使用孟德爾隨機化試驗針對歐洲英國族群與漢族，探索去脂肪體重與全身脂肪重量對乳癌的因果關係

Abstract

背景： 乳癌為女性發生率首位的癌症，其中之一的主要風險因子為肥胖，肥胖的測量指標眾多，然而在孟德爾隨機化試驗（Mendelian randomization, MR）的文獻回顧中，不同的肥胖指標對乳癌風險皆為保護因子或是不顯著，與臨床上肥胖作為乳癌的風險因子情況並不吻合。 為了探討肥胖指標對乳癌為保護因果關係的原因，我們將體重指標進行拆解，分別針對脂肪重量與去脂肪重量進行因果推論，透過拆解指標的方式來了解究竟是脂肪重量還是肌肉重量對乳癌風險有保護作用。我們希望確認肥胖與乳癌發生之間的因果關係，了解兩者之相關性不僅有助於提早找出罹患乳癌的高風險個案。在臨床意義上，更可以透過介入手段增加個體的去脂肪體重，進而作為預防措施。在本研究中，我們將使用孟德爾隨機化試驗，釐清兩者之間的因果關係。 方法： 本研究使用英國生物資料庫的女性乳癌患者資料，進行單樣本孟德爾隨機化試驗，分別篩選出164與98個變異位點作為去脂肪體重（Free fat mass, FFM）與全身肪體重（Body fat mass, BFM）的工具變數（Instrumental variables, IV），使用二階段迴歸、逆方差加權之估計方法衡量兩者間的因果關係，並額外利用加權中位數方法、MR-Egger迴歸進行穩健的因果推斷。在敏感性研究中，藉由將乳癌設定為曝露變項，去脂肪體重作為結果變項，探索兩個變數間是否存在雙向的因果關係。 在外部驗證中，我們使用從英國生物資料庫中所篩選的164個變異位點，針對去脂肪體重進行雙樣本孟德爾隨機化試驗：首先使用乳腺癌協會聯盟（Breast Cancer Association Consortium, BCAC）的匯總性全基因組關聯分析（Genome-wide association study, GWAS）資料進行驗證；接下來，使用臺灣生物資料庫的資料進行雙樣本孟德爾隨機化試驗，試圖在漢族人群中重現同樣的去脂肪體重與乳癌的因果關係。 結果： 其結果顯示（β ̂_2SLS= -0.012, OR_2SLS=0.988, P_2SLS =0.283；β ̂_IVW= -0.024, OR_IVW=0.976, P_IVW =0.012），去脂肪體重與乳癌存在負向因果關係，然而在不同統計方法上的統計效力並不穩定。在敏感性研究中（β ̂_2SLS= 0.352, OR_2SLS=1.422, P_2SLS =0.296；β ̂_IVW= 0.062, OR_IVW=1.064, P_IVW =0.175）驗證了僅有去脂肪體重增加會降低乳癌風險的因果關係，反向因果關係不存在，罹患乳癌不會影響去脂肪體重的增加或減少。 根據乳腺癌協會聯盟的外部驗證結果（β ̂_IVW= 0.002, OR_IVW=1.002, P_IVW =0.733），與臺灣生物資料庫的外部驗證結果（β ̂_IVW= -0.056, OR_IVW=0.946, P_IVW =0.448），去脂肪體重的雙樣本孟德爾隨機化試驗，並不顯著，我們認為可能由於不同資料庫間的人種差異與篩選差異所導致。 結論： 本研究使用英國生物資料庫進行的單樣本孟德爾隨機化試驗，確認了去脂肪體重為乳癌風險的保護因子，兩者存在負向的因果關係，並且不存在反向因果關係。而全身脂肪體重與乳癌風險間，則不存在任何統計顯著的因果關係。 我們針對肥胖對於乳癌，在孟德爾隨機化試驗中不尋常的保護作用，提出了可能的解釋，利用將體重拆分成去脂肪體重與全身脂肪體重，分別進行單樣本孟德爾隨機化試驗與雙樣本隨機化試驗，我們可以更清楚的了解到，究竟是脂肪重量的增加，亦或是肌肉重量的增加，導致了乳癌風險的下降。並得出了是肌肉重量（去脂肪體重）的增加降低了乳癌風險的可能。 雖然該結果於台灣族群中不具備外推性，但可能是由於乳癌在不同種族間的機轉差異導致，在未來研究中，可能需要使用漢族的去脂肪體重資料篩選出工具變數，進一步考慮乳癌族群差異造成的影響。 在本研究中，去脂肪體重作為能夠將低乳癌風險發生的保護因子，與停經後的體內激素水平息息相關，台灣女性的平均更年期為48-52歲，與台灣乳癌發生率高峰45-69歲有著很高的重合時段。站在疾病預防的角度上，應當注意女性在更年期後運動習慣是否發生改變，以及其肌肉流失的狀態，因為兩者皆有可能造成乳癌風險的增加。建議鼓勵停經期後女性進行適當的運動作為介入手段，維持體內激素水平，以降低罹患乳癌的風險。

Background: Breast cancer is the most common cancer in women, and one of the main risk factors is obesity. There are many measurements of obesity. However, in the literature review of Mendelian randomization (MR), obesity plays a different role in breast cancer. Mendelian randomization‘s results are either protective or insignificant, which are not consistent with the clinical result that obesity is a risk factor for breast cancer. To investigate the causal relationship between obesity measurements and breast cancer protection, we divide body weight into body fat weight (BFM) and free fat weight (FFM), and made causal inferences for BFM and FFM respectively. Understanding the correlation between obesity indicators and breast cancer will not only help to identify high-risk cases of breast cancer early, but also can increase the individual's fat-free body weight through interventional methods, which can be used as a preventive measure. In this study, we will use Mendelian randomization to clarify the causal relationship between obesity indicators and breast cancer. Method: This study uses the data of female breast cancer patients from the UK Biobank to conduct one-sample Mendelian randomization, extracting 164 and 98 variants as FFM’s and BFM’s instrumental variables (IV). Using two-stage least squares and inverse variance weighted estimation measures the causal relationship between obesity indicators and breast cancer. We also use the weighted median method, MR-Egger regression for the robust causal inference. In a sensitivity study, by setting breast cancer as the exposure variable and FFM as the outcome variable, we investigate whether there is a bidirectional causal relationship between the two variables. In external validation, we perform two-sample Mendelian randomization by using FFM’s 164 variants extracted from the UK Biobank. We first collect Genome-wide association study (GWAS) data from the Breast Cancer Association Consortium (BCAC). Next, two-sample Mendelian randomization is performed with data from the Taiwan Biobank to reproduce the same causal relationship between FFM and breast cancer in Han Chinese. Result: The results show （β ̂_2SLS= -0.012, OR_2SLS=0.988, P_2SLS =0.283；β ̂_IVW= -0.024, OR_IVW=0.976, P_IVW =0.012）, a negative causal relationship between FFM and breast cancer, but the statistical power is not stable in different statistical methods. The sensitivity study（β ̂_2SLS= 0.352, OR_2SLS=1.422, P_2SLS =0.296；β ̂_IVW= 0.062, OR_IVW=1.064, P_IVW =0.175）tests a causal relationship that only increased FFM is associated with a reduced risk of breast cancer, the reverse causality did not exist and having breast cancer did not affect the gain or loss of FFM. Conclusion: In this study, one-sample Mendelian randomization conducted by the UK Biobank confirms that FFM is a protective factor for breast cancer risk with the result that there exists a negative causality and no reverse causality. There is no statistically significant causal relationship between BFM and breast cancer risk. We propose possible explanations for the unusual protective effect of obesity on breast cancer in previous Mendelian randomization papers by splitting off body weight into FFM and BFM and using one-sample Mendelian randomization and two-sample randomization. Therefore, we can understand whether an increase in fat mass or an increase in muscle mass leads to a decrease in breast cancer risk more clearly. And we come to the conclusion that the increase in muscle mass (FFM) may reduce the risk of breast cancer. This result may not be significant in the Taiwanese ethnic group because of the difference in the mechanism of breast cancer among different ethnic groups. In future studies, it would be necessary to use the FFM data of the Han Chinese to extract instrumental variables and further consider breast cancer differences between ethnic. In this study, FFM, as a protective factor that can reduce the risk of breast cancer, is closely related to hormone levels after menopause. The average menopause age of Taiwanese women is 48-52 years old, which is almost matching to the peak incidence of breast cancer in Taiwan at 45-69 years old. From the perspective of disease prevention, it should be noted that postmenopausal women may increase the risk of breast cancer if their exercise habits change after menopause or their muscle loss. It is recommended for postmenopausal women to develop exercise habits as an intervention to maintain hormone levels in the body to reduce the risk of breast cancer.