Trentham-Dietz Amy, Nichols Hazel B, Egan Kathleen M, Titus-Ernstoff Linda, Hampton John M, Newcomb Polly A
University of Wisconsin Paul P. Carbone Comprehensive Cancer Center, Madison, Wisconsin 53726, USA.
Epidemiology. 2007 Sep;18(5):629-38. doi: 10.1097/EDE.0b013e318127183a.
Although the associations with cigarette smoking have been explored extensively for invasive breast cancer, the relation to in situ cancer has not previously been examined in depth.
We analyzed data from a population-based case-control study of women living in Wisconsin, Massachusetts, and New Hampshire. Eligible cases of incident breast carcinoma in situ were reported to statewide registries in 1997-2001 (n = 1878); similarly aged controls (n = 8041) were randomly selected from population lists. Smoking history and other risk factor information were collected through structured telephone interviews. Odds ratios (ORs) and 95% confidence intervals (95% CIs) were calculated from logistic regression models adjusting for potential confounders.
In multivariate models, the OR for breast carcinoma in situ among current smokers was 0.8, compared with never-smokers (95% CI = 0.7-1.0). Risk estimates increased towards the null with greater time since smoking cessation. Odds ratios were also less than 1.0 among women who initiated smoking in adolescence (OR = 0.8) or after a full-term birth (OR = 0.7), relative to women who never smoked. The reduced odds ratios associated with current smoking were strongest among women with annual screening mammograms (OR = 0.7; 95% CI = 0.6-0.9). Odds ratios were not less than 1.0 among current smokers without a recent screening mammogram (1.3; 0.9-2.0).
Our findings suggest an inverse association between current smoking and risk of breast carcinoma in situ among women undergoing breast cancer screening.
尽管对于浸润性乳腺癌与吸烟之间的关联已进行了广泛研究，但原位癌与吸烟的关系此前尚未深入探讨。
我们分析了一项基于人群的病例对照研究的数据，该研究对象为居住在威斯康星州、马萨诸塞州和新罕布什尔州的女性。1997 - 2001年期间，符合条件的原位乳腺癌新发病例被报告至各州登记处（n = 1878）；从人群名单中随机选取了年龄相仿的对照（n = 8041）。通过结构化电话访谈收集吸烟史和其他风险因素信息。根据调整潜在混杂因素的逻辑回归模型计算比值比（OR）和95%置信区间（95% CI）。
在多变量模型中，与从不吸烟者相比，当前吸烟者原位乳腺癌的OR为0.8（95% CI = 0.7 - 1.0）。自戒烟时间越长，风险估计值越接近无效值。相对于从不吸烟的女性，青春期开始吸烟（OR = 0.8）或足月分娩后开始吸烟（OR = 0.7）的女性比值比也小于1.0。在每年接受乳腺钼靶筛查的女性中，当前吸烟相关降低的比值比最为明显（OR = 0.7；95% CI = 0.6 - 0.9）。在近期未接受乳腺钼靶筛查的当前吸烟者中，比值比不小于1.0（1.3；0.9 - 2.0）。
我们的研究结果表明，在接受乳腺癌筛查的女性中，当前吸烟与原位乳腺癌风险之间存在负相关。
