. 2017 Sep 18;60(2):561–576. doi: 10.3233/JAD-170327

Table 2.

Logistic regression model result by comparing CSF with amyloid-PET measures

	SUVR_CB	SUVR_WM	V_T
Aβ_1–42	[0.788] 0.379	[0.773] 0.301	[0.692] 0.287
Aβ_1–42/Aβ_1 - 40	[0.819] 0.622	[0.862] 0.641	[0.724] 0.466
Aβ_1–42, T-tau	[0.926] 0.622	[0.903] 0.381	[0.746] 0.102
Aβ_1–42/Aβ_1–40, T-tau	[0.908] 0.526	[0.905] 0.584	[0.726] 0.466
Aβ_1–42, P-tau₁₈₁	[0.902] 0.516	[0.863] 0.145	[0.774] 0.186
Aβ_1–42/Aβ_1–40, P-tau₁₈₁	[0.894] 0.622	[0.847] 0.641	[0.794] 0.466
Aβ_1–42, T-tau, P-tau₁₈₁	[0.950] 0.754	[0.973] 0.695	[0.774] 0.186
Aβ_1–42/Aβ_1–40, T-tau, P-tau₁₈₁	[0.912] 0.604	[0.948] 0.656	[0.810] 0.466

Logistic regression models compared the amyloid-PET scans (SUVR_CB, SUVR_WM, and V_T) with either CSF Aβ_1–42 or Aβ_1–42/Aβ_1–40, or these CSF biomarkers in combination with T-tau and/or P-tau₁₈₁. [AUC] and Cohen’s κ coefficients. If T-tau or P-tau were included with CSF Aβ_1–42 and Aβ_1–42/Aβ_1–40 in the model, a higher concordance was detected when the CSF Aβ_1–42/Aβ_1–40 ratio was applied compared to CSF Aβ_1–42 alone (for V_T and both SUVR measures). Similar findings were detected in case both CSF tau concentrations were added to Aβ_1–42 alone or Aβ_1–42/Aβ_1–40 in the logistic regression models. AUC, area under the curve; CB, cerebellar grey matter; SUVR, standardized uptake value ratio; V_T, total volume of distribution; WM, whole subcortical white matter.