Table 3.

Association between short-term (up to 12 months) sNfL change and long-term MRI/clinical outcomes, using sNfL = 16 pg/mL as the threshold.

Comparison	Patients (n)	Month	Change from baseline to 4 years, LSM difference (95% CI); p
			PBVC	New T2 lesions	EDSS
No sNfL decreasea versus sNfL decreaseb	58 38	3	−0.564 (−1.440, 0.312); 0.205	1.071 (0.455, 2.521); 0.875	0.321 (−0.149, 0.791); 0.179
No sNfL decreasea versus sNfL decreaseb	42 49	6	−0.891 (−1.783, 0.002); 0.050	1.455 (0.620, 3.415); 0.389	0.237 (−0.271, 0.746); 0.357
No sNfL decreasea versus sNfL decreaseb	29 68	9	−0.882 (−1.845, 0.080); 0.072	1.839 (0.727, 4.651); 0.198	0.530 (0.019, 1.041); 0.042
No sNfL decreasea versus sNfL decreaseb	33 87	12	−1.560 (−2.401, −0.718); 0.0003	3.067 (1.342, 7.011); 0.008	0.513 (0.072, 0.954); 0.023

CI: confidence interval; EDSS: expanded disability status scale; LSM: least square means; PBVC: percent brain volume change; sNfL: serum neurofilament light.

Generalized linear regression models were used for the PBVC and EDSS outcomes, and a negative binomial regression model was used to assess T2 lesions.

^asNfL levels remained ⩾16 pg/mL.

^bsNfL levels decreased from ⩾16 pg/mL to <16 pg/mL.