Hydrocephalus in Neurosarcoidosis: Clinical Course, Radiographic Accompaniments, and Experience with Shunting.

BACKGROUND: Hydrocephalus is an uncommon manifestation of neurosarcoidosis (7-14% of reported cohorts) that poses unique challenges to patient management. Despite being a recognized complication of neurosarcoidosis, very little is known about how hydrocephalus influences its clinical course, management, and prognosis. OBJECTIVES: To characterize hydrocephalus as a clinical manifestation of neurosarcoidosis, highlight which patients required cerebrospinal fluid (CSF) diversion, understand the mediating role of immunomodulatory treatments, and report outcomes in this cohort. METHODS: Patients with a diagnosis of neurosarcoidosis seen at Emory Healthcare [01/2011-8/2021] were included if hydrocephalus was one manifestation of their disease. Means and proportions were compared between shunted and non-shunted groups using the Wilcoxon rank-sum test for continuous variables and the Fisher's exact test for categorical variables. RESULTS: Twenty-two patients with neurosarcoidosis and hydrocephalus as one disease manifestation were included (22/214, 10.3%). Hydrocephalus was communicating in 13 (13/20, 65.0%) and obstructive in 6 patients (6/20, 30.0%), with features of both seen in 1 patient (1/20, 5.0%). Chronic presentations were typical (12/22, 54.5%) with altered sensorium, gait dysfunction, headache, and weakness being present in the majority of patients. There was a rostral-to-caudal gradient in ventriculomegaly, with the lateral ventricles most affected (20/20, 100%) and the fourth ventricle the least (12/20, 60%). Meningoventricular inflammation was the most common neuroinflammatory accompaniment (18/20, 90.0%), especially infratentorial leptomeningitis (16/20, 80.0%) and fourth ventriculitis (9/20, 45.0%). Thirteen patients (13/22, 59.1%) required ventriculoperitoneal shunts (VPS). Factors associated with shunt placement were younger age at neurosarcoidosis onset (p = 0.019) and hydrocephalus onset (p = 0.015), obstructive hydrocephalus (p = 0.043), and lateral ventriculitis (p = 0.043). In the 6 patients (6/13, 46.2%) with preceding extraventricular drain (EVD) placement, all failed to wean, including 5/6 patients who received high-dose steroids while the EVD was in place. Almost all (19/20, 95.0%) were treated with steroid-sparing agents, including nine (9/20, 45.0%) with tumor necrosis factor (TNF) inhibitors. Modified Rankin Scale score at last outcome was 3.04 (range 0-6). CONCLUSION: Patients with neurosarcoidosis and hydrocephalus experience unique challenges in the management of their disease, including the potential need for CSF diversion, in addition to traditional anti-inflammatory treatments. Younger patients, those with obstructive hydrocephalus, and those with lateral ventriculitis warrant particular consideration for VPS placement, but the decision to shunt likely remains a highly individualized one. The requirement for multiple lines of immunotherapy beyond steroids and moderate disability at last follow-up suggest hydrocephalus may reflect a more severe form of neurosarcoidosis.

Using Deep Learning to Predict Treatment Response in Patients with Hepatocellular Carcinoma Treated with Y90 Radiation Segmentectomy.

Treatment of hepatocellular carcinoma (HCC) with Y90 radioembolization segmentectomy (Y90-RE) demonstrates a tumor dose-response threshold, where dose estimates are highly dependent on accurate SPECT/CT acquisition, registration, and reconstruction. Any error can result in distorted absorbed dose distributions and inaccurate estimates of treatment success. This study improves upon the voxel-based dosimetry model, one of the most accurate methods available clinically, by using a deep convolutional network ensemble to account for the spatially variable uptake of Y90 within a treated lesion. A retrospective analysis was conducted in patients with HCC who received Y90-RE at a single institution. Seventy-seven patients with 103 lesions met the inclusion criteria: three or fewer tumors, pre- and post treatment MRI, and no prior Y90-RE. Lesions were labeled as complete (n = 57) or incomplete response (n = 46) based on 3-month post treatment MRI and divided by medical record number into a 20% hold-out test set and 80% training set with 5-fold cross-validation. Slice-wise predictions were made from an average ensemble of models and thresholds from the highest accuracy epochs across all five folds. Lesion predictions were made by thresholding all slice predictions through the lesion. When compared to the voxel-based dosimetry model, our model had a higher F1-score (0.72 vs. 0.2), higher accuracy (0.65 vs. 0.60), and higher sensitivity (1.0 vs. 0.11) at predicting complete treatment response. This algorithm has the potential to identify patients with treatment failure who may benefit from earlier follow-up or additional treatment.