Department of Gynecology, Heidelberg University, Heidelberg, Germany.
Department of Symptom Research, The University of Texas MD Anderson Cancer Center, Houston, USA.
Eur J Cancer. 2021 Jan;143:134-146. doi: 10.1016/j.ejca.2020.11.006. Epub 2020 Dec 8.
Neoadjuvant systemic treatment elicits a pathologic complete response (pCR) in about 35% of women with breast cancer. In such cases, breast surgery may be considered overtreatment. We evaluated multivariate algorithms using patient, tumor, and vacuum-assisted biopsy (VAB) variables to identify patients with breast pCR.
We developed and tested four multivariate algorithms: a logistic regression with elastic net penalty, an Extreme Gradient Boosting (XGBoost) tree, Support Vector Machines (SVM), and neural network. We used data from 457 women, randomly partitioned into training and test set (2:1), enrolled in three trials with stage 1-3 breast cancer, undergoing VAB before surgery. False-negative rate (FNR) and specificity were the main outcome measures. The best performing algorithm was validated in an independent fourth trial.
In the test set (n = 152), the logistic regression with elastic net penalty, XGboost tree, SVM, and neural network revealed an FNR of 1.2% (1 of 85 patients with missed residual cancer). Specificity of the logistic regression with elastic net penalty was 52.2% (35 of 67 women with surgically confirmed breast pCR identified), of the XGBoost tree 55.2% (37 of 67), of SVM 62.7% (42 of 67), and of the neural network 67.2% (45 of 67). External validation (n = 50) of the neural network showed an FNR of 0% (0 of 27) and a specificity of 65.2% (15 of 23). Area under the ROC curve for the neural network was 0.97 (95% CI, 0.94-1.00).
A multivariate algorithm can accurately select breast cancer patients without residual cancer after neoadjuvant treatment.
新辅助全身治疗使约 35%的乳腺癌患者达到病理完全缓解（pCR）。在这种情况下，乳房手术可能被认为是过度治疗。我们评估了使用患者、肿瘤和真空辅助活检（VAB）变量的多变量算法，以确定具有乳房 pCR 的患者。
我们开发并测试了四种多变量算法：具有弹性网络惩罚的逻辑回归、极端梯度提升（XGBoost）树、支持向量机（SVM）和神经网络。我们使用了来自 457 名女性的数据，这些女性随机分为训练集和测试集（2：1），入组了三项 I-III 期乳腺癌的临床试验，在手术前进行 VAB。假阴性率（FNR）和特异性是主要的观察指标。表现最好的算法在第四个独立试验中进行了验证。
在测试集（n=152）中，具有弹性网络惩罚的逻辑回归、XGBoost 树、SVM 和神经网络的 FNR 为 1.2%（85 名患者中有 1 名患者残留癌症未被发现）。具有弹性网络惩罚的逻辑回归的特异性为 52.2%（67 名手术确认的乳房 pCR 患者中有 35 名），XGBoost 树为 55.2%（67 名中有 37 名），SVM 为 62.7%（67 名中有 42 名），神经网络为 67.2%（67 名中有 45 名）。神经网络的外部验证（n=50）显示 FNR 为 0%（27 名患者中有 0 名），特异性为 65.2%（23 名患者中有 15 名）。ROC 曲线下面积为 0.97（95%CI，0.94-1.00）。
多变量算法可以准确选择新辅助治疗后无残留癌的乳腺癌患者。
