Aims: To investigate the relationship between glycaemia and cognitive function, brain structure and incident dementia using bidirectional Mendelian randomisation (MR). Methods: UK Biobank (n~500,000) individuals, aged 40-69 years at baseline. Our exposures were genetic instruments for type-2 diabetes (163 variants) and HbA1c (52 variants) and our outcomes were reaction time (RT - milliseconds), visual memory (number of incorrect responses), hippocampal and white matter hyperintensity volumes (both mm3), Alzheimer's disease (AD). To study potential bidirectional effects, we then investigated the associations between genetic variants for RT (43 variants) and clinical type-2 diabetes and measured HbA1c. We used conventional inverse-variance weighted (IVW) MR, alongside standard MR sensitivity analyses. Results: Using IVW, genetic liability to type-2 diabetes was not associated with reaction time (exponentiatedβ=1.00, 95%CI=1.00; 1.00), visual memory (expβ=1.00, 95%CI=0.99; 1.00), white matter hyperintensity volume (expβ=0.98, 95%CI=0.93; 1.03), hippocampal volume (coefficient mm3=0.00, 95%CI=-0.01; 0.01) or risk of AD (OR 0.97, 95%CI=0.89; 1.06). HbA1c was not associated with reaction time (expβ=1.01, 95%CI=1.00; 1.01), white matter hyperintensity volume (expβ=0.88, 95%CI=0.73; 1.07), hippocampal volume (coefficient=-0.02, 95%CI=-0.10; 0.06), risk of AD (OR 0.94, 95%CI=0.47; 1.86), but HbA1c was associated with visual memory (expβ=1.06, 95%CI=1.05; 1.07) using a weighted median approach. IVW showed no evidence that reaction time was associated with diabetes (OR 0.96, 95%CI=0.63; 1.46) or HbA1c (coefficient=-0.08, 95%CI=-0.57; 0.42). MR-Egger intercept p-values indicated no major issues with unbalanced horizontal pleiotropy (all p>0.05). Conclusions: Overall, we observed little evidence of causal associations between glycaemia and cognition, structural brain and dementia phenotypes.

中文翻译：

---

血糖，认知功能，大脑结构性结局与痴呆之间的关系：英国生物库中的孟德尔随机研究

目的：使用双向孟德尔随机（MR）研究血糖与认知功能，脑结构和突发性痴呆之间的关系。方法：英国生物库（n〜500,000）个体，基线时年龄为40-69岁。我们所接触的是2型糖尿病（163个变体）和HbA1c（52个变体）的遗传工具，我们的结果是反应时间（RT-毫秒），视觉记忆（错误反应的数量），海马和白质高信号量（均为mm3） ），阿尔茨海默氏病（AD）。为了研究潜在的双向影响，我们随后研究了RT的遗传变异（43个变异）与2型糖尿病和临床HbA1c之间的关联。我们使用常规的反方差加权（IVW）MR，以及标准MR灵敏度分析。结果：使用IVW，2型糖尿病的遗传易感性与反应时间（指数β= 1.00，95％CI = 1.00; 1.00），视觉记忆（expβ= 1.00，95％CI = 0.99; 1.00），白质过高量（expβ= 0.98，95％CI = 0.93; 1.03），海马体积（系数mm3 = 0.00，95％CI = -0.01; 0.01）或患有AD的风险（OR 0.97，95％CI = 0.89; 1.06）。HbA1c与反应时间（expβ= 1.01，95％CI = 1.00; 1.01），白质过高量（expβ= 0.88，95％CI = 0.73; 1.07），海马体积（系数= -0.02、95％CI）无关= -0.10; 0.06），有AD风险（OR 0.94，95％CI = 0.47; 1.86），但使用加权中位数方法将HbA1c与视觉记忆相关（expβ= 1.06，95％CI = 1.05; 1.07）。IVW没有证据表明反应时间与糖尿病（OR 0.96，95％CI = 0.63; 1.46）或HbA1c（系数= -0.08，95％CI = -0.57; 0.42）相关。MR-Egger截距的p值表示水平多效性不平衡（无所有p> 0.05）没有重大问题。结论：总的来说，我们几乎没有观察到血糖与认知，结构性脑和痴呆表型之间因果关系的证据。