Towards development of neurofilament light chain and glial fibrillary acidic protein as precision medicine biomarkers for multiple sclerosis

Background: We have insufficient diagnostic tools to capture and anticipate the course of multiple sclerosis (MS) and to monitor treatment response. Blood-based biomarkers could provide a valuable measure to detect neurodegeneration and disease worsening in MS. Serum neurofilament light chain (sNfL) is a biomarker of neuro-axonal injury that has been investigated in its association with disease activity and disability accumulation in MS, but larger scale studies to determine the sNfL levels of healthy persons and MS patients are currently lacking. Furthermore, we lack biomarkers to discern the pathogenesis of ‘pure progression’ in MS from that due to focal inflammatory activity. Serum glial fibrillary acidic protein (sGFAP) is a marker for astrogliosis and a potential candidate biomarker that may be more strongly associated with disease progression than active inflammation in MS. Objectives: We aimed to bring sNfL closer to clinical application by establishing a reference database of sNfL levels from control persons, in order to enable the determination of pathological sNfL levels by calculation of sNfL percentiles and Z scores of MS patients. Further, we used this reference database to analyze sNfL’s ability to capture and prognosticate disease activity in patients followed in the Swiss MS Cohort (SMSC) and the Swedish MS Registry and explored the effectiveness of disease modifying therapies. Further, we assessed the value of sGFAP in addition to sNfL as a biomarker for disease progression and acute inflammation, as well as in patients under B-cell depleting therapy. Methods: We used the Single Molecule Array (SIMOA) technology (Quanterix) for the measurements of sNfL and sGFAP. For the sNfL reference database, persons with no evidence of CNS disease were included from four cohort studies in Europe and North America. A generalized additive model for location, scale and shape (GAMLSS) was used to model the distribution of sNfL concentrations in function of age and body mass index (BMI). We tested the reference database by generating sNfL percentiles and Z scores in the SMSC, and as a validation in the Swedish MS Registry. In the second study, we measured sNfL and sGFAP in three different groups of patients in the SMSC: firstly, matched patients with MS who had either stable disease or disability progression with no relapses during the entire follow-up; secondly, patients with MRI or clinical signs of acute neuroinflammation or in remission; thirdly, patients who had initiated and continued B-cell–depleting treatment (ocrelizumab or rituximab). 8 Results: In the first study we measured sNfL concentrations in 10’133 serum samples from 5’390 control persons and found an age- and BMI-related sNfL increase. We also measured 7’769 serum samples from 1’313 MS patients from the SMSC. sNfL Z scores prognosticated an increased risk for future disease activity and normalized in patients under treatment with monoclonal antibodies compared to other treatments or untreated patients. These results were validated in 4’341 samples from the Swedish MS Registry. In the second study we measured sNfL and sGFAP in 355 patients and 259 healthy controls. sGFAP concentrations in the controls increased with age and BMI and were higher in women than men. Patients with worsening progressive MS had higher levels of sGFAP than stable patients even after adjustment for sNfL. Furthermore, baseline sGFAP was associated with gray matter volume loss, but not white matter volume loss, and remained unchanged during relapses compared to remission phases. Additionally, the combination of sGFAP and sNfL Z scores could prognosticate future disability worsening and 'progression independent of relapse activity' (PIRA). Conclusion: Our reference database and the therein derivable sNfL percentiles or Z scores enable the identification of individual persons with MS at risk for future disease worsening and treatment response also in otherwise seemingly stable disease stage. Furthermore, sGFAP may be a sensitive tool to capture and prognosticate future PIRA, especially in combination with sNfL