Fibroblast activation protein and Periostin in cerebrospinal fluid: Candidate biomarkers in chronic spinal cord injury

Abstract Chronic spinal cord injury is biologically heterogeneous, and biomarkers that distinguish injury severity from longitudinal biological response are lacking. We evaluated whether fibroblast activation protein and Periostin in cerebrospinal fluid provide complementary, chronicity-dependent information on neurological severity and recovery in chronic traumatic spinal cord injury. This secondary biomarker study analyzed prospectively collected data and cerebrospinal fluid samples from two clinical trials. Seventy-two participants treated with human umbilical cord-derived mesenchymal stromal cells were classified as early chronic (2–12 months after injury; n = 35) or late chronic (> 12 months; n = 37); 18 early chronic participants received saline, and 14 non-spinal cord injury patients served as controls. Fibroblast activation protein and Periostin were measured before intervention and at month 3, with paired samples available for 62 human umbilical cord-derived mesenchymal stromal cell-treated participants. American Spinal Injury Association Impairment Scale (AIS) grade and motor, sensory, and mobility outcomes were assessed through month 15. At baseline, both biomarkers were inversely associated with AIS grade in the human umbilical cord-derived mesenchymal stromal cell cohort, with a stronger correlation for fibroblast activation protein ( ρ = .0.5730, P < 0.001) than Periostin ( ρ = .0.3125, P = 0.0075). Fibroblast activation protein discriminated AIS A from AIS B–D with an area under the curve (AUC) of 0.8522 and AIS A–B from AIS C–D with an AUC of 0.7337; corresponding AUCs for Periostin were 0.7009 and 0.6144. Associations with 15-month AIS outcomes were broader in early than late chronic spinal cord injury. In the early chronic human umbilical cord-derived mesenchymal stromal cell cohort, baseline fibroblast activation protein separated motor, pinprick, and light-touch improvement groups (AUCs, 0.7454–0.7989), and lower baseline fibroblast activation protein was associated with a higher cumulative probability of first AIS-grade improvement (log-rank P = 0.045). Early 3-month biomarker changes were not associated with motor improvement. In late chronic spinal cord injury, larger decreases in fibroblast activation protein and Periostin were associated with greater motor improvement ( P = 0.0218 and P = 0.0139), differentiated longitudinal motor, sensory, and mobility trajectories, and were associated with a higher cumulative probability of first AIS-grade improvement (both log-rank P = 0.035). Saline-cohort findings were heterogeneous and exploratory. Fibroblast activation protein and Periostin in cerebrospinal fluid showed complementary patterns across trial-defined chronicity strata in chronic spinal cord injury. Baseline fibroblast activation protein was consistently associated with injury severity and subsequent neurological status, particularly in early chronic spinal cord injury, whereas within-person changes in fibroblast activation protein and Periostin after human umbilical cord-derived mesenchymal stromal cell treatment were associated with recovery in late chronic spinal cord injury and may represent candidate response-monitoring signals. These roles require independent validation, and the treatment specificity of the dynamic biomarker changes remains to be established.