GFAP Represents Greatest Longitudinal Clinical Value Biomarker for NMOSD
Serum GFAP levels showed consistent correlations with disease activity and the largest area under the ROC curve to differentiate attacks from remissions in individuals with NMOSD.
Young-Min Lim, MD, PhD
Longitudinally, glial fibrillary acidic protein (GFAP) might be the most appropriate biomarker to monitor disease activity for patients with neuromyelitis optica spectrum disorder (NMOSD), according to results from a prospective study.
Senior author Young-Min Lim, MD, PhD, Department of Neurology, Asan Medical Center, and colleagues evaluated longitudinal changes in serum biomarkers such as neurofilament light chain (NfL), GFAP, and GFAP/NFL in 64 patients (58 women) with antiaquaporin-4 antibody NMOSD. A total of 133 samples were obtained at enrollment, after 6- to 12-months of follow-up, and at attacks. Investigators also recruited age- and sex-matched health controls (HC) who visited the Asan Medical Center during the study period.
All told, 62 patients did not develop new attacks during the main period. Patients were then classified into 2 groups depending on the baseline status of having recent attacks (< 2 months). In the baseline-attack group (n = 14), serum levels of NfL, GFAP, and GFAP/NfL quotient were significantly decreased at follow-up, as well as Expanded Disability Status Scale (EDSS) scores (baseline vs follow-up: 3.0 [range, 3.0-4.4], vs 2.8 [range, 2.0-4.1]; P = .031).
Among those in the no-baseline-attack group (n = 48), serum levels of NfL and GFAP, but not GFAP/NfL quotient, were modestly but significantly increased at follow-up. During this period, EDSS scores were constant (baseline vs follow-up: 3.3 [range, 2.0-4.0] vs 3.0 [range, 2.0-4.0]; P = .245). “These findings suggest that GFAP might be the most suitable of these serum markers to longitudinally monitor NMOSD, and warrant future confirming studies,” Lim et al wrote.
READ MORE: Optical Coherence Tomography Angiography Helps Differentiate MS and NMOSD
A total of 5 patients developed new attacks during the whole study period. Investigators found that all 3 serum markers tended to increase when attacks occurred, but with 2 exceptions. The serum level of NfL in case 2 was decreased at attack and the serum level of GFAP/NfL quotient in case 4 was decreased at attack. Despite this, the serum level of GFAP increased consistently in both cases when attacks occurred.
To differentiate clinical attacks from remission, investigators evaluated and compared 19 serum samples during the attack state and 114 during the remission state. Serum levels of all markers at attack state were higher than those of patients at remission state and those of HC; however, receiver operating characteristics (ROC) curve analysis showed that the serum level of GFAP had the largest area under the ROC (AUROC) to discriminate attack from remission (0.876; 95% CI, 0.801-0.951). The data showed a cut-off value of 218.6 pg/mL for serum GFAP, with 94.7% and 74.6% sensitivity and specificity values, respectively.
Lim and colleagues also evaluated whether aggressive immune treatment impacted the longitudinal courses of serum biomarkers by using a subgroup of patients in the nonbaseline-attack group. Patients were classified by either the use of rituximab (Rituxan; Genentech/Biogen; n = 12) or other immunosuppressants (n = 36). All of those treated in the rituximab group had already been treated with the therapy when they entered the study, whereas only 3 patients in the other immunosuppressants group had just started immunosuppressants when they entered the study.
Although a significant increase in serum biomarker levels between baseline and follow-up was only observed for serum GFAP levels in the other immunosuppressant group, all markers tended to increase over time. Additionally, Expanded Disability Status Scale scores during this period were constant in both subgroups (baseline vs follow-up: rituximab treatment, 3.8 [range, 2.6-6.0] vs 3.5 [range, 2.0-7.3]; P >.999; other immunosuppressant treatment, 3.0 [range, 2.0-4.0] vs 3.0 [range, 2.0-4.0]; P = .170).
