Search

In this study, we developed a deep learning pipeline that detects large vessel occlusion (LVO) and predicts functional outcome based on computed tomography angiography (CTA) images to improve the management of the LVO patients.A series identifier picked out 8650 LVO-protocoled studies from 2015 to 2019 at Rhode Island Hospital with an identified thin axial series that served as the data pool. Data were annotated into 2 classes: 1021 LVOs and 7629 normal. The Inception-V1 I3D architecture was applied for LVO detection. For outcome prediction, 323 patients undergoing thrombectomy were selected. A 3D convolution neural network (CNN) was used for outcome prediction (30-day mRS) with CTA volumes and embedded pre-treatment variables as inputs.For LVO-detection model, CTAs from 8,650 patients (median age 68 years, interquartile range (IQR): 58-81; 3934 females) were analyzed. The cross-validated AUC for LVO vs. not was 0.74 (95% CI: 0.72-0.75). For the mRS classification model, CTAs from 323 patients (median age 75 years, IQR: 63-84; 164 females) were analyzed. The algorithm achieved a test AUC of 0.82 (95% CI: 0.79-0.84), sensitivity of 89%, and specificity 66%. The two models were then integrated with hospital infrastructure where CTA was collected in real-time and processed by the model. If LVO was detected, interventionists were notified and provided with predicted clinical outcome information.3D CNNs based on CTA were effective in selecting LVO and predicting LVO mechanical thrombectomy short-term prognosis. End-to-end AI platform allows users to receive immediate prognosis prediction and facilitates clinical workflow.

Aims To determine if neurologic symptoms at admission can predict adverse outcomes in patients with severe acute respiratory syndrome coronavirus 2 (SARS‐CoV‐2). Methods Electronic medical records of 1053 consecutively hospitalized patients with laboratory‐confirmed infection of SARS‐CoV‐2 from one large medical center in the USA were retrospectively analyzed. Univariable and multivariable Cox regression analyses were performed with the calculation of areas under the curve (AUC) and concordance index (C‐index). Patients were stratified into subgroups based on the presence of encephalopathy and its severity using survival statistics. In sensitivity analyses, patients with mild/moderate and severe encephalopathy (defined as coma) were separately considered. Results Of 1053 patients (mean age 52.4 years, 48.0% men [n = 505]), 35.1% (n = 370) had neurologic manifestations at admission, including 10.3% (n = 108) with encephalopathy. Encephalopathy was an independent predictor for death (hazard ratio [HR] 2.617, 95% confidence interval [CI] 1.481–4.625) in multivariable Cox regression. The addition of encephalopathy to multivariable models comprising other predictors for adverse outcomes increased AUCs (mortality: 0.84–0.86, ventilation/ intensive care unit [ICU]: 0.76–0.78) and C‐index (mortality: 0.78 to 0.81, ventilation/ICU: 0.85–0.86). In sensitivity analyses, risk stratification survival curves for mortality and ventilation/ICU based on severe encephalopathy (n = 15) versus mild/moderate encephalopathy (n = 93) versus no encephalopathy (n = 945) at admission were discriminative (p, Patients with severe acute respiratory syndrome coronavirus 2 infection have a high prevalence of neurologic manifestations, including headache, encephalopathy, dizziness, taste, and smell impairment. Patients with encephalopathy at admission predict later progression to death and mechanical ventilation/ICU admission in SARS‐CoV‐2 infection, which may have important implications for risk stratification in clinical practice.

Background Coronavirus disease 2019 (COVID-19) and pneumonia of other diseases share similar CT characteristics, which contributes to the challenges in differentiating them with high accuracy. Purpose To establish and evaluate an artificial intelligence (AI) system for differentiating COVID-19 and other pneumonia at chest CT and assessing radiologist performance without and with AI assistance. Materials and Methods A total of 521 patients with positive reverse transcription polymerase chain reaction results for COVID-19 and abnormal chest CT findings were retrospectively identified from 10 hospitals from January 2020 to April 2020. A total of 665 patients with non-COVID-19 pneumonia and definite evidence of pneumonia at chest CT were retrospectively selected from three hospitals between 2017 and 2019. To classify COVID-19 versus other pneumonia for each patient, abnormal CT slices were input into the EfficientNet B4 deep neural network architecture after lung segmentation, followed by a two-layer fully connected neural network to pool slices together. The final cohort of 1186 patients (132 583 CT slices) was divided into training, validation, and test sets in a 7:2:1 and equal ratio. Independent testing was performed by evaluating model performance in separate hospitals. Studies were blindly reviewed by six radiologists without and then with AI assistance. Results The final model achieved a test accuracy of 96% (95% confidence interval [CI]: 90%, 98%), a sensitivity of 95% (95% CI: 83%, 100%), and a specificity of 96% (95% CI: 88%, 99%) with area under the receiver operating characteristic curve of 0.95 and area under the precision-recall curve of 0.90. On independent testing, this model achieved an accuracy of 87% (95% CI: 82%, 90%), a sensitivity of 89% (95% CI: 81%, 94%), and a specificity of 86% (95% CI: 80%, 90%) with area under the receiver operating characteristic curve of 0.90 and area under the precision-recall curve of 0.87. Assisted by the probabilities of the model, the radiologists achieved a higher average test accuracy (90% vs 85%, Δ = 5

BACKGROUND AND PURPOSE: Differentiating the types of pediatric posterior fossa tumors on routine imaging may help in preoperative evaluation and guide surgical resection planning. However, qualitative radiologic MR imaging review has limited performance. This study aimed to compare different machine learning approaches to classify pediatric posterior fossa tumors on routine MR imaging. MATERIALS AND METHODS: This retrospective study included preoperative MR imaging of 288 patients with pediatric posterior fossa tumors, including medulloblastoma (n = 111), ependymoma (n = 70), and pilocytic astrocytoma (n = 107). Radiomics features were extracted from T2-weighted images, contrast-enhanced T1-weighted images, and ADC maps. Models generated by standard manual optimization by a machine learning expert were compared with automatic machine learning via the Tree-Based Pipeline Optimization Tool for performance evaluation. RESULTS: For 3-way classification, the radiomics model by automatic machine learning with the Tree-Based Pipeline Optimization Tool achieved a test micro-averaged area under the curve of 0.91 with an accuracy of 0.83, while the most optimized model based on the feature-selection method χ(2) score and the Generalized Linear Model classifier achieved a test micro-averaged area under the curve of 0.92 with an accuracy of 0.74. Tree-Based Pipeline Optimization Tool models achieved significantly higher accuracy than average qualitative expert MR imaging review (0.83 versus 0.54, P

OBJECTIVE The aim of the study was to evaluate efficacy of chemoembolization and radioembolization in treating patients with recurrent hypoglycemia secondary to metastatic insulinoma. METHODS A retrospective review was performed of all patients with metastatic insulinoma treated with liver-directed therapy (LDT) at a large academic medical center from January 1998 to August 2017. Primary outcomes included blood glucose levels, occurrence of symptomatic hypoglycemic episodes, and tumor imaging response rates. RESULTS Seven patients were identified (4 male patients). The mean age at the first LDT was 60.9 (standard deviation [SD], 9.2) years. The median follow-up was 1.8 years. Thirty-three sessions of LDT were performed including 30 sessions of chemoembolization and 3 sessions of radioembolization. Technical success rate was 97% (32/33 procedures) with an initial clinical success rate of 100%, defined as absence of recurrent symptomatic hypoglycemia within 1 month after first cycle of LDT and overall clinical success rate of 85%. Random daytime glucose levels increased from 75.0 (SD, 26) mg/dL to 152.8 (SD, 52.4) mg/dL after LDT. The mean time to recurrence of intractable hypoglycemia was 21 (SD, 9) months. No severe complications were reported. CONCLUSIONS Liver-directed therapy with chemoembolization or radioembolization is effective for treatment of symptomatic hypoglycemia secondary to malignant insulinoma metastatic to liver.

Purpose: With increasing incidence of renal mass, it is important to make a pretreatment differentiation between benign renal mass and malignant tumor. We aimed to develop a deep learning model that distinguishes benign renal tumors from renal cell carcinoma (RCC) by applying a residual convolutional neural network (ResNet) on routine MR imaging. Experimental Design: Preoperative MR images (T2-weighted and T1-postcontrast sequences) of 1,162 renal lesions definitely diagnosed on pathology or imaging in a multicenter cohort were divided into training, validation, and test sets (70:20:10 split). An ensemble model based on ResNet was built combining clinical variables and T1C and T2WI MR images using a bagging classifier to predict renal tumor pathology. Final model performance was compared with expert interpretation and the most optimized radiomics model. Results: Among the 1,162 renal lesions, 655 were malignant and 507 were benign. Compared with a baseline zero rule algorithm, the ensemble deep learning model had a statistically significant higher test accuracy (0.70 vs. 0.56, P = 0.004). Compared with all experts averaged, the ensemble deep learning model had higher test accuracy (0.70 vs. 0.60, P = 0.053), sensitivity (0.92 vs. 0.80, P = 0.017), and specificity (0.41 vs. 0.35, P = 0.450). Compared with the radiomics model, the ensemble deep learning model had higher test accuracy (0.70 vs. 0.62, P = 0.081), sensitivity (0.92 vs. 0.79, P = 0.012), and specificity (0.41 vs. 0.39, P = 0.770). Conclusions: Deep learning can noninvasively distinguish benign renal tumors from RCC using conventional MR imaging in a multi-institutional dataset with good accuracy, sensitivity, and specificity comparable with experts and radiomics.

Background The benefit of adjuvant radiotherapy (RT) is uncertain in elderly patients diagnosed with World Health Organization (WHO) grade III meningiomas. Methods The US National Cancer Database was used to identify elderly patients (aged over 60) with histologically confirmed WHO grade III meningiomas who underwent surgery with or without adjuvant external beam RT. Overall survival (OS) was evaluated by Kaplan-Meier analysis, multivariable Cox proportional hazard models, and propensity score–matched analysis. Results Among the 254 patients who met inclusion criteria, 151 (58.4%) underwent surgery followed by RT and 103 (41.6%) surgery without RT. The 5-year relative survival rate was 57.8% (95% confidence interval [CI], 38.7%–85.4%) in the adjuvant RT group and 38.1% (95% CI, 21.9%–55.9%) in the group without RT when compared with age- and gender-matched US population. Adjuvant RT was associated with longer OS on both univariate (P = 0.011) and multivariable analyses (adjusted hazard ratio [AHR], 0.663; 95% CI, 0.449–0.978; P = 0.038). However, adjuvant RT did not significantly improve OS in a propensity score–matched analysis (P = 0.271) after matching the 2 cohorts on age, race, comorbidity, extent of resection, and tumor size. In patients who underwent subtotal resection or biopsy (n = 106), adjuvant RT significantly benefited OS on univariate analysis (P = 0.023), but did not show statistically significant benefit on multivariate analysis (AHR, 0.594; 95% CI, 0.337–1.046; P = 0.071). Adjuvant RT was not associated with improved OS in the elderly patients who underwent gross total resection (AHR, 0.764; 95% CI, 0.434–1.344; P = 0.350). Conclusion Adjuvant external beam RT may not provide any survival benefit for elderly patients with WHO grade III meningioma after gross total resection.

Background Use of conventional blood oxygen level-dependent functional magnetic resonance imaging (conventional-BOLD-fMRI) presents challenges in accurately identifying the hand-motor cortex when a glioma involves the ipsilateral hand-knob. Zoomed imaging technique with parallel transmission (ZOOMit)-BOLD is a novel sequence allowing high spatial resolution with a relatively small field of view that may solve this problem. Objective To compare the accuracy of ZOOMit-BOLD and conventional-BOLD in hand-motor cortex identification. Methods A total of 20 patients with gliomas involving the sensorimotor cortex were recruited to identify the hand-motor cortex by both ZOOMit-BOLD and conventional-BOLD. Based on whether the entire or partial glioma directly invaded (was located within) the hand-knob or indirectly affected it by proximity, patients were placed into the involved or uninvolved groups, respectively. Direct cortical stimulation was applied intraoperatively to verify the location of the hand-motor cortex. Overlap indices were used to evaluate the accuracy of the hand-motor cortex identification. An overlap index equal to 0, indicating lack of overlap, was classified as inaccurate classification. Results The accuracy of motor-cortex identification with ZOOMit-BOLD was 100% compared to only 65% with conventional-BOLD. The average overlap index yielded by ZOOMit-BOLD was higher than that of conventional-BOLD, regardless of whether gliomas directly invaded the hand-knob (P = .008) or not (P = .004). The overlap index in the involved group was significantly lower than that in the uninvolved group with both ZOOMit-BOLD (P = .002) and conventional-BOLD (P Conclusion ZOOMit-BOLD may potentially replace conventional-BOLD to identify the hand-motor cortex, particularly in cases in which gliomas directly invade the hand-knob.

Purpose: To explore the role of chronic liver disease (CLD) in COVID-19.Methods: A total of 1,439 consecutively hospitalized patients with COVID-19 from one large medical center in the United States from March 16, 2020 to April 23, 2020 were retrospectively identified. Clinical characteristics and outcomes were compared between patients with and without CLD. Postmortem examination of liver in 8 critically ill COVID-19 patients was performed.Results: There was no significant difference in the incidence of CLD between critical and non-critical groups (4.1% vs 2.9%, p=0.259), or COVID-19 related liver injury between patients with and without CLD (65.7% vs 49.7%, p=0.065). Postmortem examination of liver demonstrated mild liver injury associated central vein outflow obstruction and minimal to moderate portal lymphocytic infiltrate without evidence of CLD.Conclusion: Patients with CLD were not associated with a higher risk of liver injury or critical/fatal outcomes. CLD was not a significant comorbid condition for COVID-19.

Background Radiologists have difficulty distinguishing benign from malignant bone lesions because these lesions may have similar imaging appearances. The purpose of this study was to develop a deep learning algorithm that can differentiate benign and malignant bone lesions using routine magnetic resonance imaging (MRI) and patient demographics. Methods 1,060 histologically confirmed bone lesions with T1- and T2-weighted pre-operative MRI were retrospectively identified and included, with lesions from 4 institutions used for model development and internal validation, and data from a fifth institution used for external validation. Image-based models were generated using the EfficientNet-B0 architecture and a logistic regression model was trained using patient age, sex, and lesion location. A voting ensemble was created as the final model. The performance of the model was compared to classification performance by radiology experts. Findings The cohort had a mean age of 30±23 years and was 58.3% male, with 582 benign lesions and 478 malignant. Compared to a contrived expert committee result, the ensemble deep learning model achieved (ensemble vs. experts): similar accuracy (0·76 vs. 0·73, p=0·7), sensitivity (0·79 vs. 0·81, p=1·0) and specificity (0·75 vs. 0·66, p=0·48), with a ROC AUC of 0·82. On external testing, the model achieved ROC AUC of 0·79. Interpretation Deep learning can be used to distinguish benign and malignant bone lesions on par with experts. These findings could aid in the development of computer-aided diagnostic tools to reduce unnecessary referrals to specialized centers from community clinics and limit unnecessary biopsies. Funding This work was funded by a Radiological Society of North America Research Medical Student Grant (#RMS2013) and supported by the Amazon Web Services Diagnostic Development Initiative.

BACKGROUND. Although established guidelines give indications for performing staging brain MRI at initial diagnosis of non-small cell lung cancer (NSCLC), guidelines are lacking for performing surveillance brain MRI for patients without brain metastases at presentation. OBJECTIVE. The purpose of this study is to estimate the cumulative incidence of and risk factors for brain metastasis development in patients with NSCLC without brain metastases at initial presentation. METHODS. This retrospective study included 1495 patients with NSCLC (mean [± SD] age, 65 ± 10 years; 920 men and 575 women) without brain metastases at initial evaluation that included brain MRI. Follow-up brain MRI was ordered at the discretion of the referring physicians. MRI examinations were reviewed in combination with clinical records for brain metastasis development; patients not undergoing MRI were deemed to have not had metastases develop through last clinical follow-up. The cumulative incidence of brain metastases was determined, with death considered a competing risk, and was stratified by clinical stage group, cell type, and epidermal growth factor receptor (EGFR) gene mutation status. Univariable and multivariable Cox proportional hazards regression analyses were performed. RESULTS. A total of 258 of 1495 patients (17.3%) underwent follow-up brain MRI, and 72 (4.8%) had brain metastases develop at a median of 12.3 months after initial diagnosis of NSCLC. Of the 72 patients who had metastases develop, 44.4% had no neurologic symptoms, and 58.3% had stable primary thoracic disease. The cumulative incidence of brain metastases at 6, 12, 18, and 24 months after initial evaluation was 0.6%, 2.1%, 4.2%, and 6.8%, respectively. Cumulative incidence at 6, 12, 18, and 24 months was higher (p < .001) in patients with clinical stage III-IV disease (1.3%, 3.9%, 7.7%, and 10.9%, respectively) than in those with clinical stage I-II disease (0.0%, 0.8%, 1.2%, and 2.6%, respectively), and it was higher (p < .001) in patients with EGFR mutation-positive adenocarcinoma (0.7%, 2.5%, 6.3%, and 12.3%, respectively) than in those with EGFR mutation-negative adenocarcinoma (0.4%, 1.8%, 2.9%, and 4.4%, respectively). Among 1109 patients with adenocarcinoma, independent risk factors for the development of brain metastasis were clinical stage III-IV (hazard ratio [HR], 9.39; p < .001) and EGFR mutation-positive status (HR, 1.78; p = .04). The incidence of brain metastasis over the study interval was 8.7% among patients with clinical stage III-IV disease and 17.4% among those with EGFR mutation-positive adenocarcinoma. CONCLUSION. Clinical stage III-IV and EGFR mutation-positive adenocarcinoma are independent risk factors for brain metastasis development. CLINICAL IMPACT. For patients with clinical stage III-IV disease or EGFR mutation-positive adenocarcinoma, surveillance brain MRI performed 12 months after initial evaluation may be warranted.

It remains controversial if Mohs surgery is superior to surgical excision in treating localized sebaceous carcinoma.To compare Mohs surgery and surgical excision for treating patients with localized sebaceous carcinoma.The US National Cancer Database was used to identify patients with histologically confirmed Stage 0 to 2 sebaceous carcinoma from 2004 to 2014. Clinicopathologic and socioeconomic factors were compared between treatment groups using the chi-square test. Overall survival (OS) was evaluated by log-rank test, multivariable Cox proportional hazard regression, and propensity score-matched analysis. Relative survival analyses compared with age- and sex-matched US population were performed.Of 1,265 patients, 234 received Mohs surgery and 1,031 received surgical excision. Mohs surgery had a higher rate of negative margin (p = .004). On multivariate Cox regression analysis, Mohs surgery was associated with longer OS than surgical excision (HR: 0.703, 95% CI: 0.496-0.995, p = .047). The survival benefit of Mohs surgery persisted on relative survival analysis and propensity score-matched analysis (p = .0385), after matching the 2 groups on patient and disease characteristics.Patients who received Mohs surgery had significantly longer OS when compared with those who received surgical excision. Prospective clinical trials comparing these treatment paradigms are warranted.

The effect of insurance status on overall survival (OS) of patients with cutaneous T-cell lymphoma (CTCL) is unclear. We identified 11,861 patients from the US National Cancer Data Base diagnosed with CTCL from 2004-2014, of which 6088 had private insurance, 756 had Medicaid, 4536 had Medicare, and 481 are uninsured. Privately insured patients were more likely to present at an early stage (p

Objective To investigate prognostic factors of clival chordoma using the largest patient set to date. Methods Appropriate studies were identified per search criteria, data satisfying criteria were extracted, and survival analysis was performed to investigate prognostic factors of clival chordoma. Results A total of 347 patients from the literature cohort met our inclusion criteria. Of 346 cases in which extent of resection was reported, gross total resection (GTR), subtotal resection, and biopsy were achieved in 118 (34.1%), 205 (59.2%), and 21 (6.1%) cases, respectively. Two (0.6%) subjects did not undergo surgery. Of 185 cases in which surgical approach was reported, 56 (30.3%) underwent an endoscopic transoral approach, 17 (9.2%) microscopic transsphenoidal, 45 (24.3%) endoscopic or microscopic, 45 (24.3%) craniotomy, and 22 (11.9%) other approaches. There was no significant difference in GTR rates of different surgical approaches (P = 0.101). Median follow-up was 46.6 months. The 5- and 10-year rates for progression-free survival (PFS) were 59.2% and 47.9%, respectively. The 5- and 10-year rates for overall survival (OS) were 77.3% and 63.9%, respectively. On multivariate analysis for both PFS and OS, GTR demonstrated significantly improved outcomes when compared with subtotal resection (hazard ratio 0.45, 95% confidence interval 0.22–0.90, P = 0.025 for PFS; hazard ratio 0.20, confidence interval 0.06–0.65, P = 0.008 for OS). Conclusions GTR rates were comparable in different surgical approaches. GTR was a significant predictor of longer PFS and OS in clival chordoma.

In this study, we developed radiomic models that utilize a combination of imaging features and clinical variables to distinguish endometrial cancer (EC) from normal endometrium on routine computed tomography (CT).A total of 926 patients consisting of 416 endometrial cancer (EC) and 510 normal endometrium were included. The CT images of these patients were segmented manually, and divided into training, validation, testing and external testing sets. Non-texture and texture features of these images with endometrium or uterus as region of interest were extracted. The clinical feature "age" was also included in the feature set. Feature selection and machine learning classifier were applied to normalized feature set. This manual optimized combination was then compared with the best pipeline exported by Tree-Based Pipeline Optimization Tool (TPOT) on testing and external testing set. The performances of these machine learning pipelines were compared to that of radiologists.The manual expert optimized pipeline using the "reliefF" feature selection method and "Bagging" classifier on the external testing set achieved a test ROC AUC of 0.73, accuracy of 0.73 (95% CI 0.62-0.82), sensitivity of 0.64 (95% CI 0.45-0.79), and specificity of 0.78 (95% CI 0.65-0.87), while TPOT achieved a test ROC AUC of 0.79, accuracy of 0.80 (95% CI 0.70-0.87), sensitivity of 0.61 (95% CI 0.43-0.77), and specificity of 0.90 (95% CI 0.78-0.96). When compared to average radiologist performance, the TPOT achieved higher test accuracy (0.80 vs. 0.49, p 0.001) and specificity (0.90 vs. 0.51, p 0.001), with comparable sensitivity (0.61 vs. 0.46, p = 0.130).Our results demonstrate that automatic machine learning can distinguish EC from normal endometrium on routine CT imaging with higher accuracy and specificity than radiologists.

Background: To develop a deep learning model to classify primary bone tumors from preoperative radiographs and compare performance with radiologists. Methods: A total of 1356 patients (2899 images) with histologically confirmed primary bone tumors and pre-operative radiographs were identified from five institutions’ pathology databases. Manual cropping was performed by radiologists to label the lesions. Binary discriminatory capacity (benign versus not-benign and malignant versus not-malignant) and three-way classification (benign versus intermediate versus malignant) performance of our model were evaluated. The generalizability of our model was investigated on data from external test set. Final model performance was compared with interpretation from five radiologists of varying level of experience using the Permutations tests. Findings: For benign vs. not benign, model achieved area under curve (AUC) of 0•894 and 0•877 on cross-validation and external testing, respectively. For malignant vs. not malignant, model achieved AUC of 0•907 and 0•916 on cross-validation and external testing, respectively. For three-way classification, model achieved 72•1% accuracy vs. 74•6% and 72•1% for the two subspecialists on cross-validation (p = 0•03 and p = 0•52, respectively). On external testing, model achieved 73•4% accuracy vs. 69•3%, 73•4%, 73•1%, 67•9%, and 63•4% for the two subspecialists and three junior radiologists (p = 0•14, p = 0•89, p = 0•93, p = 0•02, p

Pre-treatment determination of renal cell carcinoma aggressiveness may help guide clinical decision-making. We aimed to differentiate low-grade (Fuhrman I–II) from high-grade (Fuhrman III–IV) renal cell carcinoma using radiomics features extracted from routine MRI. 482 pathologically confirmed renal cell carcinoma lesions from 2008 to 2019 in a multicenter cohort were retrospectively identified. 439 lesions with information on Fuhrman grade from 4 institutions were divided into training and test sets with an 8:2 split for model development and internal validation. Another 43 lesions from a separate institution were set aside for independent external validation. The performance of TPOT (Tree-Based Pipeline Optimization Tool), an automatic machine learning pipeline optimizer, was compared to hand-optimized machine learning pipeline. The best-performing hand-optimized pipeline was a Bayesian classifier with Fischer Score feature selection, achieving an external validation ROC AUC of 0.59 (95% CI 0.49–0.68), accuracy of 0.77 (95% CI 0.68–0.84), sensitivity of 0.38 (95% CI 0.29–0.48), and specificity of 0.86 (95% CI 0.78–0.92). The best-performing TPOT pipeline achieved an external validation ROC AUC of 0.60 (95% CI 0.50–0.69), accuracy of 0.81 (95% CI 0.72–0.88), sensitivity of 0.12 (95% CI 0.14–0.30), and specificity of 0.97 (95% CI 0.87–0.97). Automated machine learning pipelines can perform equivalent to or better than hand-optimized pipeline on an external validation test non-invasively predicting Fuhrman grade of renal cell carcinoma using conventional MRI.

Background Existing guidelines are inconsistent regarding the indications for staging brain MRI in patients with newly diagnosed, early-stage non-small cell lung cancer (NSCLC). Purpose To evaluate the diagnostic yield of staging brain MRI in the initial evaluation of lung cancer. Materials and Methods This retrospective, observational, single-institution study included patients with newly diagnosed NSCLC who underwent staging chest CT and staging brain MRI from November 2017 to October 2018. Diagnostic yield was defined as the proportion of patients with brain metastases among all patients. Yield was stratified into clinical stage groups per the eighth edition of the American Joint Committee on Cancer staging guidelines, based on staging chest CT and in adenocarcinoma with epidermal growth factor receptor

Complete biliary occlusions at different anatomic locations that are not amenable to surgical, endoscopic, or routine percutaneous methods pose a challenge to physicians. Biliary ductal neoanastomosis technique can play a very important role in solving such clinical problems. Between October 2004 and January 2018, six patients were treated in our institution for biliary drainage using biliary ductal neoanastomosis technique via sharp recanalization and a staged approach to achieve internal/external biliary drainage. All procedures were performed in an angiography suite by an interventional radiologist after multidisciplinary discussion. The mean patient age of the cohort was 54 years and 50% of the patients were female (three/six patients). The cause of obstruction was complications or disease progression after Whipple procedure in four patients, partial hepatectomy in one patient and benign biliary stricture after radioembolization in one patient. All patients were treated with percutaneous biliary neoanastomosis. All procedures were successfully performed in all six patients (100% technical success) without any major complications. All patients were successfully managed by capped drainage catheters for the duration of the follow up (one patient was transitioned to endoscopically inserted biliary stent, and in one patient the tube was eventually removed after establishment of patent neoanastomosis). Biliary ductal neoanastomosis is an effective treatment option when the anatomy has been significantly altered in the post-surgical setting and wire cannot be safely passed into target draining organ or the remainder of biliary system. When performed by an experienced operator, this technique is safe and effective with a high technical success rate.

BackgroundHemodynamic alterations post-embolization of intracranial arteriovenous malformations (AVMs) may cause delayed edema/hemorrhage in brain parenchyma adjacent to the lesion.ObjectiveTo quantify and compare cerebral perfusion changes in the peri-AVM territory pre- and post-embolization using color-coded quantitative digital subtraction angiography (q-DSA).MethodsPediatric intracranial AVM embolization procedures performed over a 5 year period were included. DSA images of all patients were retrospectively assessed using syngo iFlow. Regions of interest (ROI) were selected on anteroposterior and lateral q-DSA views: three in the peri-AVM region; two in parenchyma distant from the AVM. Time-to-peak (TTP) contrast enhancement of ROIs and ∆TTP (TTP at the selected ROI minus TTP at either the ipsilateral internal carotid/vertebral artery) were measured.Result19 pediatric patients with 19 AVMs (9 males/10 females, mean age 12 years) underwent intracranial AVM embolization: 15/19 AVMs were supplied by the anterior circulation and 4/19 by the posterior circulation. Blood flow was significantly slower post-embolization in the draining vein (19/19) (pConclusionThis study demonstrates more rapid perfusion in the peri-nidal brain parenchyma post-embolization of the AVM, which supports the theory that increased perfusion in normal tissue surrounding the AVM after embolization may underlie some post-procedural complications.

Seven patients with pulmonary arteriovenous malformations (PAVMs) not well suited to coil and/or plug treatment were treated with expanded polytetrafluoroethylene-covered stents. Mean diameter of treated arteries was 6 mm. Complete technical success was achieved in 7 of 8 PAVMs, 6 using only covered stents and 1 using both a covered and a bare stent owing to endoleak. In 1 patient, the parent vessel was sacrificed after identification of additional feeding vessels following stent graft placement. In 6 patients with median imaging follow-up of 8 months (range, 1-121 months), all stent grafts were patent, and all treated PAVMs were completely excluded without persistence.

BACKGROUND: Lower-grade gliomas (LGGs, defined as WHO grades II and III) with 1p19q codeletion have increased chemosensitivity when compared to LGGs without 1p19q codeletion, but the mechanism is currently unknown. METHODS: RNAseq data from 515 LGG patients in the Cancer Genome Atlas (TCGA) were analyzed to compare the effect of expression of the 9 DNA repair genes located on chromosome arms 1p and 19q on progression free survival (PFS) and overall survival (OS) between patients who received chemotherapy and those who did not. Chemosensitivity of cells with DNA repair genes knocked down was tested using MTS cell proliferation assay in HS683 cell line and U251 cell line. RESULTS: The expression of 9 DNA repair genes on 1p and 19q was significantly lower in 1p19q-codeleted tumors (n = 175) than in tumors without the codeletion (n = 337) (p < 0.001). In LGG patients who received chemotherapy, lower expression of LIG1, POLD1, PNKP, RAD54L and MUTYH was associated with longer PFS and OS. This difference between chemotherapy and non-chemotherapy groups in the association of gene expression with survival was not observed in non-DNA repair genes located on chromosome arms 1p and 19q. MTS assays showed that knockdown of DNA repair genes LIG1, POLD1, PNKP, RAD54L and MUTYH significantly inhibited recovery in response to temozolomide when compared with control group (p < 0.001). CONCLUSIONS: Our results suggest that reduced expression of DNA repair genes on chromosome arms 1p and 19q may account for the increased chemosensitivity of LGGs with 1p19q codeletion.

STUDY DESIGN Retrospective cohort analysis of patients with spinal astrocytoma from multi-institutional data and the literature. OBJECTIVE To determine the prognostic factors, treatment, and survival of patients. SUMMARY OF BACKGROUND DATA Our current understanding of the epidemiology, prognosis, and optimal treatment of spinal astrocytoma is limited. The literature is confined to case reports or small institutional case series. METHOD Patient demographics, tumor characteristics, treatments, and outcomes were extracted. Univariate Kaplan-Meier survival analysis was performed to identify prognostic factors followed by multivariate Cox proportional hazard analysis. Wilcoxon signed-rank test was performed on pre- and postoperational functional status as measured by McCormick score. RESULTS Ninety-four patients from four institutions and 339 patients from the literature were included. For the multi-institutional cohort, WHO grade IV tumors had shorter progression-free survival (PFS) than those of lower grades, whereas gross total resection (GTR) (hazard ratio [HR]: 0.41, 95% confidence interval [CI]: 0.14-1.27, P = 0.124) trended toward longer PFS when compared to subtotal resection (STR). Age 18 years or older, paresthesia as a presenting symptom, and higher WHO grade were associated with shorter overall survival (OS), whereas thoracic tumor location when compared to cervical tumor location, biopsy when compared to STR, and radiotherapy (HR: 0.42, 95% CI: 0.20-0.88, P = 0.022) were associated with longer OS. For the literature cohort, GTR (HR 0.43, 95% CI: 0.24-0.77, P = 0.005) was associated with longer PFS when compared to STR, whereas higher WHO grade was associated with shorter PFS. Higher WHO grade and recurrence/progression were associated with shorter OS. Postoperative McCormick score was significantly higher than preoperative score (P

It remains controversial whether concurrent or subsequent chemoradiotherapy (CRT) provides additional survival benefits when compared to radiotherapy (RT) alone in localized extranodal natural killer/T-cell lymphoma (ENKTL), nasal type We identified 248 patients from the US National Cancer Data Base who were diagnosed with localized ENKTL from 2004-2014; 68.9% received CRT and 31.1% received RT alone. Over time, the use of CRT increased, while the use of RT alone decreased (p .001). On multivariate analysis, CRT was associated with longer OS than RT alone (HR: 0.504; 95% CI: 0.338-0.751; p .001), while uninsured status and African-American race were associated with shorter OS. The survival advantages of CRT over RT alone persisted on propensity score matching for the entire cohort (p = .0014) and in a subgroup analysis of elderly patients (p .001). In conclusion, patients who received CRT had significantly longer OS than those who received RT alone. These results also apply to elderly patients.

Background The low incidence of spinal chordoma precludes a prospective study of prognostic factors with a large patient cohort. Objective To perform a comprehensive integrative analysis on the prognostic factors, treatment, and outcomes of patients with spinal chordoma using data from 2 institutions and the literature. Methods Appropriate studies were identified per search criteria. The local database was retrospectively searched to include a similar patient cohort. Results Overall, 108 studies from the literature and 30 patients from our local institution were identified, resulting in a total of 682 patients. The median age was 57 years old and 35.2% were female. The median follow-up was 46 months (range: 1-408). The median progression-free survival (PFS) and overall survival (OS) were 72 months and 115 months, respectively. Significant prognostic factors for PFS on multivariate analysis included age (pediatric vs adult, hazard ratio [HR]: 2.00-14.36), tumor location (mobile spine vs sacral spine, HR: 0.31-0.87), pathology (differentiated vs classic, HR: 2.48-10.90),and chemotherapy (HR:1.11-3.85). Significant prognostic factors for OS on multivariate analysis included age (geriatric vs adult, HR: 1.52-3.45 and pediatric vs adult, HR: 1.73-9.36), bladder or bowel dysfunction (HR: 1.27-5.43), pathology (dedifferentiated vs classic, HR: 2.38-11.09), recurrence or progression (HR: 1.72-4.48), and metastases (HR: 1.11-2.47). Conclusion In patients with spinal chordoma, young age, location in sacral spine, dedifferentiated pathology, and chemotherapy were negative predictors of PFS, while young and old age, bladder or bowel dysfunction at presentation, dedifferentiated pathology, recurrence or progression, and metastases portended a worse OS.

Purpose To develop and validate a deep learning model based on routine magnetic resonance (MR) imaging obtained before uterine fibroid embolization to predict procedure outcome. Materials and Methods Clinical data were collected on patients treated with uterine fibroid embolization at the Hospital of the University of Pennsylvania from 2007 to 2018. Fibroids for each patient were manually segmented by an abdominal radiologist on a T1-weighted contrast-enhanced (T1C) sequence and a T2-weighted sequence of MR imaging obtained before and after embolization. A residual convolutional neural network (ResNet) model to predict clinical outcome was trained using MR imaging obtained before the procedure. Results Inclusion criteria were met by 727 fibroids in 409 patients. At clinical follow-up, 85.6% (n = 350) of 409 patients (590 of 727 fibroids; 81.1%) experienced symptom resolution or improvement, and 14.4% (n = 59) of 409 patients (137 of 727 fibroids; 18.9%) had no improvement or worsening symptoms. The T1C trained model achieved a test accuracy of 0.847 (95% confidence interval [CI], 0.745–0.914), sensitivity of 0.932 (95% CI, 0.833–0.978), and specificity of 0.462 (95% CI, 0.232–0.709). In comparison, the average of 4 radiologists achieved a test accuracy of 0.722 (95% CI, 0.609–0.813), sensitivity of 0.852 (95% CI, 0.737–0.923), and specificity of 0.135 (95% CI, 0.021–0.415). Conclusions This study demonstrates that deep learning based on a ResNet model achieves good accuracy in predicting outcome of uterine fibroid embolization. If further validated, the model may help clinicians better identify patients who can most benefit from this therapy and aid clinical decision making.

To assess a novel tibial artery perfusion score (TPS) for predicting limb salvage in critical limb ischemia (CLI) patients undergoing percutaneous vascular intervention (PVI). A consecutive cohort of 115 CLI patients undergoing PVI in 144 limbs from 2011 to 2016 was analyzed. TPS comprised a 27-point scale based on: (1) patent tibial vessels following PVI, (2) severity of calcification of the tibial arteries, (3) presence of an intact pedal arch following intervention, (4) whether or not revascularization was direct or indirect based on the target angiosome, (5) presence of angiosome blush at the completion of index intervention. Limbs were stratified into (1) High [21–27 points], (2) Medium [13–20 points], and (3) Low [0–12 points] TPS. Predictive value of TPS was evaluated using logistic regression and Cox proportional hazards models. The median follow-up was 15.7 months (range 0.4–69.9 months). Limb salvage in High, Medium, and Low TPS groups was 90.6%, 85.9%, and 55.6%, respectively, as freedom from the composite outcome: (1) limb complication resulting in death, (2) tibial bypass surgery, (3) above-the-knee amputation, or (4) below-the-knee amputation in patients without supratibial disease at the time of PVI. TPS was significantly associated with limb salvage defined as freedom from both the composite outcome and major amputation. Based on this preliminary investigation, TPS was associated with limb salvage in CLI limbs, particularly in high-risk limbs. Further validation in a prospective cohort may identify patients with high-risk limbs in need of closer surveillance and earlier reintervention. Level IV, case series.

PURPOSE: Isocitrate dehydrogenase (IDH) and 1p19q codeletion status are importantin providing prognostic information as well as prediction of treatment response in gliomas. Accurate determination of the IDH mutation status and 1p19q co-deletion prior to surgery may complement invasive tissue sampling and guide treatment decisions. METHODS: Preoperative MRIs of 538 glioma patients from three institutions were used as a training cohort. Histogram, shape, and texture features were extracted from preoperative MRIs of T1 contrast enhanced and T2-FLAIR sequences. The extracted features were then integrated with age using a random forest algorithm to generate a model predictive of IDH mutation status and 1p19q codeletion. The model was then validated using MRIs from glioma patients in the Cancer Imaging Archive. RESULTS: Our model predictive of IDH achieved an area under the receiver operating characteristic curve (AUC) of 0.921 in the training cohort and 0.919 in the validation cohort. Age offered the highest predictive value, followed by shape features. Based on the top 15 features, the AUC was 0.917 and 0.916 for the training and validation cohort, respectively. The overall accuracy for 3 group prediction (IDH-wild type, IDH-mutant and 1p19q co-deletion, IDH-mutant and 1p19q non-codeletion) was 78.2% (155 correctly predicted out of 198). CONCLUSION: Using machine-learning algorithms, high accuracy was achieved in the prediction of IDH genotype in gliomas and moderate accuracy in a three-group prediction including IDH genotype and 1p19q codeletion.

Indeterminate cell histiocytosis (ICH) is a very rare histiocytic disease. We reported a case of generalized ICH presenting as eroded papules and crusts, accompanied with intermittent fever, pulmonary disorder, lymphadenectasis in the mediastinum, and the retroperitoneum. The histopathology indicated a neoplasm that was composed of small nests of large cells with ample pale cytoplasm. Immunohistochemistry results and the absence of Birbeck granules led us to the diagnosis of ICH. Four months later, the patient died of anemia, ascites, and swelling of the limbs.

Purpose To compare overall survival (OS) of ablation with no treatment for patients with advanced stage non-small cell lung cancer. Methods Patients with clinical stage IIIB (T1–4N3M0, T4N2M0) and stage IV (T1–4N0–3M1) non-small cell lung cancer, in accordance with the American Joint Committee on Cancer, 7th edition, who did not receive treatment or who received ablation as their sole primary treatment besides chemotherapy from 2004 to 2014, were identified from the National Cancer Data Base. OS was estimated using the Kaplan-Meier method and evaluated by log-rank test, univariate and multivariate Cox proportional hazard regression, and propensity score-matched analysis. Relative survival analyses comparing age- and sex-matched United States populations were performed. Results A total of 140,819 patients were included. The 1-, 2-, 3- and 5-year survival rates relative to age- and sex-matched United States population were 28%, 18%, 12%, and 10%, respectively, for ablation (n = 249); and 30%, 15%, 9%, and 5%, respectively for no treatment (n = 140,570). Propensity score matching resulted in 249 patients in the ablation group versus 498 patients in the no-treatment group. After matching, ablation was associated with longer OS than that in the no-treatment group (median, 5.9 vs 4.7 months, respectively; hazard ratio, 0.844; 95% confidence interval, 0.719–0.990; P = .037). These results persisted in patients with an initial tumor size of ≤3 cm. Conclusions Preliminary results suggest ablation may be associated with longer OS in patients with late-stage non-small cell lung cancer than survival in those who received no treatment.

To compare the overall survival (OS) outcomes of sublobar resection (SLR) with stereotactic body radiation therapy (SBRT) or ablation for patients with early stage non-small cell lung cancer (NSCLC).Patients with clinical stage I (T1-T2aN0M0) NSCLC from 2004 to 2014 who were treated with SLR, SBRT, or ablation as the sole treatment were identified from the National Cancer Database. OS was estimated using the Kaplan-Meier method and evaluated by log-rank test, univariate and multivariate Cox proportional hazard regression, and propensity score-matched analysis. Relative survival analyses compared with age- and sex-matched US population were performed.A total of 53,973 patients were identified. The 1-, 2-, 3-, and 5-year relative survival rates were 96%, 90%, 84%, and 71% for SLR (n = 30,451); 93%, 78%, 65%, and 46% for SBRT (n = 22,134); and 90%, 73%, 58%, and 37% for ablation (n = 1388). Propensity score matching resulted in 9967 patients in the SBRT group versus 9967 in the SLR group and 1062 patients in the ablation group versus 1984 in the SLR group. After matching, both SBRT (hazard ratio, 1.559; 95% confidence interval, 1.497-1.623; P .001) and ablation (hazard ratio, 1.906; 95% confidence interval, 1.730-2.101; P .001) were associated with shorter OS when compared with SLR. These results persisted in patients with tumor size ≤2 cm.Preliminary results suggest SLR may be associated with longer OS in patients with early-stage NSCLC compared with SBRT or ablation. Future prospective, randomized, controlled clinical trials comparing these treatments are needed to confirm these results.

Background The combined use of chemotherapy (CT) and radiotherapy (RT) is becoming increasingly common in the treatment of surgically resected low-grade gliomas. However, whether RT or CT is associated with improved overall survival (OS) for low-grade gliomas without surgical resection is less clear. Methods The U.S. National Cancer Data Base was used to identify patients with histologically confirmed World Health Organization (WHO) grade 2 gliomas who received either RT alone or CT alone but did not undergo surgical resection from 2004 to 2013. OS was evaluated with Kaplan-Meier analysis, multivariable Cox proportional hazard regression, and propensity score-matched analysis. Results In total, 1126 patients with WHO grade 2 gliomas were included, among whom 715 (63.5%) received RT alone and 411 (36.5%) CT alone. CT alone was associated with significantly longer OS when compared to RT alone on multivariable analysis and propensity-score matched analysis. Age less than 60 years and time from diagnosis to start of adjuvant therapy (RT or CT) greater than 30 days were also associated with longer OS. In subgroup analysis based on age, the survival advantage of CT alone over RT alone persisted in both age subgroups ( 60 years). In subgroup analyses based on histology of glioma, CT alone was associated with longer OS in the astrocytoma and oligodendroglioma group, but not in the oligoastrocytoma group. Conclusions Our results suggest that CT alone was independently associated with longer OS when compared with RT alone in patients with low-grade glioma but without surgical resection. This survival advantage persisted in both age subgroups, but it varied by histology of glioma.

Due to its rarity, current literature assessing prognostic factors and survival outcomes of hemangioblastoma is limited. Patients with histologically confirmed hemangioblastoma were identified from the US National Cancer Data Base. 1488 patients met inclusion criteria. 644 patients underwent gross total resection (GTR), 220 subtotal resection (STR)/biopsy, 60 stereotactic radiosurgery (SRS), 15 external beam radiotherapy (EBRT), 51 surgery followed by radiotherapy (SR + RT) and 498 no treatment. Independent predictors of shorter OS included age ≥ 40 (HR, 3.897; 95% CI, 2.341–6.487; p

Purpose To compare the overall survival (OS) of patients receiving cryoablation versus heat-based thermal ablation for clinical T1a renal cell carcinoma (RCC) in a large national cohort. Materials and Methods Patients with RCC from 2004 to 2014 who were treated with ablation were identified from the National Cancer Database. OS was estimated with the use of the Kaplan-Meier method and evaluated by means of log-rank test, univariate and multivariate Cox proportional hazard regression, and propensity score–matched analysis. Results A total of 3,936 patients who received cryoablation and 2,322 who received heat-based thermal ablation met the inclusion criteria. The mean age was 67 ± 12 year, and the mean size of tumors was 25 ± 8 mm. The 3-, 5-, and 10-year survival rates were, respectively, 91%, 82%, and 62% for cryoablation and 89%, 81%, and 55% for heat-based thermal ablation. After propensity score matching, cryoablation was associated with longer OS compared with heat-based thermal ablation (median 11.3 vs 10.4 years; hazard ratio 1.175, 95% CI 1.03–1.341; P = .016). For patients with tumors ≤2 cm, propensity score–matched analyses demonstrated no significant difference between the 2 treatment groups (P = .772). Conclusions Overall, cryoablation may be associated with longer OS compared with heat-based thermal ablation in cT1a RCC. No significant difference in survival rates was observed between the 2 treatments for patients with tumor sizes ≤2 cm. Owing to the inherent limitations of this study, further study with details on technology, local outcome, and complications is needed.

Purpose To compare survival outcome of radiofrequency (RF) ablation and surgical resection (SR) for treatment of hepatocellular carcinoma (HCC) ≤ 2 cm. Materials and Methods In this retrospective study, patients from the US National Cancer Database with HCC ≤ 2 cm received RF ablation or SR as sole treatment. Overall survival (OS) was compared using log-rank test, multivariable Cox proportional hazard regression, and propensity score matched analysis. Results Of 833 patients included, 620 received RF ablation and 213 received SR. The 1-, 3-, and 5-year OS rates were 90%, 64%, and 47% for RF ablation and 89%, 75%, and 62% for SR. On univariate analyses, patients who received SR had longer OS than patients who received RF ablation, but this did not achieve statistical significance (P = .113). On multivariate analyses, female sex (HR = 0.700; 95% CI, 0.501–0.979; P = .037), African American (HR = 0.611; 95% CI, 0.398–0.938; P = .024) and Asian ethnicity (HR = 0.427; 95% CI, 0.230–0.790; P = .007), and median income ≥ $48,000 (HR = 0.695; 95% CI, 0.518–0.932; P = .015) were associated with longer OS, whereas higher Model for End-stage Liver Disease (MELD) scores (HR = 1.023; 95% CI, 1.009–1.037; P = .001) were associated with shorter OS. After matching on age, sex, ethnicity, MELD score, and income, there was no significant difference in OS between the 2 treatment groups (log-rank P = .646). Conclusions There was no significant difference in OS between RF ablation and SR in treatment of HCC measuring ≤ 2 cm.

It is becoming increasingly common to incorporate chemotherapy (CT) with radiotherapy (RT) in the treatment of low-grade gliomas (LGGs) after surgical resection. However, there is a lack of literature comparing survival of patients who underwent RT or CT alone.The U.S. National Cancer Data Base was used to identify patients with histologically confirmed, World Health Organization grade 2 gliomas who received either RT alone or CT alone after surgery from 2004 to 2013. Overall survival (OS) was evaluated by Kaplan-Meier analysis, multivariable Cox proportional hazard regression, and propensity-score-matched analysis.In total, 2253 patients with World Health Organization grade 2 gliomas were included, of whom 1466 (65.1%) received RT alone and 787 (34.9%) CT alone. The median OS was 98.9 months for the RT alone group and 125.8 months for the CT alone group. On multivariable analysis, CT alone was associated with a significant OS benefit compared with RT alone (hazard ratio [HR], 0.405; 95% confidence interval, 0.277-0.592; P0.001). On subgroup analyses, the survival advantage of CT alone over RT alone persisted across all age groups, and for the subtotal resection and biopsy groups, but not in the gross total resection group. In propensity-score-matched analysis, CT alone still showed significantly improved OS compared with RT alone (HR, 0.612; 95% confidence interval, 0.506-0.741; P0.001).Our results suggest that CT alone was independently associated with longer OS compared with RT alone in patients with LGGs who underwent surgery.

Systemic anaplastic lymphoma kinase positive (ALK+) anaplastic large cell lymphoma with extranodal involvement (ALCL-E) is a rare form of non-Hodgkin lymphoma. No large study in the literature has compared the survival outcomes among different primary extranodal sites of involvement in ALK+ ALCL-E. We identified 1306 patients with ALK+ ALCL-E diagnosed between 2004 and 2014 in the US National Cancer Database, among whom 387 had primary extranodal site in the chest/abdomen/pelvis, 103 in the bone, 62 in the central nervous system, 134 in the head and neck and 620 in the cutaneous/soft tissue. Younger age, lower Charlson-Deyo score, lower clinical stage, receipt of chemotherapy and receipt of radiotherapy were predictors of longer overall survival. Patients with extranodal involvement of central nervous system and chest/abdomen/pelvis had shorter overall survival than those with involvement of head and neck, bone, and cutaneous/subcutaneous tissue after adjusting for confounding variables. We recommend treating these patients upfront with more aggressive therapy.

PURPOSE: Isocitrate dehydrogenase (IDH) mutations in glioma patients confer longer survival and may guide treatment decision-making. We aimed to predict the IDH status of gliomas from MR imaging by applying a residual convolutional neural network to pre-operative radiographic data. EXPERIMENTAL DESIGN: Preoperative imaging was acquired for 201 patients from the Hospital of University of Pennsylvania (HUP), 157 patients from Brigham and Women’s Hospital (BWH), and 138 patients from The Cancer Imaging Archive (TCIA) and divided into training, validation, and testing sets. We trained a residual convolutional neural network for each MR sequence (FLAIR, T2, T1 pre-contrast, and T1 post-contrast) and built a predictive model from the outputs. To increase the size of training set and prevent overfitting, we augmented the training set images by introducing random rotations, translations, flips, shearing, and zooming. RESULTS: With our neural network model, we achieved IDH prediction accuracies of 82.8% (AUC = 0.90), 83.0% (AUC = 0.93), and 85.7% (AUC = 0.94) within training, validation, and testing sets, respectively. When age at diagnosis was incorporated into the model, the training, validation, and testing accuracies increased to 87.3% (AUC = 0.93), 87.6% (AUC = 0.95), and 89.1% (AUC = 0.95), respectively. CONCLUSION: We developed a deep learning technique to non-invasively predict IDH genotype in grade II-IV glioma using conventional MR imaging using a multi-institutional dataset.

Background Prior studies suggest that neuroblastomas that do not accumulate metaiodobenzylguanidine (MIBG) on diagnostic imaging (MIBG non-avid) may have more favorable features compared with MIBG avid tumors. We compared clinical features, biologic features, and clinical outcomes between patients with MIBG nonavid and MIBG avid neuroblastoma. Procedure Patients had metastatic high- or intermediate-risk neuroblastoma and were treated on Children's Oncology Group protocols A3973 or A3961. Comparisons of clinical and biologic features according to MIBG avidity were made with chi-squared or Fisher exact tests. Event-free (EFS) and overall (OS) survival compared using log–rank tests and modeled using Cox models. Results Thirty of 343 patients (8.7%) had MIBG nonavid disease. Patients with nonavid tumors were less likely to have adrenal primary tumors (34.5 vs. 57.2%; P = 0.019), bone metastases (36.7 vs. 61.7%; P = 0.008), or positive urine catecholamines (66.7 vs. 91.0%; P < 0.001) compared with patients with MIBG avid tumors. Nonavid tumors were more likely to be MYCN amplified (53.8 vs. 32.6%; P = 0.030) and had lower norepinephrine transporter expression. Patients with MIBG nonavid disease had a 5-year EFS of 50.0% compared with 38.7% for patients with MIBG avid disease (P = 0.028). On multivariate testing in high-risk patients, MIBG avidity was the sole adverse prognostic factor for EFS identified (hazard ratio 1.77; 95% confidence interval 1.04–2.99; P = 0.034). Conclusions Patients with MIBG nonavid neuroblastoma have lower rates of adrenal primary tumors, bone metastasis, and catecholamine secretion. Despite being more likely to have MYCN-amplified tumors, these patients have superior outcomes compared with patients with MIBG avid disease.

Background Studies have shown contradictory results regarding the survival outcomes among white, African American, and Asian patients with mycosis fungoides (MF). Objective To evaluate the survival outcomes among white, African American, and Asian patients with MF and to determine other prognostic factors of the disease. Methods The US National Cancer Database was used to identify patients with histologically confirmed MF from 2004 to 2014. Clinicopathologic, socioeconomic, and treatment data were compared among the races by using the chi-square test. Overall survival was evaluated by using the log-rank test, multivariable Cox proportional hazard regression, and propensity score–matched analysis. Results Of 4459 patients with MF, 77.7% were white, 19.2% were African American, and 3.2% were Asian. Older age, treatment received in a community facility, government insurance, higher Charlson-Deyo score, male sex, higher clinical stage, receipt of radiotherapy or chemotherapy, and African American race were predictors of poor overall survival on multivariate analysis ( P P = .07). Limitations Retrospective analysis. Conclusion African American patients with MF demonstrated poorer survival than white patients after accounting for disease characteristics, socioeconomic factors, and types of treatment, warranting further investigation into the underlying biology of MF and prescribed treatment modalities.

To determine the risk of stroke in patients with subclavian steal syndrome (SSS). We identified 165 patients with imaging-provenSSS from two hospitals. Demographic, clinical and imaging data were retrospectively collected. Patients were followed up for stroke events. Stroke occurred in 43 patients with a median follow-up of 28months. Seven of these cases were identified prospectively and 36 cases retrospectively. On multivariate analysis, presence of symptoms at presentation (p=0.029) was a significant predictor of stroke. Presence of symptoms at presentation predicted stroke in imaging-proven SSS.

To compensate for the lack of disease-specific survival data in the study by Rae et al., the US National Cancer Data Base was used to identify patients with histologically confirmed central neurocytoma from 2004 to 2013 based on the same inclusion criteria and relative survival analysis was performed. A total of 781 patients were identified. The 5-year overall survival (OS) for the overall cohort was 87.2%, and the 5-year survival relative to the expected survival of an age-, and gender-matched US standard population was 89.5% (95% CI 86.6%-91.9%). In summary, the relative survival analysis presented here indicates that central neurocytoma was a component of the cause of mortality in the affected US population.

A 35-year-old Chinese woman presented with a 2.5-year history of facial swelling in the left lower quadrant and a 10-month history of relapsing red papules and vesicles in the perioral area resembling hydroa vacciniforme. Histologically, a tissue biopsy showed a dense infiltration of medium-sized atypical lymphocytic cells expressing CD4 and CD56. A diagnosis of cutaneous NK-/T-cell lymphoma was made. The patient was treated with alpha-interferon, valaciclovir hydrochloride, and low-dose prednisone for 2 months. Her skin lesions and lymphoadenopathy resolved initially, but she succumbed to the disease shortly after starting chemotherapy treatment 11 months later. To our knowledge, this is the first case of CD4CD56 NK-/T-cell lymphoma with clinical features resembling hydroa vacciniforme.

We read with great interest a recent paper by Mackie et al .1 The authors found that an extracapsular pattern of inflammatory involvement on whole-body, multiple-joint MRI was associated with complete glucocorticoid response, higher pretreatment C-reactive protein and serum interleukin-6. We commend the authors for studying such a challenging disease with advanced imaging and linking it to treatment response. However, we have several concerns. First, it is unclear what criteria were used to distinguish an MRI of ‘extracapsular pattern’ from that of ‘non-extracapsular pattern’. Since at least eight large joints were evaluated bilaterally, are there patients with an extracapsular pattern in certain joints but not in others? …

Nodular goiter is one of the most common benign lesions in thyroid nodule. The main treatment of the disease is still the traditional surgical resection, however there are many problems such as general anesthesia, surgical scar, postoperative thyroid or parathyroid function abnormalities, and high nodules recurrence rate in residual gland. The purpose of this study was to compare the efficacy, safety, and cost-effectiveness of 2 treatment methods, surgery and radiofrequency ablation, for the treatment of benign thyroid nodules.From May 2012 to September 2013, 200 patients with nodular goiters who underwent surgery (group A) and 200 patients treated by radiofrequency ablation (group B) were enrolled in this study. Inclusion criteria were the following: 1) cosmetic problem, 2) nodule-related symptoms, 3) hyperfunctioning nodules related to thyrotoxicosis, and 4) refusal of surgery (for group B). An internally cooled radiofrequency ablation system and an 18-ga internally cooled electrode were used. We compared the 2 groups in terms of efficacy, safety, and cost-effectiveness during a 1-year follow-up.After radiofrequency ablation, the nodule volume decreased significantly from 5.4 to 0.4 mL (P = .002) at the 12-month follow-up. The incidence of complications was significantly higher from surgery than from radiofrequency ablation (6.0% versus 1.0%, P = .002). Hypothyroidism was detected in 71.5% of patients after surgery but in none following radiofrequency ablation. The rate of residual nodules (11.9% versus 2.9%, P = .004) and hospitalization days was significantly greater after surgery (6.6 versus 2.1 days, P.001), but the cost difference was not significant.Surgical resection and radiofrequency ablation are both effective treatments of nodular goiter. Compared with surgery, the advantages of radiofrequency ablation include fewer complications, preservation of thyroid function, and fewer hospitalization days. Therefore, radiofrequency ablation should be considered a first-line treatment for benign thyroid nodules.

Medications are a major cause of acute pancreatitis; however, little is known about their influence in children. Our primary aims were to identify common comorbidities and concomitant pancreatitis etiologies in children with drug-associated pancreatitis. Our secondary aims were to identify the most commonly associated drugs in the different age groups, evaluate management practices, and compare drug-associated cases with non-drug-associated cases.In the present study, we examined children (ages 0-20 years) admitted to Yale-New Haven Children's Hospital with pancreatitis between 1994 and 2007.Of a total of 271 pediatric cases, drugs were associated with pancreatitis in 25.6% (55). The 3 most common comorbidities in children with drug-associated pancreatitis were seizure disorders, acute lymphocytic leukemia, and Crohn disease. One third of drug-associated cases had an additional pancreatitis etiology. The most commonly associated drugs were valproic acid and corticosteroids. Compared with non-drug-associated cases, children with drug-associated cases were more likely to undergo CT scanning (54.5% vs 28.4%; P0.001), stay in the hospital longer (10 vs 4 days; P0.001), and transition to parenteral nutrition from a nil per os status (37.5% vs 21.2%; P0.05). There was a higher frequency of valproic acid-associated cases in children younger than 11 years (29.4% vs 9.5% in the 11- to 20-year-old age group).Our study underscores the importance of considering drugs as a cause and a contributor to pancreatitis in children, particularly valproic acid in young children.

We read with interest a recent article by Tsivgoulis et al1 on the safety of intravenous thrombolysis (IVT) in stroke mimics (SM). Through both their own institutional data and meta-analysis of 8 published series, the authors found a low risk of symptomatic intracranial hemorrhage among SMs receiving recombinant tissue-type plasminogen activator (tPA). Because tPA does not adversely affect the favorable natural history of SMs, the authors argued that delaying IVT to rule out a stroke mimic is not warranted in terms of safety. The decision to treat a patient with IVT for acute ischemic stroke (AIS) is often based on the patient’s medical history, clinical presentation, laboratory test results, and computed tomographic (CT) findings because of the limited time window thrombolysis is feasible for AIS. As a consequence, there will always be some patients treated with tPA who have no ischemic stroke but another medical condition mimicking a stroke. Because tPA has been proven safe in these patients, can we accept SM as unavoidable when managing suspected AIS? The answer is …

We read with interest a recent paper by Almenawer et al.1 The authors performed a meta-analysis on biopsy versus partial versus gross total resection (GTR) in older patients with high-grade gliomas (HGGs). The authors compared patients undergoing overall resection (of any extent) with patients undergoing biopsy and found significantly longer overall survival and progression-free-survival, higher postoperative Karnofsky performance status (KPS), and lower morbidity and mortality. GTR was found to be superior to subtotal resection in terms of overall survival, postoperative KPS, and progression-free survival with no difference in morbidity and mortality. The authors concluded that maximum resections are safe and are associated with longer survival time, delayed tumor progression rates, and improved functional recovery. We would like to commend the authors on their efforts to examine the impact of extent of resection on survival and functional outcomes for elderly patients with HGG. The benefit of maximal resection is uncertain in this age group since older patients tend to have more comorbidities, higher complication rates, and poorer prognosis overall. In addition, fundamental differences in tumor biology may explain some of the correlation between age and outcome.1 However, we have some doubts about the methodology and the conclusions reached by the authors. First, it is quite possible that biopsy was used for diagnostic purposes only in some of the studies included in the meta-analysis. In fact, this was clearly stated in the methods section of 3 studies.2–4 We believe it is not meaningful to include these studies in the meta-analysis since the goal of the paper was to compare outcomes in elderly patients who underwent different extents of resection, not between patients who had surgery versus those who did not. Secondly, the authors admitted that the greatest limitation in their meta-analysis was the inability to account for similar distribution of patient' characteristics. We agree with the authors that variables such as patient age, brain tumor location and size, adjuvant therapy, and preoperative baseline functional status should be taken into account when performing meta-analyses. Without looking at these factors, the conclusion reached (ie, that increasing extents of tumor resection are associated with improved outcomes) suffers from selection bias. However, these variables were presented in such a way in the individual studies that it is almost impossible to incorporate them into the meta-analysis (eg, using meta-regression). Therefore, we attempted to compare age, preoperative KPS, tumor location and size, and adjuvant treatment between biopsy and resection cohorts in the individual studies via qualitative analysis (Table 1). In the 12 studies that separated the age groups for the biopsy and resection patients, 11 reported either a higher proportion of older patients undergoing biopsy,2,5–8 or a higher mean age for the biopsy group.4,9–12 The biopsy group had a lower preoperative KPS score in 5 of the 9 studies that reported this information.2,6,12–14 The resection group was associated with a larger tumor size in the 2 studies that listed tumor size separately for the resection and biopsy groups.8,13 In the 4 studies that compared tumor location between biopsy and resection groups, patients with corpus callosum, deep lesions (eg, brainstem), and bilateral lesions were more likely to undergo biopsy,8,13,15 while endangerment of the eloquent area seemed to play little role in the decision-making process.14 In the 6 studies that reported number of patients receiving chemotherapy in the 2 groups, the proportion of patients receiving chemotherapy was higher in the resection group than that in the biopsy group in all 6 studies,2,10,11,13,14,16 while the proportion of patients receiving radiotherapy was higher in only 4 studies.2,10,11,13 Table 1. Biopsy versus resection in elderly patients with high-grade glioma In summary, even among the elderly, older patients tended to undergo biopsy rather than resection of their HGGs. Tumors located in the corpus callosum, deep lesions, and bilateral lesions were more likely to be biopsied than resected. Elderly patients undergoing resection of their HGG were more likely to receive chemotherapy than those who underwent biopsy. Consequently, the favorable outcomes associated with greater extent of resection in elderly patients may be influenced by the patient selection factors listed above and should be interpreted with caution.