Objective: To evaluate CSF levels of beta-amyloid(1-42) (Abeta42) alone and in combination with CSF tau for distinguishing AD from other conditions.
Methods: At 10 centers in Europe and the United States, 150 CSF samples from AD patients were analyzed and compared with 100 CSF samples from healthy volunteers or patients with disorders not associated with pathologic conditions of the brain (CON), 84 patients with other neurologic disorders (ND), and 79 patients with non-Alzheimer types of dementia (NAD). Sandwich ELISA techniques were used on site for measuring Abeta42 and tau.
Results: Median levels of Abeta42 in CSF were significantly lower in AD (487 pg/mL) than in CON (849 pg/mL; p = 0.001), ND (643 pg/mL; p = 0.001), and NAD (603 pg/mL; p = 0.001). Discrimination of AD from CON and ND was significantly improved by the combined assessment of Abeta42 and tau. At 85% sensitivity, specificity of the combined test was 86% (95% CI: 81% to 91%) compared with 55% (95% CI: 47% to 62%) for Abeta42 alone and 65% (95% CI: 58% to 72%) for tau. The combined test at 85% sensitivity was 58% (95% CI: 47% to 69%) specific for NAD. The APOE e4 gene load was negatively correlated with Abeta42 levels not only in AD but also in NAD.
Conclusions: The combined measure of CSF Abeta42 and tau meets the requirements for clinical use in discriminating AD from normal aging and specific neurologic disorders.