The impact of increasing levels of blood C-reactive protein on the inflammatory loci SPI1 and CD33 in Alzheimer's disease

Abstract

Apolipoprotein ε4 (APOE ε4) is the most significant genetic risk factor for late-onset Alzheimer's disease (AD). Elevated blood C-reactive protein (CRP) further increases the risk of AD for people carrying the APOE ε4 allele. We hypothesized that CRP, as a key inflammatory element, could modulate the impact of other genetic variants on AD risk. We selected ten single nucleotide polymorphisms (SNPs) in reported AD risk loci encoding proteins related to inflammation. We then tested the interaction effects between these SNPs and blood CRP levels on AD incidence using the Cox proportional hazards model in UK Biobank (n = 279,176 white participants with 803 incident AD cases). The five top SNPs were tested for their interaction with different CRP cutoffs for AD incidence in the Framingham Heart Study (FHS) Generation 2 cohort (n = 3009, incident AD = 156). We found that for higher concentrations of serum CRP, the AD risk increased for SNP genotypes in 3 AD-associated genes (SPI1, CD33, and CLU). Using the Cox model in stratified genotype analysis, the hazard ratios (HRs) for the association between a higher CRP level (≥10 vs. <10 mg/L) and the risk of incident AD were 1.94 (95% CI: 1.33-2.84, p < 0.001) for the SPI1 rs1057233-AA genotype, 1.75 (95% CI: 1.20-2.55, p = 0.004) for the CD33 rs3865444-CC genotype, and 1.76 (95% CI: 1.25-2.48, p = 0.001) for the CLU rs9331896-C genotype. In contrast, these associations were not observed in the other genotypes of these genes. Finally, two SNPs were validated in 321 Alzheimer's Disease Neuroimaging (ADNI) Mild Cognitive Impairment (MCI) patients. We observed that the SPI1 and CD33 genotype effects were enhanced by elevated CRP levels for the risk of MCI to AD conversion. Furthermore, the SPI1 genotype was associated with CSF AD biomarkers, including t-Tau and p-Tau, in the ADNI cohort when the blood CRP level was increased (p < 0.01). Our findings suggest that elevated blood CRP, as a peripheral inflammatory biomarker, is an important moderator of the genetic effects of SPI1 and CD33 in addition to APOE ε4 on AD risk. Monitoring peripheral CRP levels may be helpful for precise intervention and prevention of AD for these genotype carriers.

Fig. 1. Study design and filters for UKBB, FHS and ADNI cohorts.

The following three human datasets were analyzed in this study. UKBB: 279,176 participants (mean age 60.1 ± 5.5 years) of which baseline CRP was measured, and 803 incident AD cases were identified after 8.3 ± 0.9 years of follow-up. FHS: 3009 participants (mean age 60.8 ± 9.4 years) of which baseline CRP was measured (i.e., Exam 7), and 156 incident AD cases were identified after 14.9 ± 4.0 years of follow-up. ADNI: 321 MCI patients (mean age 75.0 ± 7.0 years) of which baseline CRP was measured, and 105 MCI-to-AD converters were identified after 31.8 ± 11.2 months of follow-up.

Fig. 2. Kaplan‒Meier analysis in UKBB for AD-free probability under different CRP levels (mg/L) among genotypes in 3 SNPs.

A SPI1 rs1057233-AA vs. rs1057233-GG + GA. B CD33 rs3865444-CC vs. rs3865444-AA + AC. C CLU rs9331896-TT vs. rs9331896-CC + CT genotypes. Red: CRP < 11 mg/L, Green: CRP ≥ 11 mg/L. Raw P values are presented.

Fig. 3. Forest plots of the results from UKBB, FHS and meta-analysis for the stratified genotype analysis of 3 SNPs for the effect of CRP levels on AD incidence.

The Cox proportional hazard regression models were applied to estimate the effect of different levels of serum CRP on the incidence of AD among different genotypes of SPI1, CD33, and CLU after adjusting for age, sex, years of education, APOE ε4 and PCs. The results from UKBB are shown (a–c), and those from FHS are shown (d–f). The results from the meta-analyses of UKBB and FHS are shown (g–i). Raw P values are presented. For all CRP cutoffs from 3–12 mg/L, refer to Fig. S2.

Fig. 4. Kaplan‒Meier survival plots and forest plots using the ADNI cohort for the stratified genotype analysis for the effect of CRP levels on MCI-AD conversion in the Cox proportional hazard regression models and boxplots for CRP-SNP interaction effects on CSF biomarkers (t-Tau and p-Tau): SPI1 rs1057233 and CD33 rs3865444.

ADNI MCI participants were stratified by genotypes. Kaplan‒Meier survival plots were generated for AD free time for SPI1 rs1057233 and CD33 rs3865444 genotypes; Red: CRP < 8 mg/L, Green: CRP ≥ 8 mg/L (a, b). Forest plots with the estimated effect of different levels of serum CRP on the MCI-to-AD conversion among different genotypes after adjusting for age, sex, education and APOE ε4 (c, d). ADNI participants with measured CSF AD biomarkers were stratified by genotype. t-Tau and p-Tau measured at the last exam were log transformed and are shown in boxplots. p values of the interaction between CRP and SPI1 rs1057233/CD33 rs3865444 genotypes were calculated using linear regression analysis after adjusting for age, sex, education and APOE ε4. Red: CRP < 10 mg/L, Green: CRP ≥ 10 mg/L (e, f). Raw P values are presented.