Woldemariam Sarah R, Tang Alice S, Oskotsky Tomiko T, Yaffe Kristine, Sirota Marina
Bakar Computational Health Sciences Institute, University of California San Francisco, San Francisco, California, USA.
School of Medicine, University of California San Francisco, San Francisco, California, USA.
Commun Med (Lond). 2023 Apr 8;3(1):50. doi: 10.1038/s43856-023-00280-2.
Alzheimer's dementia (AD) is a neurodegenerative disease that is disproportionately prevalent in racially marginalized individuals. However, due to research underrepresentation, the spectrum of AD-associated comorbidities that increase AD risk or suggest AD treatment disparities in these individuals is not completely understood. We leveraged electronic medical records (EMR) to explore AD-associated comorbidities and disease networks in racialized individuals identified as Asian, Non-Latine Black, Latine, or Non-Latine White.
We performed low-dimensional embedding, differential analysis, and disease network-based analyses of 5664 patients with AD and 11,328 demographically matched controls across two EMR systems and five medical centers, with equal representation of Asian-, Non-Latine Black-, Latine-, and Non-Latine White-identified individuals. For low-dimensional embedding and disease network comparisons, Mann-Whitney U tests or Kruskal-Wallis tests followed by Dunn's tests were used to compare categories. Fisher's exact or chi-squared tests were used for differential analysis. Spearman's rank correlation coefficients were used to compare results between the two EMR systems.
Here we show that primarily established AD-associated comorbidities, such as essential hypertension and major depressive disorder, are generally similar across racialized populations. However, a few comorbidities, including respiratory diseases, may be significantly associated with AD in Black- and Latine- identified individuals.
Our study revealed similarities and differences in AD-associated comorbidities and disease networks between racialized populations. Our approach could be a starting point for hypothesis-driven studies that can further explore the relationship between these comorbidities and AD in racialized populations, potentially identifying interventions that can reduce AD health disparities.
阿尔茨海默病性痴呆（AD）是一种神经退行性疾病，在种族边缘化个体中尤为普遍。然而，由于研究代表性不足，对于增加这些个体患AD风险或提示AD治疗差异的AD相关合并症谱尚未完全了解。我们利用电子病历（EMR）来探索在被确定为亚洲人、非拉丁裔黑人、拉丁裔或非拉丁裔白人的种族化个体中的AD相关合并症和疾病网络。
我们对来自两个EMR系统和五个医疗中心的5664例AD患者和11328例人口统计学匹配的对照进行了低维嵌入、差异分析和基于疾病网络的分析，其中亚洲人、非拉丁裔黑人、拉丁裔和非拉丁裔白人身份的个体数量相等。对于低维嵌入和疾病网络比较，使用曼-惠特尼U检验或克鲁斯卡尔-沃利斯检验，随后进行邓恩检验来比较类别。使用费舍尔精确检验或卡方检验进行差异分析。使用斯皮尔曼等级相关系数来比较两个EMR系统之间的结果。
我们在此表明，主要的已确定的AD相关合并症，如原发性高血压和重度抑郁症，在种族化人群中总体相似。然而，包括呼吸系统疾病在内的一些合并症可能与被确定为黑人及拉丁裔的个体的AD显著相关。
我们的研究揭示了种族化人群之间AD相关合并症和疾病网络的异同。我们的方法可以作为假设驱动研究的起点，这些研究可以进一步探索这些合并症与种族化人群中AD之间的关系，有可能确定可以减少AD健康差异的干预措施。
