Serum glial fibrillary acidic protein predicts disease progression in multiple sclerosis

Abstract

Objective: Glial fibrillary acidic protein (GFAP) is expressed in astrocytes and may be a useful marker of non-active progressive multiple sclerosis (MS). We evaluate serum GFAP (sGFAP) in a large cohort of MS patients to determine if it predicts progression independent of relapse activity (PIRA), future gait aid, and conversion to secondary progressive disease (SPMS).

Methods: Adults with clinically isolated syndrome or any subtype of MS who were listed in the Brigham MS Center Research Database and had at least one sGFAP result were included. All clinic visits following first sample were analyzed for PIRA, future gait aid, and conversion to SPMS. Future cognitive dysfunction and fatigue were evaluated as secondary outcomes.

Results: In total, 741 patients were included (average age: 42.3, average disease duration: 3.7 years, median EDSS: 2, and median follow-up duration: 10.0 years). Of 643 patients (86.8%) without progressive disease at baseline, 15.9% developed SPMS. Among all 741, 50.5% had PIRA and 18.6% developed a gait aid requirement. sGFAP level predicted PIRA, future gait aid, and conversion to SPMS in univariable models (p < 0.001, <0.001, and 0.002). sGFAP remained predictive for PIRA and future gait aid in multivariable models in those younger than 50 (p = 0.048, 0.003). Change in sGFAP level over time was not predictive. There was no association between sGFAP and future fatigue or cognitive dysfunction.

Interpretation: sGFAP helps to predict PIRA, future gait aid, and conversion to SPMS in a large cohort of MS patients. Our data suggest that baseline levels may be more useful than the change over time.

Figure 1

(A) Scatter plot showing an increase in sGFAP with increasing age (linear regression R‐squared: 0.125; p < 0.001), with a potential nonlinear increase after approximately age 50. (B) Box plot showing an overall increase in mean sGFAP level with increasing EDSS. The number of patients at each EDSS level at FSD is listed in parentheses. (C) Box plot of sGFAP level by MS subtype at FSD showing higher levels for progressive compared to nonprogressive patients. There was no significant difference between CIS and RRMS patients or between SPMS and PPMS patients.

Figure 2

Box plots of sGFAP levels at FSD in patients with and without future PIRA (A), future gait aid requirement (B), and future conversion to SPMS (C).

Figure 3

Kaplan–Meier curves showing time to PIRA (A), time to new gait aid (B), and time to SPMS conversion (C) by sGFAP level stratified by quartile. The lowest quartile corresponds to the 25% of patients with the lowest sGFAP level at FSD (blue). For each outcome, the HR of each quartile relative to the lowest quartile is included in parentheses in the legend (p < 0.05 is denoted by an asterisk). Fewer than 1% of patients remained at risk by year 20.

Figure 4

Scatter plots showing no association between sGFAP level at FSD and future SDMT score (A) and future total MFIS score (B). For the 173 patients included in this analysis, sGFAP levels were drawn 2–5 years prior to SDMT and MFIS assessments.