Faculty Bibliography

BACKGROUND:In many parts of the world including India, the prevalence of autoimmune inflammatory diseases such as neuromyelitis optica spectrum disorders (NMOSD), myelin oligodendrocyte glycoprotein antibody-associated disease (MOGAD), and multiple sclerosis (MS) is rising. A diagnosis is often delayed due to insufficient diagnostic tools. Machine learning (ML) models have accurately differentiated eyes of patients with MS from those of healthy controls (HCs) using optical coherence tomography (OCT)-based retinal images. Examining OCT characteristics may allow for early differentiation of these conditions. The objective of this study was to determine feasibility of ML analyses to distinguish between patients with different autoimmune inflammatory diseases, other ocular diseases, and HCs based on OCT measurements of the peripapillary retinal nerve fiber layer (pRNFL), ganglion cell-inner plexiform layer (GCIPL), and inner nuclear layers (INLs). METHODS:Eyes of people with MS (n = 99 patients), NMOSD (n = 40), MOGAD (n = 74), other ocular diseases (OTHER, n = 16), and HCs (n = 54) from the Mangalore Demyelinating Disease Registry were included. Support vector machine (SVM) classification models incorporating age, pRNFL, GCIPL, and INL were performed. Data were split into training (70%) and testing (30%) data and accounted for within-patient correlations. Cross-validation was used in training to choose the best parameters for the SVM model. Accuracy and area under receiver operating characteristic curves (AUROCs) were used to assess model performance. RESULTS:The SVM models distinguished between eyes of patients with each condition (i.e., MOGAD vs NMOSD, NMOSD vs HC, MS vs OTHER, etc) with strong discriminatory power demonstrated from the AUROCs for these comparisons ranging from 0.81 to 1.00. These models also performed with moderate to high accuracy, ranging from 0.66 to 0.81, with the exception of the MS vs NMOSD comparison, which had an accuracy of 0.53. CONCLUSIONS:ML models are useful for distinguishing between autoimmune inflammatory diseases and for distinguishing these from HCs and other ocular diseases based on OCT measures. This study lays the groundwork for future deep learning studies that use analyses of raw OCT images for identifying eyes of patients with such disorders and other etiologies of optic neuropathy.

RATIONALE AND OBJECTIVES/OBJECTIVE:To evaluate the performance, stability, and decision-making behavior of large language models (LLMs) for title and abstract screening for radiology systematic reviews, with attention to prompt framing, confidence calibration, and model robustness under disagreement. MATERIALS AND METHODS/METHODS:We compared five LLMs (GPT-4o, GPT-4o mini, Gemini 1.5 Pro, Gemini 2.0 Flash, Llama 3.3 70B) on two imaging-focused systematic reviews (n = 5438 and n = 267 abstracts) using binary and ternary classification tasks, confidence scoring, and reclassification of true and synthetic disagreements. Disagreements were framed as either "LLM vs human" or "human vs human." We also piloted autonomous PubMed retrieval using OpenAI and Gemini Deep Research tools. RESULTS:LLMs achieved high specificity and variable sensitivity across reviews and tasks, with F1 scores ranging from 0.389 to 0.854. Ternary classification showed low abstention rates (<5%) and modest sensitivity gains. Confidence scores were significantly higher for correct predictions. In disagreement tasks, models more often selected the human label when disagreements were framed as "LLM vs human," consistent with authority bias. GPT-4o showed greater resistance to this effect, while others were more prone to defer to perceived human input. In the autonomous search task, OpenAI achieved moderate recall and high precision; Gemini's recall was poor but precision remained high. CONCLUSION/CONCLUSIONS:LLMs hold promise for systematic review screening tasks but require careful prompt design and circumspect human-in-the-loop oversight to ensure robust performance.

Parsonage-Turner syndrome (PTS) is a spontaneous neuropathy characterized by severe upper extremity pain and muscle denervation and is considered to be a rare disease that is under-recognized. Quantitative MRI (qMRI) characterizes muscle denervation but has not been previously assessed in a longitudinal PTS cohort. The aims of this study are to prospectively and longitudinally characterize qMRI changes in PTS patients at baseline (< 6 months' symptom onset) and at follow-up timepoints (3, 6, and 12 months), to measure associations against electromyography (EMG) and muscle strength, and to predict muscle strength at follow-up. A total of 49 subjects (age = 47.2 ± 14.0 years, 31 M/18 F) underwent 3-Tesla qMRI with T2-mapping, diffusion-based muscle fiber diameter, volumetry, and fat fraction (FF) mapping. Image segmentation of involved muscles was performed by two raters. Linear regression between qMRI metrics and days from symptom onset (DSO) was performed. Pearson's correlation quantified correlations between qMRI metrics, and Kendall's tau assessed correlations between qMRI and EMG and muscle strength. For predictive modeling of muscle strength, a generalized linear model was used, and the coefficient of determination (r²) was compared for combinations of baseline inputs. Regression detected a mean T2 increase of 0.66 ms/week and a mean muscle fiber diameter decrease of 0.96 μm/week within DSO of 100. Muscle fiber diameter correlated with muscle volume (r = 0.850). T2 correlated with EMG (|τ| = 0.34-0.78) and muscle strength (|τ| = 0.40-0.83) in most muscles that could be analyzed. Muscle fiber diameter was correlated to EMG (|τ| = 0.43-0.72) and muscle strength in some muscles (|τ| = 0.39-0.56). The addition of baseline T2 values improved the prediction of muscle strength at 3-month (from r² = 0.57 to 0.67, with -0.057 to -0.068 muscle grade per ms T2), at 6-month (r² = 0.40-0.59, -0.057 to -0.071 grade per ms), and at 12-month follow-up (r² = 0.40-0.62, -0.053 to -0.080 grade per ms). Muscle qMRI measurements in PTS depict muscle denervation and provide complementary characterization of muscle quality for diagnosis and follow-up assessment.

BACKGROUND:Primary brain calcifications are observed in several inherited diseases due to different pathogenic mechanisms, including the disruption of the neurovascular unit, mitochondrial dysfunction, and impaired nucleic acid metabolism. OBJECTIVE:The aim of the study was to identify a novel genetic cause of brain calcifications in genetically unresolved cases. METHODS:Exome sequencing data from two unrelated Pakistani patients with generalized dystonia and primary brain calcifications were analyzed. The best candidate gene (ie, RRP12) was then investigated in two large cohorts of patients with brain calcifications from France (n = 111) and China (n = 543). RRP12 loss-of-function phenotype was explored through Western blot and immunocytofluorescence studies on patient-derived fibroblasts and in a knockdown zebrafish model. RESULTS:A combined approach of exome sequencing and homozygosity mapping allowed the prioritization of a rare homozygous variant in RRP12 (c.1558C>T, p.R520C) in two apparently unrelated Pakistani patients from consanguineous families, presenting with infantile-onset generalized dystonia, spasticity, and widespread brain calcifications. Screening of two large cohorts of patients with unresolved brain calcifications revealed two affected French siblings and one unrelated Chinese individual, each carrying rare, biallelic, missense variants in the RRP12 gene (c.1429G>A, p.E477K and c.2634T>G, p.F878L, respectively). Molecular studies revealed a significant reduction in RRP12 protein and abnormal nucleolar morphology in patient'derived fibroblasts. Consistent with its essential role in RNA metabolism, rrp12 knockdown in zebrafish caused severe developmental delay, crimping, and early lethality. CONCLUSIONS:RRP12 is a novel candidate gene for autosomal recessive brain calcifications, possibly associated with a wide clinical spectrum ranging from early-onset severe forms to adult-onset paucisymptomatic presentations. © 2025 The Author(s). Movement Disorders published by Wiley Periodicals LLC on behalf of International Parkinson and Movement Disorder Society.

OBJECTIVE:To examine the associations of LRRK2 p.G2019S, GBA1 p.N409S, polygenic risk scores (PRS), and APOE E4 on PD penetrance, risk, and symptoms. METHODS:We conducted a US-based observational case-control study using data from the 23andMe Inc. and Fox Insight Genetic Substudy (FIGS) databases. The total cohort included 7,586,842 participants (n = 35,163 PD); 8791 LRRK2 p.G2019S carriers (565 with PD), 37,427 GBA1 p.N409S carriers (524 with PD), 244 dual LRRK2/GBA1 carriers (37 with PD), and 7.5 million noncarriers (34,037 with PD). PRS was calculated from the most recently published European genome-wide association study. Survival models estimated the cumulative incidence of PD. Logistic regressions estimated the relative odds of reporting motor and non-motor symptoms according to genetic exposure. RESULTS:By the age of 80 years, the cumulative incidence of PD was 30% for dual carriers, 24% for LRRK2 p.G2019S carriers, 4% for GBA1 p.N409S carriers, and 2% for noncarriers. Higher PRS was associated with increased penetrance of the variants and earlier time to PD diagnosis. GBA1 p.N409S PD was associated with the highest burden of non-motor symptoms, including REM sleep behavior disorder and cognitive/memory deficits, and LRRK2 p.G2019S with the lowest. APOE E4 dosage was associated with greater odds of reporting hallucinations and cognitive impairment in addition to carrier status. INTERPRETATION/CONCLUSIONS:Our findings support the use of genetic screening to enrich candidate selection for neuroprotective trials and better define outcome measures based on genetics.

INTRODUCTION/AIMS/OBJECTIVE:There is a lack of information about startle reflex (SR) in primary lateral sclerosis (PLS). This study examined the presence and prevalence of SR in PLS and compared findings with amyotrophic lateral sclerosis (ALS). METHODS:46 PLS and 54 ALS participants were assessed through structured interviews in this cross-sectional study. Fisher's exact test was used to compare reported SR prevalence. Multivariable linear regression was utilized to study associations between disease group and SR frequency in response to sudden stimuli. RESULTS:SR differed markedly between the two groups, with a higher prevalence in PLS (93.5%) than ALS (20.4%; p < 0.001). Among ALS patients, SR was present in all upper motor neuron (UMN)-predominant cases, which accounted for 54.5% of the SR-positive ALS group, but only 10.4% of probable/definite ALS cases. In SR-positive patients, response frequency to sudden stimuli exceeded 60% in both ALS and PLS, most often triggered by auditory stimuli. Younger age, shorter disease duration, and PLS diagnosis were associated with more frequent SR. DISCUSSION/CONCLUSIONS:SR is significantly more common in PLS than in ALS. Notably, UMN-predominant ALS, although limited in number, showed a higher prevalence of SR (6 out of 6, 100%), indicating that predominant UMN involvement may be a key determinant of SR across both conditions. These hypothesis-generating findings suggest that SR may serve as a novel clinical marker in PLS and UMN-predominant ALS, warranting further validation through prospective studies.

BACKGROUND:Deep medullary veins (DMVs) play important roles within the cerebrovascular network related to brain drainage and clearance. Although they have been previously correlated with brain volume, it is unknown whether their count is specifically correlated with subcortical or cortical volume changes. PURPOSE/OBJECTIVE:This study aims to better understand the relationship between DMVs, subcortical (lateral ventricle to intracranial volume ratio (ICV)) and cortical atrophy (cortical thickness) to identify whether DMVs can be a predictor of volume changes in these regions. METHODS:We performed a retrospective analysis of 332 cognitively healthy subjects previously followed between 2010 and 2019. Imaging and patient charts were analyzed for baseline demographic and clinical characteristics. Patients underwent a standardized cognitive interview and received a magnetic resonance imaging scan to assess DMVs, cortical thickness, lateral ventricle and global gray matter (GM) volumes, white matter lesions (WMLs) and microbleeds. RESULTS:Among 332 patients (62% female, median age 70), lateral ventricle/ICV was significantly related to DMV count (p<0.001). Similarly, sex stratified analyses confirmed that a larger lateral ventricle/ICV ratio, but not cortical thickness or global GM volumes, was associated with fewer DMVs. In the entire group, subcortical atrophy remained a significant predictor of DMVs even after accounting for baseline characteristics, WMLs, microbleeds and total gray matter volume. CONCLUSIONS:In a large cohort of cognitively unaffected people, subcortical, but not cortical, atrophy was significantly correlated with venous health as measured by DMVs. Reduced DMVs are a strong predictor of ventricular enlargement.

Antibiotic-induced microbiome injury, defined as a reduction of ecological diversity and obligate anaerobe taxa, is associated with negative health outcomes in hospitalized patients, and healthy individuals who received antibiotics in the past are at higher risk for autoimmune diseases. Postbiotics contain mixtures of bacterial fermentation metabolites and bacterial cell wall components that have the potential to modulate microbial communities. Yet, it is unknown if a fermentation-derived postbiotic can reduce antibiotic-induced microbiome injury. Here, we present the results from a single-center, randomized placebo-controlled trial involving 32 patients who received an oral, fermentation-derived postbiotic alongside oral antibiotic and probiotic therapy for non-gastrointestinal (GI) infections. At the end of the antibiotic course, patients receiving the postbiotic (n = 16) had significantly higher fecal bacterial alpha diversity (+40%, inverse Simpson index) compared to the placebo group (n = 16), and the treatment was well-tolerated. Analysis of 157 longitudinal fecal samples revealed that this increased diversity was driven by enrichment of health-associated taxa, notably obligate anaerobic Firmicutes, particularly Lachnospiraceae. In contrast, Escherichia/Shigella species, often linked to pathogenicity and antibiotic resistance, were reduced in postbiotic-treated patients at the end of antibiotic treatment and remained lower up to 10 days later. Our findings suggest that postbiotic co-administration during antibiotic therapy may augment health-associated gut microbiome composition and mitigate antibiotic-induced microbiome injury.Trial registration ISRCTN30327931 retrospectively registered.