Mutation-selection balance and compensatory mechanisms in tumour evolution.

Intratumour heterogeneity and phenotypic plasticity, sustained by a range of somatic aberrations, as well as epigenetic and metabolic adaptations, are the principal mechanisms that enable cancers to resist treatment and survive under environmental stress. A comprehensive picture of the interplay between different somatic aberrations, from point mutations to whole-genome duplications, in tumour initiation and progression is lacking. We posit that different genomic aberrations generally exhibit a temporal order, shaped by a balance between the levels of mutations and selective pressures. Repeat instability emerges first, followed by larger aberrations, with compensatory effects leading to robust tumour fitness maintained throughout the tumour progression. A better understanding of the interplay between genetic aberrations, the microenvironment, and epigenetic and metabolic cellular states is essential for early detection and prevention of cancer as well as development of efficient therapeutic strategies.

Artificial intelligence in cancer imaging: Clinical challenges and applications.

Judgement, as one of the core tenets of medicine, relies upon the integration of multilayered data with nuanced decision making. Cancer offers a unique context for medical decisions given not only its variegated forms with evolution of disease but also the need to take into account the individual condition of patients, their ability to receive treatment, and their responses to treatment. Challenges remain in the accurate detection, characterization, and monitoring of cancers despite improved technologies. Radiographic assessment of disease most commonly relies upon visual evaluations, the interpretations of which may be augmented by advanced computational analyses. In particular, artificial intelligence (AI) promises to make great strides in the qualitative interpretation of cancer imaging by expert clinicians, including volumetric delineation of tumors over time, extrapolation of the tumor genotype and biological course from its radiographic phenotype, prediction of clinical outcome, and assessment of the impact of disease and treatment on adjacent organs. AI may automate processes in the initial interpretation of images and shift the clinical workflow of radiographic detection, management decisions on whether or not to administer an intervention, and subsequent observation to a yet to be envisioned paradigm. Here, the authors review the current state of AI as applied to medical imaging of cancer and describe advances in 4 tumor types (lung, brain, breast, and prostate) to illustrate how common clinical problems are being addressed. Although most studies evaluating AI applications in oncology to date have not been vigorously validated for reproducibility and generalizability, the results do highlight increasingly concerted efforts in pushing AI technology to clinical use and to impact future directions in cancer care.

The harsh microenvironment in early breast cancer selects for a Warburg phenotype.

The harsh microenvironment of ductal carcinoma in situ (DCIS) exerts strong evolutionary selection pressures on cancer cells. We hypothesize that the poor metabolic conditions near the ductal center foment the emergence of a Warburg Effect (WE) phenotype, wherein cells rapidly ferment glucose to lactic acid, even in normoxia. To test this hypothesis, we subjected low-glycolytic breast cancer cells to different microenvironmental selection pressures using combinations of hypoxia, acidosis, low glucose, and starvation for many months and isolated single clones for metabolic and transcriptomic profiling. The two harshest conditions selected for constitutively expressed WE phenotypes. RNA sequencing analysis of WE clones identified the transcription factor KLF4 as potential inducer of the WE phenotype. In stained DCIS samples, KLF4 expression was enriched in the area with the harshest microenvironmental conditions. We simulated in vivo DCIS phenotypic evolution using a mathematical model calibrated from the in vitro results. The WE phenotype emerged in the poor metabolic conditions near the necrotic core. We propose that harsh microenvironments within DCIS select for a WE phenotype through constitutive transcriptional reprogramming, thus conferring a survival advantage and facilitating further growth and invasion.

CancerCellTracker: a brightfield time-lapse microscopy framework for cancer drug sensitivity estimation.

MOTIVATION: Time-lapse microscopy is a powerful technique that relies on images of live cells cultured ex vivo that are captured at regular intervals of time to describe and quantify their behavior under certain experimental conditions. This imaging method has great potential in advancing the field of precision oncology by quantifying the response of cancer cells to various therapies and identifying the most efficacious treatment for a given patient. Digital image processing algorithms developed so far require high-resolution images involving very few cells originating from homogeneous cell line populations. We propose a novel framework that tracks cancer cells to capture their behavior and quantify cell viability to inform clinical decisions in a high-throughput manner. RESULTS: The brightfield microscopy images a large number of patient-derived cells in an ex vivo reconstruction of the tumor microenvironment treated with 31 drugs for up to 6 days. We developed a robust and user-friendly pipeline CancerCellTracker that detects cells in co-culture, tracks these cells across time and identifies cell death events using changes in cell attributes. We validated our computational pipeline by comparing the timing of cell death estimates by CancerCellTracker from brightfield images and a fluorescent channel featuring ethidium homodimer. We benchmarked our results using a state-of-the-art algorithm implemented in ImageJ and previously published in the literature. We highlighted CancerCellTracker's efficiency in estimating the percentage of live cells in the presence of bone marrow stromal cells. AVAILABILITY AND IMPLEMENTATION: https://github.com/compbiolabucf/CancerCellTracker. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Proton export upregulates aerobic glycolysis.

INTRODUCTION: Aggressive cancers commonly ferment glucose to lactic acid at high rates, even in the presence of oxygen. This is known as aerobic glycolysis, or the "Warburg Effect." It is widely assumed that this is a consequence of the upregulation of glycolytic enzymes. Oncogenic drivers can increase the expression of most proteins in the glycolytic pathway, including the terminal step of exporting H+ equivalents from the cytoplasm. Proton exporters maintain an alkaline cytoplasmic pH, which can enhance all glycolytic enzyme activities, even in the absence of oncogene-related expression changes. Based on this observation, we hypothesized that increased uptake and fermentative metabolism of glucose could be driven by the expulsion of H+ equivalents from the cell. RESULTS: To test this hypothesis, we stably transfected lowly glycolytic MCF-7, U2-OS, and glycolytic HEK293 cells to express proton-exporting systems: either PMA1 (plasma membrane ATPase 1, a yeast H+-ATPase) or CA-IX (carbonic anhydrase 9). The expression of either exporter in vitro enhanced aerobic glycolysis as measured by glucose consumption, lactate production, and extracellular acidification rate. This resulted in an increased intracellular pH, and metabolomic analyses indicated that this was associated with an increased flux of all glycolytic enzymes upstream of pyruvate kinase. These cells also demonstrated increased migratory and invasive phenotypes in vitro, and these were recapitulated in vivo by more aggressive behavior, whereby the acid-producing cells formed higher-grade tumors with higher rates of metastases. Neutralizing tumor acidity with oral buffers reduced the metastatic burden. CONCLUSIONS: Therefore, cancer cells which increase export of H+ equivalents subsequently increase intracellular alkalization, even without oncogenic driver mutations, and this is sufficient to alter cancer metabolism towards an upregulation of aerobic glycolysis, a Warburg phenotype. Overall, we have shown that the traditional understanding of cancer cells favoring glycolysis and the subsequent extracellular acidification is not always linear. Cells which can, independent of metabolism, acidify through proton exporter activity can sufficiently drive their metabolism towards glycolysis providing an important fitness advantage for survival.

Cereblon harnesses Myc-dependent bioenergetics and activity of CD8+ T lymphocytes.

Immunomodulatory drugs, such as thalidomide and related compounds, potentiate T-cell effector functions. Cereblon (CRBN), a substrate receptor of the DDB1-cullin-RING E3 ubiquitin ligase complex, is the only molecular target for this drug class, where drug-induced, ubiquitin-dependent degradation of known "neosubstrates," such as IKAROS, AIOLOS, and CK1α, accounts for their biological activity. Far less clear is whether these CRBN E3 ligase-modulating compounds disrupt the endogenous functions of CRBN. We report that CRBN functions in a feedback loop that harnesses antigen-specific CD8+ T-cell effector responses. Specifically, Crbn deficiency in murine CD8+ T cells augments their central metabolism manifested as elevated bioenergetics, with supraphysiological levels of polyamines, secondary to enhanced glucose and amino acid transport, and with increased expression of metabolic enzymes, including the polyamine biosynthetic enzyme ornithine decarboxylase. Treatment with CRBN-modulating compounds similarly augments central metabolism of human CD8+ T cells. Notably, the metabolic control of CD8+ T cells by modulating compounds or Crbn deficiency is linked to increased and sustained expression of the master metabolic regulator MYC. Finally, Crbn-deficient T cells have augmented antigen-specific cytolytic activity vs melanoma tumor cells, ex vivo and in vivo, and drive accelerated and highly aggressive graft-versus-host disease. Therefore, CRBN functions to harness the activation of CD8+ T cells, and this phenotype can be exploited by treatment with drugs.

Integrated Biomarkers for the Management of Indeterminate Pulmonary Nodules.

Rationale: Patients with indeterminate pulmonary nodules (IPNs) at risk of cancer undergo high rates of invasive, costly, and morbid procedures. Objectives: To train and externally validate a risk prediction model that combined clinical, blood, and imaging biomarkers to improve the noninvasive management of IPNs. Methods: In this prospectively collected, retrospective blinded evaluation study, probability of cancer was calculated for 456 patient nodules using the Mayo Clinic model, and patients were categorized into low-, intermediate-, and high-risk groups. A combined biomarker model (CBM) including clinical variables, serum high sensitivity CYFRA 21-1 level, and a radiomic signature was trained in cohort 1 (n = 170) and validated in cohorts 2-4 (total n = 286). All patients were pooled to recalibrate the model for clinical implementation. The clinical utility of the CBM compared with current clinical care was evaluated in 2 cohorts. Measurements and Main Results: The CBM provided improved diagnostic accuracy over the Mayo Clinic model with an improvement in area under the curve of 0.124 (95% bootstrap confidence interval, 0.091-0.156; P < 2 × 10-16). Applying 10% and 70% risk thresholds resulted in a bias-corrected clinical reclassification index for cases and control subjects of 0.15 and 0.12, respectively. A clinical utility analysis of patient medical records estimated that a CBM-guided strategy would have reduced invasive procedures from 62.9% to 50.6% in the intermediate-risk benign population and shortened the median time to diagnosis of cancer from 60 to 21 days in intermediate-risk cancers. Conclusions: Integration of clinical, blood, and image biomarkers improves noninvasive diagnosis of patients with IPNs, potentially reducing the rate of unnecessary invasive procedures while shortening the time to diagnosis.

Radiomics predicts risk of cachexia in advanced NSCLC patients treated with immune checkpoint inhibitors.

BACKGROUND: Approximately 50% of cancer patients eventually develop a syndrome of prolonged weight loss (cachexia), which may contribute to primary resistance to immune checkpoint inhibitors (ICI). This study utilised radiomics analysis of 18F-FDG-PET/CT images to predict risk of cachexia that can be subsequently associated with clinical outcomes among advanced non-small cell lung cancer (NSCLC) patients treated with ICI. METHODS: Baseline (pre-therapy) PET/CT images and clinical data were retrospectively curated from 210 ICI-treated NSCLC patients from two institutions. A radiomics signature was developed to predict the cachexia with PET/CT images, which was further used to predict durable clinical benefit (DCB), progression-free survival (PFS) and overall survival (OS) following ICI. RESULTS: The radiomics signature predicted risk of cachexia with areas under receiver operating characteristics curves (AUCs) ≥ 0.74 in the training, test, and external test cohorts. Further, the radiomics signature could identify patients with DCB from ICI with AUCs≥0.66 in these three cohorts. PFS and OS were significantly shorter among patients with higher radiomics-based cachexia probability in all three cohorts, especially among those potentially immunotherapy sensitive patients with PD-L1-positive status (p < 0.05). CONCLUSIONS: PET/CT radiomics analysis has the potential to predict the probability of developing cachexia before the start of ICI, triggering aggressive monitoring to improve potential to achieve more clinical benefit.

Artificial selection for host resistance to tumour growth and subsequent cancer cell adaptations: an evolutionary arms race.

BACKGROUND: Cancer progression is governed by evolutionary dynamics in both the tumour population and its host. Since cancers die with the host, each new population of cancer cells must reinvent strategies to overcome the host's heritable defences. In contrast, host species evolve defence strategies over generations if tumour development limits procreation. METHODS: We investigate this "evolutionary arms race" through intentional breeding of immunodeficient SCID and immunocompetent Black/6 mice to evolve increased tumour suppression. Over 10 generations, we injected Lewis lung mouse carcinoma cells [LL/2-Luc-M38] and selectively bred the two individuals with the slowest tumour growth at day 11. Their male progeny were hosts in the subsequent round. RESULTS: The evolved SCID mice suppressed tumour growth through biomechanical restriction from increased mesenchymal proliferation, and the evolved Black/6 mice suppressed tumour growth by increasing immune-mediated killing of cancer cells. However, transcriptomic changes of multicellular tissue organisation and function genes allowed LL/2-Luc-M38 cells to adapt through increased matrix remodelling in SCID mice, and reduced angiogenesis, increased energy utilisation and accelerated proliferation in Black/6 mice. CONCLUSION: Host species can rapidly evolve both immunologic and non-immunologic tumour defences. However, cancer cell plasticity allows effective phenotypic and population-based counter strategies.

The Biological Meaning of Radiomic Features.

An earlier incorrect version appeared online. This article was corrected on February 10, 2021.

Heterogeneity analysis of MRI T2 maps for measurement of early tumor response to radiotherapy.

External beam radiotherapy (XRT) is a widely used cancer treatment, yet responses vary dramatically among patients. These differences are not accounted for in clinical practice, partly due to a lack of sensitive early response biomarkers. We hypothesize that quantitative magnetic resonance imaging (MRI) measures reflecting tumor heterogeneity can provide a sensitive and robust biomarker of early XRT response. MRI T2 mapping was performed every 72 hours following 10 Gy dose XRT in two models of pancreatic cancer propagated in the hind limb of mice. Interquartile range (IQR) of tumor T2 was presented as a potential biomarker of radiotherapy response compared with tumor growth kinetics, and biological validation was performed through quantitative histology analysis. Quantification of tumor T2 IQR showed sensitivity for detection of XRT-induced tumor changes 72 hours after treatment, outperforming T2-weighted and diffusion-weighted MRI, with very good robustness. Histological comparison revealed that T2 IQR provides a measure of spatial heterogeneity in tumor cell density, related to radiation-induced necrosis. Early IQR changes were found to correlate to subsequent tumor volume changes, indicating promise for treatment response prediction. Our preclinical findings indicate that spatial heterogeneity analysis of T2 MRI can provide a translatable method for early radiotherapy response assessment. We propose that the method may in future be applied for personalization of radiotherapy through adaptive treatment paradigms.

The role of carbonic anhydrase IX in cancer development: links to hypoxia, acidosis, and beyond.

Cancer development is a complex process that follows an intricate scenario with a dynamic interplay of selective and adaptive steps and an extensive cast of molecules and signaling pathways. Solid tumor initially grows as an avascular bulk of cells carrying oncogenic mutations until diffusion distances from the nearest functional blood vessels limit delivery of nutrients and oxygen on the one hand and removal of metabolic waste on the other one. These restrictions result in regional hypoxia and acidosis that select for adaptable tumor cells able to promote aberrant angiogenesis, remodel metabolism, acquire invasiveness and metastatic propensity, and gain therapeutic resistance. Tumor cells are thereby endowed with capability to survive and proliferate in hostile microenvironment, communicate with stroma, enter circulation, colonize secondary sites, and generate metastases. While the role of oncogenic mutations initializing and driving these processes is well established, a key contribution of non-genomic, landscaping molecular players is still less appreciated despite they can equally serve as viable targets of anticancer therapies. Carbonic anhydrase IX (CA IX) is one of these players: it is induced by hypoxia, functionally linked to acidosis, implicated in invasiveness, and correlated with therapeutic resistance. Here, we summarize the available experimental evidence supported by accumulating preclinical and clinical data that CA IX can contribute virtually to each step of cancer progression path via its enzyme activity and/or non-catalytic mechanisms. We also propose that targeting tumor cells that express CA IX may provide therapeutic benefits in various settings and combinations with both conventional and newly developed treatments.

Causes, consequences, and therapy of tumors acidosis.

While cancer is commonly described as "a disease of the genes," it is also associated with massive metabolic reprogramming that is now accepted as a disease "Hallmark." This programming is complex and often involves metabolic cooperativity between cancer cells and their surrounding stroma. Indeed, there is emerging clinical evidence that interrupting a cancer's metabolic program can improve patients' outcomes. The most commonly observed and well-studied metabolic adaptation in cancers is the fermentation of glucose to lactic acid, even in the presence of oxygen, also known as "aerobic glycolysis" or the "Warburg Effect." Much has been written about the mechanisms of the Warburg effect, and this remains a topic of great debate. However, herein, we will focus on an important sequela of this metabolic program: the acidification of the tumor microenvironment. Rather than being an epiphenomenon, it is now appreciated that this acidosis is a key player in cancer somatic evolution and progression to malignancy. Adaptation to acidosis induces and selects for malignant behaviors, such as increased invasion and metastasis, chemoresistance, and inhibition of immune surveillance. However, the metabolic reprogramming that occurs during adaptation to acidosis also introduces therapeutic vulnerabilities. Thus, tumor acidosis is a relevant therapeutic target, and we describe herein four approaches to accomplish this: (1) neutralizing acid directly with buffers, (2) targeting metabolic vulnerabilities revealed by acidosis, (3) developing acid-activatable drugs and nanomedicines, and (4) inhibiting metabolic processes responsible for generating acids in the first place.

The Image Biomarker Standardization Initiative: Standardized Quantitative Radiomics for High-Throughput Image-based Phenotyping.

Background Radiomic features may quantify characteristics present in medical imaging. However, the lack of standardized definitions and validated reference values have hampered clinical use. Purpose To standardize a set of 174 radiomic features. Materials and Methods Radiomic features were assessed in three phases. In phase I, 487 features were derived from the basic set of 174 features. Twenty-five research teams with unique radiomics software implementations computed feature values directly from a digital phantom, without any additional image processing. In phase II, 15 teams computed values for 1347 derived features using a CT image of a patient with lung cancer and predefined image processing configurations. In both phases, consensus among the teams on the validity of tentative reference values was measured through the frequency of the modal value and classified as follows: less than three matches, weak; three to five matches, moderate; six to nine matches, strong; 10 or more matches, very strong. In the final phase (phase III), a public data set of multimodality images (CT, fluorine 18 fluorodeoxyglucose PET, and T1-weighted MRI) from 51 patients with soft-tissue sarcoma was used to prospectively assess reproducibility of standardized features. Results Consensus on reference values was initially weak for 232 of 302 features (76.8%) at phase I and 703 of 1075 features (65.4%) at phase II. At the final iteration, weak consensus remained for only two of 487 features (0.4%) at phase I and 19 of 1347 features (1.4%) at phase II. Strong or better consensus was achieved for 463 of 487 features (95.1%) at phase I and 1220 of 1347 features (90.6%) at phase II. Overall, 169 of 174 features were standardized in the first two phases. In the final validation phase (phase III), most of the 169 standardized features could be excellently reproduced (166 with CT; 164 with PET; and 164 with MRI). Conclusion A set of 169 radiomics features was standardized, which enabled verification and calibration of different radiomics software. © RSNA, 2020 Online supplemental material is available for this article. See also the editorial by Kuhl and Truhn in this issue.

Radiomics of 18F-FDG PET/CT images predicts clinical benefit of advanced NSCLC patients to checkpoint blockade immunotherapy.

INTRODUCTION: Immunotherapy has improved outcomes for patients with non-small cell lung cancer (NSCLC), yet durable clinical benefit (DCB) is experienced in only a fraction of patients. Here, we test the hypothesis that radiomics features from baseline pretreatment 18F-FDG PET/CT scans can predict clinical outcomes of NSCLC patients treated with checkpoint blockade immunotherapy. METHODS: This study included 194 patients with histologically confirmed stage IIIB-IV NSCLC with pretreatment PET/CT images. Radiomics features were extracted from PET, CT, and PET+CT fusion images based on minimum Kullback-Leibler divergence (KLD) criteria. The radiomics features from 99 retrospective patients were used to train a multiparametric radiomics signature (mpRS) to predict DCB using an improved least absolute shrinkage and selection operator (LASSO) method, which was subsequently validated in both retrospective (N = 47) and prospective test cohorts (N = 48). Using these cohorts, the mpRS was also used to predict progression-free survival (PFS) and overall survival (OS) by training nomogram models using multivariable Cox regression analyses with additional clinical characteristics incorporated. RESULTS: The mpRS could predict patients who will receive DCB, with areas under receiver operating characteristic curves (AUCs) of 0.86 (95%CI 0.79-0.94), 0.83 (95%CI 0.71-0.94), and 0.81 (95%CI 0.68-0.92) in the training, retrospective test, and prospective test cohorts, respectively. In the same three cohorts, respectively, nomogram models achieved C-indices of 0.74 (95%CI 0.68-0.80), 0.74 (95%CI 0.66-0.82), and 0.77 (95%CI 0.69-0.84) to predict PFS and C-indices of 0.83 (95%CI 0.77-0.88), 0.83 (95%CI 0.71-0.94), and 0.80 (95%CI 0.69-0.91) to predict OS. CONCLUSION: PET/CT-based signature can be used prior to initiation of immunotherapy to identify NSCLC patients most likely to benefit from immunotherapy. As such, these data may be leveraged to improve more precise and individualized decision support in the treatment of patients with advanced NSCLC.

Direct and indirect assessment of cancer metabolism explored by MRI.

Magnetic resonance-based approaches to obtain metabolic information on cancer have been explored for decades. Electron paramagnetic resonance (EPR) has been developed to pursue metabolic profiling and successfully used to monitor several physiologic parameters such as pO2 , pH, and redox status. All these parameters are associated with pathophysiology of various diseases. Especially in oncology, cancer hypoxia has been intensively studied because of its relationship with metabolic alterations, acquiring treatment resistance, or a malignant phenotype. Thus, pO2 imaging leads to an indirect metabolic assessment in this regard. Proton electron double-resonance imaging (PEDRI) is an imaging technique to visualize EPR by using the Overhauser effect. Most biological parameters assessed in EPR can be visualized using PEDRI. However, EPR and PEDRI have not been evaluated sufficiently for clinical application due to limitations such as toxicity of the probes or high specific absorption rate. Hyperpolarized (HP) 13 C MRI is a novel imaging technique that can directly visualize the metabolic profile. Production of metabolites of the HP 13 C probe delivered to target tissue are evaluated in this modality. Unlike EPR or PEDRI, which require the injection of radical probes, 13 C MRI requires a probe that can be physiologically metabolized and efficiently hyperpolarized. Among several methods for hyperpolarizing probes, dissolution dynamic nuclear hyperpolarization is a widely used technique for in vivo imaging. Pyruvate is the most suitable probe for HP 13 C MRI because it is part of the glycolytic pathway and the high efficiency of pyruvate-to-lactate conversion is a distinguishing feature of cancer. Its clinical applicability also makes it a promising metabolic imaging modality. Here, we summarize the applications of these indirect and direct MR-based metabolic assessments focusing on pO2 and pyruvate-to-lactate conversion. The two parameters are strongly associated with each other, hence the acquired information is potentially interchangeable when evaluating treatment response to oxygen-dependent cancer therapies.

Hypoxia and acidosis: immune suppressors and therapeutic targets.

Due to imbalances between vascularity and cellular growth patterns, the tumour microenvironment harbours multiple metabolic stressors including hypoxia and acidosis, which have significant influences on remodelling both tumour and peritumoral tissues. These stressors are also immunosuppressive and can contribute to escape from immune surveillance. Understanding these effects and characterizing the pathways involved can identify new targets for therapy and may redefine our understanding of traditional anti-tumour therapies. In this review, the effects of hypoxia and acidosis on tumour immunity will be summarized, and how modulating these parameters and their sequelae can be a useful tool for future therapeutic interventions is discussed.

Deep learning for lung cancer prognostication: A retrospective multi-cohort radiomics study.

BACKGROUND: Non-small-cell lung cancer (NSCLC) patients often demonstrate varying clinical courses and outcomes, even within the same tumor stage. This study explores deep learning applications in medical imaging allowing for the automated quantification of radiographic characteristics and potentially improving patient stratification. METHODS AND FINDINGS: We performed an integrative analysis on 7 independent datasets across 5 institutions totaling 1,194 NSCLC patients (age median = 68.3 years [range 32.5-93.3], survival median = 1.7 years [range 0.0-11.7]). Using external validation in computed tomography (CT) data, we identified prognostic signatures using a 3D convolutional neural network (CNN) for patients treated with radiotherapy (n = 771, age median = 68.0 years [range 32.5-93.3], survival median = 1.3 years [range 0.0-11.7]). We then employed a transfer learning approach to achieve the same for surgery patients (n = 391, age median = 69.1 years [range 37.2-88.0], survival median = 3.1 years [range 0.0-8.8]). We found that the CNN predictions were significantly associated with 2-year overall survival from the start of respective treatment for radiotherapy (area under the receiver operating characteristic curve [AUC] = 0.70 [95% CI 0.63-0.78], p < 0.001) and surgery (AUC = 0.71 [95% CI 0.60-0.82], p < 0.001) patients. The CNN was also able to significantly stratify patients into low and high mortality risk groups in both the radiotherapy (p < 0.001) and surgery (p = 0.03) datasets. Additionally, the CNN was found to significantly outperform random forest models built on clinical parameters-including age, sex, and tumor node metastasis stage-as well as demonstrate high robustness against test-retest (intraclass correlation coefficient = 0.91) and inter-reader (Spearman's rank-order correlation = 0.88) variations. To gain a better understanding of the characteristics captured by the CNN, we identified regions with the most contribution towards predictions and highlighted the importance of tumor-surrounding tissue in patient stratification. We also present preliminary findings on the biological basis of the captured phenotypes as being linked to cell cycle and transcriptional processes. Limitations include the retrospective nature of this study as well as the opaque black box nature of deep learning networks. CONCLUSIONS: Our results provide evidence that deep learning networks may be used for mortality risk stratification based on standard-of-care CT images from NSCLC patients. This evidence motivates future research into better deciphering the clinical and biological basis of deep learning networks as well as validation in prospective data.

Quantitative imaging of cancer in the postgenomic era: Radio(geno)mics, deep learning, and habitats.

Although cancer often is referred to as "a disease of the genes," it is indisputable that the (epi)genetic properties of individual cancer cells are highly variable, even within the same tumor. Hence, preexisting resistant clones will emerge and proliferate after therapeutic selection that targets sensitive clones. Herein, the authors propose that quantitative image analytics, known as "radiomics," can be used to quantify and characterize this heterogeneity. Virtually every patient with cancer is imaged radiologically. Radiomics is predicated on the beliefs that these images reflect underlying pathophysiologies, and that they can be converted into mineable data for improved diagnosis, prognosis, prediction, and therapy monitoring. In the last decade, the radiomics of cancer has grown from a few laboratories to a worldwide enterprise. During this growth, radiomics has established a convention, wherein a large set of annotated image features (1-2000 features) are extracted from segmented regions of interest and used to build classifier models to separate individual patients into their appropriate class (eg, indolent vs aggressive disease). An extension of this conventional radiomics is the application of "deep learning," wherein convolutional neural networks can be used to detect the most informative regions and features without human intervention. A further extension of radiomics involves automatically segmenting informative subregions ("habitats") within tumors, which can be linked to underlying tumor pathophysiology. The goal of the radiomics enterprise is to provide informed decision support for the practice of precision oncology.