台灣族群中情緒障礙之高遺傳易感但未發病者：遺傳與非遺傳相關因子之解析

Abstract

情緒障礙，包含雙相情緒障礙症 (BD) 與重鬱症 (MDD)，常伴隨慢性病程與自殺風險，構成沉重的全球疾病負擔，而這些疾患往往同時受到遺傳與非遺傳因素影響。過去研究多著重於疾病風險與病理機制，而對於具有高遺傳風險但未發病之韌性族群，其遺傳架構與相關因素在東亞族群較缺乏探討。本研究利用台灣人體生物資料庫 (TWB) 之基線資料，探討台灣族群中對BD與MDD之韌性相關遺傳及非遺傳因素。 本研究納入TWB共134,680名年紀為20至90歲之參與者 (包含Batch 1之21,692名與Batch 2之112,988名參與者)。首先，透過PRSmix+整合62種與情緒障礙及其遺傳相關性狀之多基因風險分數 (PRS)，以提升BD與MDD的遺傳風險分層能力。接著，本研究以整合後的PRS定義前10%為高遺傳風險族群，將「高風險但未罹病者」定義為韌性個體，「高風險且罹病者」作為非韌性個體。我們於Batch 2高風險族群基因資料中，排除已知疾病風險相關變異後，針對兩種疾患各自進行韌性之全基因體關聯分析 (GWAS)，以探索獨立於疾病風險之韌性遺傳訊號，並根據此結果，於Batch 1建立多基因韌性分數 (polygenic resilience score, PRSr)。此外，本研究亦進一步分析TWB自我呈報問卷中之社會經濟、生活型態、疾病共病等因素與韌性狀態之相關性。 研究結果顯示，相較於單一PRS，PRSmix+在兩種疾病中皆展現較佳的疾病風險分層能力，辨析出韌性個體。然而，在BD與MDD之韌性GWAS中，皆未發現達全基因體顯著水準之變異位點，僅部分達提示性顯著之變異位點可能與精神或代謝等性狀有關。此外，BD與MDD PRSr在Batch 1高風險族群中，皆未能顯著區分韌性與非韌性個體，顯示遺傳訊號的偵測可能仍受限於現有樣本數與方法。在非遺傳因素方面，已婚狀態、較高教育程度、較高個人收入與規律運動與較高韌性呈正相關；相反地，二手菸暴露、嚼檳榔經驗及多項身體疾病 (如神經系統疾病) 則與較低韌性相關。 本研究成功建立高多基因風險架構，探討台灣族群情緒障礙之韌性。雖然韌性GWAS發現有限且PRSr預測表現不佳，但結果顯示社經條件、生活型態與疾病共病等可調整之非遺傳因素與情緒障礙韌性顯著相關。未來若能結合更大規模之東亞或多族群GWAS以及縱貫性資料，將有助於更精確地定義韌性並釐清其與不同因子之潛在因果路徑。

Mood disorders, including bipolar disorder (BD) and major depressive disorder (MDD), are chronic psychiatric conditions associated with substantial disease burden and suicide risk. These conditions are shaped by a complex interplay of genetic and non-genetic factors. While prior research has predominantly focused on disease risk and pathological mechanisms, the genetic architecture and correlates of “resilience”—defined as individuals who remain unaffected despite carrying a high genetic liability—remain underexplored in East Asian populations. Utilizing baseline data from the Taiwan Biobank (TWB), this study aims to investigate the genetic and non-genetic factors associated with resilience to BD and MDD in the Taiwanese population. This study included a total of 134,680 participants aged 20 to 90 (21,692 participants from TWB Batch 1 and 112,988 from Batch 2). We applied PRSmix+ to integrate 62 polygenic risk scores (PRSs) for mood disorders and genetically correlated traits to improve risk stratification for BD and MDD. Individuals in the top 10% of the integrated PRS distribution were defined as the high-genetic-risk group. Within this group, unaffected individuals were classified as resilient, whereas affected individuals were classified as non-resilient. To explore resilience-specific signals, we removed known disease-risk variants and performed a genome-wide association study (GWAS) of resilience within the Batch 2 high-risk group. Based on these GWAS summary statistics, polygenic resilience scores (PRSr) were constructed in Batch 1. Beyond genetic influences, we also analyzed the associations between resilience and socioeconomic, lifestyle, and medical comorbidity factors, using TWB questionnaires data. The results demonstrated that PRSmix+ consistently outperformed single disease PRSs in risk stratification and identifying resilient individuals for both disorders. However, no variant reached genome-wide significance in either resilience GWAS, although several suggestive loci were mapped to genes related to psychiatric or metabolic traits. Furthermore, the BD and MDD PRSr did not significantly distinguish resilient from diseased individuals the Batch 1 high-risk subgroup, suggesting limited power to detect genetic resilience signals. Regarding non-genetic factors, being married, higher educational, higher personal income, and engaging in regular exercise were positively associated with resilience, whereas secondhand smoke exposure, betel nut use, and medical comorbidities, particularly neurological diseases, were correlated with a lower odds of resilience. In conclusion, this study applied a high-polygenic-risk framework to explore resilience to mood disorders within the Taiwanese population. Although the genetic findings were limited, our results highlight the importance of considering both genetic liability and modifiable non-genetic factors in understanding resilience to mood disorders. Future studies would benefit from larger-scale East Asian or multi-ancestry GWAS cohorts to define resilience with enhanced discriminatory power, as well as longitudinal data to validate the causal pathways linking non-genetic correlates to resilience.