【作者】 张凯；

【导师】 薛付忠； 常开锋；

【作者基本信息】 山东大学 ， 公共卫生（专业学位）， 2023， 硕士

【摘要】 研究背景:随着新型冠状病毒感染(coronavirus disease 2019,COVID-19,简称“新冠感染”)大流行的持续存在,国内外许多研究发现COVID-19患者在急性感染期过后会出现心血管、神经、精神和内分泌系统等相关症状,世界卫生组织将该症状定义为新冠后遗症(post-COVID-19 condition)。目前的新冠后遗症研究多为观察性流行病学设计,其关联效应易受混杂因素干扰而导致有偏估计。此外,由于COVID-19属于新发传染病,各研究观察期较短,部分未知、潜在的后遗症症状还未显现出来。上述因素导致了新冠后遗症相关研究的结果异质性较大、证据等级较低、结局信息有限。全表型组关联研究(phenome-wide association analyses,PheWAS)是一种在全表型组范围内探究所有表型与某一暴露之间关联关系的分析方法。这种无医学假设的研究方法有助于在人类全表型范围内发现新关联,从而描绘出某暴露对整个人类表型组的关联效应图谱,为全面深入理解疾病机制提供线索。此外,为进一步探索因果效应,孟德尔随机化(Mendelianrandomization,MR)常用来验证PheWAS检验所得有统计学意义的暴露→结局关联效应,该研究策略在非传染性疾病中已多次应用。由于宿主遗传易感性的不同,人类在感染病原体后会出现不同疾病表型。当前已有诸多COVID-19相关的全基因组关联研究(genome-wide association study,GWAS)公开报道,为我们的研究提供了数据支持。因此,本研究借鉴既往探究非传染性疾病的方法(PheWAS结合MR的分析策略)系统全面的探索新冠后遗症症状,为阐明感染新冠病毒后的长远期症状提供相关科学依据。研究方法:1.PheWAS研究:GWAS汇总数据来自第7版“COVID-19宿主遗传学计划”,电子健康数据来自英国生物银行(UKBiobank,UKB)。本研究构建COVID-19的多基因风险评分(polygenic risk score,PRS)作为暴露,构建人类全表型数据库作为结局。基于个体数据的PheWAS分析本质是回归分析,因此采用logistic回归进行分析。此外,本研究使用错误发现率(false discovery rate,FDR)法对P值进行多重校正以控制Ⅰ类错误概率,当P<0.05视为具有潜在的统计学意义的结果,通过FDR校正视为稳健的结果。2.单变量MR:以逆方差加权(inverse-variance weighted,IVW)法为主分析方法,简单中位数法、加权中位数法、MR-Egger回归法和孟德尔随机化多效性残差和离群值法作为敏感性方法。多种单变量MR方法来进一步验证PheWAS检验所得有统计学意义的暴露→结局关联效应,为COVID-19与相关人类疾病表型提供稳健的因果证据。3.多变量孟德尔随机化(multivariate Mendelian randomization,MVMR):采用MVMR方法检验其关联效应是否受吸烟、饮酒、身体质量指数(body mass index,BMI)等潜在协变量的干扰,得出更加稳健的因果证据。研究结果:1.PheWAS分析结果:普通型COVID-19导致发病风险升高的疾病种类最多,共有13个系统51种疾病,其中以心血管、肿瘤、内分泌/代谢系统尤为显著。住院型COVID-19和危重型COVID-19分别会导致36种疾病(13个系统)和35种疾病(15个系统)的发病风险升高(P<0.05)。2.单变量MR检验结果:IVW检验结果表明危重型、住院型、普通型COVID-19分别对21种、18种和36种疾病有不利的影响(P<0.05),主要集中在心脑血管、癌症、神经精神和皮肤等系统。经过其余单变量MR法等敏感性分析方法检验后,较为稳健的结果有:危重型COVID-19与高脂血症、乳腺恶性肿瘤(HER2阴性)、卵巢癌和食道癌;住院型COVID-19与房室传导阻滞、高脂血症、对疾病的控制力、行为和情感的混合障碍;普通型COVID-19与冠状动脉心脏疾病、肠道憩室疾病、甲状腺功能减退症、佩吉特氏病、情绪不稳定的人格障碍。3.MVMR检验结果:危重型COVID-19对于非毒性弥漫性甲状腺肿、甲状腺毒症伴弥漫性甲状腺肿和乳腺恶性肿瘤(HER2阴性)的效应值较单变量MR均增大。住院型COVID-19对于乳腺恶性肿瘤(HER阳性)和乳腺恶性肿瘤(HER2阴性)的关系效应较单变量MR分析增加了,而与胃食管反流病的效应在校正协变量后降低了。普通型COVID-19对于缺血性卒中(大动脉粥样硬化型)、恶性程度低的黏液性卵巢癌、乳腺恶性肿瘤(HER阳性)、父亲有阿尔兹海默症的病史、儿童失神癫痫均较单变量MR分析的危害效应增加,而与十二指肠溃疡的效应较单变量MR方法降低。研究结论:本研究基于大规模GWAS汇总数据和电子健康数据,识别出COVID-19对人体各个系统均会造成危害,尤其是普通型COVID-19与心血管、肿瘤和内分泌/代谢系统的关联效应较为显著。此外,危重型、住院型和普通型COVID-19均对乳腺恶性肿瘤具有潜在危害效应。本研究为进一步全面认识新冠后遗症的危害性及开展防治措施提供了重要科学依据。更多还原

【Abstract】 Background:As the pandemic of coronavirus disease 2019(COVID-19)persists,many studies in China and abroad have found that patients with COVID-19 develop cardiovascular,neurological/psychiatric,and endocrine symptoms after the acute infection period.The World Health Organization has defined this condition as the post-COVID-19 condition.Most of the current post-COVID-19 condition studies are observational designs,and the association effects are susceptible to confounding factors that may lead to biased estimates.In addition,because COVID-19 is a kind of emerging infectious disease,the observation period is short,and some unknown and potential sequelae symptoms have not yet been revealed.The above-mentioned factors contributed to the high heterogeneity of results,low level of evidence,and limited outcome information in studies related to post-COVID-19 condition.Phenome-wide association analyses(PheWAS)is a method that explores the association between all phenotypes and a given exposure within the context of a phenome-wide group.This hypothesis-free approach helps to identify new associations across the human phenome,thus mapping the association effects of exposure on the entire human phenome and providing clues for a comprehensive understanding of disease mechanisms.In addition,to further explore causal effects,Mendelian randomization(MR)is commonly used to validate statistically significant exposure→outcome association effects from the PheWAS test(this research strategy has been applied in non-communicable diseases(NCDs).Due to differences in host genetic susceptibility,humans can develop different disease phenotypes upon infection with pathogens.Several genome-wide association studies(GWAS)related to COVID-19 have been publicly reported,providing summary-level data to support our study.Therefore,in this study,we used the previous approach to explore NCDs(PheWAS combined with MR analysis strategy)to systematically and comprehensively explore the symptoms of COVID-19 and to provide a scientific basis for elucidating the long-term symptoms of COVID-19.Methods:1.PheWAS study:GWAS summary data were obtained from the 7th edition of the COVID-19 Host Genetics Program and electronic health data from the UK Biobank(UKB).In this study,a polygenic risk score(PRS)for COVID-19 was constructed as the exposure,and a human full phenotype database was constructed as the outcome.The PheWAS analysis based on individual data is essentially a regression analysis,so logistic regression was used for the analysis.In addition,this study used the False discovery rate(FDR)method for multiple corrections of P-values to control for Type Ⅰ error probability,and when P<0.05 was considered as a potentially statistically significant result,it was considered as a robust result by FDR correction.2.Univariate MR:Inverse-variance weighted(IVW)method was used as the main analysis method,and the simple median method,weighted median method,MR-Egger regression method,and Mendelian randomized multi-effect residuals and outliers method were used as sensitivity methods.Multiple univariate MR methods were used to further validate the statistically significant exposure→outcome association effects obtained from the PheWAS test and to provide robust causal evidence for the association between COVID-19 and related human disease phenotypes.3.Multivariate Mendelian randomization(MVMR):The MVMR method was used to test whether the association effects were confounded by potential covariates such as smoking,alcohol consumption,and body mass index(BMI),yielding more robust causal evidence.Results:1.Results of PheWAS analysis:general type COVID-19 resulted in the greatest number of diseases with an elevated risk of morbidity,with 51 diseases in 13 systems,of which cardiovascular,oncological,and endocrine/metabolic systems were particularly significant.Hospitalized COVID-19 and critically ill COVID-19 caused an increased risk of 36 diseases(13 systems)and 35 diseases(15 systems),respectively(P<0.05).2.Univariate MR test results:IVW test results showed that critical,hospitalized,and general COVID-19 adversely affected 21,18,and 36 diseases,respectively(P<0.05),mainly in the cardiovascular,cancer,neuropsychiatric,and cutaneous systems.After the remaining univariate MR method and other sensitivity analysis methods were tested,the more robust results were:critical COVID-19 and hyperlipidemia,breast malignancy(HER2 negative),ovarian cancer and esophageal cancer;hospitalized COVID-19 and atrioventricular block,hyperlipidemia,mixed disorders of control over disease,behavior,and emotion;general COVID-19 and coronary artery cardiac disease,intestinal diverticular disease,hypothyroidism,Paget’s disease,and emotionally unstable personality disorders.3.MVMR test results:the effect values of critical COVID-19 were increased for non-toxic diffuse goiter,thyrotoxicosis with diffuse goiter,and breast malignancy(HER2 negative)compared to univariate MR.The effect of hospitalized COVID-19 for breast malignancy(HERpositive)and breast malignancy(HER2-negative)increased in univariate MR analysis,while the effect with gastroesophageal reflux disease(GERD)decreased after correction for covariates.The hazard effect of COVID-19 for ischemic stroke(large atherosclerotic),mucinous ovarian cancer with low malignancy,breast malignancy(HER positive),father’s history of Alzheimer’s disease,and childhood anhedonic epilepsy was increased compared to univariate MR analysis,whereas the effect with duodenal ulcer was decreased compared to univariate MR method.Conclusions:Based on large-scale GWAS summary data and electronic health data,this study identified that COVID-19 can cause harmful effects on all human systems,especially the association effects of general COVID-19 with cardiovascular,oncological,and endocrine/metabolic systems were more significant.In addition,critical,hospitalized,and general COVID-19 all have potentially harmful effects on breast malignancies.This study provides an important scientific basis for a further comprehensive understanding of the harmful effects of post-CO VID-19 condition and the development of preventive and curative measures.更多还原