Citations

Background and Purpose— We evaluated deep learning algorithms’ segmentation of acute ischemic lesions on heterogeneous multi-center clinical diffusion-weighted magnetic resonance imaging (MRI) data sets and explored the potential role of this tool for phenotyping acute ischemic stroke. Methods— Ischemic stroke data sets from the MRI-GENIE (MRI-Genetics Interface Exploration) repository consisting of 12 international genetic research centers were retrospectively analyzed using an automated deep learning segmentation algorithm consisting of an ensemble of 3-dimensional convolutional neural networks. Three ensembles were trained using data from the following: (1) 267 patients from an independent single-center cohort, (2) 267 patients from MRI-GENIE, and (3) mixture of (1) and (2). The algorithms’ performances were compared against manual outlines from a separate 383 patient subset from MRI-GENIE. Univariable and multivariable logistic regression with respect to demographics, stroke subtypes, and vascular risk factors were performed to identify phenotypes associated with large acute diffusion-weighted MRI volumes and greater stroke severity in 2770 MRI-GENIE patients. Stroke topography was investigated. Results— The ensemble consisting of a mixture of MRI-GENIE and single-center convolutional neural networks performed best. Subset analysis comparing automated and manual lesion volumes in 383 patients found excellent correlation (ρ=0.92; P <0.0001). Median (interquartile range) diffusion-weighted MRI lesion volumes from 2770 patients were 3.7 cm 3 (0.9–16.6 cm 3 ). Patients with small artery occlusion stroke subtype had smaller lesion volumes ( P <0.0001) and different topography compared with other stroke subtypes. Conclusions— Automated accurate clinical diffusion-weighted MRI lesion segmentation using deep learning algorithms trained with multi-center and diverse data is feasible. Both lesion volume and topography can provide insight into stroke subtypes with sufficient sample size from big heterogeneous multi-center clinical imaging phenotype data sets.

OBJECTIVES: Nosocomial infections (NIs) can be a major cause of morbidity and mortality in cirrhosis. This study aims to define the determinants of NI development and its impact on 30-day outcomes among hospitalized patients with cirrhosis. METHODS: North American Consortium for the Study of End-Stage Liver Disease enrolled patients with cirrhosis who were admitted nonelectively. Admission variables and 30-day outcomes were compared between patients with and without NI. These were also compared based on whether there was an isolated admission infection, NI, or both. Models were created for NI development using admission variables and for 30-day mortality. RESULTS: The study included 2,864 patients; of which, 15% (n = 436) developed NI. When comparing NI vs no NI, 1,866 patients were found to be infection free, whereas 562 had admission infections only, 228 had only NI, and 208 had both infections. At admission, patients with NI were more likely to be infected and have advanced cirrhosis. NIs were associated with higher rates of acute-on-chronic liver failure, death, and transplant regardless of admission infections. Patients with NI had higher respiratory infection, urinary tract infection, Clostridium difficile infection, fungal infections, and infection with vancomycin-resistant enterococci compared with patients without NI. Risk factors for NIs were admission infections, model for end-stage liver disease (MELD) > 20, systemic inflammatory response syndrome criteria, proton pump inhibitor, rifaximin, and lactulose use, but the regression model (sensitivity, 0.67; specificity, 0.63) was not robust. Age, alcohol etiology, admission MELD score, lactulose use, acute-on-chronic liver failure, acute kidney injury, intensive care unit, and NI increased the risk of death, whereas rifaximin decreased the risk of death. DISCUSSION: NIs are prevalent in hospitalized patients with cirrhosis and are associated with poor outcomes. Although higher MELD scores and systemic inflammatory response syndrome are associated with NI, all hospitalized patients with cirrhosis require vigilance and preventive strategies.

Abstract Many factors contribute to the poor survival of malignant brain tumor patients, some of which are not easily remedied. However, one contributor to the lack of progress that may be modifiable is poor clinical trial accrual. Surveys of brain tumor patients and neuro-oncology providers suggest that clinicians do a poor job of discussing clinical trials with patients and referring patients for clinical trials. Yet, data from the Cancer Action Network of the American Cancer Society suggest that most eligible oncology patients asked to enroll on a clinical trial will agree to do so. To this end, the Society for Neuro-Oncology (SNO) in collaboration with the Response Assessment in Neuro-Oncology (RANO) Working Group, patient advocacy groups, clinical trial cooperative groups, including the Adult Brain Tumor Consortium (ABTC), and other partners are working together with the intent to double clinical trial accrual over the next 5 years. Here we describe the factors contributing to poor clinical trial accrual in neuro-oncology and offer possible solutions.

PURPOSE In patients with diffuse large B-cell lymphoma (DLBCL), most relapses occur within the first 2 years of diagnosis. We sought to define the rate and outcome of late relapses that occurred after achieving event-free survival at 24 months (EFS24). METHODS We prospectively followed 1,324 patients with newly diagnosed DLBCL from 2002 to 2015 and treated with immunochemotherapy. Cumulative incidences of late DLBCL and indolent lymphoma relapses were analyzed as competing events. Postrelapse survival was defined as time from first relapse to death from any cause. RESULTS In 847 patients who achieved EFS24, the cumulative incidence of late relapse was 6.9% at 3 years, 9.3% at 5 years, and 10.3% at 8 years after EFS24. The incidence of DLBCL relapse was similar in patients with DLBCL alone at diagnosis (6.3% at 5 years), compared with patients with concurrent indolent lymphoma at diagnosis (5.2%; P = .46). However, the rate of indolent lymphoma relapse was higher in patients with concurrent indolent lymphoma (7.4% v 2.1% at 5 years; P < .01). In patients with DLBCL alone, the rate of DLBCL relapse was similar in the germinal center B-cell–like (GCB) (4.1% at 5 years) and non-GCB (4.0%; P = .71) subtypes, whereas the rate of indolent lymphoma relapse was higher in patients with the GCB subtype (3.9% v 0.0% at 5 years; P = .02). Postrelapse survival was inferior for patients who relapsed with DLBCL than for those who relapsed with indolent lymphoma (median 29.9 months v unreached; P < .01). CONCLUSION Patients with DLBCL with a concurrent indolent lymphoma and those with the GCB subtype had a higher rate of late relapse, owing to increased relapses with indolent lymphoma. Patients who relapsed with DLBCL had a worse prognosis than those who relapsed with indolent lymphoma.

Purpose To examine how qualitative narrative comments and quantitative ratings from end-of-rotation assessments change for a cohort of residents from entry to graduation, and explore associations between comments and ratings. Method The authors obtained end-of-rotation quantitative ratings and narrative comments for 1 cohort of internal medicine residents at the University of Illinois at Chicago College of Medicine from July 2013–June 2016. They inductively identified themes in comments, coded orientation (praising/critical) and relevance (specificity and actionability) of feedback, examined associations between codes and ratings, and evaluated changes in themes and ratings across years. Results Data comprised 1,869 assessments (828 comments) on 33 residents. Five themes aligned with ACGME competencies (interpersonal and communication skills, professionalism, medical knowledge, patient care, and systems-based practice), and 3 did not (personal attributes, summative judgment, and comparison to training level). Work ethic was the most frequent subtheme. Comments emphasized medical knowledge more in year 1 and focused more on autonomy, leadership, and teaching in later years. Most comments (714/828 [86%]) contained high praise, and 412/828 (50%) were very relevant. Average ratings correlated positively with orientation (β = 0.46, P < .001) and negatively with relevance (β = −0.09, P = .01). Ratings increased significantly with each training year (year 1, mean [standard deviation]: 5.31 [0.59]; year 2: 5.58 [0.47]; year 3: 5.86 [0.43]; P < .001). Conclusions Narrative comments address resident attributes beyond the ACGME competencies and change as residents progress. Lower quantitative ratings are associated with more specific and actionable feedback.

6594 Background: Given the significant short-term adverse effects of CAR-T cell therapy, it is important to evaluate its impact on QOL of patients in addition to efficacy, compared with established forms of cellular therapy like SCT. Methods: QOL was evaluated prospectively in patients undergoing CAR-T therapy, autoSCT & alloSCT for hematologic malignancies. QOL was assessed with FACT-G at baseline, 2 weeks and monthly for 6 months thereafter. Functional well-being (FWB), physical WB (PWB) emotional WB (EWB) & social WB (SWB) and change over time were compared across groups. Results: 45 patients were recruited (CAR-T: 10; Auto SCT: 22; Allo SCT: 13) with follow up for 2 weeks & 1 month available for 23 &15 patients, respectively (Table). There was no statistically significant difference in baseline total QOL scores (p=0.13), though scores were lower in the alloSCT group (85,84,68). EWB &FWB were numerically higher in the CAR-T group, followed by autoSCT group. At 2 weeks, overall QOL decreased by only 2 points in CAR-T group vs. 22 & 18 points in auto & alloSCT groups (p=0.09). Change in PWB vs. baseline was less pronounced in the CAR-T group (-1, -9, -13, p=0.03). At 1 month, overall QOL was 6 points lower than baseline in CAR-T group vs. 3 and 14 points lower in auto & alloSCT groups, respectively (p=0.34). Importantly, PWB had at least returned to baseline in the CAR-T group. Conclusions: Preliminary data show that patients undergoing CAR-T cell therapy do not experience a more significant decline in QOL compared with auto & allo SCT, and may experience fewer physical side effects in the short-term. Accrual & follow-up are ongoing. [Table: see text]

3599 Background: Existing tools for post-treatment CRC surveillance and monitoring are insensitive and expensive. MDMs are broadly informative for early detection of CRC but have not been extensively studied for disease monitoring. As a first step, we sought to assess the concordance of novel CRC-associated MDMs in pCRC and mCRC. Methods: A panel of 14 MDMs previously identified to be highly discriminant for pCRC was selected on the basis of high median fold-change of MDM levels relative to buffy coat (682 (IQR: 132-19347)). Surgically resected pCRC and paired mCRC were identified from institutional pathology databases. Quantitative methylation-specific PCR was used to assay MDMs. 30 paired samples per metastatic subtype were calculated to be sufficient. MDM levels were compared using two-sample and paired Wilcoxon rank sum tests. Results: 87 patients with paired pCRC and mCRC including 57 synchronous and 30 metachronous metastases were included. 41/87 (47%) had neoadjuvant and 59/87 (68%) had adjuvant chemotherapy. All synchronous metastases were to liver. Metachronous metastases were to liver in 19/30 (63%) and to lung in 11/30 (37%). The levels of 14 selected MDMs were remarkably similar between paired pCRC and mCRC (Table). Individual MDM levels and the average level of all MDMs combined were not significantly different ( p> 0.0018 by Bonferroni correction). Conclusions: MDM levels are highly concordant in pCRC and mCRC. Thus, MDMs discovered from pCRC should be further studied for non-invasive surveillance after surgical resection and monitoring of treatment response in mCRC. [Table: see text]