乳腺癌新辅助化疗疗效相关因素的探索及疗效预测模型的建立

发布时间：2018-08-06 19:52

【摘要】：背景及目的：贫血是一种常见的恶性肿瘤合并症,既往研究提示贫血可能是乳腺癌化疗后不良结局的预测因子,但是大多数研究主要关注贫血与术后辅助化疗结局的关系,这种情况下常不能直观评价疗效,很难准确评价贫血与化疗的相关性。新辅助化疗又称术前化疗,是乳腺癌的重要治疗手段,可提供肿瘤化疗体内疗效的直接证据,在新辅助化疗中可更好地探究贫血对化疗疗效的真实影响。治疗前贫血是否与乳腺癌新辅助化疗结局相关尚无定论,本研究拟评价治疗前贫血作为临床因素对乳腺癌新辅助化疗病理学疗效及远期生存的影响,即明确治疗前贫血是否为乳腺癌新辅助化疗的疗效预测因素和预后因素。方法：本研究共纳入可手术或局部晚期乳腺癌(Ⅱa-Ⅲc期)患者655人,所有患者均在1999年1月至2011年12月期间于我院接受含蒽环类联合紫杉类方案的新辅助化疗。根据治疗前血红蛋白(hemoglobin, Hb)水平将患者分为贫血组(Hb12.0g/dL)和非贫血组(Hb≥12.0g/dL)。主要研究终点为病理学完全缓解(pathological complete remission, pCR),次要研究终点为无复发生存(relapse-free survival RFS)、肿瘤特异性生存(cancer-specific survival, CSS)和总生存(overall survival, OS)。对比贫血组和非贫血组的pCR率、RFS、CSS和OS,并通过多因素回归分析明确包括贫血在内的多种临床因素是否为乳腺癌新辅助化疗疗效预测因素和预后因素。结果：166人(25.3%)合并治疗前贫血。贫血组pCR率明显低于非贫血组(5.4% vs.11.7%；p=0.024)；治疗前贫血为pCR的独立预测因素,且与pCR呈负相关(OR 0.428,95% C10.198-0.927,p=0.031)；基线Hb水平作为连续变量与新辅助化疗病理学疗效之间无显著相关性(OR 1.007,95% CI 0.990-1.024, p=0.452);年龄(OR 4.805,95% CI 1.845-12.515, p=0.001)和ER表达(OR 4.199,95% CI 1.904-9.261, p0.001)也是pCR的独立预测因素。与非贫血组相比,贫血组的10年RFS(59.1% vs 66.0%,p=0.022)、OS(75.3% vs 90.9%,p0.001)和CSS(82.4% vs 94.4%,p0.001)均明显下降；多因素分析提示治疗前贫血是乳腺癌新辅助化疗后RFS(HR 1.453,95% CI 1.077-1.962, p=0.015)、OS(HR 2.873,95% CI 1.757-4.699, p0.001)和CSS(HR 2.961, 95% CI 1.679-5.222, p0.001)的独立预后因素；将基线Hb水平作为连续变量进行评价时发现基线Hb水平高者CSS (HR 0.979,95% CI 0.961-0.997, p=0.021)和OS(HR 0.977,95% CI 0.962-0.992, p=0.003)优于Hb水平低者,基线Hb水平亦为乳腺癌预后因素。结论：本研究证明治疗前贫血与乳腺癌新辅助化疗后pCR呈负相关,合并贫血是新辅助化疗病理学疗效的独立预测因素,治疗前贫血也与乳腺癌新辅助化疗后远期生存显著相关,合并贫血是乳腺癌的不良预后因素。现有的纠正贫血的治疗措施不良反应多,今后需进一步优化贫血干预措施,以期改善合并贫血的乳腺癌患者的临床结局。背景及目的：新辅助化疗是局部晚期乳腺癌的重要治疗手段,近年来在早期可手术的浸润性乳腺癌中的应用也逐渐增多。通过新辅助化疗获得病理学完全缓解的患者其预后更好,化疗后无缓解甚至是疾病进展的患者生存状况较差,因此亟需一种可以预测乳腺癌化疗敏感性的生物标记物来筛选可能会从新辅助化疗中获益的患者。循环微小RNA (microRNA, miRNA)是指存在于体液如血浆、血清中的miRNA,探讨其作为肿瘤诊断、疗效预测和监测、预后指标的研究日益增多。本研究拟在接受新辅助化疗的乳腺癌患者中动态监测其血浆miRNA的变化,并探讨这种动态变化与疗效的相关性,以期发现可早期预测乳腺癌新辅助化疗敏感性的分子标记物。方法：本研究基于一项对比不同亚型乳腺癌新辅助化疗方案疗效的前瞻性临床试验进行,共纳入109名可手术或局部晚期乳腺癌患者,在我院接受4-6周期表阿霉素联合紫杉醇(ET方案)新辅助化疗。根据激素受体(hormone receptor, HR)和人类表皮生长因子受体2(human epidermal growth factor receptor 2, HER2)表达状况将患者分成HR阳性/HER2阴性、HER2阳性和三阴性乳腺癌(triple-negative breast cancer, TNBC)组,在每个亚组中分别进行研究。采集患者基线、2周期化疗后(C2)和术前的外周血标本并进行血浆miRNA检测。根据临床疗效将患者分为化疗敏感(完全/部分缓解)和不敏感(疾病稳定/进展)者。先通过TaqMan miRNA芯片检测筛选出其动态变化与化疗敏感性可能相关的候选miRNA;在基线、C2和术前标本中用实时定量PCR法检测待验证的miRNA,并通过logistic回归分析评价基线/C2miRNA表达水平是否可预测化疗敏感性。结果：化疗不敏感者占31.2%(34/109)。通过miRNA芯片检测发现两种血浆miRNA波动趋势可能与疗效相关：1、基线时在敏感组和不敏感组之间表达明显差异,术前血浆中这种差异进一步扩大；2、基线时两组间表达相近,化疗后呈相反的变化趋势。在HR+/HER2-亚型(n=51)中选择了4个血浆miRNA (miR222、miR-20a、 miR-451、miR-9)进一步验证：与敏感组基线相比,不敏感组基线miR-222表达显著升高(2.065倍,p=0.047),化疗2周期后继续升高(4.870倍,p0.001),而在敏感组中未观察到上述化疗诱导的波动趋势(C2 vs.基线,0.977倍,p=0.826)；基线时miR-20a在敏感和不敏感组间表达无明显差异(p=0.218),化疗后不敏感组miR-20a表达上调(C2 vs.基线,2.637倍,p=0.008),敏感组血浆miR-20a表达无明显改变(C2 vs.基线,0.986倍,p=0.882)；血浆miR-451波动趋势与miR-20a大致相反(基线vs.基线,p=0.673；不敏感组C2 vs.基线,0.762倍,p=0.014；敏感组C2 vs.基线,1.194倍,p=0.060)。在3个亚型中均发现血浆miR-34a的动态变化与化疗敏感性相关：基线时血浆miR-34a在两组间表达相近,不敏感者化疗后miR-34a水平下降。基线miR-222 (OR=6.422,p=0.049)、C2 miR-20a (OR=0.144, p=0.021)、 C2 miR-451 (OR=8.213,p=0.012)水平是HR阳性/HER2阴性乳腺癌新辅助化疗临床疗效的预测因素。结论：本研究通过在接受ET方案新辅助化疗的乳腺癌患者中动态监测血浆miRNA的变化趋势,发现基线miR-222高表达、2周期化疗后miR-20a表达上调和miR-451表达下降是HR阳性/HER2阴性乳腺癌化疗敏感性的预测因素。此外,2周期化疗后血浆miR-34a表达下降与化疗不敏感相关,提示miR-34a是一种潜在的早期疗效预测标记物。背景及目的：既往关于乳腺癌新辅助化疗疗效预测因素的研究结果多不一致,提示单一因素预测疗效的能力不足,将多种潜在的疗效相关因素组合起来建立模型或可提高预测的准确度。除了常见的临床病理特征外,治疗前贫血也是乳腺癌新辅助化疗疗效的预测因素；生物学因素如血浆miRNA的动态变化是潜在的疗效预测分子标记物,其中血浆miR-34a化疗后表达下调与临床疗效不佳相关。列线图是一种用来估算个体发生特定临床结局事件的可能性或风险的统计学工具。本研究拟在接受新辅助化疗的乳腺癌患者中探究疗效相关的临床和生物学因素,并建立可早期预测新辅助化疗疗效的列线图,用来估计乳腺癌患者从新辅助化疗中获益的可能性。方法：本研究共纳入149名患者,其中109人来自一项对比不同亚型乳腺癌新辅助化疗方案疗效的前瞻性临床试验,接受4-6周期表阿霉素联合紫杉醇(ET方案)新辅助化疗,该群体定义为训练集；另有40名患者于我院接受ET方案新辅助化疗,但未参加上述临床试验,该群体定义为验证集。根据临床疗效将患者定义为化疗敏感(完全／部分缓解)或不敏感(疾病稳定／进展)。在训练集(n=109)中收集患者的临床病理信息以及C2血浆miR-34a相对表达量,筛选与临床疗效(化疗不敏感)相关的预测因子并建模,通过列线图的方式展示所建立的预测模型；在上述训练集中对预测模型的区分度(用曲线下面积(area under curve, AUC)衡量)和校正进行考核,也就是内部验证；在验证集(n=40)中对建立的预测模型进行外部验证,评价其区分度和校正情况。结果：在训练集中,结合logistic回归结果和临床实际意义确定基线Ki-67、治疗前贫血、C2 miR-34a表达、HER2表达状况和临床N分期为潜在的乳腺癌新辅助化疗疗效预测因素,基于这5个因素建立预测模型并描绘列线图。在训练集中从区分度和校正情况两方面评价模型的预测能力,模型的AUC为0.765,通过bootstrap法进行内部验证,校正后的AUC为0.726；模型的校正情况良好,预测概率与实际发生率之间无明显差异(Hosmer-Lemeshow检验p=0.835)。在验证集中预测模型的AUC为0.751,说明模型的区分度较好；模型预测概率与实际观察到的发生率较为一致,无明显差异,说明校正情况良好(U统计量对应的p值0.99)。结论：本研究基于临床生物学因素提出了一种可早期预测乳腺癌新辅助化疗疗效的模型,并用列线图进行展示,通过列线图可以快速估算个体从ET方案新辅助化疗中获益的概率,但是该模型的预测能力仍有待在大样本、多中心研究中进一步验证,今后可进一步优化该模型,扩大其适用范围。
[Abstract]:Background and purpose: anemia is a common malignant tumor complication. Previous studies suggest that anemia may be a predictor of adverse outcomes after chemotherapy for breast cancer. However, most studies focus on the relationship between anemia and postoperative adjuvant chemotherapy outcomes. In this case, it is often difficult to evaluate the curative effect intuitively, and it is difficult to accurately evaluate anemia and chemotherapy. Correlation. Neoadjuvant chemotherapy, also known as preoperative chemotherapy, is an important treatment for breast cancer, which provides direct evidence for the effect of tumor chemotherapy in vivo. In the new adjuvant chemotherapy, it can better explore the true influence of anemia on the therapeutic effect of chemotherapy. Pre treatment anemia as a clinical factor on the pathological effect and long-term survival of neoadjuvant chemotherapy for breast cancer, that is, to determine whether anemia is a predictive factor and prognostic factor for neoadjuvant chemotherapy for breast cancer before treatment. Methods: This study included 655 patients with surgical or locally advanced breast cancer (stage II a- III C), all of which were 199 A new adjuvant chemotherapy with anthracycline combined paclitaxel regimen was accepted in our hospital from January to December 2011 9. The patients were divided into anemia group (Hb12.0g/dL) and non anemia group (Hb > 12.0g/dL) according to the pre treatment hemoglobin (hemoglobin, Hb) level. The main end point was complete pathological remission (pathological complete remission, pCR), and secondary end point. The study endpoint was relapse-free survival RFS, tumor specific survival (cancer-specific survival, CSS) and total survival (overall survival, OS). Compare the pCR rates of the anemia and non anemia groups, RFS, CSS, and other factors, and to determine whether various clinical factors including anemia were new adjuvant to breast cancer by multiple regression analysis. Prognostic factors and prognostic factors. Results: 166 people (25.3%) combined with anemia before treatment. The pCR rate in anemia group was significantly lower than non anemia group (5.4% vs.11.7%; p=0.024); pre treatment anemia was an independent predictor of pCR and was negatively correlated with pCR (OR 0.428,95% C10.198-0.927, p=0.031); baseline Hb level as a continuous variable and a new auxiliary There was no significant correlation between the curative effects of chemotherapy and chemotherapy (OR 1.007,95% CI 0.990-1.024, p=0.452); age (OR 4.805,95% CI 1.845-12.515, p=0.001) and ER expression were also independent predictors. Compared with the non anemia group, the 10 years of anemia group (59.1%, 66%, 90.9%) 001) and CSS (82.4% vs 94.4%, p0.001) decreased significantly. Multivariate analysis suggested that pre treatment anemia was an independent prognostic factor of RFS (HR 1.453,95% CI 1.077-1.962, p=0.015), OS (HR 2.873,95%, 2.961, 95%). It was found that CSS (HR 0.979,95% CI 0.961-0.997, p=0.021) and OS (HR 0.977,95% CI 0.962-0.992) were superior to those with low levels. The baseline level was also a prognostic factor for breast cancer. Conclusion: This study showed that anemia before treatment was negatively correlated with breast cancer neoadjuvant chemotherapy and anemia was a new supplement. The independent predictors for the therapeutic effect of chemotherapy, anemia before treatment are also significantly related to the long-term survival of breast cancer after neoadjuvant chemotherapy. Anemia is a bad prognostic factor for breast cancer. There are many adverse reactions to the treatment of anemia, and further optimization of anemia intervention should be made in the future in order to improve the incidence of anemia associated with breast cancer. Background and objective: neoadjuvant chemotherapy is an important treatment for local advanced breast cancer. In recent years, the application in early operable invasive breast cancer is increasing. The prognosis of patients with complete pathologic remission through neoadjuvant chemotherapy is better, and the patients who have no remission or even the progression of the disease after chemotherapy are born. Poor survival, so there is a need for a biomarker that can predict chemosensitivity to breast cancer to screen patients who may benefit from neoadjuvant chemotherapy. Circulating minute RNA (microRNA, miRNA) refers to the presence of miRNA in body fluids such as plasma and serum, to explore the diagnosis, prognosis, and monitoring of the tumor as a swelling tumor, and the study day of prognostic indicators. This study intends to dynamically monitor changes in plasma miRNA in breast cancer patients receiving neoadjuvant chemotherapy and explore the correlation between this dynamic change and efficacy in order to find molecular markers for early prediction of breast cancer sensitivity to neoadjuvant chemotherapy. Method: This study was based on a new contrast of different subtypes of breast cancer. A prospective clinical trial of chemotherapeutic regimens was carried out in 109 patients with surgical or locally advanced breast cancer, with 4-6 cycles of epirubicin combined with paclitaxel (ET) neoadjuvant chemotherapy in our hospital. According to the hormone receptor (hormone receptor, HR) and human epidermal growth factor receptor 2 (human epidermal growth factor receptor 2, H) ER2) was divided into HR positive /HER2 negative, HER2 positive and three negative breast cancer (triple-negative breast cancer, TNBC) groups in each subgroup. The patients were studied in each subgroup. The patient baseline, 2 cycles of chemotherapy (C2) and preoperative peripheral blood specimens and plasma miRNA tests were carried out. The patients were divided into chemotherapeutic sensitivity according to the clinical effect. Complete / partial remission) and insensitivity (disease stability / progression). First, TaqMan miRNA chip detection was used to screen candidate miRNA for possible dynamic changes associated with chemosensitivity; miRNA was detected by real-time quantitative PCR in baseline, C2 and preoperative specimens, and the baseline /C2miRNA expression level was evaluated by logistic regression analysis. Results: chemosensitivity was predicted. Results: 31.2% (34/109) was not sensitive to chemotherapy. The miRNA chip detection showed that the trend of miRNA fluctuation in two plasma may be related to the curative effect: 1, the difference between the sensitive group and the insensitive group at baseline was significantly different, and the difference of plasma in the preoperative plasma was increased step by step; 2, the expression of the two groups was similar in the baseline. 4 plasma miRNA (miR222, miR-20a, miR-451, miR-9) were selected in the HR+/HER2- subtype (n=51) to further verify that the baseline miR-222 expression in the insensitive group was significantly increased (2.065 times, p=0.047) compared with the sensitive group baseline, and continued to increase (4.870 times, p0.001) after 2 weeks of chemotherapy, but not in the sensitive group. The trend of chemotherapy induced fluctuation (C2 vs. baseline, 0.977 times, p=0.826), and no significant difference (p=0.218) between sensitive and insensitive groups at baseline (p=0.218), the expression of miR-20a in the non sensitive group after chemotherapy was up (C2 vs. baseline, 2.637 times, p=0.008), and there was no significant change in the expression of miR-20a in the sensitive group (C2 vs. baseline, 0.986 times,); 1 the fluctuation trend was roughly the opposite of miR-20a (baseline vs. baseline, p=0.673; insensitive group C2 vs. baseline, 0.762 times, p=0.014; sensitive group C2 vs. baseline, 1.194 times, p=0.060). In the 3 subtypes, the dynamic changes of plasma miR-34a were found to be related to chemosensitivity: the plasma miR-34a in the baseline was similar between the two groups, and the insensitive patients after chemotherapy miR-3 4A level decreased. Baseline miR-222 (OR=6.422, p=0.049), C2 miR-20a (OR=0.144, p=0.021), C2 miR-451 (OR=8.213, p=0.012) were predictors of the clinical efficacy of neoadjuvant chemotherapy for positive negative breast cancer. Conclusion: This study monitored the changes of plasma levels dynamically in breast cancer patients receiving neoadjuvant chemotherapy. Trend, the high expression of baseline miR-222, up regulation of miR-20a expression and decrease of miR-451 expression after 2 cycles of chemotherapy are predictors for chemosensitivity of HR positive /HER2 negative breast cancer. In addition, the decrease of miR-34a expression after 2 cycles of chemotherapy is associated with chemotherapy insensitivity, suggesting that miR-34a is a potential prognostic marker for early efficacy. The results of previous studies on the predictive factors of neoadjuvant chemotherapy for breast cancer are often inconsistent, suggesting that the single factor is not able to predict the curative effect, combining various potential therapeutic factors together to establish a model or to improve the accuracy of the prediction. Prognostic factors for adjuvant chemotherapy; biological factors such as dynamic changes in plasma miRNA are potential prognostic molecular markers, in which the down-regulation of plasma miR-34a after chemotherapy is associated with poor clinical efficacy. A column map is a statistical tool used to estimate the possibility or risk of individual occurrence of a specific clinical outcome. To explore the clinical and biological factors related to the effectiveness of the breast cancer patients receiving neoadjuvant chemotherapy and to establish a line map that can predict the efficacy of neoadjuvant chemotherapy early to estimate the possibility of breast cancer patients benefiting from the neoadjuvant chemotherapy. Methods: This study included 149 patients, of which 109 were from a different contrast. A prospective clinical trial of neoadjuvant chemotherapy for subtype breast cancer received 4-6 cycles of epirubicin combined with paclitaxel (ET) neoadjuvant chemotherapy. The group was defined as a training set; another 40 patients received neoadjuvant chemotherapy in our hospital, but the group was not involved in the clinical trials. The group was defined as a validation set. According to clinical efficacy, the group was defined as a test set. The patient is defined as chemotherapy sensitive (complete / partial remission) or insensitivity (disease stability / progression). In the training set (n=109), the clinicopathological information of the patient and the relative expression of the C2 plasma miR-34a are collected, and the predictive factors associated with the clinical efficacy (insensitivity to chemotherapy) are screened and modeled, and the prediction of the prediction is presented through the line diagram. Model; in the above training center, the area indexing of the prediction model (area under curve, AUC) is measured and the correction is examined, that is, internal verification. In the verification set (n=40), the established prediction model is verified externally, and the degree and correction of the area are evaluated. Results: in the training concentration, combined with the logistic regression knot. Determine baseline Ki-67, pre treatment anemia, C2 miR-34a expression, HER2 expression, and clinical N staging as potential prognostic factors for neoadjuvant chemotherapy for breast cancer. Based on these 5 factors, a predictive model is established and a line map is depicted. The predictive ability of the model is evaluated in the two aspects of the training centralization and correction. The AUC of the model is 0.765, the internal verification by bootstrap method and the corrected AUC are 0.726. The correction of the model is good, there is no obvious difference between the prediction probability and the actual occurrence rate (Hosmer-Lemeshow test p=0.835). In the verification of the centralized prediction model, the AUC is 0.751, indicating the good distinction of the model; the model prediction probability and the actual view. The rate of detection was consistent and no significant difference was found, indicating that the correction was good (the corresponding p value of U statistics 0.99). Conclusion: Based on the clinical biological factors, a new model for predicting the efficacy of new adjuvant chemotherapy for breast cancer was proposed in this study, and a column diagram was used to predict the new individual from the ET scheme. The probability of benefiting in adjuvant chemotherapy, but the prediction ability of the model still needs to be further verified in the large sample and multi center research. In the future, the model can be further optimized and its scope of application can be expanded.
【学位授予单位】：北京协和医学院
【学位级别】：博士
【学位授予年份】：2016
【分类号】：R737.9

【相似文献】