Using CSF biomarkers to replicate genetic associations in Alzheimer's disease.

Defining cases and controls on the basis of biomarkers rather than clinical diagnosis may reduce sample sizes required for genetic studies. The aim of this study was to assess whether characterizing case/control status on the basis of cerebrospinal fluid (CSF) profile would increase power to replicate known genetic associations for Alzheimer's disease (AD). Independent of clinical diagnosis, Alzheimer's Disease Neuroimaging Initiative (ADNI) subjects with 2 CSF biomarkers for AD (Aβ1-42 < 192 pg/mL and tau phosphorylated at threonine 181 (p-tau) > 23 pg/mL, "CSF-positive") were compared with those without CSF evidence for AD (Aβ1-42 > 192 pg/mL and 181-phosphorylated tau < 23 pg/mL, "CSF-negative"). Minor allele frequency (MAF) and odds ratios (ORs) between these 2 groups were calculated for 7 single-nucleotide polymorphisms (SNPs) of interest. Two hundred thirty-two individuals were CSF-positive and 94 CSF-negative. There were no differences in age (74.7 ± 7.2 vs. 75.0 ± 6.5 years, p = 0.7), but significant differences in Mini Mental State Examination (MMSE) (25.9 ± 2.6 vs. 28.2 ± 1.7, p < 0.001) between the CSF-positive and CSF-negative groups. Significant differences in MAF (p < 0.05, uncorrected) were seen for CR1 (rs1408077; OR, 1.59; 95% confidence interval [CI], 1.01-2.49), PICALM (rs541458; OR, 0.68, 95% CI, 0.47-0.98), TOMM40 (rs2075650; OR, 4.30; 95% CI, 2.61-7.06); and possession of 1 or more APOE ε4 alleles (OR, 9.84; 95% CI, 5.48-17.67). These results suggest that using biomarkers of AD pathology to define case and control status may increase power in genetic association studies.