Centre for Translational Microbiome Research (CTMR), Dept. of Microbiology, Tumor and Cell Biology, Karolinska Institutet, Biomedicum Kvarter 8A, Tomtebodavägen 16, SE-171 65, Stockholm, Sweden; Science for Life Laboratory (SciLifeLab), SE-171 21 Stockholm, Sweden.
Department of Epidemiology, Harvard T.H. Chan School of Public Health, Boston, MA, USA; Department of Healthcare Policy and Research, Weill Cornell Medicine, New York, NY, USA.
Pharmacol Res. 2020 Oct;160:105072. doi: 10.1016/j.phrs.2020.105072. Epub 2020 Jul 15.
Oral antibiotics are posed as a possible risk factor for breast cancer. Evidence is insufficient to determine whether the choice of antibiotic class could effect this potential association, and non-linearity has not been studied. We aimed to fill these important knowledge gaps.
PubMed, Web of Science, Embase and a trial registry were searched from inception until January 2020, without any restrictions. Additionally, extensive manual searches were undertaken. Random-effects meta-analyses provided pooled risk estimates with 95 % confidence intervals (CI). Dose-response analyses modeling the relationship between number of antibiotic prescriptions and breast cancer risk were extended to non-linear models. Heterogeneity, publication bias and small-study effects were assessed.
Of 7805 identified publications ten were eligible, including 3,719,383 individuals and 84,485 breast cancer cases. The pooled breast cancer risk was modestly increased among individuals who ever used antibiotics (relative risk RR = 1.18, 95 %CI 1.08-1.29), also after excluding the last year prior diagnosis. This excess risk was seen among penicillin (RR = 1.09, 95 %CI 1.01-1.18), tetracycline (RR = 1.13, 95 %CI 1.04-1.24) and nitrofuran users (RR = 1.26, 95 %CI 1.05-1.52), whilst nitroimidazole and metronidazole use (RR = 1.05, 95 %CI 1.00-1.11) indicated for marginal association. No apparent association was found for other antibiotics. Data suggested for a non-linear dose-dependent relationship, with a seemingly protective effect after at least 35 prescriptions. However, these findings might partly be explained by limited power of dose-response analyses.
The association of antibiotics with breast cancer risk appears to differ between the various antibiotic classes. Whether this association is causal remains unclear, requiring further clarification and mechanistic studies.
口服抗生素被认为是乳腺癌的一个潜在危险因素。目前还没有足够的证据来确定抗生素类别选择是否会影响这种潜在的关联，并且还没有研究过非线性关系。我们旨在填补这些重要的知识空白。
从建库到 2020 年 1 月，我们在 PubMed、Web of Science、Embase 和一个试验注册中心进行了无限制的搜索，此外还进行了广泛的手动搜索。我们采用随机效应荟萃分析提供了合并的风险估计值及其 95%置信区间（CI）。我们将抗生素处方数量与乳腺癌风险之间的关系进行了剂量-反应分析，并扩展到非线性模型。我们评估了异质性、发表偏倚和小样本效应。
在 7805 篇已确定的出版物中，有 10 篇符合纳入标准，包括 3719383 名个体和 84485 例乳腺癌病例。与从未使用过抗生素的人相比，曾经使用过抗生素的个体乳腺癌风险适度增加（相对风险 RR = 1.18，95%CI 1.08-1.29），即使排除了诊断前的最后一年。在青霉素（RR = 1.09，95%CI 1.01-1.18）、四环素（RR = 1.13，95%CI 1.04-1.24）和硝基呋喃使用者（RR = 1.26，95%CI 1.05-1.52）中观察到这种超额风险，而硝基咪唑和甲硝唑的使用（RR = 1.05，95%CI 1.00-1.11）则表明相关性较小。其他抗生素未发现明显的相关性。数据提示存在非线性剂量依赖性关系，至少使用 35 剂后呈现出保护作用。然而，这些发现可能部分归因于剂量-反应分析的能力有限。
抗生素与乳腺癌风险之间的关联似乎因抗生素类别而异。这种关联是否具有因果关系尚不清楚，需要进一步澄清和机制研究。
