Thiruvengadam, Nikhil Ravi

Abstract Clinicians struggle to accurately classify biliary strictures as benign or malignant. Current endoscopic retrograde cholangiopancreatography (ERCP)-based sampling modalities including brush cytology and forceps biopsy have poor sensitivity for pathologic confirmation of malignancy. Cholangioscopy allows for direct visualization and sampling of biliary pathology; however, this technology is also associated with inaccurate classification of biliary disease. Previously, an artificial intelligence (AI) system that analyzes cholangioscopy footage was found to be more accurate in diagnosing biliary malignancy than ERCP sampling techniques. The aim of this study was to validate this AI system on a new series of examinations. Three academic centers collected all available unedited cholangioscopy recordings. The videos were processed by the cholangioscopy AI system. After analyzing videos, the AI system provided predictions as to whether malignancy was present. AI performance in classifying strictures was compared with the performance of brush cytology and forceps biopsy. 112 cholangioscopy examinations (containing 4 817 081 images) were generated from 99 patients. Of those examinations, 61 (54.5%) were for investigation of biliary strictures (31 [50.8%] benign, 30 [49.2%] malignant). For the correct classification of strictures, the AI system was 80.0% sensitive and 90.3% specific. It was also significantly more accurate for stricture classification (85.2%) than brush cytology (52.5%; P<0.001), forceps biopsy (68.2%; P=0.04), and the combination of brush cytology and forceps biopsy (66.7%; P=0.02). A previously developed cholangioscopy AI system was found to continually outperform standard ERCP sampling modalities for accurate identification of malignancy, without additional retraining, in a multicenter validation cohort.

Background: Two National Cancer Institute-funded pharmacovigilance programs have identified 49 serious hematology-related ADRs since 1998. The programs, called RADAR (Research on Adverse Drug events and Reports) and then SONAR (Southern Network on Adverse Reactions) involve &gt; 50 institutions and &gt; 75 co-investigators worldwide. The work builds on three prior studies evaluating “Davids and Goliaths” in medical oncology and in cardiology (PLOS One, eClinicalMedicine, and Journal of Scientific and Professional Integrity). Here we review the hematology focused safety investigations of RADAR/SONAR. Methods: Data were based on initial safety signals being identified by a RADAR/SONAR co-investigator. Then, case-series were derived. Collaborations with basic scientists allowed for basic science correlative studies. Results were disseminated primarily as Brief Report in the literature and Black Box warnings on the related drug. In some instances, studies identified hematologic toxicities that prevent FDA from approving a submitted drug application. Results: sADRs were frequently identified based on very small case series, including ticlopidine-and clopidogrel associated thrombotic thrombocytopenic purpura (22 and 10 patients, respectively), thalidomide- and lenalidomide-associated venous thromboembolism (9 and 5 patients, respectively), rituximab-associated progressive multi-focal leukoencephalopathy (22 patients), peginesatide-associated fatal anaphylaxis (5 patients), and COVID-19 vaccine associated immune thrombocytopenia (1 patient). Meta-analyses provided data for epoetin- and darbepoetin-associated mortality among cancer patients and lenalidomide- and thalidomide-associated venous thromboembolism. Following the onset of the COVID-19 pandemic, RADAR/SONAR investigated social media and pre-prints (COVID-19 vaccine associated cerebral vein thrombosis and immune thrombocytopenia). Time from FDA approval to sADR discovery was a median of 5 years (range, 0 months (thalidomide-associated venous thromboembolism) to 34 years (ciprofloxacin-associated neuropsychiatric toxicity). Basic science correlative studies identified ADAMTS13 autoantibodies (ticlopidine), leachates that developed in a multi-dose vial (peginesatide), anti-red blood cell antibodies (epoetin-associated pure red cell aplasia), and high-risk genes (fluoroquinolones and rituximab). Overall, RADAR/SONAR studies are estimated to have saved over 1 million lives and also resulted in overall payments to the Department of Justice of $1.5 billion (related to marketing of unsafe drugs). Discussion: RADAR/SONAR has proven to be a very important adjunct to FDA and pharmaceutical manufacturer-led safety investigations for hematology, paralleling the success in medical oncology and in cardiology. Going forward, independent centers of excellence for safety-focused investigations such as the CERSI network and the SENTINEL network (both are NIH funded) should be broadened to include a hematology-focused safety center.

Objective: We report the development and validation of a combined DNA/RNA next-generation sequencing (NGS) platform to improve the evaluation of pancreatic cysts. Background and Aims: Despite a multidisciplinary approach, pancreatic cyst classification, such as a cystic precursor neoplasm, and the detection of high-grade dysplasia and early adenocarcinoma (advanced neoplasia) can be challenging. NGS of preoperative pancreatic cyst fluid improves the clinical evaluation of pancreatic cysts, but the recent identification of novel genomic alterations necessitates the creation of a comprehensive panel and the development of a genomic classifier to integrate the complex molecular results. Methods: An updated and unique 74-gene DNA/RNA-targeted NGS panel (PancreaSeq Genomic Classifier) was created to evaluate 5 classes of genomic alterations to include gene mutations (e.g., KRAS, GNAS, etc.), gene fusions and gene expression. Further, CEA mRNA (CEACAM5) was integrated into the assay using RT-qPCR. Separate multi-institutional cohorts for training (n=108) and validation (n=77) were tested, and diagnostic performance was compared to clinical, imaging, cytopathologic, and guideline data. Results: Upon creation of a genomic classifier system, PancreaSeq GC yielded a 95% sensitivity and 100% specificity for a cystic precursor neoplasm, and the sensitivity and specificity for advanced neoplasia were 82% and 100%, respectively. Associated symptoms, cyst size, duct dilatation, a mural nodule, increasing cyst size, and malignant cytopathology had lower sensitivities (41–59%) and lower specificities (56–96%) for advanced neoplasia. This test also increased the sensitivity of current pancreatic cyst guidelines (IAP/Fukuoka and AGA) by >10% and maintained their inherent specificity. Conclusions: PancreaSeq GC was not only accurate in predicting pancreatic cyst type and advanced neoplasia but also improved the sensitivity of current pancreatic cyst guidelines.

AbstractBackground and study aims Endoscopic ultrasound (EUS)-guided tissue sampling is the standard of care for diagnosing solid pancreatic lesions. While many two-way comparisons between needle types have been made in randomized controlled trials (RCTs), it is unclear which size and type of needle offers the best probability of diagnosis. We therefore performed a network meta-analysis (NMA) to compare different sized and shaped needles to rank the diagnostic performance of each needle.Methods We searched MEDLINE, EMBASE and Cochrane Library databases through August, 2020 for RCTs that compared the diagnostic accuracy of EUS fine-needle aspiration (FNA) and biopsy (FNB) needles in solid pancreatic masses. Using a random-effects NMA under the frequentist framework, RCTs were analyzed to identify the best needle type and sampling technique. Performance scores (P-scores) were used to rank the different needles based on pooled diagnostic accuracy. The NMA model was used to calculate pairwise relative risk (RR) with 95 % confidence intervals.Results Review of 2577 studies yielded 29 RCTs for quantitative synthesis, comparing 13 different needle types. All 22G FNB needles had an RR > 1 compared to the reference 22G FNA (Cook) needle. The highest P-scores were seen with the 22G Medtronic FNB needle (0.9279), followed by the 22G Olympus FNB needle (0.8962) and the 22G Boston Scientific FNB needle (0.8739). Diagnostic accuracy was not significantly different between needles with or without suction.Conclusions In comparison to FNA needles, FNB needles offer the highest diagnostic performance in sampling pancreatic masses, particularly with 22G FNB needles.