Serum levels of neurofilament light-chain (NfL), a biomarker of axonal and synaptic damage, predict 5-year outcome in acutely ill schizophrenia patients

Abstract

Background: Disruptions of neuronal macro- and microcircuits in schizophrenia (SCZ) have been attributed to pathologies of neuronal connectivity. We hypothesized that during acute psychosis structural damage to axons and synapses occurs in a subset of patients, rendering them more susceptible to poor treatment outcomes or relapse. Methods: Neurofilament light-chain protein (NfL), a marker of axonal and synaptic damage, was quantified in sera of acutely ill unmedicated schizophrenia patients (SCZ, n=98) at hospital admission (T0) and after 6 weeks of inpatient treatment (T6), compared to healthy controls (HC, n=132). Additionally, we analyzed blood cell counts, the levels of 171 serum biomarkers, clinical symptomatology and functioning relapse occurrences requiring hospitalisation during the 5-year follow-up. Univariate analysis and support vector machine learning were employed to identify SCZ with increased NfL, interrogate the relationship between biological factors and NfL and examine whether NfL and associated biological factors can predict treatment outcome or likelihood of relapse. Results: NfL was increased in SCZ versus HC (p=0.014), correlated positively with positive and negative syndrome scale-positive scores (rho=0.265, P=0.019) and inversely with global assessment of functioning-scores (rho=-0.365, P<0.001) at T0. Decreased NfL at T6 (P<0.001) did not correlate with administered cumulative antipsychotic dosage. NfL levels were predicted using a serum biomarker set with a balanced accuracy of 67.2 % (P5000=0.0008). Higher levels of insulin-like growth factor binding protein 2 (IGFBP-2), granulocytes, leukocytes, thrombocytes, Creactive protein (CRP) and mean corpuscular volume (MCV) predicted elevated NfL, whereas erythrocytes, hematocrit, HbA1c and hemoglobin predicted normal NfL levels. NfL and these associated factors also predicted psychotic relapse during 5-year follow-up with a balanced accuracy of 66.3% (P5000=0.0382). Conclusions: NfL is embedded in a circuitry of neurobiological factors related to synaptic plasticity (IGFBP-2), innate immune response (granulocytes, leukocytes, CRP), glucose metabolism (IGFBP-2, glucose, HbA1c), and haemodynamics (MCV, hematocrit, thrombocytes, haemoglobin). Approximately 25% of the SCZ patients showed increased NfL levels. The identified associated network of biological factors was predictive for long-term treatment outcome in schizophrenia in terms of the likelihood of future relapses thus pointing towards innate immunity–based acutely destabilized neuronal circuits as an endophenotype for future therapy and disease monitoring in schizophrenia.