Table 3.
Association between plasma biomarkers and conversion to AD dementia.
| Model | Hazard ratio | AUC [95% CI] | Ref: basic model | |||
|---|---|---|---|---|---|---|
| Plasma Aβ42/Aβ40 | Plasma P-tau217 | Plasma NfL | P value | AICΔ | ||
| ATN |
1.83 [1.23, 2.74] (P = 0.0031) |
2.97 [1.62, 5.46] (P = 0.0004) |
1.07 [0.67, 1.71] (P = 0.7898) |
0.82 [0.77, 0.91] | <0.0001 | −25 |
| A |
2.00 [1.39, 2.89] (P = 0.0002) |
0.77 [0.70, 0.86] | 0.0003 | −11 | ||
| T |
3.54 [1.98, 6.31] (P < 0.0001) |
0.77 [0.68, 0.86] | <0.0001 | −21 | ||
| N |
1.51 [0.96, 2.35] (P = 0.0719) |
0.67 [0.60, 0.78] | 0.0753 | −1 | ||
This table shows the results from fitting Cox regression models with conversion to AD as an outcome and plasma biomarkers added separately or all together to a basic model consisting of age, sex, and education. Hazard ratios are presented in terms of “increased risk of converting to AD for each standard deviation change in biomarker value.” AUC values were evaluated and confidence intervals were calculated using 1000 bootstrapped samples. The basic model consisting of only demographics had AUC = 0.64 (95% CI [0.55, 0.77]) and AIC = 274. P values represent an ANOVA comparison to the basic model; AICΔ values represent the change in AIC compared to the basic model and an AICΔ value of −2 or lower implies a better fit than the basic model. All statistical tests were two-sided with no adjustment for multiple comparisons.