Classification Of Breast Cancer Patients Research Articles

A comparison of machine learning techniques for survival prediction in breast cancer

BackgroundThe ability to accurately classify cancer patients into risk classes, i.e. to predict the outcome of the pathology on an individual basis, is a key ingredient in making therapeutic decisions. In recent years gene expression data have been successfully used to complement the clinical and histological criteria traditionally used in such prediction. Many "gene expression signatures" have been developed, i.e. sets of genes whose expression values in a tumor can be used to predict the outcome of the pathology. Here we investigate the use of several machine learning techniques to classify breast cancer patients using one of such signatures, the well established 70-gene signature.ResultsWe show that Genetic Programming performs significantly better than Support Vector Machines, Multilayered Perceptrons and Random Forests in classifying patients from the NKI breast cancer dataset, and comparably to the scoring-based method originally proposed by the authors of the 70-gene signature. Furthermore, Genetic Programming is able to perform an automatic feature selection.ConclusionsSince the performance of Genetic Programming is likely to be improvable compared to the out-of-the-box approach used here, and given the biological insight potentially provided by the Genetic Programming solutions, we conclude that Genetic Programming methods are worth further investigation as a tool for cancer patient classification based on gene expression data.

A Signature Inferred from Drosophila Mitotic Genes Predicts Survival of Breast Cancer Patients

IntroductionThe classification of breast cancer patients into risk groups provides a powerful tool for the identification of patients who will benefit from aggressive systemic therapy. The analysis of microarray data has generated several gene expression signatures that improve diagnosis and allow risk assessment. There is also evidence that cell proliferation-related genes have a high predictive power within these signatures.MethodsWe thus constructed a gene expression signature (the DM signature) using the human orthologues of 108 Drosophila melanogaster genes required for either the maintenance of chromosome integrity (36 genes) or mitotic division (72 genes).ResultsThe DM signature has minimal overlap with the extant signatures and is highly predictive of survival in 5 large breast cancer datasets. In addition, we show that the DM signature outperforms many widely used breast cancer signatures in predictive power, and performs comparably to other proliferation-based signatures. For most genes of the DM signature, an increased expression is negatively correlated with patient survival. The genes that provide the highest contribution to the predictive power of the DM signature are those involved in cytokinesis.ConclusionThis finding highlights cytokinesis as an important marker in breast cancer prognosis and as a possible target for antimitotic therapies.

Neutropenia Prediction Based on First-Cycle Blood Counts Using a FOS-3NN Classifier

Background. Delivery of full doses of adjuvant chemotherapy on schedule is key to optimal breast cancer outcomes. Neutropenia is a serious complication of chemotherapy and a common barrier to this goal, leading to dose reductions or delays in treatment. While past research has observed correlations between complete blood count data and neutropenic events, a reliable method of classifying breast cancer patients into low- and high-risk groups remains elusive. Patients and Methods. Thirty-five patients receiving adjuvant chemotherapy for early-stage breast cancer under the care of a single oncologist are examined in this study. FOS-3NN stratifies patient risk based on complete blood count data after the first cycle of treatment. All classifications are independent of breast cancer subtype and clinical markers, with risk level determined by the kinetics of patient blood count response to the first cycle of treatment. Results. In an independent test set of patients unseen by FOS-3NN, 19 out of 21 patients were correctly classified (Fisher's exact test probability P < 0.00023 [2 tailed], Matthews' correlation coefficient +0.83). Conclusions. We have developed a model that accurately predicts neutropenic events in a population treated with adjuvant chemotherapy in the first cycle of a 6-cycle treatment.

Significance of Serum Tumor Necrosis Factor-Alpha and its Combination with Her-2 Codon 655 Polymorphism in the Diagnosis and Prognosis of Breast Cancer

The present study was conducted to clarify the diagnostic and prognostic significance of TNF-alpha and its combination with HER-2 Ile655Val SNP in breast cancer. In this case-control study, 56 consecutive patients with primary breast cancer were prospectively evaluated. The control group consisted of 45 healthy women. Serum concentrations of TNF-alpha were measured by quantitative sandwich enzyme immunoassay (ELISA). HER-2 SNP was genotyped using the PCR-RFLP method. Serum TNF-alpha was significantly increased in patients compared to controls. ROC analysis indicated a cutoff point of 11.00 pg/mL to classify breast cancer patients (sensitivity, 86%; specificity, 71%). Elevated TNF-alpha levels were associated with larger, poorly differentiated, invasive and advanced-stage tumors, and >3 positive lymph nodes. Regarding HER-2 SNP, patients with Ile-Val and Val-Val genotypes had significant TNF-alpha elevation compared with homozygous Ile-Ile patients. In multivariate analysis, high serum TNF-alpha remained an independent prognostic factor of worse overall survival; its combination with Val-Val genotype predicted a worse prognosis than high TNF-alpha alone. Serum TNF-a could be used clinically as a useful tumor marker for diagnosis, disease extent and outcome of breast cancer. The negative impact on survival seems to be enhanced through the interaction with HER-2 Ile655Val SNP.

Comparative Study on the Use of Analytical Software to Identify the Different Stages of Breast Cancer Using Discrete Temperature Data

Breast cancer is the second leading cause of death in women. It occurs when cells in the breast begin to grow out of control and invade nearby tissues or spread throughout the body. The limitations of mammography as a screening modality, especially in young women with denser breasts, necessitated the development of novel and more effective screening strategies with acceptable sensitivity and specificity. The aim of this study was to develop a feasible interpretive software system which was able to detect and classify breast cancer patients by employing techniques of different analytical software. The protocol described uses 6,000 pieces of thermal data collected from 16-sensors, eight placed on the surface of each breast. Data was collected every 5 min for the duration of the test period. Placement of sensors was accomplished with the use of a template design from information provided by the national tumor registry to insure that the information was collected in areas of the breast where most breast cancers develop. Data in this study was collected from 90 individuals exhibiting four different breast conditions, namely: normal, benign, cancer and suspected-cancer. The temperature data collected from these 16 sensors placed on the surface of each breast were fed as inputs to the classifiers. Comparisons were made on five different kinds of classifiers: back-propagation algorithm, probabilistic neural network, fuzzy (Sugeno-type), Gaussian mixture modeland support vector machine. These classifiers were able to attain approximately 80% accuracy in classifying the four different diagnoses (normal, benign, cancer and suspected-cancer). Gaussian mixture model was the most sensitive classifier, achieving the highest sensitivity of 94.8%. Support vector machine was considered the best classifier as it was able to produce the most specific and accurate results. Based on these evaluations, this current effort shows the feasibility of applying analytical software techniques together with the real-time functional thermal analysis to develop a potential tool for the detection and classification of breast cancer.

Breast cancer patients with micrometastases only: Is a basis provided for tailored treatment?

Women with breast cancer and micrometastases only constitute a treatment dilemma. If only a micrometastasis is found in a sentinel lymph node, an axillary lymph node dissection may be considered to be overtreatment and perhaps could be avoided. However, studies have shown decreased survival in patients with micrometastases only. This paper focuses on the pathological work-up behind the classification of breast cancer patients having micrometastases only and on the most recent literature concerning prognosis for breast cancer patients with micrometastases. The goal was to determine if studies to date have been able to define a population of breast cancer patients with micrometastases where the size of the metastasis could indicate whether an axillary lymph node dissection should be undertaken. Tailored surgical treatment of breast cancer patients with micrometastases only is not possible at the present time, due to lack of standardization in the pathological work-up of lymph nodes, which implies that this group of breast cancer patients cannot be delimited with sufficient precision. Tailored systemic therapy is also impossible due to lack of a precisely defined target for a feasible therapy.

KERNEL-BASED NAIVE BAYES CLASSIFIER FOR BREAST CANCER PREDICTION

The classification of breast cancer patients is of great importance in cancer diagnosis. Most classical cancer classification methods are clinical-based and have limited diagnostic ability. The recent advances in machine learning technique has made a great impact in cancer diagnosis. In this research, we develop a new algorithm: Kernel-Based Naive Bayes (KBNB) to classify breast cancer tumor based on memography data. The performance of the proposed algorithm is compared with that of classical navie bayes algorithm and kernel-based decision tree algorithm C4.5. The proposed algorithm is found to outperform in the both cases. We recommend the proposed algorithm could be used as a tool to classify the breast patient for early cancer diagnosis.

Gene Expression Signature in Organized and Growth-Arrested Mammary Acini Predicts Good Outcome in Breast Cancer

Nonmalignant human mammary epithelial cells (HMEC) seeded in laminin-rich extracellular matrix (lrECM) form polarized acini and, in doing so, transit from a disorganized proliferating state to an organized growth-arrested state. We hypothesized that the gene expression pattern of organized and growth-arrested HMECs would share similarities with breast tumors with good prognoses. Using Affymetrix HG-U133A microarrays, we analyzed the expression of 22,283 gene transcripts in 184 (finite life span) and HMT3522 S1 (immortal nonmalignant) HMECs on successive days after seeding in a lrECM assay. Both HMECs underwent growth arrest in G0-G1 and differentiated into polarized acini between days 5 and 7. We identified gene expression changes with the same temporal pattern in both lines and examined the expression of these genes in a previously published panel of microarray data for 295 breast cancer samples. We show that genes that are significantly lower in the organized, growth-arrested HMEC than in their proliferating counterparts can be used to classify breast cancer patients into poor and good prognosis groups with high accuracy. This study represents a novel unsupervised approach to identifying breast cancer markers that may be of use clinically.

Establishment of a predictive classifier of node-positive breast cancer patients treated by FEC100 chemotherapy using gene expression profiling

13004 Background: In breast cancer treatment, biomarkers providing information about chemotherapy sensitivity are needed. FEC100 combination, frequently prescribed in Europe, is still applied empirically to patients. Our study’s goal was to establish a classifier of sensitivity to this regimen using gene expression data and classical clinicopathologic parameters. Methods: The study retrospectively included 151 patients belonging to the FEC100 arms of two multicentric phase III clinical trials: PACS01 (n = 128) and PEGASE01 (n = 23) (FNCLCC). Patients had unilateral breast cancer, showed no evidence of distant metastasis, were node-positive, aged less than 65-year-old. Median follow-up was 5 years. The number of relapses were respectively 23 and 10. We used cDNA nylon microarrays containing 7,000 genes to analyze gene expression profiles of tumor. Patients were split into a training set and a test set. A 3-step analysis based on Cox regression was applied: 1) elimination of non discriminant genes, 2) robust (resampling) univariate selection of discriminant genes and 3) development of multivariate models combining the discriminant genes, the Nottingham Prognostic Index (NPI) (developed in 2 binary variables) and the hormonal receptors. As a final step, after dichotomization of the retained genes, a risk score was built and applied first on the test set, and then on the whole cohort. Kaplan-Meier curves and logrank tests were performed to assess the new risk score. Results: The new risk score (one gene [G6224] and NPI) permitted to separate patients from the test set in 3 significantly different groups; it was also an improvement on NPI (test set: NPI, p = 0.0005; risk score p = 0.0001 - whole cohort: NPI, p = 2.10−5; risk score, p = 1.10−10). Conclusions: We identified G6224 whose expression combined with NPI showed a good capacity for classifying breast cancer patients who received FEC100 chemotherapy. Our results strengthen the interest of G6224 since it was previously described in various solid tumors as a prognostic biomarker with a tumor-suppressor activity. No significant financial relationships to disclose.

Prognostic Utility of a Gene-Expression Signature in Untreated Node-Negative Breast Cancer

Purpose: Gene expression profiling has the potential to lead to a more accurate prognostic classification of breast cancer patients. Recently, a set of 70 genes had been identified that allowed risk classification of node-negative patients. We analysed the prognostic utility of a subset of these genes and compared it with established prognostic factors in a collective of 150 untreated node-negative breast cancer patients. Methods: cRNA of 150 untreated node-negative breast cancer patients was hybridized on the Affymetrix HG-U133A array and probe sets corresponding to the set of 70 genes were identified. A good and poor prognosis profile was generated for the expression of 52 genes. The prognostic utility of this gene-expression signature was then compared with established prognostic factors (i.e. histological grade, tumor size and steroid hormone receptor status). The prognostic utility for disease-free survival (DFS) was evaluated using univariate and multivariate statistical analyses. Results: 33 (22%) of the tumors were assigned to the good and 117 (78%) to the poor prognosis group based on the gene-expression signature. This led to a sensitivity of 93% with a specificity of 26%. 5-year DFS was 94% in the good and 78% in the poor prognosis group, respectively. However, this difference diminished with longer follow-up (p=0.167). Of the prognostic factors analysed, only histological grade was associated with DFS (p<0,0005). Conclusion: histological grade had a higher prognostic utility than the gene-expression signature in our cohort of untreated node-negative breast cancer patients.

00034 PROGNOSTIC UTILITY OF A GENE-EXPRESSION SIGNATURE IN UNTREATED NODE-NEGATIVE BREAST CANCER

Purpose: Gene expression profiling has the potential to lead to a more accurate prognostic classification of breast cancer patients. Recently, a set of 70 genes had been identified that allowed risk classification of node-negative patients. We analysed the prognostic utility of a subset of these genes and compared it with established prognostic factors in a collective of 150 untreated node-negative breast cancer patients. Methods: cRNA of 150 untreated node-negative breast cancer patients was hybridized on the Affymetrix HG-U133A array and probe sets corresponding to the set of 70 genes were identified. A good and poor prognosis profile was generated for the expression of 52 genes. The prognostic utility of this gene-expression signature was then compared with established prognostic factors (i.e. histological grade, tumor size and steroid hormone receptor status). The prognostic utility for disease-free survival (DFS) was evaluated using univariate and multivariate statistical analyses. Results: 33 (22%) of the tumors were assigned to the good and 117 (78%) to the poor prognosis group based on the gene-expression signature. This led to a sensitivity of 93% with a specificity of 26%. 5-year DFS was 94% in the good and 78% in the poor prognosis group, respectively. However, this difference diminished with longer follow-up (p=0.167). Of the prognostic factors analysed, only histological grade was associated with DFS (p Conclusion: histological grade had a higher prognostic utility than the gene-expression signature in our cohort of untreated node-negative breast cancer patients

Pooled analysis of prognostic impact of uPA and PAI-1 in breast cancer patients.

In this report we present an extension of the pooled analysis of the prognostic impact of urokinase-type plasminogen activator (uPA) and its inhibitor PAI-I in breast cancer patients. We analyzed a different endpoint, metastasis-free survival (MFS). We checked the consistency of the estimates for uPA and PAI-1 for relapse-free survival (RFS) and MFS exploring possible sources of heterogeneity. Nodal status, the most important prognostic factor for breast cancer, introduced heterogeneity in the uPA/PAI-1 survival analyses, reflecting the interaction between nodal status and uPA/PAI-1. The estimates for uPA and PAI-1 were found to be consistent, even when a different transformation of their values was used. The heterogeneity of the separate data sets decreased if the levels of uPA and PAI-1 were ranked, data sets were pooled, and the analyses corrected for the base model that included all traditional prognostic factors, and stratified by data set. We conclude that uPA and PAI-1 are ready to be used in the clinic to help classify breast cancer patients into high and low risk groups.

Survival-Time Classification of Breast Cancer Patients

The identification of breast cancer patients for whom chemotherapy could prolong survival time is treated here as a data mining problem. This identification is achieved by clustering 253 breast cancer patients into three prognostic groups: Good, Poor and Intermediate. Each of the three groups has a significantly distinct Kaplan-Meier survival curve. Of particular significance is the Intermediate group, because patients with chemotherapy in this group do better than those without chemotherapy in the same group. This is the reverse case to that of the overall population of 253 patients for which patients undergoing chemotherapy have worse survival than those who do not. We also prescribe a procedure that utilizes three nonlinear smooth support vector machines (SSVMs) for classifying breast cancer patients into the three above prognostic groups. These results suggest that the patients in the Good group should not receive chemotherapy while those in the Intermediate group should receive chemotherapy based on our survival curve analysis. To our knowledge this is the first instance of a classifiable group of breast cancer patients for which chemotherapy can possibly enhance survival.

Molecular classification of breast cancer patients by gene expression profiling

Molecular classification of breast cancer patients by gene expression profiling

Molecular classification of breast cancer patients by gene expression profiling.

For many tumors, pathological subclasses exist which have to be further defined by genetic markers to improve therapy and follow-up strategies. In this study, cDNA array analyses of breast cancers have been performed to classify tumors into categories based on expression patterns. Comparing purified normal ductal epithelial cells and corresponding tumour tissues, the expression of only a small fraction of genes was found to be significantly changed. A subset of genes repeatedly found to be differentially expressed in breast cancers was subsequently employed to perform a classification of 82 normal and malignant breast specimens by cluster analysis. This analysis identifies a subgroup of transcriptionally related tumours, designated class A, which can be further subdivided into A1 and A2. Correlation with classical clinicopathological parameters revealed that subgroup A1 was characterized by a high number of node-positive tumours (14 of 16). In this subgroup there was a disproportionate number of patients who had already developed distant metastases at the time of diagnosis (25% in this subgroup, compared with 5% among the rest of the samples). Taken together, the use of these differentially expressed marker genes in conjunction with sample clustering algorithms provides a novel molecular classification of breast cancer specimens, which facilitates the identification of patients with a higher risk of recurrence.

Prognosis in breast cancer utilizing histologic characteristics of the primary tumor.

A study was made of the inter-relationships and prognostic significance of structural characteristics found in primary breast cancers and their associated axillary lymph nodes. The prognostically favorable characteristics included the following. For the primary tumor: nuclear differentiation of the cancer cells, diffuse lymphoid cell infiltrations (LI) and perivenous lymphoid cell infiltrations (PVI). For the axillary lymph nodes: sinus histiocytosis (SH). Perivenous lymphoid cell infiltrations (PVI) in the primary tumor are found to be as important a prognostic factor as SH in the axillary lymph nodes; these two characteristics are found to be positively associated. Evaluation of the nuclear grade (NG), LI, and PVI in the primary tumor allows for the definition of association with cancer cells having a low (anaplastic) nuclear grade, and to be positively associated with follicular hyperplasia (FH) in the lymph nodes. We also found a positive association between the cellular responses to areas of in situ carcinoma and the cellular responses to accompanying invasive breast breast cancer tissue. They also provide a prognostic system for classifying breast cancer patients on the basis of the microscopic characteristics of the primary tumor and surrounding breast tissue. The latter system should be of value in comparing the therapeutic benefits of various treatments.