Discovery and Clinical Proof-of-Concept of RLY-2608, a First-in-Class Mutant-Selective Allosteric PI3Kα Inhibitor That Decouples Antitumor Activity from Hyperinsulinemia.

PIK3CA (PI3Kα) is a lipid kinase commonly mutated in cancer, including ∼40% of hormone receptor-positive breast cancer. The most frequently observed mutants occur in the kinase and helical domains. Orthosteric PI3Kα inhibitors suffer from poor selectivity leading to undesirable side effects, most prominently hyperglycemia due to inhibition of wild-type (WT) PI3Kα. Here, we used molecular dynamics simulations and cryo-electron microscopy to identify an allosteric network that provides an explanation for how mutations favor PI3Kα activation. A DNA-encoded library screen leveraging electron microscopy-optimized constructs, differential enrichment, and an orthosteric-blocking compound led to the identification of RLY-2608, a first-in-class allosteric mutant-selective inhibitor of PI3Kα. RLY-2608 inhibited tumor growth in PIK3CA-mutant xenograft models with minimal impact on insulin, a marker of dysregulated glucose homeostasis. RLY-2608 elicited objective tumor responses in two patients diagnosed with advanced hormone receptor-positive breast cancer with kinase or helical domain PIK3CA mutations, with no observed WT PI3Kα-related toxicities. SIGNIFICANCE: Treatments for PIK3CA-mutant cancers are limited by toxicities associated with the inhibition of WT PI3Kα. Molecular dynamics, cryo-electron microscopy, and DNA-encoded libraries were used to develop RLY-2608, a first-in-class inhibitor that demonstrates mutant selectivity in patients. This marks the advance of clinical mutant-selective inhibition that overcomes limitations of orthosteric PI3Kα inhibitors. See related commentary by Gong and Vanhaesebroeck, p. 204 . See related article by Varkaris et al., p. 227 . This article is featured in Selected Articles from This Issue, p. 201.

18F-Fluciclovine PET/CT: A Potential Imaging Biomarker for the Evaluation of Multiple Myeloma.

ABSTRACT: Incidental concomitant second primary malignancy may be detected on PET/CT imaging. We present an 18F-fluciclovine PET/CT of a patient undergoing evaluation of biochemically recurrent prostate cancer with incidental radiotracer uptake within lytic osseous lesions confirmed to be multiple myeloma. We present the 18F-fluciclovine PET/CT images of an 83-year-old man with prostate cancer treated in 2005 who presented with back pain and a CT scan revealing multiple lytic osseous lesions concerning for metastases versus a plasma cell neoplasm. Prostate-specific antigen at the time of evaluation was 0.1 ng/mL.

18F-FDG PET/CT for Response Assessment in Lung Cancer.

Treatment response assessment in lung cancer is crucial in the management strategy and outcome of patients. Accurate treatment response assessment can guide the treating physicians and improve patient survival. Anatomic and metabolic tumor response assessments have been evaluated extensively, showing a positive impact in the management of these patients. 18F-FDG PET/CT provides early and more specific treatment response assessments, preceding anatomic changes in these tumors. Familiarity with the different treatment response assessment algorithms, criteria, time intervals, imaging pitfalls is essential for treating physicians and nuclear radiologists to provide accurate response assessments. Artificial Intelligence is being more frequently explored for this purpose and can assist physicians in providing prompt and accurate treatment response assessments.

Utility of 18F-Fluciclovine PET/CT for Detecting Prostate Cancer Recurrence in Patients With Low (< 1 ng/mL) or Very Low (< 0.3 ng/mL) Prostate-Specific Antigen Levels.

OBJECTIVE. We reviewed a retrospective series of 126 18F-fluciclovine PET/CT studies of patients with biochemically recurrent prostate cancer at low (< 1 ng/mL) and very low (< 0.3 ng/mL) prostate-specific antigen (PSA) levels. CONCLUSION. The rate of PET/CT positivity was 33% (15/46) in patients with low PSA levels and 0% (0/17) in patients with very low PSA levels. Our results suggest that 18F-fluciclovine PET/CT can be helpful for localizing recurrence in patients with PSA levels between 0.3 and 1 ng/mL and that 18F-fluciclovine PET/CT is not recommended in patients with PSA levels less than 0.3 ng/mL.

Positron emission tomography as an adjuvant diagnostic test in the evaluation of checkpoint inhibitor-associated acute interstitial nephritis.

BACKGROUND: Acute interstitial nephritis is an immune-related adverse event that can occur in patients receiving immune checkpoint inhibitor therapy. Differentiating checkpoint inhibitor-associated acute interstitial nephritis from other causes of acute kidney injury in patients with cancer is challenging and can lead to diagnostic delays and/or unwarranted immunosuppression. In this case report, we assess the use of 18F-flourodeoxyglucose positron-emission tomography imaging as an alternative diagnostic modality in the evaluation of potential acute interstitial nephritis. CASE PRESENTATION: A 55-year-old woman with metastatic vulvar melanoma underwent treatment with two cycles of ipilimumab plus nivolumab, followed by seven cycles of nivolumab combined with radiation therapy. During her treatment, she developed non-oliguric acute kidney injury to a creatinine of 4.5 mg/dL from a baseline of 0.5 mg/dL. A clinical diagnosis of acute interstitial nephritis was made, and steroids were initiated, with rapid improvement of her acute kidney injury. Retrospectively, four positron-emission tomography scans obtained for cancer staging purposes were reviewed. We found a markedly increased 18F-flourodeoxyglucose uptake in the renal cortex at the time acute interstitial nephritis was diagnosed compared to baseline. In three cases of acute kidney injury due to alternative causes there was no increase in 18F-flourodeoxyglucose uptake from baseline. CONCLUSIONS: To our knowledge, this is the first report describing increased 18F-flourodeoxyglucose uptake in the renal cortex in a patient with checkpoint inhibitor-associated acute interstitial nephritis. Our findings suggest that 18F-flourodeoxyglucose positron-emission tomography may be a valuable test for diagnosing immune-mediated nephritis, particularly in patients where timely kidney biopsy is not feasible.

Machine Friendly Machine Learning: Interpretation of Computed Tomography Without Image Reconstruction.

Recent advancements in deep learning for automated image processing and classification have accelerated many new applications for medical image analysis. However, most deep learning algorithms have been developed using reconstructed, human-interpretable medical images. While image reconstruction from raw sensor data is required for the creation of medical images, the reconstruction process only uses a partial representation of all the data acquired. Here, we report the development of a system to directly process raw computed tomography (CT) data in sinogram-space, bypassing the intermediary step of image reconstruction. Two classification tasks were evaluated for their feasibility of sinogram-space machine learning: body region identification and intracranial hemorrhage (ICH) detection. Our proposed SinoNet, a convolutional neural network optimized for interpreting sinograms, performed favorably compared to conventional reconstructed image-space-based systems for both tasks, regardless of scanning geometries in terms of projections or detectors. Further, SinoNet performed significantly better when using sparsely sampled sinograms than conventional networks operating in image-space. As a result, sinogram-space algorithms could be used in field settings for triage (presence of ICH), especially where low radiation dose is desired. These findings also demonstrate another strength of deep learning where it can analyze and interpret sinograms that are virtually impossible for human experts.

An explainable deep-learning algorithm for the detection of acute intracranial haemorrhage from small datasets.

Owing to improvements in image recognition via deep learning, machine-learning algorithms could eventually be applied to automated medical diagnoses that can guide clinical decision-making. However, these algorithms remain a 'black box' in terms of how they generate the predictions from the input data. Also, high-performance deep learning requires large, high-quality training datasets. Here, we report the development of an understandable deep-learning system that detects acute intracranial haemorrhage (ICH) and classifies five ICH subtypes from unenhanced head computed-tomography scans. By using a dataset of only 904 cases for algorithm training, the system achieved a performance similar to that of expert radiologists in two independent test datasets containing 200 cases (sensitivity of 98% and specificity of 95%) and 196 cases (sensitivity of 92% and specificity of 95%). The system includes an attention map and a prediction basis retrieved from training data to enhance explainability, and an iterative process that mimics the workflow of radiologists. Our approach to algorithm development can facilitate the development of deep-learning systems for a variety of clinical applications and accelerate their adoption into clinical practice.

Mandated health insurance increases rates of elective knee surgery.

BACKGROUND: The recent federal ruling to against Affordable Care Act (ACA), specifically the mandate requiring people to buy insurance, has once again brought the healthcare reform debate to the spotlight. The ACA increased the number of insured Americans through the development of subsidized healthcare plans and health insurance exchanges. Insurance-based differences in the rate of upper extremity elective orthopaedic surgery have been described before and after healthcare reform in Massachusetts, where a similar mandate was put into place years before the ACA was passed. However, no comprehensive study has evaluated insurance-based differences of knee elective surgery before and after reform. AIM: To investigate how an individual mandate to purchase health insurance affects rates of knee surgery. METHODS: A retrospective review was performed within an orthopaedic surgery department at a tertiary-care, academic medical center in Massachusetts. The rate of elective knee surgery performed before and after the healthcare reform (2005-2006 and 2007-2010, respectively) was calculated. The patients were categorized by insurance type (Commonwealth Care, Medicare, Medicaid, private insurance, Workers' Compensation, TriCare, and Uninsured). Using χ 2 testing, differences in rates of surgery between the pre-reform and post-reform period and among insurance subgroups were calculated. RESULTS: Rate of surgery increased in the post-reform period (pre-reform 8.07% (95%CI: 7.03%-9.11%), post-reform 9.38% (95%CI: 8.74%-10.03%) (P = 0.04) and was statistically significant. When the insurance groups and insurance types were compared, the rates of surgery are not significantly different before or after reform. CONCLUSION: The increase in the rate of elective knee surgery in the post-reform period suggests that health care reform in Massachusetts has been successful in decreasing the uninsured population and may increase health care expenditures. This is a hypothesis generating study that suggests further avenues of study on how mandated coverage may change healthcare utilization and cost.

Artificial intelligence-assisted interpretation of bone age radiographs improves accuracy and decreases variability.

OBJECTIVE: Radiographic bone age assessment (BAA) is used in the evaluation of pediatric endocrine and metabolic disorders. We previously developed an automated artificial intelligence (AI) deep learning algorithm to perform BAA using convolutional neural networks. We compared the BAA performance of a cohort of pediatric radiologists with and without AI assistance. MATERIALS AND METHODS: Six board-certified, subspecialty trained pediatric radiologists interpreted 280 age- and gender-matched bone age radiographs ranging from 5 to 18 years. Three of those radiologists then performed BAA with AI assistance. Bone age accuracy and root mean squared error (RMSE) were used as measures of accuracy. Intraclass correlation coefficient evaluated inter-rater variation. RESULTS: AI BAA accuracy was 68.2% overall and 98.6% within 1 year, and the mean six-reader cohort accuracy was 63.6 and 97.4% within 1 year. AI RMSE was 0.601 years, while mean single-reader RMSE was 0.661 years. Pooled RMSE decreased from 0.661 to 0.508 years, all individually decreasing with AI assistance. ICC without AI was 0.9914 and with AI was 0.9951. CONCLUSIONS: AI improves radiologist's bone age assessment by increasing accuracy and decreasing variability and RMSE. The utilization of AI by radiologists improves performance compared to AI alone, a radiologist alone, or a pooled cohort of experts. This suggests that AI may optimally be utilized as an adjunct to radiologist interpretation of imaging studies to improve performance.

Beyond Human Perception: Sexual Dimorphism in Hand and Wrist Radiographs Is Discernible by a Deep Learning Model.

Despite the well-established impact of sex and sex hormones on bone structure and density, there has been limited description of sexual dimorphism in the hand and wrist in the literature. We developed a deep convolutional neural network (CNN) model to predict sex based on hand radiographs of children and adults aged between 5 and 70 years. Of the 1531 radiographs tested, the algorithm predicted sex correctly in 95.9% (κ = 0.92) of the cases. Two human radiologists achieved 58% (κ = 0.15) and 46% (κ = - 0.07) accuracy. The class activation maps (CAM) showed that the model mostly focused on the 2nd and 3rd metacarpal base or thumb sesamoid in women, and distal radioulnar joint, distal radial physis and epiphysis, or 3rd metacarpophalangeal joint in men. The radiologists reviewed 70 cases (35 females and 35 males) labeled with sex along with heat maps generated by CAM, but they could not find any patterns that distinguish the two sexes. A small sample of patients (n = 44) with sexual developmental disorders or transgender identity was selected for a preliminary exploration of application of the model. The model prediction agreed with phenotypic sex in only 77.8% (κ = 0.54) of these cases. To the best of our knowledge, this is the first study that demonstrated a machine learning model to perform a task in which human experts could not fulfill.

Toward Augmented Radiologists: Changes in Radiology Education in the Era of Machine Learning and Artificial Intelligence.

Radiology practice will be altered by the coming of artificial intelligence, and the process of learning in radiology will be similarly affected. In the short term, radiologists will need to understand the first wave of artificially intelligent tools, how they can help them improve their practice, and be able to effectively supervise their use. Radiology training programs will need to develop curricula to help trainees acquire the knowledge to carry out this new supervisory duty of radiologists. In the longer term, artificially intelligent software assistants could have a transformative effect on the training of residents and fellows, and offer new opportunities to bring learning into the ongoing practice of attending radiologists.

Delayed-type hypersensitivity to vaccine aluminum adjuvant causing subcutaneous leg mass and urticaria in a child.

A 3-year-old girl presented with a 7-month history of a waxing and waning left thigh mass associated with pruritus and erythema at the site of two previous DTaP-HepB-IPV vaccinations. Patch testing was positive to aluminum chloride, supporting a diagnosis of vaccine granuloma secondary to aluminum allergy; her symptoms had been well controlled with antihistamines and topical steroids. Injection site granulomas are a benign but potentially bothersome reaction to aluminum-containing immunizations that can be supportively managed, and we encourage strict adherence to the recommended vaccine schedule in this setting. Patch testing is a sensitive, noninvasive diagnostic tool for patients presenting with this clinical finding, and dermatologist awareness can prevent unnecessary medical examination and provide reassurance.

A Deep-Learning System for Fully-Automated Peripherally Inserted Central Catheter (PICC) Tip Detection.

A peripherally inserted central catheter (PICC) is a thin catheter that is inserted via arm veins and threaded near the heart, providing intravenous access. The final catheter tip position is always confirmed on a chest radiograph (CXR) immediately after insertion since malpositioned PICCs can cause potentially life-threatening complications. Although radiologists interpret PICC tip location with high accuracy, delays in interpretation can be significant. In this study, we proposed a fully-automated, deep-learning system with a cascading segmentation AI system containing two fully convolutional neural networks for detecting a PICC line and its tip location. A preprocessing module performed image quality and dimension normalization, and a post-processing module found the PICC tip accurately by pruning false positives. Our best model, trained on 400 training cases and selectively tuned on 50 validation cases, obtained absolute distances from ground truth with a mean of 3.10 mm, a standard deviation of 2.03 mm, and a root mean squares error (RMSE) of 3.71 mm on 150 held-out test cases. This system could help speed confirmation of PICC position and further be generalized to include other types of vascular access and therapeutic support devices.

Pixel-Level Deep Segmentation: Artificial Intelligence Quantifies Muscle on Computed Tomography for Body Morphometric Analysis.

Pretreatment risk stratification is key for personalized medicine. While many physicians rely on an "eyeball test" to assess whether patients will tolerate major surgery or chemotherapy, "eyeballing" is inherently subjective and difficult to quantify. The concept of morphometric age derived from cross-sectional imaging has been found to correlate well with outcomes such as length of stay, morbidity, and mortality. However, the determination of the morphometric age is time intensive and requires highly trained experts. In this study, we propose a fully automated deep learning system for the segmentation of skeletal muscle cross-sectional area (CSA) on an axial computed tomography image taken at the third lumbar vertebra. We utilized a fully automated deep segmentation model derived from an extended implementation of a fully convolutional network with weight initialization of an ImageNet pre-trained model, followed by post processing to eliminate intramuscular fat for a more accurate analysis. This experiment was conducted by varying window level (WL), window width (WW), and bit resolutions in order to better understand the effects of the parameters on the model performance. Our best model, fine-tuned on 250 training images and ground truth labels, achieves 0.93 ± 0.02 Dice similarity coefficient (DSC) and 3.68 ± 2.29% difference between predicted and ground truth muscle CSA on 150 held-out test cases. Ultimately, the fully automated segmentation system can be embedded into the clinical environment to accelerate the quantification of muscle and expanded to volume analysis of 3D datasets.

Impact of Clinical Decision Support on Radiography for Acute Ankle Injuries: A Randomized Trial.

INTRODUCTION: While only 15-20% of patients with foot and ankle injuries presenting to urgent care centers have clinically significant fractures, most undergo radiography. We examined the impact of electronic point-of-care clinical decision support (CDS) on adherence to the Ottawa Ankle Rules (OAR), as well as use and yield of foot and ankle radiographs in patients with acute ankle injury. METHODS: We obtained institutional review board approval for this randomized controlled study performed April 18, 2012-December 15, 2013. All ordering providers credentialed at an urgent care affiliated with a quaternary care academic hospital were randomized to either receive or not receive CDS, based on the OAR and integrated into the physician order-entry system, with feedback at the time of imaging order. If the patient met OAR low-risk criteria, providers were advised against imaging and could either cancel the order or ignore the alert. We identified patients with foot and ankle complaints via ICD-9 billing codes and electronic health records and radiology reports reviewed for those who were eligible. Chi-square was used to compare adherence to the OAR (primary outcome), radiography utilization rate and radiography yield of foot and ankle imaging (secondary outcomes) between the intervention and control groups. RESULTS: Of 14,642 patients seen at urgent care during the study period, 613 (4.2%, representing 632 visits) presented with acute ankle injury and were eligible for application of the OAR; 374 (59.2%) of these were seen by control-group providers. In the intervention group, CDS adherence was higher for both ankle (239/258=92.6% vs. 231/374=61.8%, p=0.02) and foot radiography (209/258=81.0% vs. 238/374=63.6%; p<0.01). However, ankle radiography use was higher in the intervention group (166/258=64.3% vs. 183/374=48.9%; p<0.01), while foot radiography use (141/258=54.6% vs. 202/374=54.0%; p=0.95) was not. Radiography yield was also higher in the intervention group (26/307=8.5% vs. 18/385=4.7%; p=0.04). CONCLUSION: Clinical decision support, previously demonstrated to improve guideline adherence for high-cost imaging, can also improve guideline adherence for radiography - as demonstrated by increased OAR adherence and increased imaging yield.

Fully Automated Deep Learning System for Bone Age Assessment.

Skeletal maturity progresses through discrete phases, a fact that is used routinely in pediatrics where bone age assessments (BAAs) are compared to chronological age in the evaluation of endocrine and metabolic disorders. While central to many disease evaluations, little has changed to improve the tedious process since its introduction in 1950. In this study, we propose a fully automated deep learning pipeline to segment a region of interest, standardize and preprocess input radiographs, and perform BAA. Our models use an ImageNet pretrained, fine-tuned convolutional neural network (CNN) to achieve 57.32 and 61.40% accuracies for the female and male cohorts on our held-out test images. Female test radiographs were assigned a BAA within 1 year 90.39% and within 2 years 98.11% of the time. Male test radiographs were assigned 94.18% within 1 year and 99.00% within 2 years. Using the input occlusion method, attention maps were created which reveal what features the trained model uses to perform BAA. These correspond to what human experts look at when manually performing BAA. Finally, the fully automated BAA system was deployed in the clinical environment as a decision supporting system for more accurate and efficient BAAs at much faster interpretation time (<2 s) than the conventional method.

Impact of a clinical decision support tool on adherence to the Ottawa Ankle Rules.

OBJECTIVE: The objective of the study is to determine impact of a clinical decision support (CDS) tool on documented adherence to the Ottawa Ankle Rules (OAR) and utilization and yield of ankle/foot radiography, for emergency department patients with acute ankle injury. METHODS: This is a before-and-after intervention study conducted at a 793-bed, quaternary care, academic hospital from August 2012 to October 2013. Emergency department visits from adults with acute ankle injury 6 months before and 8 months after the intervention were included. The intervention embedded the OAR into a CDS tool integrated with a computerized physician order entry system, which had data capture capability and provided feedback at the time of ankle/foot radiography order. Primary outcome was rate of documented adherence to OAR. Secondary outcomes were utilization and yield (clinically significant fracture rates among patients with acute ankle injuries) of ankle/foot radiography. RESULTS: The study population included 460 visits; 205 (44.6%) occurred preintervention. After intervention, documented OAR adherence increased from 55.9% (229/410) to 95.7% (488/510; P < .001). Utilization remained stable for ankle (77.5%; P = .800) and foot (48.6%; P = .514) radiography. Yield remained stable for ankle (17.8%; P = .891) and foot (19.8%; P = .889) radiography. DISCUSSION: Lack of documentation of key clinical data may hamper provider communication, delay care coordination, and result in legal liability. By embedding the OAR into a CDS tool, we achieved the same rate of documented adherence as previous onerous educational implementations while automating data collection/retrieval. In summary, implementation of the OAR into a CDS tool was associated with an increase in documented adherence to the OAR.

Lower extremity venous reflux.

Venous incompetence in the lower extremity is a common clinical problem. Basic understanding of venous anatomy, pathophysiologic mechanisms of venous reflux is essential for choosing the appropriate treatment strategy. The complex interplay of venous pressure, abdominal pressure, venous valvular function and gravitational force determine the venous incompetence. This review is intended to provide a succinct review of the pathophysiology of venous incompetence and the current role of imaging in its management.