Citations

ObjectiveTo compare the diagnostic performance of an artificial intelligence deep learning system with that of expert neuro‐ophthalmologists in classifying optic disc appearance.MethodsThe deep learning system was previously trained and validated on 14,341 ocular fundus photographs from 19 international centers. The performance of the system was evaluated on 800 new fundus photographs (400 normal optic discs, 201 papilledema [disc edema from elevated intracranial pressure], 199 other optic disc abnormalities) and compared with that of 2 expert neuro‐ophthalmologists who independently reviewed the same randomly presented images without clinical information. Area under the receiver operating characteristic curve, accuracy, sensitivity, and specificity were calculated.ResultsThe system correctly classified 678 of 800 (84.7%) photographs, compared with 675 of 800 (84.4%) for Expert 1 and 641 of 800 (80.1%) for Expert 2. The system yielded areas under the receiver operating characteristic curve of 0.97 (95% confidence interval [CI] = 0.96–0.98), 0.96 (95% CI = 0.94–0.97), and 0.89 (95% CI = 0.87–0.92) for the detection of normal discs, papilledema, and other disc abnormalities, respectively. The accuracy, sensitivity, and specificity of the system's classification of optic discs were similar to or better than the 2 experts. Intergrader agreement at the eye level was 0.71 (95% CI = 0.67–0.76) between Expert 1 and Expert 2, 0.72 (95% CI = 0.68–0.76) between the system and Expert 1, and 0.65 (95% CI = 0.61–0.70) between the system and Expert 2.InterpretationThe performance of this deep learning system at classifying optic disc abnormalities was at least as good as 2 expert neuro‐ophthalmologists. Future prospective studies are needed to validate this system as a diagnostic aid in relevant clinical settings. ANN NEUROL 2020;88:785–795

Objectives: To characterize the association between the use of physiologic assessment (central venous pressure, pulmonary artery occlusion pressure, stroke volume variation, pulse pressure variation, passive leg raise test, and critical care ultrasound) with fluid and vasopressor administration 24 hours after shock onset and with in-hospital mortality. Design: Multicenter prospective cohort study between September 2017 and February 2018. Settings: Thirty-four hospitals in the United States and Jordan. Patients: Consecutive adult patients requiring admission to the ICU with systolic blood pressure less than or equal to 90 mm Hg, mean arterial blood pressure less than or equal to 65 mm Hg, or need for vasopressor. Interventions: None. Measurement and Main Results: Of 1,639 patients enrolled, 39% had physiologic assessments. Use of physiologic assessment was not associated with cumulative fluid administered within 24 hours of shock onset, after accounting for baseline characteristics, etiology and location of shock, ICU types, Acute Physiology and Chronic Health Evaluation III, and hospital (beta coefficient, 0.04; 95% CI, −0.07 to 0.15). In multivariate analysis, the use of physiologic assessment was associated with a higher likelihood of vasopressor use (adjusted odds ratio, 1.98; 95% CI, 1.45−2.71) and higher 24-hour cumulative vasopressor dosing as norepinephrine equivalent (beta coefficient, 0.37; 95% CI, 0.19−0.55). The use of vasopressor was associated with increased odds of in-hospital mortality (adjusted odds ratio, 1.88; 95% CI, 1.27−2.78). In-hospital mortality was not associated with the use of physiologic assessment (adjusted odds ratio, 0.86; 95% CI, 0.63−1.18). Conclusions: The use of physiologic assessment in the 24 hours after shock onset is associated with increased use of vasopressor but not with fluid administration.

Objectives: High throughput pre-treatment imaging features may predict radiation treatment outcome and guide individualized treatment in radiotherapy (RT). Given relatively small patient sample (as compared with high dimensional imaging features), identifying potential prognostic imaging biomarkers is typically challenging. We aimed to develop robust machine learning methods for patient survival prediction using pre-treatment quantitative CT image features for a subgroup of head-and-neck cancer patients. Methods: Three neural network models, including back propagation (BP), Genetic Algorithm-Back Propagation (GA-BP), and Probabilistic Genetic Algorithm-Back Propagation (PGA-BP) neural networks were trained to simulate association between patient survival and radiomics data in radiotherapy. To evaluate the models, a subgroup of 59 head-and-neck patients with primary cancers in oral tongue area were utilized. Quantitative image features were extracted from planning CT images, a novel t-Distributed Stochastic Neighbor Embedding (t-SNE) method was used to remove irrelevant and redundant image features before fed into the network models. 80% patients were used to train the models, and remaining 20% were used for evaluation. Results: Of the three supervised machine-learning methods studied, PGA-BP yielded the best predictive performance. The reported actual patient survival interval of 30.5 ± 21.3 months, the predicted survival times were 47.3 ± 38.8, 38.5 ± 13.5 and 29.9 ± 15.3 months using the traditional PCA. Combining with the novel t-SNE dimensionality reduction algorithm, the predicted survival intervals are 35.8 ± 15.2, 32.3 ± 13.1 and 31.6 ± 15.8 months for the BP, GA-BP and PGA-BP neural network models, respectively. Conclusion: The work demonstrated that the proposed probabilistic genetic algorithm optimized neural network models, integrating with the t-SNE dimensionality reduction algorithm, achieved accurate prediction of patient survival. Advances in knowledge: The proposed PGA-BP neural network, integrating with an advanced dimensionality reduction algorithm (t-SNE), improved patient survival prediction accuracy using pre-treatment quantitative CT image features of head-and-neck cancer patients.

Background Lymphedema treatment is an ongoing challenge. It impacts quality of life due to pain, loss of range of motion of the extremity, and repeated episodes of cellulitis. Different modalities have been used to evaluate lymphedema; some are more error-prone and some are more invasive. However, these measurements are poorly standardized, and intrarater and interrater reliabilities are difficult to achieve. This pilot study aims to assess the feasibility of ultrasound vibroelastography for assessing patients with extremity lymphedema via measuring shear wave speeds of subcutaneous tissues. Methods Patients with clinical and lymphoscintigraphic diagnosis of secondary lymphedema in the extremities without prior surgical treatment were included. A 0.1-s harmonic vibration was generated at three frequencies (100, 150, and 200 Hz) by the indenter of a handheld shaker on the skin. An ultrasound probe was used for noninvasively capturing of wave propagation in the subcutaneous tissue. Wave speeds were measured in the subcutaneous tissues of both the control and affected extremities. Results A total of 11 female patients with secondary lymphedema in the extremities were enrolled in this study. The magnitudes of the wave speeds of the region of interest in the subcutaneous tissue at lymphedema sites in the upper extremity (3.9 ± 0.17 m/s, 5.96 ± 0.67 m/s, and 7.41 ± 1.09 m/s) were statistically higher than those of the control sites (2.1 ± 0.27 m/s, 2.93 ± 0.57 m/s, and 3.56 ± 0.76 m/s) at 100, 150, and 200 Hz (P < 0.05), and at 100 and 200 Hz (P < 0.05) between lymphedema (4.33 ± 0.35 m/s, 4.17 ± 1.00 m/s, and 4.56 ± 0.37 m/s) and controls sites (2.48 ± 0.43 m/s, 2.77 ± 0.55 m/s, and 3.06 ± 0.29 m/s) in the lower extremity. Conclusions These preliminary data suggest that ultrasound vibroelastography may be useful in the evaluation of secondary lymphedema and can be a valuable tool to noninvasively track treatment progress.

Mutations in the transcription factor FOXC2 are predominately associated with lymphedema. Herein, we demonstrate a key role for related factor FOXC1, in addition to FOXC2, in regulating cytoskeletal activity in lymphatic valves. FOXC1 is induced by laminar, but not oscillatory, shear and inducible, endothelial-specific deletion impaired postnatal lymphatic valve maturation in mice. However, deletion of Foxc2 induced valve degeneration, which is exacerbated in Foxc1; Foxc2 mutants. FOXC1 knockdown (KD) in human lymphatic endothelial cells increased focal adhesions and actin stress fibers whereas FOXC2-KD increased focal adherens and disrupted cell junctions, mediated by increased ROCK activation. ROCK inhibition rescued cytoskeletal or junctional integrity changes induced by inactivation of FOXC1 and FOXC2 invitro and vivo respectively, but only ameliorated valve degeneration in Foxc2 mutants. These results identify both FOXC1 and FOXC2 as mediators of mechanotransduction in the postnatal lymphatic vasculature and posit cytoskeletal signaling as a therapeutic target in lymphatic pathologies.

Background: Brugada syndrome (BrS) is an oligogenic arrhythmic disease with increased risk of sudden cardiac arrest. Several BrS or ECG traits-related single-nucleotide polymorphisms (SNPs) were identified through previous genome-wide association studies in white patients. We aimed to validate these SNPs in BrS patients in the Taiwanese population, assessing the cumulative effect of risk alleles and the BrS-polygenic risk score in predicting cardiac events. Methods: We genotyped 190 unrelated BrS patients using the TWB Array, and Taiwan Biobank was used as controls. SNPs not included in the array were imputed by IMPUTE2. Cox proportional hazards model was used to evaluate the associations between each particular SNP, the collective BrS-polygenic risk score, and clinical outcomes. Results: Of the 88 previously reported SNPs, 22 were validated in Taiwanese BrS patients ( P <0.05). Of the 22 SNPs, 2 (rs10428132 and rs9388451) were linked with susceptibility to BrS, 10 were SNPs previously reaching genome-wide significance, and 10 were SNPs associated with ECG traits. For the 3 most commonly reported SNPs, disease risk increased consistently with the number of risk alleles (odds ratio, 3.54; P trend =1.38×10 −9 for 5 risk alleles versus 1). Similar patterns were observed in both SCN5A mutation+ (odds ratio, 3.66; P trend =0.049) and SCN5A mutation− (odds ratio, 3.75; P trend =8.54×10 −9 ) subgroups. Furthermore, BrS patients without SCN5A mutations had more risk alleles than BrS patients with SCN5A mutations regardless of the range of polygenic risk scores. Three SNPs (rs4687718, rs7784776, and rs2968863) showed significant associations with the composite outcome (sudden cardiac arrest plus syncope, hazard ratio, 2.13, 1.48, and 0.41; P =0.02, 0.006, and 0.008, respectively). Conclusions: Our findings suggested that some SNPs associated with BrS or ECG traits exist across multiple populations. The cumulative risk of the BrS-related SNPs is similar to that in white BrS patients, but it appears to correlate with the absence of SCN5A mutations.

e16771 Background: AbGn-107 is an ADC directed against AG-7 antigen, a Lewis A-like glycol-epitope expressed in 24-61% of gastric (G), colorectal (CRC), pancreatic (PDA), and biliary (BIL) cancers. Based on promising antitumor activity of AbGn-107 in both in vitro and in vivo preclinical studies, we performed a Phase Ia trial in patients with the aforementioned GI malignancies. Methods: Standard 3+3 dose escalation was used. Key eligibility included locally advanced or metastatic G, CRC, PDA, or BIL cancer, previously treated, ECOG PS 0-1; positive AG-7 expression was not required. Two dosing intervals were tested: AbGn-107 administered i.v. Q4 weeks (at doses ranging from 0.1-1.2 mg/kg) and Q2 weeks (at doses from 0.8-1.0 mg/kg). Dose limiting toxicities (DLT) were based on grade 3/4 hematologic and non-heme AEs occurring during the initial 4-week rx window. Patients were treated until disease progression or unacceptable toxicity, with tumor assessments Q8 weeks. 1o objectives: safety and MTD; 2o objectives: PK, immunogenicity, and efficacy defined by ORR (RECIST 1.1). Results: 35 patients were enrolled across 6 dose levels (median age 61.5 yo (range 40 – 81); G (0)/CRC (12)/PDA (20)/BIL (3); median # lines of prior rx = 3 (range 1-7). Safety: 5 patients experienced Grade 3 or 4 neutropenia, all at higher dose levels, inc. 1 episode of febrile neutropenia. Other frequent drug-related AEs, mostly grade 1/2, inc. fatigue (29%), nausea (20%), and diarrhea (14%). DLTs included grade 4 CK elevation (n = 1) at 0.8 mg/kg Q4W and grade 3 arthralgias (n = 1) at 1.2 mg/kg Q4W. Based on safety profile and PK data, 1.0 mg/kg Q2W was selected as the dose schedule for cohort expansion phase. Efficacy: Median duration of treatment = 56 days (range, 8 – 225 days). Six pts have demonstrated a minor response by RECIST (range, -1.3 to -21.1%); 4 pts (all PDA) have had durable dz control > 180 days. Conclusions: Overall, AbGn-107 appears well-tolerated with encouraging prelim signs of efficacy in unselected pts with heavily pre-treated advanced GI cancers. Updated safety and efficacy data will be provided at the time of the meeting for the cohort expansion phase of this study (currently open across multiple sites in U.S. and Taiwan), in which subjects with G, CRC, PDA, and BIL cancers are pre-screened for high levels of AG-7 tumor expression. Clinical trial information: NCT02908451 .

Abstract Background Inflammatory bowel disease (IBD) is associated with a reduced quality of life. Minority patients with IBD specifically report more impairing symptoms compared with nonminority patients. Sleep quality, a key component of quality of life, is significantly compromised in minority patients compared with nonminority patients. Nevertheless, subjective and objective sleep assessments in minority patients with IBD have not explicitly been assessed. The purpose of this prospective cohort study is to assess and compare objective sleep parameters utilizing wrist actigraphy between minority and nonminority IBD patients. Methods In this institutional review board approved study, 74 patients with IBD were recruited and stratified into 2 cohorts by self-identification: white nonminority patients and minority patients. Patients in the minority cohort included black and Hispanic individuals (black and nonblack). Exclusion criteria included significant comorbidity, a history of an underlying sleep disorder, or patients who did not self-identify into categorized cohorts. Sleep was measured not only through wrist-based actigraphy but also with sleep surveys. Sleep parameters were compared between minority and nonminority cohorts. Regression analyses were performed to assess for factors independently associated with parameters of poor sleep quality. Results Sixty-four patients (86.4%) were included in the final analysis. Thirty-one individuals (48.4%) were categorized into the nonminority cohort, and 33 (51.6%) patients were in the minority cohort. A significantly higher number of minority patients had poorer sleep efficiency and fragmented sleep compared with nonminority patients (90.9% vs 67.7%; P = 0.03 and 87.8% vs 61.3%; P = 0.02). In the adjusted analysis, minority status was independently associated with poor sleep efficiency (odds ratio = 6.41; 95% confidence interval, 1.48-28.17; P = 0.0139) and fragmented sleep (odds ratio = 4.98; 95% confidence interval, 1.09-22.89; P = 0.0389). Conclusions Minority patients with IBD were shown to have poorer objective measures of sleep as assessed through wrist actigraphy compared to nonminority patients. Cultural competency in the care of minority patients with IBD, specifically focusing on the management of psychosocial issues, is needed to address these disparities in sleep. The inclusion of minority patients with IBD in studies investigating sleep and other psychosocial issues are warranted not only to assess potential disparities in disease course but also to determine the etiologies of poor sleep in minority patients with IBD.