TY - JOUR
T1 - Agreement between Published Reference Resources for Neurofilament Light Chain Levels in People with Multiple Sclerosis
AU - Sotirchos, Elias S.
AU - Hu, Chen
AU - Smith, Matthew D.
AU - Lord, Hannah Noelle
AU - Duval, Anna L.
AU - Arrambide, Georgina
AU - Montalban, Xavier
AU - Akgun, Katja
AU - Ziemssen, Tjalf
AU - Naismith, Robert T.
AU - Hersh, Carrie M.
AU - Hyland, Megan
AU - Krupp, Lauren B.
AU - Nicholas, Jacqueline A.
AU - Bermel, Robert A.
AU - Mowry, Ellen M.
AU - Calabresi, Peter A.
AU - Fitzgerald, Kathryn C.
N1 - Publisher Copyright:
© American Academy of Neurology.
PY - 2023/12/5
Y1 - 2023/12/5
N2 - ObjectivesTo examine the agreement between published reference resources for neurofilament light chain (NfL) applied to a large population of people with multiple sclerosis (MS).MethodsSix published reference resources were used to classify NfL in participants in the Multiple Sclerosis Partners Advancing Technology and Health Solutions (MS PATHS) network as elevated or normal and to derive age-specific NfL Z-scores. NfL values were classified as elevated if they exceeded the >95th percentile (i.e., Z-score >1.645) of the age-specific reference range. Furthermore, age-specific NfL Z-scores could be derived for 4 of 6 reference resources.ResultsNfL measurements were assessed from 12,855 visits of 6,687 people with MS (median 2 samples per individual [range 1-7]). The mean ± SD age was 47.1 ± 11.7 years, 72.1% of participants were female, disease duration was 15.0 ± 10.6 years, body mass index was 28.6 ± 6.9 kg/m2, and serum NfL was 12.87 ± 12.86 pg/mL. Depending on the selection of the reference resource, the proportion of NfL measurements classified as elevated varied from 3.7% to 30.9%. The kappa coefficient across the 6 reference resources used was 0.576 (95% CI 0.571-0.580) indicating moderate agreement. Spearman correlations between Z-scores derived from the various reference resources exceeded 0.90; however, concordance coefficients were lower, ranging from 0.72 to 0.89.DiscussionInterpretation of blood NfL values may vary markedly depending on the selection of the reference resource. Borderline elevated values should be interpreted with caution, and future studies should focus on standardizing NfL measurement and reporting across laboratories/platforms, better characterizing the effects of confounding/influencing factors, and defining the performance of NfL (including as part of multimodal predictive algorithms) for prediction of disease-specific outcomes.
AB - ObjectivesTo examine the agreement between published reference resources for neurofilament light chain (NfL) applied to a large population of people with multiple sclerosis (MS).MethodsSix published reference resources were used to classify NfL in participants in the Multiple Sclerosis Partners Advancing Technology and Health Solutions (MS PATHS) network as elevated or normal and to derive age-specific NfL Z-scores. NfL values were classified as elevated if they exceeded the >95th percentile (i.e., Z-score >1.645) of the age-specific reference range. Furthermore, age-specific NfL Z-scores could be derived for 4 of 6 reference resources.ResultsNfL measurements were assessed from 12,855 visits of 6,687 people with MS (median 2 samples per individual [range 1-7]). The mean ± SD age was 47.1 ± 11.7 years, 72.1% of participants were female, disease duration was 15.0 ± 10.6 years, body mass index was 28.6 ± 6.9 kg/m2, and serum NfL was 12.87 ± 12.86 pg/mL. Depending on the selection of the reference resource, the proportion of NfL measurements classified as elevated varied from 3.7% to 30.9%. The kappa coefficient across the 6 reference resources used was 0.576 (95% CI 0.571-0.580) indicating moderate agreement. Spearman correlations between Z-scores derived from the various reference resources exceeded 0.90; however, concordance coefficients were lower, ranging from 0.72 to 0.89.DiscussionInterpretation of blood NfL values may vary markedly depending on the selection of the reference resource. Borderline elevated values should be interpreted with caution, and future studies should focus on standardizing NfL measurement and reporting across laboratories/platforms, better characterizing the effects of confounding/influencing factors, and defining the performance of NfL (including as part of multimodal predictive algorithms) for prediction of disease-specific outcomes.
UR - http://www.scopus.com/inward/record.url?scp=85178665613&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85178665613&partnerID=8YFLogxK
U2 - 10.1212/WNL.0000000000207957
DO - 10.1212/WNL.0000000000207957
M3 - Article
C2 - 37816633
AN - SCOPUS:85178665613
SN - 0028-3878
VL - 101
SP - E2448-E2453
JO - Neurology
JF - Neurology
IS - 23
ER -