Dataset Information

A personalized committee classification approach to improving prediction of breast cancer metastasis.

ABSTRACT: Metastasis prediction is a well-known problem in breast cancer research. As breast cancer is a complex and heterogeneous disease with many molecular subtypes, predictive models trained for one cohort often perform poorly on other cohorts, and a combined model may be suboptimal for individual patients. Furthermore, attempting to develop subtype-specific models is hindered by the ambiguity and stereotypical definitions of subtypes.Here, we propose a personalized approach by relaxing the definition of breast cancer subtypes. We assume that each patient belongs to a distinct subtype, defined implicitly by a set of patients with similar molecular characteristics, and construct a different predictive model for each patient, using as training data, only the patients defining the subtype. To increase robustness, we also develop a committee-based prediction method by pooling together multiple personalized models. Using both intra- and inter-dataset validations, we show that our approach can significantly improve the prediction accuracy of breast cancer metastasis compared with several popular approaches, especially on those hard-to-learn cases. Furthermore, we find that breast cancer patients belonging to different canonical subtypes tend to have different predictive models and gene signatures, suggesting that metastasis in different canonical subtypes are likely governed by different molecular mechanisms.Source code implemented in MATLAB and Java available at www.cs.utsa.edu/?jruan/PCC/.

SUBMITTER: Jahid MJ

PROVIDER: S-EPMC4071199 | biostudies-literature | 2014 Jul

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

A personalized committee classification approach to improving prediction of breast cancer metastasis.

Jahid Md Jamiul MJ Huang Tim H TH Ruan Jianhua J

Bioinformatics (Oxford, England) 20140310 13

<h4>Motivation</h4>Metastasis prediction is a well-known problem in breast cancer research. As breast cancer is a complex and heterogeneous disease with many molecular subtypes, predictive models trained for one cohort often perform poorly on other cohorts, and a combined model may be suboptimal for individual patients. Furthermore, attempting to develop subtype-specific models is hindered by the ambiguity and stereotypical definitions of subtypes.<h4>Results</h4>Here, we propose a personalized ...[more]

PMID: 24618465

Similar Datasets

Project description:ImportanceThe Cancer Genome Atlas (TCGA) classification is a newly emerging method for prediction of uveal melanoma (UM)-related metastasis and death. Limited information is available regarding the accuracy of the TCGA classification for prediction of metastasis in patients with UM.ObjectiveTo investigate the accuracy of the TCGA classification for predicting UM-related metastasis compared with the American Joint Committee on Cancer (AJCC) classification.Design, setting, and participantsIn this retrospective cohort study, patients with UM treated with plaque radiotherapy at a tertiary referral center from October 1, 2008, to December 31, 2018, were evaluated. All patients with tumors classified according to the American Joint Committee on Cancer Staging Manual, 8th Edition, and who completed pretreatment fine-needle aspiration biopsy sampling for genetic analysis of chromosomes 3 and 8 for TCGA classification were included. Tumors were classified into 4 categories, 17 subcategories, and 4 stages using AJCC classification and further grouped into 4 classes using TCGA classification.Main outcomes and measuresValue of TCGA classification vs AJCC classification for predicting UM-related metastasis.ResultsOf 642 included patients, 331 (51.6%) were women, and the mean (SD) age was 58.0 (13.8) years. There were 642 tumors from 642 patients classified according to both AJCC and TCGA classifications. The mean (range) follow-up time for the entire cohort was 43.7 (1.4-159.2) months. At 5 years, TCGA classification showed higher value for prediction of metastasis (4 TCGA classes: Wald statistic, 94.8; hazard ratio [HR], 2.8; 95% CI, 2.3-3.5; P < .001; 4 AJCC categories: Wald statistic, 67.5; HR, 2.6; 95% CI, 2.1-3.2; P < .001; 17 AJCC subcategories: Wald statistic, 74.3; HR, 1.3; 95% CI, 1.2-1.3; P < .001; 4 AJCC stages: Wald statistic, 67.0; HR, not applicable; P < .001). After entering TCGA and AJCC classifications into a multivariate model, TCGA classification still showed higher value for prediction of metastasis (TCGA classification: Wald statistic, 61.5; HR, 2.4; 95% CI, 1.9-2.9; P < .001; AJCC classification: Wald statistic, 35.5; HR, 1.9; 95% CI, 1.5-2.4; P < .001).Conclusions and relevanceThese results suggest that TCGA classification provides accuracy that is superior to AJCC categories, subcategories, and stages for predicting metastasis from UM. When genetic testing is available, TCGA classification can provide a more accurate way to identify patients at high risk of metastasis who might benefit from adjuvant therapy.

Project description:Objectives: To evaluate whether incorporating the radiomics, genomics, and clinical features allows prediction of metastasis in colorectal cancer (CRC) and to develop a preoperative nomogram for predicting metastasis. Methods: We retrospectively analyzed radiomics features of computed tomography (CT) images in 134 patients (62 in the primary cohort, 28 in the validation cohort, and 44 in the independent-test cohort) clinicopathologically diagnosed with CRC at Dazhou Central Hospital from February 2018 to October 2019. Tumor tissues were collected from all patients for RNA sequencing, and clinical data were obtained from medical records. A total of 854 radiomics features were extracted from enhanced venous-phase CT of CRC. Least absolute shrinkage and selection operator regression analysis was utilized for data dimension reduction, feature screen, and radiomics signature development. Multivariable logistic regression analysis was performed to build a multiscale predicting model incorporating the radiomics, genomics, and clinical features. The receiver operating characteristic curve, calibration curve, and decision curve were conducted to evaluate the performance of the nomogram. Results: The radiomics signature based on 16 selected radiomics features showed good performance in metastasis assessment in both primary [area under the curve (AUC) = 0.945, 95% confidence interval (CI) 0.892-0.998] and validation cohorts (AUC = 0.754, 95% CI 0.570-0.938). The multiscale nomogram model contained radiomics features signatures, four-gene expression related to cell cycle pathway, and CA 19-9 level. The multiscale model showed good discrimination performance in the primary cohort (AUC = 0.981, 95% CI 0.953-1.000), the validation cohort (AUC = 0.822, 95% CI 0.635-1.000), and the independent-test cohort (AUC = 0.752, 95% CI 0.608-0.896) and good calibration. Decision curve analysis confirmed the clinical application value of the multiscale model. Conclusion: This study presented a multiscale model that incorporated the radiological eigenvalues, genomics features, and CA 19-9, which could be conveniently utilized to facilitate the individualized preoperatively assessing metastasis in CRC patients.

Project description:Bitter pit (BP) is one of the most relevant post-harvest disorders for apple industry worldwide, which is often related to calcium (Ca) deficiency at the calyx end of the fruit. Its occurrence takes place along with an imbalance with other minerals, such as potassium (K). Although the K/Ca ratio is considered a valuable indicator of BP, a high variability in the levels of these elements occurs within the fruit, between fruits of the same plant, and between plants and orchards. Prediction systems based on the content of elements in fruit have a high variability because they are determined in samples composed of various fruits. With X-ray fluorescence (XRF) spectrometry, it is possible to characterize non-destructively the signal intensity for several mineral elements at a given position in individual fruit and thus, the complete signal of the mineral composition can be used to perform a predictive model to determine the incidence of bitter pit. Therefore, it was hypothesized that using a multivariate modeling approach, other elements beyond the K and Ca could be found that could improve the current clutter prediction capability. Two studies were carried out: on the first one an experiment was conducted to determine the K/Ca and the whole spectrum using XRF of a balanced sample of affected and non-affected 'Granny Smith' apples. On the second study apples of three cultivars ('Granny Smith', 'Brookfield' and 'Fuji'), were harvested from two commercial orchards to evaluate the use of XRF to predict BP. With data from the first study a multivariate classification system was trained (balanced database of healthy and BP fruit, consisting in 176 from each group) and then the model was applied on the second study to fruit from two orchards with a history of BP. Results show that when dimensionality reduction was performed on the XRF spectra (1.5 - 8 KeV) of 'Granny Smith' apples, comparing fruit with and without BP, along with K and Ca, four other elements (i.e., Cl, Si, P, and S) were found to be deterministic. However, the PCA revealed that the classification between samples (BP vs. non-BP fruit) was not possible by univariate analysis (individual elements or the K/Ca ratio).Therefore, a multivariate classification approach was applied, and the classification measures (sensitivity, specificity, and balanced precision) of the PLS-DA models for all cultivars evaluated ('Granny Smith', 'Fuji' and 'Brookfield') on the full training samples and with both validation procedures (Venetian and Monte Carlo), ranged from 0.76 to 0.92. The results of this work indicate that using this technology at the individual fruit level is essential to understand the factors that determine this disorder and can improve BP prediction of intact fruit.

Dataset Information

A personalized committee classification approach to improving prediction of breast cancer metastasis.

Publications

A personalized committee classification approach to improving prediction of breast cancer metastasis.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets