Longitudinal Assessment of Tumor-Infiltrating Lymphocytes in Primary Breast Cancer Following Neoadjuvant Radiation Therapy.

PURPOSE: Tumor-infiltrating lymphocytes (TILs) have prognostic significance in several cancers, including breast cancer. Despite interest in combining radiation therapy with immunotherapy, little is known about the effect of radiation therapy itself on the tumor-immune microenvironment, including TILs. Here, we interrogated longitudinal dynamics of TILs and systemic lymphocytes in patient samples taken before, during, and after neoadjuvant radiation therapy (NART) from PRADA and Neo-RT breast clinical trials. METHODS AND MATERIALS: We manually scored stromal TILs (sTILs) from longitudinal tumor samples using standardized guidelines as well as deep learning-based scores at cell-level (cTIL) and cell- and tissue-level combination analyses (SuperTIL). In parallel, we interrogated absolute lymphocyte counts from routine blood tests at corresponding time points during treatment. Exploratory analyses studied the relationship between TILs and pathologic complete response (pCR) and long-term outcomes. RESULTS: Patients receiving NART experienced a significant and uniform decrease in sTILs that did not recover at the time of surgery (P < .0001). This lymphodepletive effect was also mirrored in peripheral blood. Our SuperTIL deep learning score showed good concordance with manual sTILs and importantly performed comparably to manual scores in predicting pCR from diagnostic biopsies. The analysis suggested an association between baseline sTILs and pCR, as well as sTILs at surgery and relapse, in patients receiving NART. CONCLUSIONS: This study provides novel insights into TIL dynamics in the context of NART in breast cancer and demonstrates the potential for artificial intelligence to assist routine pathology. We have identified trends that warrant further interrogation and have a bearing on future radioimmunotherapy trials.

Auto-segmentation of pelvic organs at risk on 0.35T MRI using 2D and 3D Generative Adversarial Network models.

PURPOSE: Manual recontouring of targets and Organs At Risk (OARs) is a time-consuming and operator-dependent task. We explored the potential of Generative Adversarial Networks (GAN) to auto-segment the rectum, bladder and femoral heads on 0.35T MRIs to accelerate the online MRI-guided-Radiotherapy (MRIgRT) workflow. METHODS: 3D planning MRIs from 60 prostate cancer patients treated with 0.35T MR-Linac were collected. A 3D GAN architecture and its equivalent 2D version were trained, validated and tested on 40, 10 and 10 patients respectively. The volumetric Dice Similarity Coefficient (DSC) and 95th percentile Hausdorff Distance (HD95th) were computed against expert drawn ground-truth delineations. The networks were also validated on an independent external dataset of 16 patients. RESULTS: In the internal test set, the 3D and 2D GANs showed DSC/HD95th of 0.83/9.72 mm and 0.81/10.65 mm for the rectum, 0.92/5.91 mm and 0.85/15.72 mm for the bladder, and 0.94/3.62 mm and 0.90/9.49 mm for the femoral heads. In the external test set, the performance was 0.74/31.13 mm and 0.72/25.07 mm for the rectum, 0.92/9.46 mm and 0.88/11.28 mm for the bladder, and 0.89/7.00 mm and 0.88/10.06 mm for the femoral heads. The 3D and 2D GANs required on average 1.44 s and 6.59 s respectively to generate the OARs' volumetric segmentation for a single patient. CONCLUSIONS: The proposed 3D GAN auto-segments pelvic OARs with high accuracy on 0.35T, in both the internal and the external test sets, outperforming its 2D equivalent in both segmentation robustness and volume generation time.

Radiomics systematic review in cervical cancer: gynecological oncologists' perspective.

OBJECTIVE: Radiomics is the process of extracting quantitative features from radiological images, and represents a relatively new field in gynecological cancers. Cervical cancer has been the most studied gynecological tumor for what concerns radiomics analysis. The aim of this study was to report on the clinical applications of radiomics combined and/or compared with clinical-pathological variables in patients with cervical cancer. METHODS: A systematic review of the literature from inception to February 2023 was performed, including studies on cervical cancer analysing a predictive/prognostic radiomics model, which was combined and/or compared with a radiological or a clinical-pathological model. RESULTS: A total of 57 of 334 (17.1%) screened studies met inclusion criteria. The majority of studies used magnetic resonance imaging (MRI), but positron emission tomography (PET)/computed tomography (CT) scan, CT scan, and ultrasound scan also underwent radiomics analysis. In apparent early-stage disease, the majority of studies (16/27, 59.3%) analysed the role of radiomics signature in predicting lymph node metastasis; six (22.2%) investigated the prediction of radiomics to detect lymphovascular space involvement, one (3.7%) investigated depth of stromal infiltration, and one investigated (3.7%) parametrial infiltration. Survival prediction was evaluated both in early-stage and locally advanced settings. No study focused on the application of radiomics in metastatic or recurrent disease. CONCLUSION: Radiomics signatures were predictive of pathological and oncological outcomes, particularly if combined with clinical variables. These may be integrated in a model using different clinical-pathological and translational characteristics, with the aim to tailor and personalize the treatment of each patient with cervical cancer.

Radiomics and Radiogenomics of Ovarian Cancer: Implications for Treatment Monitoring and Clinical Management.

Ovarian cancer, one of the deadliest gynecologic malignancies, is characterized by high intra- and inter-site genomic and phenotypic heterogeneity. The traditional information provided by the conventional interpretation of diagnostic imaging studies cannot adequately represent this heterogeneity. Radiomics analyses can capture the complex patterns related to the microstructure of the tissues and provide quantitative information about them. This review outlines how radiomics and its integration with other quantitative biological information, like genomics and proteomics, can impact the clinical management of ovarian cancer.

Investigating the contribution of hyaluronan to the breast tumour microenvironment using multiparametric MRI and MR elastography.

Hyaluronan (HA) is a key component of the dense extracellular matrix in breast cancer, and its accumulation is associated with poor prognosis and metastasis. Pegvorhyaluronidase alfa (PEGPH20) enzymatically degrades HA and can enhance drug delivery and treatment response in preclinical tumour models. Clinical development of stromal-targeted therapies would be accelerated by imaging biomarkers that inform on therapeutic efficacy in vivo. Here, PEGPH20 response was assessed by multiparametric magnetic resonance imaging (MRI) in three orthotopic breast tumour models. Treatment of 4T1/HAS3 tumours, the model with the highest HA accumulation, reduced T1 and T2 relaxation times and the apparent diffusion coefficient (ADC), and increased the magnetisation transfer ratio, consistent with lower tissue water content and collapse of the extracellular space. The transverse relaxation rate R2 * increased, consistent with greater erythrocyte accessibility following vascular decompression. Treatment of MDA-MB-231 LM2-4 tumours reduced ADC and dramatically increased tumour viscoelasticity measured by MR elastography. Correlation matrix analyses of data from all models identified ADC as having the strongest correlation with HA accumulation, suggesting that ADC is the most sensitive imaging biomarker of tumour response to PEGPH20.

Deep-learned estimation of uncertainty in measurements of apparent diffusion coefficient from whole-body diffusion-weighted MRI.

PURPOSE: To use deep learning to calculate the uncertainty in apparent diffusion coefficient (σADC) voxel-wise measurements to clinically impact the monitoring of treatment response and improve the quality of ADC maps. MATERIALS AND METHODS: We use a uniquely designed diffusion-weighted imaging (DWI) acquisition protocol that provides gold-standard measurements of σADC to train a deep learning model on two separate cohorts: 16 patients with prostate cancer and 28 patients with mesothelioma. Our network was trained with a novel cost function, which incorporates a perception metric and a b-value regularisation term, on ADC maps calculated by combinations of 2 or 3 b-values (e.g. 50/600/900, 50/900, 50/600, 600/900 s/mm2). We compare the accuracy of the deep-learning based approach for estimation of σADC with gold-standard measurements. RESULTS: The model accurately predicted the σADC for every b-value combination in both cohorts. Mean values of σADC within areas of active disease deviated from those measured by the gold-standard by 4.3% (range, 2.87-6.13%) for the prostate and 3.7% (range, 3.06-4.54%) for the mesothelioma cohort. We also showed that the model can easily be adapted for a different DWI protocol and field-of-view with only a few images (as little as a single patient) using transfer learning. CONCLUSION: Deep learning produces maps of σADC from standard clinical diffusion-weighted images (DWI) when 2 or more b-values are available.

Accelerating Whole-Body Diffusion-weighted MRI with Deep Learning-based Denoising Image Filters.

PURPOSE: To use deep learning to improve the image quality of subsampled images (number of acquisitions = 1 [NOA1]) to reduce whole-body diffusion-weighted MRI (WBDWI) acquisition times. MATERIALS AND METHODS: Both retrospective and prospective patient groups were used to develop a deep learning-based denoising image filter (DNIF) model. For initial model training and validation, 17 patients with metastatic prostate cancer with acquired WBDWI NOA1 and NOA9 images (acquisition period, 2015-2017) were retrospectively included. An additional 22 prospective patients with advanced prostate cancer, myeloma, and advanced breast cancer were used for model testing (2019), and the radiologic quality of DNIF-processed NOA1 (NOA1-DNIF) images were compared with NOA1 images and clinical NOA16 images by using a three-point Likert scale (good, average, or poor; statistical significance was calculated by using a Wilcoxon signed ranked test). The model was also retrained and tested in 28 patients with malignant pleural mesothelioma (MPM) who underwent lung MRI (2015-2017) to demonstrate feasibility in other body regions. RESULTS: The model visually improved the quality of NOA1 images in all test patients, with the majority of NOA1-DNIF and NOA16 images being graded as either "average" or "good" across all image-quality criteria. From validation data, the mean apparent diffusion coefficient (ADC) values within NOA1-DNIF images of bone disease deviated from those within NOA9 images by an average of 1.9% (range, 1.1%-2.6%). The model was also successfully applied in the context of MPM; the mean ADCs from NOA1-DNIF images of MPM deviated from those measured by using clinical-standard images (NOA12) by 3.7% (range, 0.2%-10.6%). CONCLUSION: Clinical-standard images were generated from subsampled images by using a DNIF.Keywords: Image Postprocessing, MR-Diffusion-weighted Imaging, Neural Networks, Oncology, Whole-Body Imaging, Supervised Learning, MR-Functional Imaging, Metastases, Prostate, Lung Supplemental material is available for this article. Published under a CC BY 4.0 license.

Noninvasive MRI Native T1 Mapping Detects Response to MYCN-targeted Therapies in the Th-MYCN Model of Neuroblastoma.

Noninvasive early indicators of treatment response are crucial to the successful delivery of precision medicine in children with cancer. Neuroblastoma is a common solid tumor of young children that arises from anomalies in neural crest development. Therapeutic approaches aiming to destabilize MYCN protein, such as small-molecule inhibitors of Aurora A and mTOR, are currently being evaluated in early phase clinical trials in children with high-risk MYCN-driven disease, with limited ability to evaluate conventional pharmacodynamic biomarkers of response. T1 mapping is an MRI scan that measures the proton spin-lattice relaxation time T1. Using a multiparametric MRI-pathologic cross-correlative approach and computational pathology methodologies including a machine learning-based algorithm for the automatic detection and classification of neuroblasts, we show here that T1 mapping is sensitive to the rich histopathologic heterogeneity of neuroblastoma in the Th-MYCN transgenic model. Regions with high native T1 corresponded to regions dense in proliferative undifferentiated neuroblasts, whereas regions characterized by low T1 were rich in apoptotic or differentiating neuroblasts. Reductions in tumor-native T1 represented a sensitive biomarker of response to treatment-induced apoptosis with two MYCN-targeted small-molecule inhibitors, Aurora A kinase inhibitor alisertib (MLN8237) and mTOR inhibitor vistusertib (AZD2014). Overall, we demonstrate the potential of T1 mapping, a scan readily available on most clinical MRI scanners, to assess response to therapy and guide clinical trials for children with neuroblastoma. The study reinforces the potential role of MRI-based functional imaging in delivering precision medicine to children with neuroblastoma. SIGNIFICANCE: This study shows that MRI-based functional imaging can detect apoptotic responses to MYCN-targeted small-molecule inhibitors in a genetically engineered murine model of MYCN-driven neuroblastoma.

SuperHistopath: A Deep Learning Pipeline for Mapping Tumor Heterogeneity on Low-Resolution Whole-Slide Digital Histopathology Images.

High computational cost associated with digital pathology image analysis approaches is a challenge towards their translation in routine pathology clinic. Here, we propose a computationally efficient framework (SuperHistopath), designed to map global context features reflecting the rich tumor morphological heterogeneity. SuperHistopath efficiently combines i) a segmentation approach using the linear iterative clustering (SLIC) superpixels algorithm applied directly on the whole-slide images at low resolution (5x magnification) to adhere to region boundaries and form homogeneous spatial units at tissue-level, followed by ii) classification of superpixels using a convolution neural network (CNN). To demonstrate how versatile SuperHistopath was in accomplishing histopathology tasks, we classified tumor tissue, stroma, necrosis, lymphocytes clusters, differentiating regions, fat, hemorrhage and normal tissue, in 127 melanomas, 23 triple-negative breast cancers, and 73 samples from transgenic mouse models of high-risk childhood neuroblastoma with high accuracy (98.8%, 93.1% and 98.3% respectively). Furthermore, SuperHistopath enabled discovery of significant differences in tumor phenotype of neuroblastoma mouse models emulating genomic variants of high-risk disease, and stratification of melanoma patients (high ratio of lymphocyte-to-tumor superpixels (p = 0.015) and low stroma-to-tumor ratio (p = 0.028) were associated with a favorable prognosis). Finally, SuperHistopath is efficient for annotation of ground-truth datasets (as there is no need of boundary delineation), training and application (~5 min for classifying a whole-slide image and as low as ~30 min for network training). These attributes make SuperHistopath particularly attractive for research in rich datasets and could also facilitate its adoption in the clinic to accelerate pathologist workflow with the quantification of phenotypes, predictive/prognosis markers.

Superpixel-Based Conditional Random Fields (SuperCRF): Incorporating Global and Local Context for Enhanced Deep Learning in Melanoma Histopathology.

Computational pathology-based cell classification algorithms are revolutionizing the study of the tumor microenvironment and can provide novel predictive/prognosis biomarkers crucial for the delivery of precision oncology. Current algorithms used on hematoxylin and eosin slides are based on individual cell nuclei morphology with limited local context features. Here, we propose a novel multi-resolution hierarchical framework (SuperCRF) inspired by the way pathologists perceive regional tissue architecture to improve cell classification and demonstrate its clinical applications. We develop SuperCRF by training a state-of-art deep learning spatially constrained- convolution neural network (SC-CNN) to detect and classify cells from 105 high-resolution (20×) H&E-stained slides of The Cancer Genome Atlas melanoma dataset and subsequently, a conditional random field (CRF) by combining cellular neighborhood with tumor regional classification from lower resolution images (5, 1.25×) given by a superpixel-based machine learning framework. SuperCRF led to an 11.85% overall improvement in the accuracy of the state-of-art deep learning SC-CNN cell classifier. Consistent with a stroma-mediated immune suppressive microenvironment, SuperCRF demonstrated that (i) a high ratio of lymphocytes to all lymphocytes within the stromal compartment (p = 0.026) and (ii) a high ratio of stromal cells to all cells (p < 0.0001 compared to p = 0.039 for SC-CNN only) are associated with poor survival in patients with melanoma. SuperCRF improves cell classification by introducing global and local context-based information and can be implemented in combination with any single-cell classifier. SuperCRF provides valuable tools to study the tumor microenvironment and identify predictors of survival and response to therapy.

Investigating the Contribution of Collagen to the Tumor Biomechanical Phenotype with Noninvasive Magnetic Resonance Elastography.

Increased stiffness in the extracellular matrix (ECM) contributes to tumor progression and metastasis. Therefore, stromal modulating therapies and accompanying biomarkers are being developed to target ECM stiffness. Magnetic resonance (MR) elastography can noninvasively and quantitatively map the viscoelastic properties of tumors in vivo and thus has clear clinical applications. Herein, we used MR elastography, coupled with computational histopathology, to interrogate the contribution of collagen to the tumor biomechanical phenotype and to evaluate its sensitivity to collagenase-induced stromal modulation. Elasticity (G d) and viscosity (G l) were significantly greater for orthotopic BT-474 (G d = 5.9 ± 0.2 kPa, G l = 4.7 ± 0.2 kPa, n = 7) and luc-MDA-MB-231-LM2-4 (G d = 7.9 ± 0.4 kPa, G l = 6.0 ± 0.2 kPa, n = 6) breast cancer xenografts, and luc-PANC1 (G d = 6.9 ± 0.3 kPa, G l = 6.2 ± 0.2 kPa, n = 7) pancreatic cancer xenografts, compared with tumors associated with the nervous system, including GTML/Trp53KI/KI medulloblastoma (G d = 3.5 ± 0.2 kPa, G l = 2.3 ± 0.2 kPa, n = 7), orthotopic luc-D-212-MG (G d = 3.5 ± 0.2 kPa, G l = 2.3 ± 0.2 kPa, n = 7), luc-RG2 (G d = 3.5 ± 0.2 kPa, G l = 2.3 ± 0.2 kPa, n = 5), and luc-U-87-MG (G d = 3.5 ± 0.2 kPa, G l = 2.3 ± 0.2 kPa, n = 8) glioblastoma xenografts, intracranially propagated luc-MDA-MB-231-LM2-4 (G d = 3.7 ± 0.2 kPa, G l = 2.2 ± 0.1 kPa, n = 7) breast cancer xenografts, and Th-MYCN neuroblastomas (G d = 3.5 ± 0.2 kPa, G l = 2.3 ± 0.2 kPa, n = 5). Positive correlations between both elasticity (r = 0.72, P < 0.0001) and viscosity (r = 0.78, P < 0.0001) were determined with collagen fraction, but not with cellular or vascular density. Treatment with collagenase significantly reduced G d (P = 0.002) and G l (P = 0.0006) in orthotopic breast tumors. Texture analysis of extracted images of picrosirius red staining revealed significant negative correlations of entropy with G d (r = -0.69, P < 0.0001) and G l (r = -0.76, P < 0.0001), and positive correlations of fractal dimension with G d (r = 0.75, P < 0.0001) and G l (r = 0.78, P < 0.0001). MR elastography can thus provide sensitive imaging biomarkers of tumor collagen deposition and its therapeutic modulation. SIGNIFICANCE: MR elastography enables noninvasive detection of tumor stiffness and will aid in the development of ECM-targeting therapies.

MRI Imaging of the Hemodynamic Vasculature of Neuroblastoma Predicts Response to Antiangiogenic Treatment.

Childhood neuroblastoma is a hypervascular tumor of neural origin, for which antiangiogenic drugs are currently being evaluated; however, predictive biomarkers of treatment response, crucial for successful delivery of precision therapeutics, are lacking. We describe an MRI-pathologic cross-correlative approach using intrinsic susceptibility (IS) and susceptibility contrast (SC) MRI to noninvasively map the vascular phenotype in neuroblastoma Th-MYCN transgenic mice treated with the vascular endothelial growth factor receptor inhibitor cediranib. We showed that the transverse MRI relaxation rate R 2* (second-1) and fractional blood volume (fBV, %) were sensitive imaging biomarkers of hemorrhage and vascular density, respectively, and were also predictive biomarkers of response to cediranib. Comparison with MRI and pathology from patients with MYCN-amplified neuroblastoma confirmed the high degree to which the Th-MYCN model vascular phenotype recapitulated that of the clinical phenotype, thereby supporting further evaluation of IS- and SC-MRI in the clinic. This study reinforces the potential role of functional MRI in delivering precision medicine to children with neuroblastoma. SIGNIFICANCE: This study shows that functional MRI predicts response to vascular-targeted therapy in a genetically engineered murine model of neuroblastoma.

Non-Invasive Prostate Cancer Characterization with Diffusion-Weighted MRI: Insight from In silico Studies of a Transgenic Mouse Model.

Diffusion-weighted magnetic resonance imaging (DWI) enables non-invasive, quantitative staging of prostate cancer via measurement of the apparent diffusion coefficient (ADC) of water within tissues. In cancer, more advanced disease is often characterized by higher cellular density (cellularity), which is generally accepted to correspond to a lower measured ADC. A quantitative relationship between tissue structure and in vivo measurements of ADC has yet to be determined for prostate cancer. In this study, we establish a theoretical framework for relating ADC measurements with tissue cellularity and the proportion of space occupied by prostate lumina, both of which are estimated through automatic image processing of whole-slide digital histology samples taken from a cohort of six healthy mice and nine transgenic adenocarcinoma of the mouse prostate (TRAMP) mice. We demonstrate that a significant inverse relationship exists between ADC and tissue cellularity that is well characterized by our model, and that a decrease of the luminal space within the prostate is associated with a decrease in ADC and more aggressive tumor subtype. The parameters estimated from our model in this mouse cohort predict the diffusion coefficient of water within the prostate-tissue to be 2.18 × 10-3 mm2/s (95% CI: 1.90, 2.55). This value is significantly lower than the diffusion coefficient of free water at body temperature suggesting that the presence of organelles and macromolecules within tissues can drastically hinder the random motion of water molecules within prostate tissue. We validate the assumptions made by our model using novel in silico analysis of whole-slide histology to provide the simulated ADC (sADC); this is demonstrated to have a significant positive correlation with in vivo measured ADC (r2 = 0.55) in our mouse population. The estimation of the structural properties of prostate tissue is vital for predicting and staging cancer aggressiveness, but prostate tissue biopsies are painful, invasive, and are prone to complications such as sepsis. The developments made in this study provide the possibility of estimating the structural properties of prostate tissue via non-invasive virtual biopsies from MRI, minimizing the need for multiple tissue biopsies and allowing sequential measurements to be made for prostate cancer monitoring.