(C-205) Integrate Real-World Data and Genetics: A Causal Machine Learning Approach to Assess Heterogeneous Treatment Effect of Statins on Alzheimer's Disease

Background: The impact of statins on Alzheimer’s disease (AD) and related dementia (ADRD) has been debated for over 2 decades. Population differences may modify statins effects, contributing to inconsistent findings. Approaches to identify patients most likely to benefit are needed.

Objectives: Apply iterative causal forest (iCF) causal machine learning subgrouping algorithm in emulated trials in UK Biobank (UKB)’s real-world and genetic data, to assess heterogeneous treatment effects (HTEs) of statins on AD.

Methods: We conducted an emulated target trial using 36 monthly sequential trials (2009–2011) with linked UKB primary care data comparing statins initiators and non-initiators aged 55+. Incident AD was identified using UKB’s algorithmically-defined outcomes and Read Version 2 codes (positive predictive value 74.1%). Patients were followed from baseline until dementia, death, loss to follow-up, or end of primary care (5/31/2016–8/31/2017). Confounders and potential effect modifiers include genetics, socioeconomic status, lifestyle, comorbidities, and medications. Genetic modifiers included Apolipoprotein E (APOE) genotype, the polygenic risk score (PRS) for AD excluding the APOE region (nonAPOE-PRS), and the PRS for statins response predicting low-density lipoprotein cholesterol reduction after statins initiation (statin-PRS). We performed propensity score (PS)-matching on the 36 (stacked) trials, censored upon treatment deviation in per-protocol analysis, accounting for selection using inverse probability of censoring weights. Next, we applied iCF to PS-matched trials to identify subgroups with HTEs and estimated 5-year adjusted risk difference (aRD%) by inverse probability treatment and censoring weights in each identified subgroup.

Results: The overall aRD% was 0.01 (-0.15, 0.24) with risks of 0.46% in 9175 statins initiators vs. 0.45% in 9175 matched non-statins initiators. The iCF showed HTEs (p = 0.0008): patients with a high (> second tertile) nonAPOE-PRS benefited from statins (aRD% -0.44 [-0.79, -0.01], risk 0.46% vs. 0.90%, 3069 statins initiators vs. 3057 non initiators), and this was even more pronounced in men (aRD% -0.72 [-1.14, -0.11], risk 0.27% vs. 0.99%, 1820 statins initiators vs. 1815 non initiators), while patients with low (≤ second tertile) nonAPOE-PRS showed no benefit (aRD% 0.23 [0.02, 0.46]; risk 0.48 vs. 0.26).

Conclusions: Statins showed no beneficial effect on AD in the overall population. However, iCF analysis integrating real-word and genetic data identified a subpopulation with high nonAPOE-PRS (predictive of AD) may benefit from statins, especially men. This suggests that genetic factors beyond APOE genotype could modify statins' effects on ADRD outcomes. Future studies are needed to test our findings.