Influence of age as a continuous variable on the prognosis of patients with pT1-2N1 breast cancer.

PURPOSE: To assess the influence of age as a continuous variable on the prognosis of pT1-2N1 breast cancer and examine its decision-making value for postmastectomy radiotherapy (PMRT). METHODS: We retrospectively evaluated 5438 patients with pT1-2N1 breast cancer after mastectomy in 11 hospitals. A multivariable Cox proportional hazards regression model with penalized splines was used to examine the relationship between age and oncologic outcomes. RESULTS: The median follow-up was 67.0 months. After adjustments for confounding characteristics, nonsignificant downward trend in locoregional recurrence (LRR) risk was observed with increasing age (P-non-linear association = 0.640; P-linear association = 0.078). A significant non-linear association was found between age and disease-free survival (DFS) and overall survival (OS) (P-non-linear association <0.05; P-linear association >0.05, respectively). The DFS and OS exhibited U-shaped relationships, with the hazard ratios (HRs), reaching a nadir at 50 years old. A decreased risk of LRR with PMRT vs. no PMRT (HR = 0.304, 95% CI: 0.204-0.454) was maintained in all ages. The HR of PMRT vs. no PMRT for DFS and OS gradually increased with age. In patients ≤50 years old, PMRT was independently associated with favorable LRR, DFS, and OS, all P < 0.05). In patients >50 years old, PMRT was independently associated with reduced LRR (P = 0.004), but had no effect on DFS or OS. CONCLUSIONS: Age was an independent prognostic factor for pT1-2N1 breast cancer; PMRT provided survival benefits for patients ≤50 years old, but not for patients >50 years old.

Effect of postmastectomy radiotherapy on pT1-2N1 breast cancer patients with different molecular subtypes.

OBJECTIVE: To clarify the effect of postmastectomy radiotherapy (PMRT) on pT1-2N1 breast cancer patients with different molecular subtypes. METHODS: We retrospectively analyzed the data of 5442 patients with pT1-2N1 breast cancer treated using modified radical mastectomy in 11 hospitals in China. Univariate, multivariate, and propensity score matching (PSM) analyses were used to evaluate the effect of PMRT on locoregional recurrence (LRR). RESULTS: With a median follow-up duration of 63.8 months, the 5-year LRR rates were 4.0% and 7.7% among patients treated with and without PMRT, respectively (p < 0.001). PMRT was independently associated with reduced LRR after adjustments for confounders (p < 0.001). After grouping the patients according to the molecular subtype of cancer and conducting PSM, we found that the 5-year LRR rates among patients treated with and without PMRT (in that order) were as follows: luminal HER2-negative cancer, 1.9% and 6.5% (p < 0.001); luminal HER2-positive cancer, 3.8% and 13.7% (p = 0.041); HER2-overexpressing cancer, 10.2% and 15.5% (p = 0.236); and triple-negative cancer, 4.6% and 15.9% (p = 0.002). Among patients with HER2-overexpressing and triple-negative cancers, the LRR hazard rate displayed a dominant early peak, and was extremely low after 5 years. However, patients with luminal cancer continued to have a long-lasting high annual LRR hazard rate during follow-up. CONCLUSION: PMRT significantly reduced the LRR risk in patients with pT1-2N1 luminal and triple-negative breast cancers, but had no effect on the LRR risk in patients with HER2-overexpressing cancer. Patients with different molecular subtypes displayed different annual LRR patterns, and the late recurrence of the luminal subtype suggests the necessity of long-term follow-up to evaluate the efficacy of PMRT.

A novel nomogram for predicting locoregional recurrence risk in breast cancer patients treated with neoadjuvant chemotherapy and mastectomy.

BACKGROUND: This study aimed to establish a nomogram for predicting locoregional recurrence (LRR) in breast cancer patients treated with neoadjuvant chemotherapy (NAC) and mastectomy. METHODS: A total of 2368 patients who received NAC and mastectomy between 2000 and 2014 from 12 grade A tertiary hospitals in China were analyzed retrospectively. The nomogram was developed based on the patients treated in three cancer hospitals (training set, n = 1629) and validated based on patients from the other nine general hospitals (validation set, n = 739). Factors identified from Fine and Gray's competing risk analysis were used to establish the nomogram. The predictive performance of the nomogram model was compared with the cTNM stage, ypTNM stage, and the Neo-Bioscore model by using the area under the time dependent receiver operating characteristic curves (tAUC), calibration curve, and decision curve analysis (DCA). RESULTS: The nomogram incorporated six risk factors derived from multivariable analysis of the training set including age, ypT stage, ypN stage, lymph node ratio, postmastectomy radiotherapy, and endocrine therapy. In the training set, the AUC of the nomogram was 0.792, which was higher than the values of the cTNM stage (0.582), ypTNM stage (0.737), and the Neo-Bioscore prognosis model (0.658). In the validation set, the AUC of the cTNM (0.619); ypTNM (0.636); and Neo-Bioscore staging system (0.584) were also significantly lower than the AUC of the nomogram (0.705). Both in the training and validation sets, the calibration curve showed good agreement between the nomogram-based predictions and the actual observations. CONCLUSION: The novel nomogram provides a more accurate evaluation of LRR for breast cancer patients treated with NAC and mastectomy.

Nomogram predicting survival as a selection criterion for postmastectomy radiotherapy in patients with T1 to T2 breast cancer with 1 to 3 positive lymph nodes.

BACKGROUND: The role of postmastectomy radiotherapy (PMRT) in women with pT1-T2N1 breast cancer is controversial. The authors developed a nomogram that was predictive for overall survival (OS) and identified patients who derived no benefit from PMRT. METHODS: The authors retrospectively evaluated 4869 patients with pT1-T2N1 breast cancer who were treated with mastectomy between 2000 and 2014 in 11 Chinese hospitals. Rates of locoregional recurrence and distant metastasis were calculated using competing risk analysis, and disease-free survival and OS rates were calculated using the Kaplan-Meier method. Based on the risk factors identified from Cox regression analysis in 3298 unirradiated patients, a nomogram predicting OS was developed. The benefit of PMRT was evaluated in different risk groups stratified by the nomogram model. RESULTS: After a median follow-up of 65.9 months, the 5-year OS, disease-free survival, locoregional recurrence, and distant metastasis rates were 93.3%, 84.3%, 5.2%, and 8.3%, respectively. A total of 1571 patients (32.3%) underwent PMRT. On multivariable analyses, PMRT was found to increase OS significantly (hazard ratio, 0.61; P = .002). An OS prediction nomogram evaluated the effect of age; tumor location; tumor size; positive lymph node ratio; estrogen receptor, progesterone receptor, and human epidermal growth factor receptor 2 status; and treatment with trastuzumab. Based on nomogram scores, the entire patient cohort was classified into 3 risk groups. PMRT significantly improved the OS of patients in the intermediate-risk (P < .001) and high-risk groups (P = .004), but not in the low-risk group (P = .728). CONCLUSIONS: The authors developed a nomogram that is predictive of OS among women with pT1-T2N1 breast cancer after mastectomy. This nomogram may help to select a subgroup of patients with a good prognosis who will not benefit from PMRT.

Postmastectomy Radiation Therapy Based on Pathologic Nodal Status in Clinical Node-Positive Stage II to III Breast Cancer Treated with Neoadjuvant Chemotherapy.

PURPOSE: The present study aimed to evaluate the effect of postmastectomy radiation therapy (PMRT) after neoadjuvant chemotherapy in patients with node-positive stage II to III (cT1-4N1-2M0) breast cancer. METHODS AND MATERIALS: A total of 1813 patients from 12 institutions were retrospectively reviewed. Patients were classified into 1 of 3 groups based on the pathologic lymph node status after neoadjuvant chemotherapy: ypN0, ypN1, and ypN2-3. The role of PMRT was separately evaluated in each group. Locoregional control, disease-free survival, and overall survival (OS) were estimated using the Kaplan-Meier method. The effect of PMRT was assessed by propensity score-matched analyses and multivariate Cox analyses. RESULTS: With a median follow-up of 72.9 months, 5-year locoregional control, disease-free survival, and OS rates were 86.3%, 68.4%, and 83.1% for the entire cohort, respectively. There were 490 (27.0%), 567 (31.3%), and 756 (41.7%) patients in the ypN0, ypN1, and ypN2-3 groups, respectively. PMRT significantly improved 5-year OS in the ypN2-3 group (74.2% vs 55.9%; P < .001) but had no effect on 5-year OS in the ypN0 group (93.1% vs 95.5%; P = .517) and ypN1 group (88.4% vs 87.8%; P = .549). CONCLUSIONS: With modern systemic therapy, PMRT significantly improved OS in the ypN2-3 group but not in the ypN0 and ypN1 groups. Whether PMRT can be safely omitted in the ypN0 and ypN1 groups should be addressed prospectively.

A prediction model of substrates and non-substrates of breast cancer resistance protein (BCRP) developed by GA-CG-SVM method.

Breast cancer resistance protein (BCRP) is one of the key multi-drug resistance proteins, which significantly influences the therapeutic effects of many drugs, particularly anti-cancer drugs. Thus, distinguishing between substrates and non-substrates of BCRP is important not only for clinical use but also for drug discovery and development. In this study, a prediction model of the substrates and non-substrates of BCRP was developed using a modified support vector machine (SVM) method, namely GA-CG-SVM. The overall prediction accuracy of the established GA-CG-SVM model is 91.3% for the training set and 85.0% for an independent validation set. For comparison, two other machine learning methods, namely, C4.5 DT and k-NN, were also adopted to build prediction models. The results show that the GA-CG-SVM model is significantly superior to C4.5 DT and k-NN models in terms of the prediction accuracy. To sum up, the prediction model of BCRP substrates and non-substrates generated by the GA-CG-SVM method is sufficiently good and could be used as a screening tool for identifying the substrates and non-substrates of BCRP.

Three-class classification models of logS and logP derived by using GA-CG-SVM approach.

In this investigation, three-class classification models of aqueous solubility (logS) and lipophilicity (logP) have been developed by using a support vector machine (SVM) method combined with a genetic algorithm (GA) for feature selection and a conjugate gradient method (CG) for parameter optimization. A 5-fold cross-validation and an independent test set method were used to evaluate the SVM classification models. For logS, the overall prediction accuracy is 87.1% for training set and 90.0% for test set. For logP, the overall prediction accuracy is 81.0% for training set and 82.0% for test set. In general, for both logS and logP, the prediction accuracies of three-class models are slightly lower by several percent than those of two-class models. A comparison between the performance of GA-CG-SVM models and that of GA-SVM models shows that the SVM parameter optimization has a significant impact on the quality of SVM classification model.

In silico prediction of mitochondrial toxicity by using GA-CG-SVM approach.

Drug-induced mitochondrial toxicity has become one of the key reasons for which some drugs fail to enter market or are withdrawn from market. Thus early identification of new chemical entities that injure mitochondrial function grows to be very necessary to produce safer drugs and directly reduce attrition rate in later stages of drug development. In this study, support vector machine (SVM) method combined with genetic algorithm (GA) for feature selection and conjugate gradient method (CG) for parameter optimization (GA-CG-SVM), has been employed to develop prediction model of mitochondrial toxicity. We firstly collected 288 compounds, including 171 MT+ and 117 MT-, from different literature resources. Then these compounds were randomly separated into a training set (253 compounds) and a test set (35 compounds). The overall prediction accuracy for the training set by means of 5-fold cross-validation is 84.59%. Further, the SVM model was evaluated by using the independent test set. The overall prediction accuracy for the test set is 77.14%. These clearly indicate that the mitochondrial toxicity is predictable. Meanwhile impacts of the feature selection and SVM parameter optimization on the quality of SVM model were also examined and discussed. The results implicate the potential of the proposed GA-CG-SVM in facilitating the prediction of mitochondrial toxicity.

An integrated scheme for feature selection and parameter setting in the support vector machine modeling and its application to the prediction of pharmacokinetic properties of drugs.

OBJECTIVE: Support vector machine (SVM), a statistical learning method, has recently been evaluated in the prediction of absorption, distribution, metabolism, and excretion properties, as well as toxicity (ADMET) of new drugs. However, two problems still remain in SVM modeling, namely feature selection and parameter setting. The two problems have been shown to have an important impact on the efficiency and accuracy of SVM classification. In particular, the feature subset choice and optimal SVM parameter settings influence each other; this suggested that they should be dealt with simultaneously. In this paper, we propose an integrated scheme to account for both feature subset choice and SVM parameter settings in concert. METHOD: In the proposed scheme, a genetic algorithm (GA) is used for the feature selection and the conjugate gradient (CG) method for the parameter optimization. Several classification models of ADMET related properties have been built for assessing and testing the integrated GA-CG-SVM scheme. They include: (1) identification of P-glycoprotein substrates and nonsubstrates, (2) prediction of human intestinal absorption, (3) prediction of compounds inducing torsades de pointes, and (4) prediction of blood-brain barrier penetration. RESULTS: Compared with the results of previous SVM studies, our GA-CG-SVM approach significantly improves the overall prediction accuracy and has fewer input features. CONCLUSIONS: Our results indicate that considering feature selection and parameter optimization simultaneously, in SVM modeling, can help to develop better predictive models for the ADMET properties of drugs.

Prediction models of human plasma protein binding rate and oral bioavailability derived by using GA-CG-SVM method.

In this study, support vector machine (SVM) method combined with genetic algorithm (GA) for feature selection and conjugate gradient (CG) method for parameter optimization (GA-CG-SVM), has been employed to develop prediction models of human plasma protein binding rate (PPBR) and oral bioavailability (BIO). The advantage of the GA-CG-SVM is that it can deal with feature selection and SVM parameter optimization simultaneously. Five-fold cross-validation as well as independent test set method were used to validate the prediction models. For the PPBR, a total of 692 compounds were used to train and test the prediction model. The prediction accuracy by means of 5-fold cross-validation is 86% and that for the independent test set (161 compounds) is 81%. These accuracies are markedly higher over that of the best model currently available in literature. The number of descriptors selected is 29. For the BIO, the training set is composed of 690 compounds and external 76 compounds form an independent validation set. The prediction accuracy for the training set by using 5-fold cross-validation and that for the independent test set are 80% and 86%, respectively, which are better than or comparable to those of other classification models in literature. The number of descriptors selected is 25. For both the PPBR and BIO, the descriptors selected by GA-CG method cover a large range of molecular properties which imply that the PPBR and BIO of a drug might be affected by many complicated factors.