Association and multimodal model of retinal and blood-based biomarkers for detection of preclinical Alzheimer’s disease

Table 2 Univariate and multivariate joint mixed models for predicting putative retinal gliosis from blood-based biomarkers

	Fixed effects			Random effects	Effect size
Variables	β-estimate (S.E)	t-test statistic	p-value	ICC	Pseudo-R² (fixed effects/total)
Model 1 – Univariate mixed model (p-tau217); Observations = 141, Groups = 82
(Intercept)	1.48(0.58)	2.58	0.012	0.53	0.085/0.57
p-tau217	0.48(0.17)	2.79	0.007^#
Age	-0.014(0.009)	-1.63	0.11
Model 2 – Univariate mixed model (Aβ42/40); Observations = 133, Groups = 78
(Intercept)	1.21(0.62)	1.96	0.05	0.56	0.007/0.56
Aβ42/40	0.002(0.05)	0.046	0.96
Age	-0.007(0.009)	-0.77	0.44
Model 3 – Univariate mixed model (p-tau181); Observations = 135, Groups = 79
(Intercept)	1.40(061)	2.30	0.024	0.56	0.022/0.57
p-tau181	0.003(0.003)	1.04	0.30
Age	-0.011(0.009)	-1.16	0.25
Model 4 – Univariate mixed model (NfL); Observations = 133, Groups = 78
(Intercept)	1.17(0.63)	1.56	0.07	0.56	0.008/0.56
NfL	-0.003(0.009)	-0.30	0.76
Age	-0.006(0.01)	-0.58	0.56
Model 5 – Multivariate mixed model; Observations = 133, Groups = 78
(Intercept)	1.29(0.62)	2.10	0.039	0.54	0.079/0.58
p-tau217	0.43(0.18)	2.35	0.022
Aβ42/40	0.004(0.06)	0.069	0.95
p-tau181	0.002(0.004)	0.59	0.55
NfL	-0.008(0.01)	-0.74	0.46
Age	-0.001(0.01)	-1.00	0.32

RNFL – retinal nerve fiber layer; p-tau – phosphorylated tau; Aβ – amyloid beta; ICC – intraclass correlation coefficient; S.E - standard error; ^#p < 0.01 (α_Bonferroni)

ISSN: 1758-9193