Framework for Radiation Oncology Department-wide Evaluation and Implementation of Commercial Artificial Intelligence Autocontouring.

PURPOSE: Artificial intelligence (AI)-based autocontouring in radiation oncology has potential benefits such as standardization and time savings. However, commercial AI solutions require careful evaluation before clinical integration. We developed a multidimensional evaluation method to test pretrained AI-based automated contouring solutions across a network of clinics. METHODS AND MATERIALS: Curated data included 121 patient planning computed tomography (CT) scans with a total of 859 clinically approved contours used for treatment from 4 clinics. Regions of interest (ROIs) were generated with 3 commercial AI-based automated contouring software solutions (AI1, AI2, AI3) spanning the following disease sites: brain, head and neck (H&N), thorax, abdomen, and pelvis. Quantitative agreement between AI-generated and clinical contours was measured by Dice similarity coefficient (DSC) and Hausdorff distance (HD). Qualitative assessment was performed by multiple experts scoring blinded AI-contours using a Likert scale. Workflow and usability surveying was also conducted. RESULTS: AI1, AI2, and AI3 contours had high quantitative agreement in 27.8%, 32.8%, and 34.1% of cases (DSC >0.9), performing well in pelvis (median DSC = 0.86/0.88/0.91) and thorax (median DSC = 0.91/0.89/0.91). All 3 solutions had low quantitative agreement in 7.4%, 8.8%, and 6.1% of cases (DSC <0.5), performing worse in brain (median DSC = 0.65/0.78/0.75) and H&N (median DSC = 0.76/0.80/0.81). Qualitatively, AI1 and AI2 contours were acceptable (rated 1-2) with at most minor edits in 70.7% and 74.6% of ROIs (2906 ratings), higher for abdomen (AI1: 79.2%) and thorax (AI2: 90.2%), and lower for H&N (29.0/35.6%). An end-user survey showed strong user preference for full automation and mixed preferences for accuracy versus total number of structures generated. CONCLUSIONS: Our evaluation method provided a comprehensive analysis of both quantitative and qualitative measures of commercially available pretrained AI autocontouring algorithms. The evaluation framework served as a roadmap for clinical integration that aligned with user workflow preference.

Integrating Artificial Intelligence and Machine Learning Into Cancer Clinical Trials.

The practice of oncology requires analyzing and synthesizing abundant data. From the patient's workup to determine eligibility to the therapies received to the post-treatment surveillance, practitioners must constantly juggle, evaluate, and weigh decision-making based on their best understanding of information at hand. These complex, multifactorial decisions have a tremendous opportunity to benefit from data-driven machine learning (ML) methods to drive opportunities in artificial intelligence (AI). Within the past 5 years, we have seen AI move from simply a promising opportunity to being used in prospective trials. Here, we review recent efforts of AI in clinical trials that have moved the needle towards improved prediction of actionable outcomes, such as predicting acute care visits, short term mortality, and pathologic extranodal extension. We then pause and reflect on how these AI models ask a different question than traditional statistics models that readers may be more familiar with; how then should readers conceptualize and interpret AI models that they are not as familiar with. We end with what we believe are promising future opportunities for AI in oncology, with an eye towards allowing the data to inform us through unsupervised learning and generative models, rather than asking AI to perform specific functions.

Predicted Inferior Outcomes for Lung SBRT With Treatment Planning Systems That Fail Independent Phantom-Based Audits.

PURPOSE: More than 15% of radiation therapy clinics fail external audits with anthropomorphic phantoms conducted by Imaging and Radiation Oncology Core-Houston (IROC-H) while passing other industry-standard quality assurance (QA) tests. We seek to evaluate the predicted effect of such failed plans on outcomes for patients treated with stereotactic body radiation therapy (SBRT) for lung tumors. METHODS AND MATERIALS: We conducted a retrospective study of 55 patients treated with SBRT for lung tumors with a prescription biologically equivalent dose (BED)10 ≥100 Gy using a treatment planning system (TPS) that passed IROC-H phantom audits. Sample linear accelerator beam models with introduced errors were commissioned by varying the multileaf collimator leaf-tip offset parameter (ie, dosimetric leaf gap) over the range ±1.0 mm relative to the validated model. These models mimic TPS that pass internal QA measures but fail IROC-H tests. Patient plans were recalculated on sample beam models. The predicted tumor control probability (TCP) and normal tissue complication probability (NTCP) were calculated based on published dose-response models. RESULTS: A leaf-tip offset value of -1.0 mm decreased the fraction of plans receiving a planning treatment volume of BED10 ≥100 Gy from 95% to 27%. This translated to a significant decrease in 2-year TCP of 4.8% (95% CI: 2.0%-5.5%) with a decrease in TCP up to 21%. Conversely, a leaf-tip offset of +1.0 mm resulted in 36% of patients exceeding previously met organs at risk (OAR) constraints, including 2 instances of spinal cord and brachial plexus overdoses and a small increase in chest wall NTCP of 0.7%, (95% CI: 0.5%-0.8%). CONCLUSIONS: Simulated treatment plans with modest MLC leaf offsets result in lung SBRT plans that significantly underdose tumor or exceed OAR constraints. These dosimetric endpoints translate to significant detriments in TCP. These simulated plans mimic planning systems that pass internal QA measures but fail independent phantom-based tests, underscoring the need for enhanced quality assurance including external audits of TPS.

Techniques for characterizing mechanical properties of soft tissues.

The characterization of soft tissues remains a vital need for various bioengineering and medical fields. Developing areas such as regenerative medicine, robot-aided surgery, and surgical simulations all require accurate knowledge about the mechanical properties of soft tissues to replicate their mechanics. Mechanical properties can be characterized through several different characterization techniques such as atomic force microscopy, compression testing, and tensile testing. However, many of these methods contain considerable differences in ability to accurately characterize the mechanical properties of soft tissues. As a result of these variations, there are often discrepancies in the reported values for numerous studies. This paper reviews common characterization methods that have been applied to obtain the mechanical properties of soft tissues and highlights their advantages as well as disadvantages. The limitations, accuracies, repeatability, in-vivo testing capability, and types of properties measurable for each method are also discussed.

Evaluation of the TruSight Oncology 500 Assay for Routine Clinical Testing of Tumor Mutational Burden and Clinical Utility for Predicting Response to Pembrolizumab.

Pembrolizumab is approved for treating patients with unresectable or metastatic solid tumors with high tumor mutational burden (TMB), as assessed by the Food and Drug Administration-approved companion diagnostic FoundationOneCDx, after progression on prior treatment. To expand TMB assessment for enriching response to pembrolizumab, TMB measurement from TruSight Oncology 500 (TSO500) was evaluated in archival pan-tumor samples from 294 patients enrolled in eight pembrolizumab monotherapy studies. TSO500 is a panel-based next-generation sequencing assay with broad availability, quick turnaround time, and a standardized bioinformatics pipeline. TSO500 TMB was evaluated for correlation and concordance with two reference methods: FoundationOneCDx and whole-exome sequencing. The TSO500 cut-off for TMB-high was selected based on the receiver-operating characteristic curve analysis against each reference method's cut-off for TMB-high. Clinical utility of the selected TSO500 cut-off (10 mutations/Mb) was assessed by calculating the sensitivity, specificity, positive and negative predictive values, and objective response rate enrichment. There was high correlation and concordance of TSO500 TMB with both reference methods. TSO500 TMB was associated significantly with the best overall response, and the selected cut-off had comparable clinical utility with respective cut-offs for the reference methods in predicting response to pembrolizumab. As a commercial assay with global kit distribution complete with an off-the-shelf software package, TSO500 may provide additional access to immunotherapy for patients with tumors with TMB ≥10 mutations/Mb.

National Cancer Institute Workshop on Artificial Intelligence in Radiation Oncology: Training the Next Generation.

PURPOSE: Artificial intelligence (AI) is about to touch every aspect of radiation therapy, from consultation to treatment planning, quality assurance, therapy delivery, and outcomes modeling. There is an urgent need to train radiation oncologists and medical physicists in data science to help shepherd AI solutions into clinical practice. Poorly trained personnel may do more harm than good when attempting to apply rapidly developing and complex technologies. As the amount of AI research expands in our field, the radiation oncology community needs to discuss how to educate future generations in this area. METHODS AND MATERIALS: The National Cancer Institute (NCI) Workshop on AI in Radiation Oncology (Shady Grove, MD, April 4-5, 2019) was the first of 2 data science workshops in radiation oncology hosted by the NCI in 2019. During this workshop, the Training and Education Working Group was formed by volunteers among the invited attendees. Its members represent radiation oncology, medical physics, radiology, computer science, industry, and the NCI. RESULTS: In this perspective article written by members of the Training and Education Working Group, we provide and discuss action points relevant for future trainees interested in radiation oncology AI: (1) creating AI awareness and responsible conduct; (2) implementing a practical didactic curriculum; (3) creating a publicly available database of training resources; and (4) accelerating learning and funding opportunities. CONCLUSION: Together, these action points can facilitate the translation of AI into clinical practice.

Genomics models in radiotherapy: From mechanistic to machine learning.

Machine learning (ML) provides a broad framework for addressing high-dimensional prediction problems in classification and regression. While ML is often applied for imaging problems in medical physics, there are many efforts to apply these principles to biological data toward questions of radiation biology. Here, we provide a review of radiogenomics modeling frameworks and efforts toward genomically guided radiotherapy. We first discuss medical oncology efforts to develop precision biomarkers. We next discuss similar efforts to create clinical assays for normal tissue or tumor radiosensitivity. We then discuss modeling frameworks for radiosensitivity and the evolution of ML to create predictive models for radiogenomics.

Stereotactic body radiotherapy in patients with multiple lung tumors: a focus on lung dosimetric constraints.

Introduction: Lung dosimetric constraints with stereotactic body/ablative radiotherapy (SBRT/SABR) for multiple lung lesions are not well-characterized in published literature. Classically, the lung is considered a 'parallel' organ, for which injury to functional subunits could result in partially compromised function of that organ/tissue. Therefore, with SBRT/SABR for >1 thoracic target (especially involving both lungs), lung dosimetry requires special consideration.Areas covered: Current cooperative group and multi-institutional studies of SBRT/SABR for oligometastases rely on lung constraints from expert opinion, including constraints of exposure (i.e., volume of lung receiving more than a threshold dose or mean lung dose) and/or critical volume (i.e. volume of lung receiving less than a threshold dose; also termed complementary volume). For radiation pneumonitis, which reflects inflammatory lung injury, it remains unclear which type of constraint is more predictive of toxicity risks.Expert opinion: With SBRT/SABR for multiple lung lesions, it is prudent to use both exposure and critical volume constraints. Treatment on alternate days (for radiation plans with separate treatment fields) or staging treatment may also lower lung toxicity risks. Further study on lung normal tissue complication probability in the setting of multiple lung targets is urgently needed, particularly analyses of critical volume metrics, which are relatively poorly studied.

Is It Coincidence or Consequence for a Case with Antiphospholipid Antibody Syndrome Overlapping SLE to Develop an Immune Complex Nephropathy Followed by a Nonimmune Complex Podocytopathy?

Antiphospholipid antibody syndrome (APS) may occur in a primary form or in association with SLE and seldom presents with nephrotic syndrome (NS). We present a case with APS who developed recurrent NS 6 years apart. The first episode of NS occurred with biopsy findings consistent with lupus nephritis (LN) class V (membranous) with no clear evidence of SLE, and responded to a remission with steroids and MMF. On the 2nd episode, the biopsy revealed negative immunofluorescent (IF) study for immune complexes and EM findings of complete effacement of foot processes and acellular debris in thickened capillary walls, compatible with healed previous episode of membranous LN and minimal change disease (MCD), a nonimmune complex podocytopathy. The 2nd episode responded to a partial remission, primarily with a short-term steroid therapy, and subsequently developed serologic evidence of SLE. Now there is growing evidence that a subset of SLE patients with NS are found to have MCD, likely due to podocyte injury caused by nonimmune complex pathway, called lupus podocytopathy. In LN, serial kidney biopsies often show transformation from one to another class of immune complex-induced glomerular lesions; however there are rare reports describing transformation of an immune complex to a nonimmune complex LN. Since the pathogenic mechanism of lupus podocytopathy is not delineated, and so far there are no reports on transformation of membranous LN, an immune complex nephropathy, to a nonimmune complex lupus podocytopathy, it still remains as a question whether our case with APS overlapping SLE had a concomitant membranous LN and lupus podocytopathy, or consequential membranous LN and lupus podocytopathy 6 years apart.

Machine Learning and Radiogenomics: Lessons Learned and Future Directions.

Due to the rapid increase in the availability of patient data, there is significant interest in precision medicine that could facilitate the development of a personalized treatment plan for each patient on an individual basis. Radiation oncology is particularly suited for predictive machine learning (ML) models due to the enormous amount of diagnostic data used as input and therapeutic data generated as output. An emerging field in precision radiation oncology that can take advantage of ML approaches is radiogenomics, which is the study of the impact of genomic variations on the sensitivity of normal and tumor tissue to radiation. Currently, patients undergoing radiotherapy are treated using uniform dose constraints specific to the tumor and surrounding normal tissues. This is suboptimal in many ways. First, the dose that can be delivered to the target volume may be insufficient for control but is constrained by the surrounding normal tissue, as dose escalation can lead to significant morbidity and rare. Second, two patients with nearly identical dose distributions can have substantially different acute and late toxicities, resulting in lengthy treatment breaks and suboptimal control, or chronic morbidities leading to poor quality of life. Despite significant advances in radiogenomics, the magnitude of the genetic contribution to radiation response far exceeds our current understanding of individual risk variants. In the field of genomics, ML methods are being used to extract harder-to-detect knowledge, but these methods have yet to fully penetrate radiogenomics. Hence, the goal of this publication is to provide an overview of ML as it applies to radiogenomics. We begin with a brief history of radiogenomics and its relationship to precision medicine. We then introduce ML and compare it to statistical hypothesis testing to reflect on shared lessons and to avoid common pitfalls. Current ML approaches to genome-wide association studies are examined. The application of ML specifically to radiogenomics is next presented. We end with important lessons for the proper integration of ML into radiogenomics.

Recombinant adeno-associated viral (rAAV) vectors mediate efficient gene transduction in cultured neonatal and adult microglia.

Microglia are a specialized population of myeloid cells that mediate CNS innate immune responses. Efforts to identify the cellular and molecular mechanisms that regulate microglia behaviors have been hampered by the lack of effective tools for manipulating gene expression. Cultured microglia are refractory to most chemical and electrical transfection methods, yielding little or no gene delivery and causing toxicity and/or inflammatory activation. Recombinant adeno-associated viral (rAAVs) vectors are non-enveloped, single-stranded DNA vectors commonly used to transduce many primary cell types and tissues. In this study, we evaluated the feasibility and efficiency of utilizing rAAV serotype 2 (rAAV2) to modulate gene expression in cultured microglia. rAAV2 yields high transduction and causes minimal toxicity or inflammatory response in both neonatal and adult microglia. To demonstrate that rAAV transduction can induce functional protein expression, we used rAAV2 expressing Cre recombinase to successfully excise a LoxP-flanked miR155 gene in cultured microglia. We further evaluated rAAV serotypes 5, 6, 8, and 9, and observed that all efficiently transduced cultured microglia to varying degrees of success and caused little or no alteration in inflammatory gene expression. These results provide strong encouragement for the application of rAAV-mediated gene expression in microglia for mechanistic and therapeutic purposes. Neonatal microglia are functionally distinct from adult microglia, although the majority of in vitro studies utilize rodent neonatal microglia cultures because of difficulties of culturing adult cells. In addition, cultured microglia are refractory to most methods for modifying gene expression. Here, we developed a novel protocol for culturing adult microglia and evaluated the feasibility and efficiency of utilizing Recombinant Adeno-Associated Virus (rAAV) to modulate gene expression in cultured microglia.

Machine Learning Approaches for Predicting Radiation Therapy Outcomes: A Clinician's Perspective.

Radiation oncology has always been deeply rooted in modeling, from the early days of isoeffect curves to the contemporary Quantitative Analysis of Normal Tissue Effects in the Clinic (QUANTEC) initiative. In recent years, medical modeling for both prognostic and therapeutic purposes has exploded thanks to increasing availability of electronic data and genomics. One promising direction that medical modeling is moving toward is adopting the same machine learning methods used by companies such as Google and Facebook to combat disease. Broadly defined, machine learning is a branch of computer science that deals with making predictions from complex data through statistical models. These methods serve to uncover patterns in data and are actively used in areas such as speech recognition, handwriting recognition, face recognition, "spam" filtering (junk email), and targeted advertising. Although multiple radiation oncology research groups have shown the value of applied machine learning (ML), clinical adoption has been slow due to the high barrier to understanding these complex models by clinicians. Here, we present a review of the use of ML to predict radiation therapy outcomes from the clinician's point of view with the hope that it lowers the "barrier to entry" for those without formal training in ML. We begin by describing 7 principles that one should consider when evaluating (or creating) an ML model in radiation oncology. We next introduce 3 popular ML methods--logistic regression (LR), support vector machine (SVM), and artificial neural network (ANN)--and critique 3 seminal papers in the context of these principles. Although current studies are in exploratory stages, the overall methodology has progressively matured, and the field is ready for larger-scale further investigation.

Cervical cancer outcome prediction to high-dose rate brachytherapy using quantitative magnetic resonance imaging analysis of tumor response to external beam radiotherapy.

BACKGROUND AND PURPOSE: In order to assess tumor regression and outcomes, a volumetric analysis was conducted for cervical cancer patients treated with magnetic resonance imaging (MRI)-based image-guided brachytherapy (IGBT). MATERIALS AND METHODS: Consecutive patients with FIGO stage IB1-IVA cervical cancer receiving chemoradiation from 2007 to 2013 were identified, excluding patients with perineal template-based interstitial brachytherapy or without undergoing MRI. A ring and tandem applicator±interstitial needles was used. T2-weighted imaging was completed following applicator insertion. Gross tumor volumes (GTVs) were retrospectively contoured: initial GTV (GTV(Pre-EBRT)), GTV at first brachytherapy (GTV(IGBT)) and percent residual GTV at first brachytherapy (% GTV(Residual)). RESULTS: Eighty-four patients were identified. With 20.8-month median follow-up, two-year estimates of local control (LC), disease-free survival (DFS) and overall survival (OS) were 91.3, 79.8, and 85.0%, respectively. Multivariate Cox regression revealed adenocarcinoma (HR 5.88, p=0.03) and GTV(IGBT) (HR 1.17, p<0.01) as predictors for local failure. GTV(IGBT)>7.5 cc was associated with inferior 2-year LC (75.0 vs. 96.6%, p<0.01), DFS (42.6 vs. 91.6%, p<0.01) and OS (65.2 vs. 91.5%, p<0.01). No difference in mean HRCTV D(90) EQD(2) was seen between the groups (p=0.61). CONCLUSION: Aside from known benefits of IGBT, MRI-based planning allows for assessment of tumor regression and prognosticates patients.

The role of CD38 in Fcγ receptor (FcγR)-mediated phagocytosis in murine macrophages.

Phagocytosis is a crucial event in the immune system that allows cells to engulf and eliminate pathogens. This is mediated through the action of immunoglobulin (IgG)-opsonized microbes acting on Fcγ receptors (FcγR) on macrophages, which results in sustained levels of intracellular Ca(2+) through the mobilization of Ca(2+) second messengers. It is known that the ADP-ribosyl cyclase is responsible for the rise in Ca(2+) levels after FcγR activation. However, it is unclear whether and how CD38 is involved in FcγR-mediated phagocytosis. Here we show that CD38 is recruited to the forming phagosomes during phagocytosis of IgG-opsonized particles and produces cyclic-ADP-ribose, which acts on ER Ca(2+) stores, thus allowing an increase in FcγR activation-mediated phagocytosis. Ca(2+) data show that pretreatment of J774A.1 macrophages with 8-bromo-cADPR, ryanodine, blebbistatin, and various store-operated Ca(2+) inhibitors prevented the long-lasting Ca(2+) signal, which significantly reduced the number of ingested opsonized particles. Ex vivo data with macrophages extracted from CD38(-/-) mice also shows a reduced Ca(2+) signaling and phagocytic index. Furthermore, a significantly reduced phagocytic index of Mycobacterium bovis BCG was shown in macrophages from CD38(-/-) mice in vivo. This study suggests a crucial role of CD38 in FcγR-mediated phagocytosis through its recruitment to the phagosome and mobilization of cADPR-induced intracellular Ca(2+) and store-operated extracellular Ca(2+) influx.