Bioinformatics

Kira Trares - Bioinformatics M.Sc.
Network Aging Research (NAR)

Bergheimer Straße 20
D-69115 Heidelberg

Phone: +49 (0)6221 54 8143
trares(at)nar.uni-heidelberg.de

Fellow: Dr. Ben Schöttker, PhD

Improved risk prediction of all-cause dementia, Alzheimer’s disease, and vascular dementia by biomarkers of oxidative stress and inflammation

Summary of PhD thesis

Dementia is a major challenge for global public health and social care systems. With increasing life expectancy and constantly rising numbers of dementia cases, preventing or delaying the onset of dementia are of tremendous importance. Therefore, it is indispensable to identify individuals at risk of dementia early. This dissertation aimed to combine established dementia risk factors with newly identified biomarkers from the field of oxidative stress and inflammation in a dementia risk prediction model.

First, to assess the associations between oxidative stress and dementia, the gold-standard biomarker 8-iso-prostaglandin F2α was utilized and measured in urinary samples from 5,853 participants aged between 50 and 75 years of the German population-based ESTHER study. Over 14 years of follow-up, 365 all-cause dementia cases were diagnosed, including 127 vascular dementia and 109 Alzheimer’s disease cases. Participants in the top compared to the bottom 8-iso-prostaglandin F2α tertile had a 45% increased risk of all-cause dementia incidence. Furthermore, continuously modelled, logarithmized 8-iso-prostaglandin F2α levels were statistically significantly associated with Alzheimer’s disease incidence. Moreover, an interaction between high 8-iso-prostaglandin F2α levels and the APOE ε4/ε4 genotype was detected. Participants with both risk factors had an almost 9-fold increased risk of dementia.

Next, I put my own study results in context with the pre-existing literature and conducted a systematic review. In a random-effects model meta-analysis of 25 cross-sectional studies, F2-isoprostane levels were statistically significantly associated with Alzheimer’s disease (Hedge’s g [95% confidence interval]: 1.00 [0.69-1.32]). In addition, when studies were grouped by biomarker and sample specimen, F2-isoprostanes and 8-iso-prostaglandin F2α levels were statistically significantly elevated in tissue samples of the frontal lobe of Alzheimer’s disease patients. Moreover, F2-isoprostane levels in cerebrospinal fluid and 8-iso-prostaglandin F2α levels in blood samples of Alzheimer’s disease patients were significantly increased. The 4 longitudinal studies did not yield significant associations in meta-analyses raising doubts about the causality of the association found in the cross-sectional studies. In addition, none of the cross-sectional studies was adjusted for potential confounders. Thus, the quality of evidence in this field is poor and further adjusted longitudinal studies are required to reinforce results.

After examining an oxidative stress biomarker in dementia, the Olink Target 96 Inflammation panel was used in the ESTHER study to evaluate whether inflammation-related biomarkers are associated with incident all-cause dementia, Alzheimer’s disease, and vascular dementia cases ascertained until the 17-year follow-up. In a case-cohort study design, biomarker levels were measured in serum samples of 504 all-cause dementia cases (including 163 Alzheimer’s disease and 195 vascular dementia cases) and 1,278 controls. After correction for multiple testing, 58 out of 72 tested (80.6%) biomarkers were statistically significantly associated with all-cause dementia, 22 with Alzheimer’s disease, and 33 with vascular dementia incidence. Four biomarker clusters were identified. Among those, the strongest representatives, CX3CL1, EN-RAGE, LAP TGF-beta-1 and VEGF-A, were significantly associated with dementia endpoints independently from other inflammation-related proteins. In addition, all named associations were stronger among APOE ε4 negative subjects.

Finally, to create the intended dementia risk prediction model, the established and well-validated Cardiovascular Risk Factors, Aging and Dementia (CAIDE) model was applied to the ESTHER case-cohort sample described above. I aimed to improve it by adding the investigated biomarkers using LASSO regression. Different models for all-cause dementia, Alzheimer’s disease, and vascular dementia were created. The oxidative stress biomarker 8-iso-prostaglandin F2α did not improve dementia risk prediction. However, adding 16 biomarkers from the Olink Target 96 Inflammation to the CAIDE model version, including APOE ε4 significantly improved the predictive ability for all-cause dementia (area under the curve (AUC) increase by 0.032 increments) and resulted in an AUC of 0.776. The AUC increase of prediction models for Alzheimer’s disease and vascular dementia was of similar magnitude but not statistically significant due to lower case numbers. The CAIDE model generally performed better in mid-life (50-64 years) than in late-life (65-75 years) subsamples. All AUC increases by inflammation-related biomarkers were larger in the mid-life subsample.

Overall, this work showed that oxidative stress is involved in dementia pathogenesis, but the representative biomarker 8-iso-prostaglandin F2α does not improve dementia risk prediction models. In contrast, the majority of the studied inflammation-related biomarkers were associated with all-cause dementia and additionally significantly improved the predictive ability of the established CAIDE model. Especially the predictive abilities of these blood-based biomarkers in mid-life are of clinical relevance to guide early preventive measures against dementia.

Publications

• Trares, K., Chen, L.-J., & Schöttker, B. (2022). Association of F2-isoprostane levels with Alzheimer’s disease in observational studies: A systematic review and meta-analysis. Ageing Research Reviews, 74, 101552. https://doi.org/https://doi.org/10.1016/j.arr.2021.101552.
• Trares K, Gào X, Perna L, Rujescu D, Stocker H, Möllers T, Beyreuther K, Brenner H, Schöttker B. Associations of urinary 8-iso-prostaglandin F(2α) levels with all-cause dementia, Alzheimer's disease, and vascular dementia incidence: results from a prospective cohort study. Alzheimers Dement. 2020;16(5):804-13. https://doi.org/10.1002/alz.12081.
• Trares, K., Bhardwaj, M., Perna, L. et al. Association of the inflammation-related proteome with dementia development at older age: results from a large, prospective, population-based cohort study. Alz Res Therapy 14, 128 (2022). https://doi.org/10.1186/s13195-022-01063-y.
• Trares, K., Wiesenfarth, M., Stocker, H., Perna, L., Beyreuther, K., Brenner, H. and Schöttker, B. (2022). Improved risk prediction of all-cause dementia, Alzheimer’s disease, and vascular dementia by inflammation-related proteins. [Submitted].

Vita