radiomics. The performance of a fully interpretable model was similar to that of comprehensive radiomics models. Conclusions: CT-based radiomics allows for predicting response to NACT in a timely manner and without the need for abdominal surgery. Adding pre-NACT radiomics to volumetry improved model performance for predictions of response to NACT in HGSOC and was robust to external testing. A radiomic signature based on five robust predictive features provides improved clinical interpretability and may thus facilitate clinical acceptance and application.

Hyperpolarized 13C-Pyruvate Metabolism as a Surrogate for Tumor Grade and Poor Outcome in Renal Cell Carcinoma-A Proof of Principle Study.

Differentiating aggressive clear cell renal cell carcinoma (ccRCC) from indolent lesions is challenging using conventional imaging. This work prospectively compared the metabolic imaging phenotype of renal tumors using carbon-13 MRI following injection of hyperpolarized [1-13C]pyruvate (HP-13C-MRI) and validated these findings with histopathology. Nine patients with treatment-naïve renal tumors (6 ccRCCs, 1 liposarcoma, 1 pheochromocytoma, 1 oncocytoma) underwent pre-operative HP-13C-MRI and conventional proton (1H) MRI. Multi-regional tissue samples were collected using patient-specific 3D-printed tumor molds for spatial registration between imaging and molecular analysis. The apparent exchange rate constant (kPL) between 13C-pyruvate and 13C-lactate was calculated. Immunohistochemistry for the pyruvate transporter (MCT1) from 44 multi-regional samples, as well as associations between MCT1 expression and outcome in the TCGA-KIRC dataset, were investigated. Increasing kPL in ccRCC was correlated with increasing overall tumor grade (ρ = 0.92, p = 0.009) and MCT1 expression (r = 0.89, p = 0.016), with similar results acquired from the multi-regional analysis. Conventional 1H-MRI parameters did not discriminate tumor grades. The correlation between MCT1 and ccRCC grade was confirmed within a TCGA dataset (p < 0.001), where MCT1 expression was a predictor of overall and disease-free survival. In conclusion, metabolic imaging using HP-13C-MRI differentiates tumor aggressiveness in ccRCC and correlates with the expression of MCT1, a predictor of survival. HP-13C-MRI may non-invasively characterize metabolic phenotypes within renal cancer.

Triage-driven diagnosis of Barrett's esophagus for early detection of esophageal adenocarcinoma using deep learning.

Deep learning methods have been shown to achieve excellent performance on diagnostic tasks, but how to optimally combine them with expert knowledge and existing clinical decision pathways is still an open challenge. This question is particularly important for the early detection of cancer, where high-volume workflows may benefit from (semi-)automated analysis. Here we present a deep learning framework to analyze samples of the Cytosponge-TFF3 test, a minimally invasive alternative to endoscopy, for detecting Barrett's esophagus, which is the main precursor of esophageal adenocarcinoma. We trained and independently validated the framework on data from two clinical trials, analyzing a combined total of 4,662 pathology slides from 2,331 patients. Our approach exploits decision patterns of gastrointestinal pathologists to define eight triage classes of varying priority for manual expert review. By substituting manual review with automated review in low-priority classes, we can reduce pathologist workload by 57% while matching the diagnostic performance of experienced pathologists.

Investigating the relationship between diffusion kurtosis tensor imaging (DKTI) and histology within the normal human brain.

Measurements of water diffusion with MRI have been used as a biomarker of tissue microstructure and heterogeneity. In this study, diffusion kurtosis tensor imaging (DKTI) of the brain was undertaken in 10 healthy volunteers at a clinical field strength of 3 T. Diffusion and kurtosis metrics were measured in regions-of-interest on the resulting maps and compared with quantitative analysis of normal post-mortem tissue histology from separate age-matched donors. White matter regions showed low diffusion (0.60 ± 0.04 × 10-3 mm2/s) and high kurtosis (1.17 ± 0.06), consistent with a structured heterogeneous environment comprising parallel neuronal fibres. Grey matter showed intermediate diffusion (0.80 ± 0.02 × 10-3 mm2/s) and kurtosis (0.82 ± 0.05) values. An important finding is that the subcortical regions investigated (thalamus, caudate and putamen) showed similar diffusion and kurtosis properties to white matter. Histological staining of the subcortical nuclei demonstrated that the predominant grey matter was permeated by small white matter bundles, which could account for the similar kurtosis to white matter. Quantitative histological analysis demonstrated higher mean tissue kurtosis and vector standard deviation values for white matter (1.08 and 0.81) compared to the subcortical regions (0.34 and 0.59). Mean diffusion on DKTI was positively correlated with tissue kurtosis (r = 0.82, p < 0.05) and negatively correlated with vector standard deviation (r = -0.69, p < 0.05). This study demonstrates how DKTI can be used to study regional structural variations in the cerebral tissue microenvironment and could be used to probe microstructural changes within diseased tissue in the future.

Assessing Histology Structures by Ex Vivo MR Microscopy and Exploring the Link Between MRM-Derived Radiomic Features and Histopathology in Ovarian Cancer.

The value of MR radiomic features at a microscopic scale has not been explored in ovarian cancer. The objective of this study was to probe the associations of MR microscopy (MRM) images and MRM-derived radiomic maps with histopathology in high-grade serous ovarian cancer (HGSOC). Nine peritoneal implants from 9 patients with HGSOC were imaged ex vivo with MRM using a 9.4-T MR scanner. All MRM images and computed pixel-wise radiomics maps were correlated with the slice-matched stroma and tumor proportion maps derived from whole histopathologic slide images (WHSI) of corresponding peritoneal implants. Automated MRM-derived segmentation maps of tumor and stroma were constructed using holdout test data and validated against the histopathologic gold standard. Excellent correlation between MRM images and WHSI was observed (Dice index = 0.77). Entropy, correlation, difference entropy, and sum entropy radiomic features were positively associated with high stromal proportion (r = 0.97,0.88, 0.81, and 0.96 respectively, p < 0.05). MR signal intensity, energy, homogeneity, auto correlation, difference variance, and sum average were negatively associated with low stromal proportion (r = -0.91, -0.93, -0.94, -0.9, -0.89, -0.89, respectively, p < 0.05). Using the automated model, MRM predicted stromal proportion with an accuracy ranging from 61.4% to 71.9%. In this hypothesis-generating study, we showed that it is feasible to resolve histologic structures in HGSOC using ex vivo MRM at 9.4 T and radiomics.

Tissue-specific and interpretable sub-segmentation of whole tumour burden on CT images by unsupervised fuzzy clustering.

BACKGROUND: Cancer typically exhibits genotypic and phenotypic heterogeneity, which can have prognostic significance and influence therapy response. Computed Tomography (CT)-based radiomic approaches calculate quantitative features of tumour heterogeneity at a mesoscopic level, regardless of macroscopic areas of hypo-dense (i.e., cystic/necrotic), hyper-dense (i.e., calcified), or intermediately dense (i.e., soft tissue) portions. METHOD: With the goal of achieving the automated sub-segmentation of these three tissue types, we present here a two-stage computational framework based on unsupervised Fuzzy C-Means Clustering (FCM) techniques. No existing approach has specifically addressed this task so far. Our tissue-specific image sub-segmentation was tested on ovarian cancer (pelvic/ovarian and omental disease) and renal cell carcinoma CT datasets using both overlap-based and distance-based metrics for evaluation. RESULTS: On all tested sub-segmentation tasks, our two-stage segmentation approach outperformed conventional segmentation techniques: fixed multi-thresholding, the Otsu method, and automatic cluster number selection heuristics for the K-means clustering algorithm. In addition, experiments showed that the integration of the spatial information into the FCM algorithm generally achieves more accurate segmentation results, whilst the kernelised FCM versions are not beneficial. The best spatial FCM configuration achieved average Dice similarity coefficient values starting from 81.94±4.76 and 83.43±3.81 for hyper-dense and hypo-dense components, respectively, for the investigated sub-segmentation tasks. CONCLUSIONS: The proposed intelligent framework could be readily integrated into clinical research environments and provides robust tools for future radiomic biomarker validation.

Integrative radiogenomics for virtual biopsy and treatment monitoring in ovarian cancer.

BACKGROUND: Ovarian cancer survival rates have not changed in the last 20 years. The majority of cases are High-grade serous ovarian carcinomas (HGSOCs), which are typically diagnosed at an advanced stage with multiple metastatic lesions. Taking biopsies of all sites of disease is infeasible, which challenges the implementation of stratification tools based on molecular profiling. MAIN BODY: In this review, we describe how these challenges might be overcome by integrating quantitative features extracted from medical imaging with the analysis of paired genomic profiles, a combined approach called radiogenomics, to generate virtual biopsies. Radiomic studies have been used to model different imaging phenotypes, and some radiomic signatures have been associated with paired molecular profiles to monitor spatiotemporal changes in the heterogeneity of tumours. We describe different strategies to integrate radiogenomic information in a global and local manner, the latter by targeted sampling of tumour habitats, defined as regions with distinct radiomic phenotypes. CONCLUSION: Linking radiomics and biological correlates in a targeted manner could potentially improve the clinical management of ovarian cancer. Radiogenomic signatures could be used to monitor tumours during the course of therapy, offering additional information for clinical decision making. In summary, radiogenomics may pave the way to virtual biopsies and treatment monitoring tools for integrative tumour analysis.

Integrated Multi-Tumor Radio-Genomic Marker of Outcomes in Patients with High Serous Ovarian Carcinoma.

Purpose: Develop an integrated intra-site and inter-site radiomics-clinical-genomic marker of high grade serous ovarian cancer (HGSOC) outcomes and explore the biological basis of radiomics with respect to molecular signaling pathways and the tumor microenvironment (TME). Method: Seventy-five stage III-IV HGSOC patients from internal (N = 40) and external factors via the Cancer Imaging Archive (TCGA) (N = 35) with pre-operative contrast enhanced CT, attempted primary cytoreduction, at least two disease sites, and molecular analysis performed within TCGA were retrospectively analyzed. An intra-site and inter-site radiomics (cluDiss) measure was combined with clinical-genomic variables (iRCG) and compared against conventional (volume and number of sites) and average radiomics (N = 75) for prognosticating progression-free survival (PFS) and platinum resistance. Correlation with molecular signaling and TME derived using a single sample gene set enrichment that was measured. Results: The iRCG model had the best platinum resistance classification accuracy (AUROC of 0.78 [95% CI 0.77 to 0.80]). CluDiss was associated with PFS (HR 1.03 [95% CI: 1.01 to 1.05], p = 0.002), negatively correlated with Wnt signaling, and positively to immune TME. Conclusions: CluDiss and the iRCG prognosticated HGSOC outcomes better than conventional and average radiomic measures and could better stratify patient outcomes if validated on larger multi-center trials.

Three-Dimensional Printed Molds for Image-Guided Surgical Biopsies: An Open Source Computational Platform.

PURPOSE: Spatial heterogeneity of tumors is a major challenge in precision oncology. The relationship between molecular and imaging heterogeneity is still poorly understood because it relies on the accurate coregistration of medical images and tissue biopsies. Tumor molds can guide the localization of biopsies, but their creation is time consuming, technologically challenging, and difficult to interface with routine clinical practice. These hurdles have so far hindered the progress in the area of multiscale integration of tumor heterogeneity data. METHODS: We have developed an open-source computational framework to automatically produce patient-specific 3-dimensional-printed molds that can be used in the clinical setting. Our approach achieves accurate coregistration of sampling location between tissue and imaging, and integrates seamlessly with clinical, imaging, and pathology workflows. RESULTS: We applied our framework to patients with renal cancer undergoing radical nephrectomy. We created personalized molds for 6 patients, obtaining Dice similarity coefficients between imaging and tissue sections ranging from 0.86 to 0.96 for tumor regions and between 0.70 and 0.76 for healthy kidneys. The framework required minimal manual intervention, producing the final mold design in just minutes, while automatically taking into account clinical considerations such as a preference for specific cutting planes. CONCLUSION: Our work provides a robust and automated interface between imaging and tissue samples, enabling the development of clinical studies to probe tumor heterogeneity on multiple spatial scales.

Non-invasive imaging prediction of tumor hypoxia: A novel developed and externally validated CT and FDG-PET-based radiomic signatures.

BACKGROUND: Tumor hypoxia increases resistance to radiotherapy and systemic therapy. Our aim was to develop and validate a disease-agnostic and disease-specific CT (+FDG-PET) based radiomics hypoxia classification signature. MATERIAL AND METHODS: A total of 808 patients with imaging data were included: N = 100 training/N = 183 external validation cases for a disease-agnostic CT hypoxia classification signature, N = 76 training/N = 39 validation cases for the H&N CT signature and N = 62 training/N = 36 validation cases for the Lung CT signature. The primary gross tumor volumes (GTV) were manually defined by experts on CT. In order to dichotomize between hypoxic/well-oxygenated tumors a threshold of 20% was used for the [18F]-HX4-derived hypoxic fractions (HF). A random forest (RF)-based machine-learning classifier/regressor was trained to classify patients as hypoxia-positive/ negative based on radiomic features. RESULTS: A 11 feature "disease-agnostic CT model" reached AUC's of respectively 0.78 (95% confidence interval [CI], 0.62-0.94), 0.82 (95% CI, 0.67-0.96) and 0.78 (95% CI, 0.67-0.89) in three external validation datasets. A "disease-agnostic FDG-PET model" reached an AUC of 0.73 (0.95% CI, 0.49-0.97) in validation by combining 5 features. The highest "lung-specific CT model" reached an AUC of 0.80 (0.95% CI, 0.65-0.95) in validation with 4 CT features, while the "H&N-specific CT model" reached an AUC of 0.84 (0.95% CI, 0.64-1.00) in validation with 15 CT features. A tumor volume-alone model was unable to significantly classify patients as hypoxia-positive/ negative. A significant survival split (P = 0.037) was found between CT-classified hypoxia strata in an external H&N cohort (n = 517), while 117 significant hypoxia gene-CT signature feature associations were found in an external lung cohort (n = 80). CONCLUSION: The disease-specific radiomics signatures perform better than the disease agnostic ones. By identifying hypoxic patients our signatures have the potential to enrich interventional hypoxia-targeting trials.