Deep learning-based predictive model for pathological complete response to neoadjuvant chemotherapy in breast cancer from biopsy pathological images: a multicenter study.

Introduction: Early predictive pathological complete response (pCR) is beneficial for optimizing neoadjuvant chemotherapy (NAC) strategies for breast cancer. The hematoxylin and eosin (HE)-stained slices of biopsy tissues contain a large amount of information on tumor epithelial cells and stromal. The fusion of pathological image features and clinicopathological features is expected to build a model to predict pCR of NAC in breast cancer. Methods: We retrospectively collected a total of 440 breast cancer patients from three hospitals who underwent NAC. HE-stained slices of biopsy tissues were scanned to form whole-slide images (WSIs), and pathological images of representative regions of interest (ROI) of each WSI were selected at different magnifications. Based on several different deep learning models, we propose a novel feature extraction method on pathological images with different magnifications. Further, fused with clinicopathological features, a multimodal breast cancer NAC pCR prediction model based on a support vector machine (SVM) classifier was developed and validated with two additional validation cohorts (VCs). Results: Through experimental validation of several different deep learning models, we found that the breast cancer pCR prediction model based on the SVM classifier, which uses the VGG16 model for feature extraction of pathological images at ×20 magnification, has the best prediction efficacy. The area under the curve (AUC) of deep learning pathological model (DPM) were 0.79, 0.73, and 0.71 for TC, VC1, and VC2, respectively, all of which exceeded 0.70. The AUCs of clinical model (CM), a clinical prediction model established by using clinicopathological features, were 0.79 for TC, 0.73 for VC1, and 0.71 for VC2, respectively. The multimodal deep learning clinicopathological model (DPCM) established by fusing pathological images and clinicopathological features improved the AUC of TC from 0.79 to 0.84. The AUC of VC2 improved from 0.71 to 0.78. Conclusion: Our study reveals that pathological images of HE-stained slices of pre-NAC biopsy tissues can be used to build a pCR prediction model. Combining pathological images and clinicopathological features can further enhance the predictive efficacy of the model.

A Predictive Model for Nonsentinel Node Status after Sentinel Lymph Node Biopsy in Sentinel Lymph Node-Positive Chinese Women with Early Breast Cancer.

BACKGROUND: Axial lymph node dissection (ALND) is needed in patients with positive sentinel lymph node (SLN). ALND is easy to cause upper limb edema. Therefore, accurate prediction of nonsentinel lymph nodes (non-SLN) which may not need ALND can avoid excessive dissection and reduce complications. We constructed a new prognostic model to predict the non-SLN metastasis of Chinese breast cancer patients. METHODS: We enrolled 736 patients who underwent sentinel lymph node biopsy (SLNB); 228 (30.98%) were diagnosed with SLNB metastasis which was determined by intraoperative pathological detection and further accepted ALND. We constructed a prediction model by univariate analysis, multivariate analysis, "R" language, and binary logistic regression in the abovementioned 228 patients and verified this prediction model in 60 patients. RESULTS: Based on univariate analysis using α = 0.05 as the significance level for type I error, we found that age (P=0.045), tumor size (P=0.006), multifocality (P=0.011), lymphovascular invasion (P=0.003), positive SLN number (P=0.009), and negative SLN number (P=0.034) were statistically significant. Age was excluded in multivariate analysis, and we constructed a predictive equation to assess the risk of non-SLN metastasis: Logit(P)=Ln(P/1 - P)=0.267∗a+1.443∗b+1.078∗c+0.471∗d - 0.618∗e - 2.541 (where "a" represents tumor size, "b" represents multifocality, "c" represents lymphovascular invasion, "d" represents the number of metastasis of SLN, and "e" represents the number of SLNs without metastasis). AUCs for the training group and validation group were 0.715 and 0.744, respectively. When setting the risk value below 22.3%, as per the prediction equation's low-risk interval, our model predicted that about 4% of patients could avoid ALND. CONCLUSIONS: This study established a model which demonstrated good prognostic performance in assessing the risk of non-SLN metastasis in Chinese patients with positive SLNs.

Mutant p53 and Twist1 Co-Expression Predicts Poor Prognosis and Is an Independent Prognostic Factor in Breast Cancer.

PURPOSE: The basic helix-loop-helix transcription factor (bHLH) transcription factor Twist1 plays a key role in embryonic development and tumorigenesis. p53 is a frequently mutated tumor suppressor in cancer. Both proteins play a key and significant role in breast cancer tumorigenesis. However, the regulatory mechanism and clinical significance of their co-expression in this disease remain unclear. The purpose of this study was to analyze the expression patterns of p53 and Twist1 and determine their association with patient prognosis in breast cancer. We also investigated whether their co-expression could be a potential marker for predicting patient prognosis in this disease. METHODS: Twist1 and mutant p53 expression in 408 breast cancer patient samples were evaluated by immunohistochemistry. Kaplan-Meier Plotter was used to analyze the correlation between co-expression of Twist1 and wild-type or mutant p53 and prognosis for recurrence-free survival (RFS) and overall survival (OS). Univariate analysis, multivariate analysis, and nomograms were used to explore the independent prognostic factors in disease-free survival (DFS) and OS in this cohort. RESULTS: Of the 408 patients enrolled, 237 (58%) had high mutant p53 expression. Two-hundred twenty patients (53.9%) stained positive for Twist1, and 188 cases were Twist1-negative. Furthermore, patients that co-expressed Twist1 and mutant p53 (T+P+) had significantly advanced-stage breast cancer [stage III, 61/89 T+P+ (68.5%) vs. 28/89 T-P- (31.5%); stage II, 63/104 T+P+ (60.6%)vs. 41/104 T-P- (39.4%)]. Co-expression was negatively related to early clinical stage (i.e., stages 0 and I; P = 0.039). T+P+ breast cancer patients also had worse DFS (95% CI = 1.217-7.499, P = 0.017) and OS (95% CI = 1.009-9.272, P = 0.048). Elevated Twist1 and mutant p53 expression predicted shorter RFS in basal-like patients. Univariate and multivariate analysis identified three variables (i.e., lymph node involvement, larger tumor, and T+P+) as independent prognostic factors for DFS. Lymph node involvement and T+P+ were also independent factors for OS in this cohort. The total risk scores and nomograms were reliable for predicting DFS and OS in breast cancer patients. CONCLUSIONS: Our results revealed that co-expression of mutant p53 and Twist1 was associated with advanced clinical stage, triple negative breast cancer (TNBC) subtype, distant metastasis, and shorter DFS and OS in breast cancer patients. Furthermore, lymph nodes status and co-expression of Twist1 and mutant p53 were classified as independent factors for DFS and OS in this cohort. Co-evaluation of mutant p53 and Twist1 might be an appropriate tool for predicting breast cancer patient outcome.

Comparison of indocyanine green and methylene blue use for axillary reverse mapping during axillary lymph node dissection.

Axillary reverse mapping (ARM) is a technique to identify arm lymphatic drainage during axillary lymph node dissection (ALND). This study compared the feasibility of ARM using indocyanine green (ICG) or methylene blue (MB), and accessed the oncologic safety of the procedure. Overall, 158 patients qualified for ALND were enrolled. The characteristics of ARM-identified nodes were recorded with ICG (n = 78) or MB (n = 80) visualization. Fine-needle aspiration cytology (FNAC) of the nodes were performed and validated by histologic analysis. The nodal identification rate in the ICG group significantly surpassed that of the MB group (87.2% vs 52.5%, P < .05) with fewer complications. Note that 10.9% of the patients had metastatic involvement of the ARM-identified nodes. Also 80% of the positive nodes were found in areas B and D, while the ARM-identified nodes mainly located in area A. All the 51 nodes diagnosed as negative of malignancy by FNAC were free of metastasis. Nodal metastasis was significantly correlated with extensive nodel involvement, advanced disease, and the characteristics of identified nodes. In conclusion, ICG appears superior to MB for ARM nodes identification. FNAC, together with the features of primary tumors and ARM nodes, can delineate which nodes could be preserved during ALND.

Detection of Sentinel Lymph Nodes with Near-Infrared Imaging in Malignancies.

Optical molecular imaging, a highly sensitive and noninvasive technique which is simple to operate, inexpensive, and has the real-time capability, is increasingly being used in the diagnosis and treatment of carcinomas. The near-infrared fluorescence dye indocyanine green (ICG) is widely used in optical imaging for the dynamic detection of sentinel lymph nodes (SLNs) in real time improving the detection rate and accuracy. ICG has the advantages of low scattering in tissue absorbance, low auto-fluorescence, and high signal-to-background ratio. The detection rate of axillary sentinel lymph nodes biopsy (SLNB) in breast cancers with ICG was more than 95 %, the false-negative rate was lower than 10 %, and the average detected number ranged from 1.75 to 3.8. The combined use of ICG with nuclein or blue dye resulted in a lower false-negative rate. ICG is also being used for the sentinel node detection in other malignant cancers such as head and neck, gastrointestinal, and gynecological carcinomas. In this article, we provide an overview of numerous studies that used the near-infrared fluorescence imaging to detect the sentinel lymph nodes in breast carcinoma and other malignant cancers. It is expected that with improvements in the optical imaging systems together with the use of a combination of multiple dyes and verification in large clinical trials, optical molecular imaging will become an essential tool for SLN detection and image-guided precise resection.

Validation and update of a lymph node metastasis prediction model for breast cancer.

PURPOSE: This study aimed to validate and update a model for predicting the risk of axillary lymph node (ALN) metastasis for assisting clinical decision-making. METHODS: We included breast cancer patients diagnosed at six Dutch hospitals between 2011 and 2015 to validate the original model which includes six variables: clinical tumor size, tumor grade, estrogen receptor status, lymph node longest axis, cortical thickness and hilum status as detected by ultrasonography. Subsequently, we updated the original model using generalized linear model (GLM) tree analysis and by adjusting its intercept and slope. The area under the receiver operator characteristic curve (AUC) and calibration curve were used to assess the original and updated models. Clinical usefulness of the model was evaluated by false-negative rates (FNRs) at different cut-off points for the predictive probability. RESULTS: Data from 1416 patients were analyzed. The AUC for the original model was 0.774. Patients were classified into four risk groups by GLM analysis, for which four updated models were created. The AUC for the updated models was 0.812. The calibration curves showed that the updated model predictions were better in agreement with actual observations than the original model predictions. FNRs of the updated models were lower than the preset 10% at all cut-off points when the predictive probability was less than 12.0%. CONCLUSIONS: The original model showed good performance in the Dutch validation population. The updated models resulted in more accurate ALN metastasis prediction and could be useful preoperative tools in selecting low-risk patients for omission of axillary surgery.

A nomogram to predict the probability of axillary lymph node metastasis in early breast cancer patients with positive axillary ultrasound.

Among patients with a preoperative positive axillary ultrasound, around 40% of them are pathologically proved to be free from axillary lymph node (ALN) metastasis. We aimed to develop and validate a model to predict the probability of ALN metastasis as a preoperative tool to support clinical decision-making. Clinicopathological features of 322 early breast cancer patients with positive axillary ultrasound findings were analyzed. Multivariate logistic regression analysis was performed to identify independent predictors of ALN metastasis. A model was created from the logistic regression analysis, comprising lymph node transverse diameter, cortex thickness, hilum status, clinical tumour size, histological grade and estrogen receptor, and it was subsequently validated in another 234 patients. Coefficient of determination (R(2)) and the area under the ROC curve (AUC) were calculated to be 0.9375 and 0.864, showing good calibration and discrimination of the model, respectively. The false-negative rates of the model were 0% and 5.3% for the predicted probability cut-off points of 7.1% and 13.8%, respectively. This means that omission of axillary surgery may be safe for patients with a predictive probability of less than 13.8%. After further validation in clinical practice, this model may support increasingly limited surgical approaches to the axilla in breast cancer.