Table 2.

Cox regression analysis identifiying κ-FLC index and sNfL Z score as predictors for time to second clinical attack.

	Coefficient	Standard error	Hazard ratio	95%-CI	P value
Age (years)	−0.051	0.032	0.950	0.892–1.012	0.109
Sex (ref: male)	0.462	0.650	1.587	0.443–5.677	0.478
Disease duration (days)	−0.024	0.010	0.976	0.957–0.996	0.018
Follow-up duration (months)	0.204	0.079	1.226	1.050–1.433	0.010
Number of T2 hyperintense lesions	−0.021	0.013	0.979	0.955–1.004	0.099
Number of T1 contrast-enhancing lesions	−0.096	0.219	0.909	0.592–1.395	0.661
DMT administration	0.427	0.732	1.532	0.365–6.438	0.560
κ-FLC index	0.021	0.006	1.021	1.010–1.032	<0.001
sNfL Z score	0.784	0.268	2.191	1.300–3.706	0.003

Cox and Snell's pseudo R² = 0.481.

Concordance = 0.834 (SE = 0.047).

Disease duration was the time between symptom onset and lumbar puncture. Age was determined at the time of lumbar puncture. Number of MRI lesions were also determined at baseline. Follow-up duration was the time between disease onset and the last clinical visit. DMT administration was determined until occurrence of second clinical attack or end of follow-up, respectively.

Model quality: Covariates were properly included (no necessity for non-linearity). No influential observations were detected (Jackknife procedure revealed that after excluding each patient once all estimates were within the 95%-CI of the estimates of original patient cohort). Proportional hazards assumption was met (χ² = 11.267, p = 0.258).

CI, confidence interval; DMT, disease-modifying treatment; FLC, free light chain; MRI, magnetic resonance imaging; sNfL, serum neurofilament light.