Endovascular Alternatives to Transhepatic Shunt Creation.

Portomesenteric decompression is often necessary to treat patients with refractory symptoms of portal hypertension. When transjugular or direct intrahepatic portosystemic shunt creation is not feasible or is inadequate, surgical portosystemic shunt creation is considered, which carries significant morbidity and mortality in these high-risk patients. Surgery is further complicated in patients with portomesenteric thrombosis who require concurrent thrombectomy and long-term anticoagulation. In this article, we outline the technique for performing advanced endovascular alternatives to intrahepatic portosystemic shunt creation including mesocaval and splenorenal shunting. We will also discuss some of the clinical considerations for treating these patients with symptomatic portal hypertension and portomesenteric thrombosis.

Advanced Techniques for Accessing the Portal Vein during Transjugular Intrahepatic Portosystemic Shunt Creation.

Transjugular portosystemic shunt (TIPS) placement is a widely accepted method for reducing portal pressures in conditions including symptomatic varices, refractory ascites, hepatic hydrothorax, portomesenteric thrombosis, and Budd-Chiari syndrome. The traditional method for performing TIPS involves "blind" access into a portal vein branch from a hepatic vein, typically right portal vein to right hepatic vein, using preoperative imaging and intraoperative wedged portography for guidance. However, standard access technique may not always be feasible due to occluded portal or hepatic veins, distorted anatomy, or prior TIPS. In this article we discuss alternative techniques for accessing the portal vein for TIPS placement when standard methods are not sufficient.

Integrated use of PD-1 inhibition and transarterial chemoembolization for hepatocellular carcinoma: evaluation of safety and efficacy in a retrospective, propensity score-matched study.

BACKGROUND: Immune checkpoint inhibitors (ICIs) have revolutionized treatment of advanced hepatocellular carcinoma. Integrated use of transarterial chemoembolization (TACE), a locoregional inducer of immunogenic cell death, with ICI has not been formally assessed for safety and efficacy outcomes. METHODS: From a retrospective multicenter dataset of 323 patients treated with ICI, we identified 31 patients who underwent >1 TACE 60 days before or concurrently, with nivolumab at a single center. We derived a propensity score-matched cohort of 104 patients based on Child-Pugh Score, portal vein thrombosis, extrahepatic metastasis and alpha fetoprotein (AFP) who received nivolumab monotherapy. We described overall survival (OS), progression-free survival (PFS), objective responses according to modified RECIST criteria and safety in the multimodal arm in comparison to monotherapy. RESULTS: Over a median follow-up of 9.3 (IQR 4.0-16.4) months, patients undergoing multimodal immunotherapy with TACE achieved a significantly longer median (95% CI) PFS of 8.8 (6.2-23.2) vs 3.7 (2.7-5.4) months (log-rank 0.15, p<0.01) in the monotherapy group. Multimodal immunotherapy with TACE demonstrated a numerically longer OS compared with ICI monotherapy with a median 35.1 (16.1-Not Evaluable) vs 16.6 (15.7-32.6) months (log-rank 0.41, p=0.12). In the multimodal treatment group, there were three (10%) grade 3 or higher adverse events (AEs) attributed to immunotherapy compared with seven (6.7%) in the matched ICI monotherapy arm. There were no AEs grade 3 or higher attributed to TACE in the multimodal treatment arm. At 3 months following each TACE in the multimodal arm, there was an overall objective response rate of 84%. There were no significant changes in liver functional reserve 1 month following each TACE. Four patients undergoing multimodal treatment were successfully bridged to transplant. CONCLUSIONS: TACE can be safely integrated with programmed cell death 1 blockade and may lead to a significant delay in tumor progression and disease downstaging in selected patients.

Machine Learning Offers Exciting Potential for Predicting Postprocedural Outcomes: A Framework for Developing Random Forest Models in IR.

PURPOSE: To demonstrate that random forest models trained on a large national sample can accurately predict relevant outcomes and may ultimately contribute to future clinical decision support tools in IR. MATERIALS AND METHODS: Patient data from years 2012-2014 of the National Inpatient Sample were used to develop random forest machine learning models to predict iatrogenic pneumothorax after computed tomography-guided transthoracic biopsy (TTB), in-hospital mortality after transjugular intrahepatic portosystemic shunt (TIPS), and length of stay > 3 days after uterine artery embolization (UAE). Model performance was evaluated with area under the receiver operating characteristic curve (AUROC) and maximum F1 score. The threshold for AUROC significance was set at 0.75. RESULTS: AUROC was 0.913 for the TTB model, 0.788 for the TIPS model, and 0.879 for the UAE model. Maximum F1 score was 0.532 for the TTB model, 0.357 for the TIPS model, and 0.700 for the UAE model. The TTB model had the highest AUROC, while the UAE model had the highest F1 score. All models met the criteria for AUROC significance. CONCLUSIONS: This study demonstrates that machine learning models may suitably predict a variety of different clinically relevant outcomes, including procedure-specific complications, mortality, and length of stay. Performance of these models will improve as more high-quality IR data become available.

A Novel Machine Learning Model Developed to Assist in Patient Selection for Outpatient Total Shoulder Arthroplasty.

INTRODUCTION: Patient selection for outpatient total shoulder arthroplasty (TSA) is important to optimizing patient outcomes. This study aims to develop a machine learning tool that may aid in patient selection for outpatient total should arthroplasty based on medical comorbidities and demographic factors. METHODS: Patients undergoing elective TSA from 2011 to 2016 in the American College of Surgeons National Surgical Quality Improvement Program were queried. A random forest machine learning model was used to predict which patients had a length of stay of 1 day or less (short stay). A multivariable logistic regression was then used to identify which variables were significantly correlated with a short or long stay. RESULTS: From 2011 to 2016, 4,500 patients were identified as having undergone elective TSA and having the necessary predictive features and outcomes recorded. The machine learning model was able to successfully identify short stay patients, producing an area under the receiver operator curve of 0.77. The multivariate logistic regression identified numerous variables associated with a short stay including age less than 70 years and male sex as well as variables associated with a longer stay including diabetes, chronic obstructive pulmonary disease, and American Society of Anesthesiologists class greater than 2. CONCLUSIONS: Machine learning may be used to predict which patients are suitable candidates for short stay or outpatient TSA based on their medical comorbidities and demographic profile.