Artificial Intelligence-based Radiomics in the Era of Immuno-oncology.

The recent, rapid advances in immuno-oncology have revolutionized cancer treatment and spurred further research into tumor biology. Yet, cancer patients respond variably to immunotherapy despite mounting evidence to support its efficacy. Current methods for predicting immunotherapy response are unreliable, as these tests cannot fully account for tumor heterogeneity and microenvironment. An improved method for predicting response to immunotherapy is needed. Recent studies have proposed radiomics-the process of converting medical images into quantitative data (features) that can be processed using machine learning algorithms to identify complex patterns and trends-for predicting response to immunotherapy. Because patients undergo numerous imaging procedures throughout the course of the disease, there exists a wealth of radiological imaging data available for training radiomics models. And because radiomic features reflect cancer biology, such as tumor heterogeneity and microenvironment, these models have enormous potential to predict immunotherapy response more accurately than current methods. Models trained on preexisting biomarkers and/or clinical outcomes have demonstrated potential to improve patient stratification and treatment outcomes. In this review, we discuss current applications of radiomics in oncology, followed by a discussion on recent studies that use radiomics to predict immunotherapy response and toxicity.

Redirecting abiraterone metabolism to fine-tune prostate cancer anti-androgen therapy.

Abiraterone blocks androgen synthesis and prolongs survival in patients with castration-resistant prostate cancer, which is otherwise driven by intratumoral androgen synthesis. Abiraterone is metabolized in patients to Δ(4)-abiraterone (D4A), which has even greater anti-tumour activity and is structurally similar to endogenous steroidal 5α-reductase substrates, such as testosterone. Here, we show that D4A is converted to at least three 5α-reduced and three 5ß-reduced metabolites in human serum. The initial 5α-reduced metabolite, 3-keto-5α-abiraterone, is present at higher concentrations than D4A in patients with prostate cancer taking abiraterone, and is an androgen receptor agonist, which promotes prostate cancer progression. In a clinical trial of abiraterone alone, followed by abiraterone plus dutasteride (a 5α-reductase inhibitor), 3-keto-5α-abiraterone and downstream metabolites were depleted by the addition of dutasteride, while D4A concentrations rose, showing that dutasteride effectively blocks production of a tumour-promoting metabolite and permits D4A accumulation. Furthermore, dutasteride did not deplete the three 5ß-reduced metabolites, which were also clinically detectable, demonstrating the specific biochemical effects of pharmacological 5α-reductase inhibition on abiraterone metabolism. Our findings suggest a previously unappreciated and biochemically specific method of clinically fine-tuning abiraterone metabolism to optimize therapy.

Drug resistance profiling of a new triple negative breast cancer patient-derived xenograft model.

BACKGROUND: Triple-negative breast cancer (TNBC) represents an aggressive subtype with limited therapeutic options. Experimental preclinical models that recapitulate their tumors of origin can accelerate target identification, thereby potentially improving therapeutic efficacy. Patient-derived xenografts (PDXs), due to their genomic and transcriptomic fidelity to the tumors from which they are derived, are poised to improve the preclinical testing of drug-target combinations in translational models. Despite the previous development of breast and TNBC PDX models, those derived from patients with demonstrated health-disparities are lacking. METHODS: We use an aggressive TNBC PDX model propagated in SCID/Beige mice that was established from an African-American woman, TU-BcX-2 K1, and assess its metastatic potential and drug sensitivities under distinct in vitro conditions. Cellular derivatives of the primary tumor or the PDX were grown in 2D culture conditions or grown in mammospheres 3D culture. Flow cytometry and fluorescence staining was used to quantify cancer stem cell-like populations. qRT-PCR was used to describe the mesenchymal gene signature of the tumor. The sensitivity of TU-BcX-2 K1-derived cells to anti-neoplastic oncology drugs was compared in adherent cells and mammospheres. Drug response was evaluated using a live/dead staining kit and crystal violet staining. RESULTS: TU-BcX-2 K1 has a low propensity for metastasis, reflects a mesenchymal state, and contains a large burden of cancer stem cells. We show that TU-BcX-2 K1 cells have differential responses to cytotoxic and targeted therapies in 2D compared to 3D culture conditions insofar as several drug classes conferred sensitivity in 2D but not in 3D culture, or cells grown as mammospheres. CONCLUSIONS: Here we introduce a new TNBC PDX model and demonstrate the differences in evaluating drug sensitivity in adherent cells compared to mammosphere, or suspension, culture.

Potential systematic uncertainties in IGRT when FBCT reference images are used for pancreatic tumors.

The purpose of this study was to quantify the systematic uncertainties resulting from using free breathing computed tomography (FBCT) as a reference image for image-guided radiation therapy (IGRT) for patients with pancreatic tumors, and to quantify the associated dosimetric impact that resulted from using FBCT as reference for IGRT. Fifteen patients with implanted fiducial markers were selected for this study. For each patient, a FBCT and an average intensity projection computed tomography (AIP) created from four-dimensional computed tomography (4D CT) were acquired at the simulation. The treatment plan was created based on the FBCT. Seventy-five weekly kilovoltage (kV) cone-beam computed tomography (CBCT) images (five for each patient) were selected for this study. Bony alignment without rotation correction was performed 1) between the FBCT and CBCT, 2) between the AIP and CBCT, and 3) between the AIP and FBCT. The contours of the fiducials from the FBCT and AIP were transferred to the corresponding CBCT and were compared. Among the 75 CBCTs, 20 that had > 3 mm differences in centers of mass (COMs) in any directions between the FBCT and AIP were chosen for further dosimetric analysis. These COM discrepancies were converted into isocenter shifts in the corresponding planning FBCT, and dose was recalculated and compared to the initial FBCT plans. For the 75 CBCTs studied, the mean absolute differences in the COMs of the fiducial markers between the FBCT and CBCTs were 3.3 mm ± 2.5 mm, 3.5 mm ± 2.4 mm, and 5.8 mm ± 4.4 mm in the right-left (RL), anterior-posterior (AP), and superior-inferior (SI) directions, respectively. Between the AIP and CBCTs, the mean absolute differences were 3.2 mm ± 2.2mm, 3.3 mm ± 2.3 mm, and 6.3 mm ± 5.4 mm. The absolute mean discrepancies in these COMs shifts between FBCT/CBCT and AIP/CBCT were 1.1 mm ± 0.8 mm, 1.3 mm ± 0.9 mm, and 3.3 mm ± 2.6 mm in RL, AP, and SI, respectively. This represented a potential systematic error. For the 20 CBCTs that had COM discrepancies > 3 mm in any direction, the average reduction in planning target volume (PTV) coverage (PTV volume receiving 100% of prescription dose) was 5.3% ± 3.1% (range: 0.7%-12.8%). Using FBCT as a reference for IGRT may introduce potential interfractional systematic COM shifts if the FBCT is acquired at a different breathing phase than the average breathing phase. The potential systematic error could be significant in the SI direction and varied among patients for the other directions. AIP is a better choice of reference image set for IGRT in order to correct interfractional variations due to respiratory motion and nonrespiratory organ displacement.

Quality of Large Language Model Responses to Radiation Oncology Patient Care Questions.

Importance: Artificial intelligence (AI) large language models (LLMs) demonstrate potential in simulating human-like dialogue. Their efficacy in accurate patient-clinician communication within radiation oncology has yet to be explored. Objective: To determine an LLM's quality of responses to radiation oncology patient care questions using both domain-specific expertise and domain-agnostic metrics. Design, Setting, and Participants: This cross-sectional study retrieved questions and answers from websites (accessed February 1 to March 20, 2023) affiliated with the National Cancer Institute and the Radiological Society of North America. These questions were used as queries for an AI LLM, ChatGPT version 3.5 (accessed February 20 to April 20, 2023), to prompt LLM-generated responses. Three radiation oncologists and 3 radiation physicists ranked the LLM-generated responses for relative factual correctness, relative completeness, and relative conciseness compared with online expert answers. Statistical analysis was performed from July to October 2023. Main Outcomes and Measures: The LLM's responses were ranked by experts using domain-specific metrics such as relative correctness, conciseness, completeness, and potential harm compared with online expert answers on a 5-point Likert scale. Domain-agnostic metrics encompassing cosine similarity scores, readability scores, word count, lexicon, and syllable counts were computed as independent quality checks for LLM-generated responses. Results: Of the 115 radiation oncology questions retrieved from 4 professional society websites, the LLM performed the same or better in 108 responses (94%) for relative correctness, 89 responses (77%) for completeness, and 105 responses (91%) for conciseness compared with expert answers. Only 2 LLM responses were ranked as having potential harm. The mean (SD) readability consensus score for expert answers was 10.63 (3.17) vs 13.64 (2.22) for LLM answers (P < .001), indicating 10th grade and college reading levels, respectively. The mean (SD) number of syllables was 327.35 (277.15) for expert vs 376.21 (107.89) for LLM answers (P = .07), the mean (SD) word count was 226.33 (191.92) for expert vs 246.26 (69.36) for LLM answers (P = .27), and the mean (SD) lexicon score was 200.15 (171.28) for expert vs 219.10 (61.59) for LLM answers (P = .24). Conclusions and Relevance: In this cross-sectional study, the LLM generated accurate, comprehensive, and concise responses with minimal risk of harm, using language similar to human experts but at a higher reading level. These findings suggest the LLM's potential, with some retraining, as a valuable resource for patient queries in radiation oncology and other medical fields.

Determining risk and predictors of head and neck cancer treatment-related lymphedema: A clinicopathologic and dosimetric data mining approach using interpretable machine learning and ensemble feature selection.

Background and purpose: The ability to determine the risk and predictors of lymphedema is vital in improving the quality of life for head and neck (HN) cancer patients. However, selecting robust features is challenging due to the multicollinearity and high dimensionality of radiotherapy (RT) data. This study aims to overcome these challenges using an ensemble feature selection technique with machine learning (ML). Materials and methods: Thirty organs-at-risk, including bilateral cervical lymph node levels, were contoured, and dose-volume data were extracted from 76 HN treatment plans. Clinicopathologic data was collected. Ensemble feature selection was used to reduce the number of features. Using the reduced features as input to ML and competing risk models, internal and external lymphedema prediction capability was evaluated with the ML models, and time to lymphedema event and risk stratification were estimated using the risk models. Results: Two ML models, XGBoost and random forest, exhibited robust prediction performance. They achieved average F1-scores and AUCs of 84 ± 3.3 % and 79 ± 11.9 % (external lymphedema), and 64 ± 12 % and 78 ± 7.9 % (internal lymphedema). Predictive ML and risk models identified common predictors, including bulky node involvement, high dose to various lymph node levels, and lymph nodes removed during surgery. At 180 days, removing 0-25, 26-50, and > 50 lymph nodes increased external lymphedema risk to 72.1 %, 95.6 %, and 57.7 % respectively (p = 0.01). Conclusion: Our approach, involving the reduction of HN RT data dimensionality, resulted in effective ML models for HN lymphedema prediction. Predictive dosimetric features emerged from both predictive and competing risk models. Consistency with clinicopathologic features from other studies supports our methodology.

An Efficient Multi-Scale Fusion Network for 3D Organs at Risk (OARs) Segmentation.

Accurate segmentation of organs-at-risks (OARs) is a precursor for optimizing radiation therapy planning. Existing deep learning-based multi-scale fusion architectures have demonstrated a tremendous capacity for 2D medical image segmentation. The key to their success is aggregating global context and maintaining high resolution representations. However, when translated into 3D segmentation problems, existing multi-scale fusion architectures might underperform due to their heavy computation overhead and substantial data diet. To address this issue, we propose a new OAR segmentation framework, called OARFocalFuseNet, which fuses multi-scale features and employs focal modulation for capturing global-local context across multiple scales. Each resolution stream is enriched with features from different resolution scales, and multi-scale information is aggregated to model diverse contextual ranges. As a result, feature representations are further boosted. The comprehensive comparisons in our experimental setup with OAR segmentation as well as multi-organ segmentation show that our proposed OARFocalFuseNet outperforms the recent state-of-the-art methods on publicly available OpenKBP datasets and Synapse multi-organ segmentation. Both of the proposed methods (3D-MSF and OARFocalFuseNet) showed promising performance in terms of standard evaluation metrics. Our best performing method (OARFocalFuseNet) obtained a dice coefficient of 0.7995 and hausdorff distance of 5.1435 on OpenKBP datasets and dice coefficient of 0.8137 on Synapse multi-organ segmentation dataset. Our code is available at https://github.com/NoviceMAn-prog/OARFocalFuse.

Image-Based Deep Neural Network for Individualizing Radiotherapy Dose Is Transportable Across Health Systems.

PURPOSE: We developed a deep neural network that queries the lung computed tomography-derived feature space to identify radiation sensitivity parameters that can predict treatment failures and hence guide the individualization of radiotherapy dose. In this article, we examine the transportability of this model across health systems. METHODS: This multicenter cohort-based registry included 1,120 patients with cancer in the lung treated with stereotactic body radiotherapy. Pretherapy lung computed tomography images from the internal study cohort (n = 849) were input into a multitask deep neural network to generate an image fingerprint score that predicts time to local failure. Deep learning (DL) scores were input into a regression model to derive iGray, an individualized radiation dose estimate that projects a treatment failure probability of < 5% at 24 months. We validated our findings in an external, holdout cohort (n = 271). RESULTS: There were substantive differences in the baseline patient characteristics of the two study populations, permitting an assessment of model transportability. In the external cohort, radiation treatments in patients with high DL scores failed at a significantly higher rate with 3-year cumulative incidences of local failure of 28.5% (95% CI, 19.8 to 37.8) versus 10.2% (95% CI, 5.9 to 16.2; hazard ratio, 3.3 [95% CI, 1.74 to 6.49]; P < .001). A model that included DL score alone predicted treatment failures with a concordance index of 0.68 (95% CI, 0.59 to 0.77), which had a similar performance to a nested model derived from within the internal cohort (0.70 [0.64 to 0.75]). External cohort patients with iGray values that exceeded the delivered doses had proportionately higher rates of local failure (P < .001). CONCLUSION: Our results support the development and implementation of new DL-guided treatment guidance tools in the image-replete and highly standardized discipline of radiation oncology.

Dose prescription and reporting in stereotactic body radiotherapy: A multi-institutional study.

BACKGROUND AND PURPOSE: Radiation dose prescriptions are foundational for optimizing treatment efficacy and limiting treatment-related toxicity. We sought to assess the lack of standardization of SBRT dose prescriptions across institutions. MATERIALS & METHODS: Dosimetric data from 1298 patients from 9 academic institutions treated with IMRT and VMAT were collected. Dose parameters D100, D98, D95, D50, and D2 were used to assess dosimetric variability. RESULTS: Disease sites included lung (48.3 %) followed by liver (29.7 %), prostate (7.5 %), spine (6.8 %), brain (4.1 %), and pancreas (2.5 %). The PTV volume in lung varied widely with bimodality into two main groups (22.0-28.7 cm3) and (48.0-67.1 cm3). A hot spot ranging from 120-150 % was noted in nearly half of the patients, with significant variation across institutions. A D50 ≥ 110 % was found in nearly half of the institutions. There was significant dosimetric variation across institutions. CONCLUSIONS: The SBRT prescriptions in the literature or in treatment guidelines currently lack nuance and hence there is significant variation in dose prescriptions across academic institutions. These findings add greater importance to the identification of dose parameters associated with improved clinical outcome comparisons as we move towards more hypofractionated treatments. There is a need for standardized reporting to help institutions in adapting treatment protocols based on the outcome of clinical trials. Dosimetric parameters are subsequently needed for uniformity and thereby standardizing planning guidelines to maximize efficacy, mitigate toxicity, and reduce treatment disparities are urgently needed.

Multivalent state transitions shape the intratumoral composition of small cell lung carcinoma.

Studies to date have not resolved how diverse transcriptional programs contribute to the intratumoral heterogeneity of small cell lung carcinoma (SCLC), an aggressive tumor associated with a dismal prognosis. Here, we identify distinct and commutable transcriptional states that confer discrete functional attributes in individual SCLC tumors. We combine an integrative approach comprising the transcriptomes of 52,975 single cells, high-resolution measurement of cell state dynamics at the single-cell level, and functional and correlative studies using treatment naïve xenografts with associated clinical outcomes. We show that individual SCLC tumors contain distinctive proportions of stable cellular states that are governed by bidirectional cell state transitions. Using drugs that target the epigenome, we reconfigure tumor state composition in part by altering individual state transition rates. Our results reveal new insights into how single-cell transition behaviors promote cell state equilibrium in SCLC and suggest that facile plasticity underlies its resistance to therapy and lethality.

The Mutational Landscape of Cancer's Vulnerability to Ionizing Radiation.

PURPOSE: Large-scale sequencing efforts have established that cancer-associated genetic alterations are highly diverse, posing a challenge to the identification of variants that regulate complex phenotypes like radiation sensitivity. The impact of the vast majority of rare or common genetic variants on the sensitivity of cancers to radiotherapy remains largely unknown. EXPERIMENTAL DESIGN: We developed a scalable gene editing and irradiation platform to assess the role of categories of variants in cells. Variants were prioritized on the basis of genotype-phenotype associations from a previously completed large-scale cancer cell line radiation profiling study. Altogether, 488 alleles (396 unique single-nucleotide variants) from 92 genes were generated and profiled in an immortalized lung cell line, BEAS-2B. We validated our results in other cell lines (TRT-HU1 and NCI-H520), in vivo via the use of both cell line and patient-derived murine xenografts, and in clinical cohorts. RESULTS: We show that resistance to radiation is characterized by substantial inter- and intra-gene allelic variation. Some genes (e.g., KEAP1) demonstrated significant intragenic allelic variation in the magnitude of conferred resistance and other genes (e.g., CTNNB1) displayed both resistance and sensitivity in a protein domain-dependent manner. We combined results from our platform with gene expression and metabolite features and identified the upregulation of amino acid transporters that facilitate oxidative reductive capacity and cell-cycle deregulation as key regulators of radiation sensitivity. CONCLUSIONS: Our results reveal new insights into the genetic determinants of tumor sensitivity to radiotherapy and nominate a multitude of cancer mutations that are predicted to impact treatment efficacy.

Deterministic small-scale undulations of image-based risk predictions from the deep learning of lung tumors in motion.

INTRODUCTION: Deep learning (DL) models that use medical images to predict clinical outcomes are poised for clinical translation. For tumors that reside in organs that move, however, the impact of motion (i.e., degenerated object appearance or blur) on DL model accuracy remains unclear. We examine the impact of tumor motion on an image-based DL framework that predicts local failure risk after lung stereotactic body radiotherapy (SBRT). METHODS: We input pre-therapy free breathing (FB) computed tomography (CT) images from 849 patients treated with lung SBRT into a multitask deep neural network to generate an image fingerprint signature (or DL score) that predicts time-to-event local failure outcomes. The network includes a convolutional neural network encoder for extracting imaging features and building a task-specific fingerprint, a decoder for estimating handcrafted radiomic features, and a task-specific network for generating image signature for radiotherapy outcome prediction. The impact of tumor motion on the DL scores was then examined for a holdout set of 468 images from 39 patients comprising: (1) FB CT, (2) four-dimensional (4D) CT, and (3) maximum-intensity projection (MIP) images. Tumor motion was estimated using a 3D vector of the maximum distance traveled, and its association with DL score variance was assessed by linear regression. FINDINGS: The variance and amplitude in 4D CT image-derived DL scores were associated with tumor motion (R2  = 0.48 and 0.46, respectively). Specifically, DL score variance was deterministic and represented by sinusoidal undulations in phase with the respiratory cycle. DL scores, but not tumor volumes, peaked near end-exhalation. The mean of the scores derived from 4D CT images and the score obtained from FB CT images were highly associated (Pearson r = 0.99). MIP-derived DL scores were significantly higher than 4D- or FB-derived risk scores (p < 0.0001). INTERPRETATION: An image-based DL risk score derived from a series of 4D CT images varies in a deterministic, sinusoidal trajectory in a phase with the respiratory cycle. These results indicate that DL models of tumors in motion can be robust to fluctuations in object appearance due to movement and can guide standardization processes in the clinical translation of DL models for patients with lung cancer.

Fractionated Proton Radiation Therapy and Hearing Preservation for Vestibular Schwannoma: Preliminary Analysis of a Prospective Phase 2 Clinical Trial.

BACKGROUND: Local management for vestibular schwannoma (VS) is associated with excellent local control with focus on preserving long-term serviceable hearing. Fractionated proton radiation therapy (FPRT) may be associated with greater hearing preservation because of unique dosimetric properties of proton radiotherapy. OBJECTIVE: To investigate hearing preservation rates of FPRT in adults with VS and secondarily assess local control and treatment-related toxicity. METHODS: A prospective, single-arm, phase 2 clinical trial was conducted of patients with VS from 2010 to 2019. All patients had serviceable hearing at baseline and received FPRT to a total dose of 50.4 to 54 Gy relative biological effectiveness (RBE) over 28 to 30 fractions. Serviceable hearing preservation was defined as a Gardner-Robertson score of 1 to 2, measured by a pure tone average (PTA) of ≤50 dB and a word recognition score (WRS) of ≥50%. RESULTS: Twenty patients had a median follow-up of 4.0 years (range 1.0-5.0 years). Local control at 4 years was 100%. Serviceable hearing preservation at 1 year was 53% (95% CI 29%-76%), and primary end point was not yet reached. Median PTA and median WRS both worsened 1 year after FPRT (P < .0001). WRS plateaued after 6 months, whereas PTA continued to worsen up to 1 year after FPRT. Median cochlea D90 was lower in patients with serviceable hearing at 1 year (40.6 Gy [RBE] vs 46.9 Gy [RBE]), trending toward Wilcoxon rank-sum test statistical significance (P = .0863). Treatment was well-tolerated, with one grade 1 cranial nerve V dysfunction and no grade 2+ cranial nerve dysfunction. CONCLUSION: FPRT for VS did not meet the goal of serviceable hearing preservation. Higher cochlea doses trended to worsening hearing preservation, suggesting that dose to cochlea correlates with hearing preservation independent of treatment modality.

Targeting the Epidermal Growth Factor Receptor in EGFR-Mutated Lung Cancer: Current and Emerging Therapies.

Epidermal growth factor receptor-targeting tyrosine kinase inhibitors (EGFR TKIs) are the standard of care for patients with EGFR-mutated metastatic lung cancer. While EGFR TKIs have initially high response rates, inherent and acquired resistance constitute a major challenge to the longitudinal treatment. Ongoing work is aimed at understanding the molecular basis of these resistance mechanisms, with exciting new studies evaluating novel agents and combination therapies to improve control of tumors with all forms of EGFR mutation. In this review, we first provide a discussion of EGFR-mutated lung cancer and the efficacy of available EGFR TKIs in the clinical setting against both common and rare EGFR mutations. Second, we discuss common resistance mechanisms that lead to therapy failure during treatment with EGFR TKIs. Third, we review novel approaches aimed at improving outcomes and overcoming resistance to EGFR TKIs. Finally, we highlight recent breakthroughs in the use of EGFR TKIs in non-metastatic EGFR-mutated lung cancer.

A Therapeutic Strategy for Preferential Targeting of TET2 Mutant and TET-dioxygenase Deficient Cells in Myeloid Neoplasms.

TET2 is frequently mutated in myeloid neoplasms. Genetic TET2 deficiency leads to skewed myeloid differentiation and clonal expansion, but minimal residual TET activity is critical for survival of neoplastic progenitor and stem cells. Consistent with mutual exclusivity of TET2 and neomorphic IDH1/2 mutations, here we report that IDH1/2 mutant-derived 2-hydroxyglutarate is synthetically lethal to TET-dioxygenase deficient cells. In addition, a TET-selective small molecule inhibitor decreased cytosine hydroxymethylation and restricted clonal outgrowth of TET2 mutant, but not normal hematopoietic precursor cells in vitro and in vivo. While TET-inhibitor phenocopied somatic TET2 mutations, its pharmacologic effects on normal stem cells were, unlike mutations, reversible. Treatment with TET inhibitor suppressed the clonal evolution of TET2 mutant cells in murine models and TET2-mutated human leukemia xenografts. These results suggest that TET inhibitors may constitute a new class of targeted agents in TET2 mutant neoplasia.

Therapy-Induced Senescence: Opportunities to Improve Anticancer Therapy.

Cellular senescence is an essential tumor suppressive mechanism that prevents the propagation of oncogenically activated, genetically unstable, and/or damaged cells. Induction of tumor cell senescence is also one of the underlying mechanisms by which cancer therapies exert antitumor activity. However, an increasing body of evidence from preclinical studies demonstrates that radiation and chemotherapy cause accumulation of senescent cells (SnCs) both in tumor and normal tissue. SnCs in tumors can, paradoxically, promote tumor relapse, metastasis, and resistance to therapy, in part, through expression of the senescence-associated secretory phenotype. In addition, SnCs in normal tissue can contribute to certain radiation- and chemotherapy-induced side effects. Because of its multiple roles, cellular senescence could serve as an important target in the fight against cancer. This commentary provides a summary of the discussion at the National Cancer Institute Workshop on Radiation, Senescence, and Cancer (August 10-11, 2020, National Cancer Institute, Bethesda, MD) regarding the current status of senescence research, heterogeneity of therapy-induced senescence, current status of senotherapeutics and molecular biomarkers, a concept of "one-two punch" cancer therapy (consisting of therapeutics to induce tumor cell senescence followed by selective clearance of SnCs), and its integration with personalized adaptive tumor therapy. It also identifies key knowledge gaps and outlines future directions in this emerging field to improve treatment outcomes for cancer patients.

RNA expression and risk of venous thromboembolism in lung cancer.

BACKGROUND: The propensity to develop venous thromboembolism (VTE) on the basis of individual tumor biological features remains unknown. OBJECTIVES: We conducted a whole transcriptome RNA sequencing strategy, focusing on a single cancer type (lung cancer), to identify biomarkers of cancer-associated VTE. METHODS: Twelve propensity-matched patients, 6 each with or without VTE, were identified from a prospective institutional review board-approved registry at the Cleveland Clinic with available tissue from surgical excision of a primary lung mass between 2010 and 2015. Patients were propensity matched based on age, sex, race, history of prior cancer, date of cancer diagnosis, stage, histology, number of lines of chemotherapy, and length of follow-up. RNA sequencing was performed on tumor tissue, and gene set enrichment analysis (GSEA) was performed on differentially expressed genes. RESULTS: We identified 1037 genes with differential expression. In patients with VTE, 869 genes were overexpressed and 168 were underexpressed compared to patients without VTE. Of these, 276 overexpressed and 35 underexpressed were significantly different (Q < 0.05). GSEA revealed upregulation of genes in complement, inflammation, and KRAS signaling pathways in tumors from patients with VTE. CONCLUSIONS: These differentially expressed genes and associated pathways provide biologic insights into cancer-associated VTE and may provide insignts to develop new risk stratification schemes, prevention, or treatment strategies.

Resistance to targeted therapies as a multifactorial, gradual adaptation to inhibitor specific selective pressures.

Despite high initial efficacy, targeted therapies eventually fail in advanced cancers, as tumors develop resistance and relapse. In contrast to the substantial body of research on the molecular mechanisms of resistance, understanding of how resistance evolves remains limited. Using an experimental model of ALK positive NSCLC, we explored the evolution of resistance to different clinical ALK inhibitors. We found that resistance can originate from heterogeneous, weakly resistant subpopulations with variable sensitivity to different ALK inhibitors. Instead of the commonly assumed stochastic single hit (epi) mutational transition, or drug-induced reprogramming, we found evidence for a hybrid scenario involving the gradual, multifactorial adaptation to the inhibitors through acquisition of multiple cooperating genetic and epigenetic adaptive changes. Additionally, we found that during this adaptation tumor cells might present unique, temporally restricted collateral sensitivities, absent in therapy naïve or fully resistant cells, suggesting the potential for new therapeutic interventions, directed against evolving resistance.

An image-based deep learning framework for individualizing radiotherapy dose.

Background: Radiotherapy continues to be delivered uniformly without consideration of individual tumor characteristics. To advance toward more precise treatments in radiotherapy, we queried the lung computed tomography (CT)-derived feature space to identify radiation sensitivity parameters that can predict treatment failure and hence guide the individualization of radiotherapy dose. Methods: We used a cohort-based registry of 849 patients with cancer in the lung treated with high dose radiotherapy using stereotactic body radiotherapy. We input pre-therapy lung CT images into a multi-task deep neural network, Deep Profiler, to generate an image fingerprint that primarily predicts time to event treatment outcomes and secondarily approximates classical radiomic features. We validated our findings in an independent study population (n = 95). Deep Profiler was combined with clinical variables to derive iGray, an individualized dose that estimates treatment failure probability to be <5%. Findings: Radiation treatments in patients with high Deep Profiler scores fail at a significantly higher rate than in those with low scores. The 3-year cumulative incidences of local failure were 20.3% (95% CI: 16.0-24.9) and 5.7% (95% CI: 3.5-8.8), respectively. Deep Profiler independently predicted local failure (hazard ratio 1.65, 95% 1.02-2.66, p = 0.04). Models that included Deep Profiler and clinical variables predicted treatment failures with a concordance index of 0.72 (95% CI: 0.67-0.77), a significant improvement compared to classical radiomics or clinical variables alone (p = <0.001 and <0.001, respectively). Deep Profiler performed well in an external study population (n = 95), accurately predicting treatment failures across diverse clinical settings and CT scanner types (concordance index = 0.77 [95% CI: 0.69-0.92]). iGray had a wide dose range (21.1-277 Gy, BED), suggested dose reductions in 23.3% of patients and can be safely delivered in the majority of cases. Interpretation: Our results indicate that there are image-distinct subpopulations that have differential sensitivity to radiotherapy. The image-based deep learning framework proposed herein is the first opportunity to use medical images to individualize radiotherapy dose.

Cellular and Genetic Determinants of the Sensitivity of Cancer to α-Particle Irradiation.

Targeted α-particle-emitting radionuclides have great potential for the treatment of a broad range of cancers at different stages of progression. A platform that accurately measures cancer cellular sensitivity to α-particle irradiation could guide and accelerate clinical translation. Here, we performed high-content profiling of cellular survival following exposure to α-particles emitted from radium-223 (223Ra) using 28 genetically diverse human tumor cell lines. Significant variation in cellular sensitivity across tumor cells was observed. 223Ra was significantly more potent than sparsely ionizing irradiation, with a median relative biological effectiveness of 10.4 (IQR: 8.4-14.3). Cells that are the most resistant to γ radiation, such as Nrf2 gain-of-function mutant cells, were sensitive to α-particles. Combining these profiling results with genetic features, we identified several somatic copy-number alterations, gene mutations, and the basal expression of gene sets that correlated with radiation survival. Activating mutations in PIK3CA, a frequent event in cancer, decreased sensitivity to 223Ra. The identification of cellular and genetic determinants of sensitivity to 223Ra may guide the clinical incorporation of targeted α-particle emitters in the treatment of several cancer types. SIGNIFICANCE: These findings address limitations in the preclinical guidance and prediction of radionuclide tumor sensitivity by identifying intrinsic cellular and genetic determinants of cancer cell survival following exposure to α-particle irradiation.See related commentary by Sgouros, p. 5479.