【摘要】 目的 系统分析和比较美国国家健康与营养调查(National Health and Nutrition Examination Survey, NHANES)和上海郊区自然人群队列(Shanghai Suburban Adult Cohort and Biobank, SSACB)女性乳腺癌相关因素，并评估相关因素与地区之间的效应异质性，为了解中国女性乳腺癌风险特征提供参考。方法 基于NHANES 2013年-2020年3月数据及SSACB基线数据开展横断面比较研究。纳入NHANES女性8 608例(乳腺癌226例),SSACB女性39 683例(乳腺癌249例)。采用标准化均数差(standardized mean difference, SMD)评价两队列基线特征可比性，队列内乳腺癌组与非乳腺癌组比较采用t检验、Wilcoxon秩和检验或χ~2检验。采用Firth校正logistic回归模型分别估计两队列内乳腺癌相关因素，并合并数据构建含“地区×相关因素”交互项的加权Firth回归模型，通过似然比检验与Wald检验评估效应异质性，计算分层平均边际效应(average marginal effect, AME)以量化不同亚组中地区间绝对风险差异。结果 两队列在年龄、教育水平、生殖因素及生活方式等多数变量上存在差异(SMD>0.2)。多因素分析显示，年龄在NHANES(OR=1.11,95%CI:1.08～1.13)与SSACB(OR=1.02,95%CI:1.00～1.04)队列中均与乳腺癌风险呈正相关。NHANES队列中，本科及以上学历(OR=2.20,95%CI:1.24～4.11)、一日静坐时长增加(OR=1.04,95%CI:1.00～1.08)与乳腺癌风险升高相关，而怀孕次数增加(OR=0.92,95%CI:0.85～1.00)、有雌激素服用史(OR=0.59,95%CI:0.40～0.84)及高血压(OR=0.72,95%CI:0.53～0.98)与风险降低相关。SSACB队列中，自然绝经(OR=4.89,95%CI:2.48～10.39)、人工绝经(OR=16.38,95%CI:8.04～35.58)、有雌激素服用史(OR=3.00,95%CI:1.43～5.55)、高血糖(OR=1.61,95%CI:1.19～2.16)及轻度焦虑抑郁(OR=2.19,95%CI:1.45～3.19)均与乳腺癌风险升高相关，而初次月经年龄增加(OR=0.92,95%CI:0.86～0.99)与风险降低相关。合并分析显示，完全调整协变量后SSACB女性乳腺癌预测绝对风险较NHANES女性平均低19.0人/千人(95%CI:-24.0～-14.8人/千人)。绝经状态(P=0.001)、雌激素服用史(P<0.001)及焦虑抑郁程度(P=0.012)与地区存在显著交互作用。分层AME结果显示，在未使用雌激素人群中，SSACB绝对风险显著低于NHANES(AME=-21.1人/千人，95%CI:-26.2～-16.7人/千人),而在曾使用雌激素亚组中地区间绝对风险差异呈逆转趋势(AME=6.0人/千人，95%CI:-7.2～32.4人/千人);轻度焦虑抑郁亚组中，SSACB相较于NHANES的低风险优势显著收窄(AME=-12.4人/千人，95%CI:-20.9～-1.5人/千人)。结论 中美女性乳腺癌相关因素及关联强度存在人群异质性，其中雌激素服用史、轻度焦虑抑郁状态、人工绝经在SSACB女性中表现出更强的正向关联。更多还原

【Abstract】 Objective To systematically analyze and compare breast cancer-related factors among women in the National Health and Nutrition Examination Survey(NHANES) and the Shanghai Suburban Adult Cohort and Biobank(SSACB),and to evaluate effect heterogeneity between related factors across regions,thereby providing a reference for understanding the risk characteristics of breast cancer among Chinese women.Methods A cross-sectional comparative study was conducted using data from NHANES(2013-March 2020) and SSACB baseline data.The study included 8 608 women from NHANES(226 breast cancer cases) and 39 683 women from SSACB(249 breast cancer cases).Standardized mean differences(SMD) were employed to assess the comparability of baseline characteristics between the two cohorts.Within each cohort comparisons between the breast cancer and non-breast cancer groups were performed using t-tests,Wilcoxon rank-sum tests,or χ~2 tests,as appropriate.Firth-corrected logistic regression models were used to estimate breast cancerrelated factors within each cohort separately.Subsequently,pooled data were utilized to construct a weighted Firth regression model incorporating " region×factor" interaction terms.Effect heterogeneity was evaluated via likelihood ratio tests and "Wald tests,and stratified average marginal effects(AME) were calculated to quantify absolute risk differences between regions across different subgroups.Results The two cohorts showed substantial differences in most variables,including age,education level,reproductive factors,and lifestyle factors(SMD>0.2).Multivariable analysis showed that age was positively associated with breast cancer risk in both the NHANES cohort(OR=1.11,95% CI:1.08-1.13) and the SSACB cohort(OR=1.02,95% CI:1.00-1.04).In the NHANES cohort,bachelor’ s degree or above(OR=2.20,95% CI:1.24-4.11) and longer daily sedentary time(OR=1.04,95% CI:1.00-1.08) were associated with an increased risk of breast cancer,whereas a greater number of pregnancies(OR=0.92,95 % CI:0.85-1.00),history of estrogen use(OR=0.59,95% CI:0.40-0.84),and hypertension(OR=0.72,95% CI:0.53-0.98) were associated with a reduced risk.In the SSACB cohort,natural menopause(OR=4.89,95 % CI:2.48-10.39),artificial menopause(OR=16.38,95% CI:8.04-35.58),history of estrogen use(OR=3.00,95% CI:1.43-5.55),hyperglycemia(OR=1.61,95 % CI:1.19-2.16),and mild anxiety/depression(OR=2.19,95% CI:1.45-3.19) were all associated with an increased risk of breast cancer,whereas older age at menarche(OR=0.92,95% CI:0.86-0.99) was associated with a reduced risk.Pooled analysis showed that,after full adjustment,the average predicted absolute risk of breast cancer in SSACB women was 19.0 per 1 000 lower than that in NHANES women(95% CI:-24.0 to-14.8 per 1 000).Significant interactions with region were observed for menopausal status(P=0.001),history of estrogen use(P <0.001),and anxiety/depression status(P=0.012).Stratified AME analysis showed that among women without estrogen use,the absolute risk in SSACB was significantly lower than that in NHANES(AME=-21.1 per 1 000,95% CI:-26.2 to-16.7per 1 000),whereas in the subgroup with a history of estrogen use,the regional difference in absolute risk tended to reverse(AME=6.0 per 1 000,95 % CI:-7.2 to 32.4 per 1 000).In the subgroup with mild anxiety/depression,the lowrisk advantage of SSACB relative to NHANES was markedly attenuated(AME=-12.4 per 1 000,95 % CI:-20.9 to-1.5 per 1 000).Conclusion There was population heterogeneity in the spectrum and strength of associations of breast cancer-related factors between Chinese and American women.In particular,history of estrogen use,mild anxiety/depression,and artificial menopause showed stronger positive associations among women in suburban Shanghai.更多还原