To evaluate the diagnostic and prognostic value of cerebrospinal fluid (CSF) chitinase-3-like protein 2 (CHI3L2) in amyotrophic lateral sclerosis (ALS).

ALS is a progressive neurodegenerative disease with variable clinical course. Reliable biomarkers are needed to improve early diagnosis and predict disease progression.

In the discovery cohort, we performed proteomic analysis of CSF in the Second Affiliated Hospital of Zhejiang University. ALS was diagnosed according to the El Escorial criteria and classified as Rapid-ALS (ΔFS ≥0.75) or Slow-ALS (ΔFS <0.75) based on ALSFRS-R. Feature selection and the XGBoost model were applied to identify key proteins and assess diagnostic performance. In the validation cohort, CHI3L2 levels were measured by ELISA.

A total of 90 individuals (mean age 59 years; 71.7% men) were included in the discovery cohort, involving 60 ALS patients (30 Rapid-ALS, 30 Slow-ALS) and 30 healthy controls. Using XGBoost model, CSF CHI3L2 was identified as an important feature protein distinguishing ALS from healthy controls and Rapid-ALS from Slow-ALS. A CHI3L2-based machine learning model achieved good diagnostic performance (ALS vs. controls: AUC = 0.88; Rapid-ALS vs. Slow-ALS: AUC = 0.82). The result was consistent in the validation cohort (37 ALS, 18 controls). After adjusting for age and sex, higher CSF CHI3L2 was associated with increased ALS rish(OR = 1.281, 95% CI: 1.110–1.477, p <0.001) and faster disease progression (OR = 1.206, 95% CI: 1.064–1.368, p = 0.003). CHI3L2 also distinguished ALS from healthy controls (AUC =0.863) and Rapid-ALS from Slow-ALS (AUC =0.826).

CSF CHI3L2 is elevated in ALS and correlates with disease progression rate, supporting its potential as an early biomarker for identifying patients at risk of faster disability progression.