Ahmed Negida

Low Ch4 gray-matter density marks a more severe, faster-declining PD-MCI group

Using voxel-based morphometry and probabilistic Ch4 maps in 162 PPMI participants with PD-MCI, we identified 40 with low Ch4 density (below −1 SD). They had worse MDS-UPDRS, autonomic, and olfactory scores and reached cognitive milestones sooner (log-rank p = 0.002). The classifier is released as an open research tool.

Parkinsonism & Related Disorders, 2025 →

Illustration: Julich-Brain Ch4 on the MNI152 template (not a figure from the article).

Ch4 atrophy and free water independently predict cognition in PD

We combined T1-weighted and free-water diffusion MRI in 92 participants (controls, PD-NC, PD-MCI, PDD). Ch4 gray-matter density was lowest and free water highest in PDD. Only these two Ch4 measures independently predicted cognition, and together they separated PDD from PD without dementia (exploratory AUC 0.93).

medRxiv preprint, 2026 →

Ch4–cingulo-opercular connectivity links Ch4 degeneration to cognition

Resting-state fMRI extends the multimodal Ch4 signature. Ch4 connectivity with the attention-related cingulo-opercular network statistically mediated the relationship between Ch4 degeneration and cognition, supporting a candidate circuit link between Ch4 degeneration and cognitive impairment.

Slower resting EEG rhythms flag cognitive fluctuations in Lewy body disease

Eyes-closed resting EEG in 53 patients across the Lewy body spectrum. Patients with fluctuations had lower dominant frequency, dominant-alpha prevalence, and individual alpha peak frequency after adjusting for age, sex, and MoCA; an elastic-net model detected fluctuations with AUC 0.86. The topographic maps show mean dominant frequency at every electrode: about 6–7 Hz in patients with fluctuations versus about 8 Hz in those without.

Alzheimer's Research & Therapy, 2026 → CC BY 4.0

Diffuse-malignant PD progresses fastest and could halve trial sample sizes

Over 10 years of PPMI follow-up, we compared clinical, pathological, and data-driven subtypes on milestones across six functional domains using Cox regression. Diffuse-malignant patients reached 63.0% of milestones (PIGD 55.6%, body-first 54.0%), and in trial simulations enrolling them cut the required sample size by about 50%.

npj Parkinson's Disease, 2026 → CC BY 4.0 · cropped

Year-specific percentiles enable real-time PD subtyping for trial enrollment

Percentile thresholds for motor, cognitive, RBD, and autonomic scores, set for each of the first 5 years after diagnosis (PPMI, N = 1,030), classify patients prospectively each year. Diffuse-malignant PD reached a first progression milestone fastest (HR 3.03, 95% CI 2.30–3.97, vs mild motor-predominant).

Mov Disord Clin Pract, 2026 → CC BY-NC-ND 4.0

LRRK2 and GBA1 map onto different PD subtypes

Variant frequencies were compared across four subtyping frameworks in 1,390 genotyped PPMI patients with FDR correction. LRRK2 variants were enriched in α‑synuclein SAA-negative PD (37.1% vs 10.2%) and GBA1 variants in diffuse-malignant PD (14.0% vs 5.9% in mild motor-predominant); the tremor/gait split showed no signal that survived correction.

Genes (Basel), 2026 → CC BY 4.0

TBI mortality across 29 countries does not follow a development gradient

This prospective cohort enrolled 2,165 patients with operative or nonoperative TBI at 100 hospitals. In inverse-probability-weighted mixed-effects models, mortality showed no dose-response with the Human Development Index; injury mechanism, demographics, and prehospital access drove the disparities. Role: co-founder of the Global NeuroSurg Research Collaborative; co-principal investigator, steering committee, and writing group.

Nature Medicine, 2026 →