Dataset Information

Breast cancer risk prediction with heterogeneous risk profiles according to breast cancer tumor markers.

ABSTRACT: Relationships between some risk factors and breast cancer incidence are known to vary by tumor subtype. However, breast tumors can be classified according to a number of markers, which may be correlated, making it difficult to identify heterogeneity of risk factors with specific tumor markers when using standard competing-risk survival analysis. In this paper, we propose a constrained competing-risk survival model that allows for assessment of heterogeneity of risk factor associations according to specific tumor markers while controlling for other markers. These methods are applied to Nurses' Health Study data from 1980-2006, during which 3,398 incident invasive breast cancers occurred over 1.4 million person-years of follow-up. Results suggested that when estrogen receptor (ER) and progesterone receptor (PR) status are mutually considered, some risk factors thought to be characteristic of "estrogen-positive tumors," such as high body mass index during postmenopause and increased height, are actually significantly associated with PR-positive tumors but not ER-positive tumors, while other risk factors thought to be characteristic of "estrogen-negative tumors," such as late age at first birth, are actually significantly associated with PR-negative rather than ER-negative breast cancer. This approach provides a strategy for evaluating heterogeneity of risk factor associations by tumor marker levels while controlling for additional tumor markers.

SUBMITTER: Rosner B

PROVIDER: S-EPMC3816337 | biostudies-literature | 2013 Jul

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Breast cancer risk prediction with heterogeneous risk profiles according to breast cancer tumor markers.

Rosner Bernard B Glynn Robert J RJ Tamimi Rulla M RM Chen Wendy Y WY Colditz Graham A GA Willett Walter C WC Hankinson Susan E SE

American journal of epidemiology 20130503 2

Relationships between some risk factors and breast cancer incidence are known to vary by tumor subtype. However, breast tumors can be classified according to a number of markers, which may be correlated, making it difficult to identify heterogeneity of risk factors with specific tumor markers when using standard competing-risk survival analysis. In this paper, we propose a constrained competing-risk survival model that allows for assessment of heterogeneity of risk factor associations according ...[more]

PMID: 23645624

Similar Datasets

Project description:BACKGROUND: For over two decades, the Nottingham Prognostic Index (NPI) has been used in the United Kingdom to calculate risk scores and inform management about breast cancer patients. It is derived using just three clinical variables - nodal involvement, tumour size and grade. New scientific methods now make cost-effective measurement of many biological characteristics of tumour tissue from breast cancer biopsy samples possible. However, the number of potential explanatory variables to be considered presents a statistical challenge. The aim of this study was to investigate whether in ER+ tamoxifen-treated breast cancer patients, biological variables can add value to NPI predictors, to provide improved prognostic stratification in terms of overall recurrence-free survival (RFS) and also in terms of remaining recurrence free while on tamoxifen treatment (RFoT). A particular goal was to enable the discrimination of patients with a very low risk of recurrence. METHODS: Tissue samples of 401 cases were analysed by microarray technology, providing biomarker data for 72 variables in total, from AKT, BAD, HER, MTOR, PgR, MAPK and RAS families. Only biomarkers screened as potentially informative (i.e., exhibiting univariate association with recurrence) were offered to the multivariate model. The multiple imputation method was used to deal with missing values, and bootstrap sampling was used to assess internal validity and refine the model. RESULTS: Neither the RFS nor RFoT models derived included Grade, but both had better predictive and discrimination ability than NPI. A slight difference was observed between models in terms of biomarkers included, and, in particular, the RFoT model alone included HER2. The estimated 7-year RFS rates in the lowest-risk groups by RFS and RFoT models were 95 and 97%, respectively, whereas the corresponding rate for the lowest-risk group of NPI was 89%. CONCLUSION: The findings demonstrate considerable potential for improved prognostic modelling by incorporation of biological variables into risk prediction. In particular, the ability to identify a low-risk group with minimal risk of recurrence is likely to have clinical appeal. With larger data sets and longer follow-up, this modelling approach has the potential to enhance an understanding of the interplay of biological characteristics, treatment and cancer recurrence.

Project description:BackgroundHigher circulating prolactin has been associated with increased breast cancer risk. Prolactin binding to the prolactin receptor (PRLR) can activate the transcription factor STAT5, thus, we examined the association between plasma prolactin and breast cancer risk by tumor expression of PRLR, STAT5, and the upstream kinase JAK2.MethodsUsing data from 745 cases and 2454 matched controls in the Nurses' Health Study, we conducted polytomous logistic regression to examine the association between prolactin (> 11 ng/mL vs. ≤ 11 ng/mL) measured within 10 years of diagnosis and breast cancer risk by PRLR (nuclear [N], cytoplasmic [C]), phosphorylated STAT5 (pSTAT5; N, C), and phosphorylated JAK2 (pJAK2; C) tumor expression. Analyses were conducted separately in premenopausal (n = 168 cases, 765 controls) and postmenopausal women (n = 577 cases, 1689 controls).ResultsIn premenopausal women, prolactin levels > 11 ng/mL were positively associated with risk of tumors positive for pSTAT5-N (OR 2.30, 95% CI 1.02-5.22) and pSTAT5-C (OR 1.64, 95% CI 1.01-2.65), but not tumors that were negative for these markers (OR 0.98, 95% CI 0.65-1.46 and OR 0.73, 95% CI 0.43-1.25; p-heterogeneity = 0.06 and 0.02, respectively). This was stronger when tumors were positive for both pSTAT5-N and pSTAT5-C (OR 2.88, 95% CI 1.14-7.25). No association was observed for PRLR or pJAK2 (positive or negative) and breast cancer risk among premenopausal women. Among postmenopausal women, plasma prolactin levels were positively associated with breast cancer risk irrespective of PRLR, pSTAT5, or pJAK2 expression (all p-heterogeneity ≥ 0.21).ConclusionWe did not observe clear differences in the association between plasma prolactin and breast cancer risk by tumor expression of PRLR or pJAK2, although associations for premenopausal women were observed for pSTAT5 positive tumors only. While additional studies are needed, this suggests that prolactin may act on human breast tumor development through alternative pathways.

Project description:BackgroundMajor depressive disorder (MDD) has become a leading cause of disability worldwide. However, the diagnosis of the disorder is dependent on clinical experience and inventory. At present, there are no reliable biomarkers to help with diagnosis and treatment. DNA methylation patterns may be a promising approach for elucidating the etiology of MDD and predicting patient susceptibility. Our overarching aim was to identify biomarkers based on DNA methylation, and then use it to propose a methylation prediction score for MDD, which we hope will help us evaluate the risk of breast cancer.MethodsMethylation data from 533 samples were extracted from the Gene Expression Omnibus (GEO) database, of which, 324 individuals were diagnosed with MDD. Statistical difference of DNA Methylation between Promoter and Other body region (SIMPO) score for each gene was calculated based on the DNA methylation data. Based on SIMPO scores, we selected the top genes that showed a correlation with MDD in random resampling, then proposed a methylation-derived Depression Index (mDI) by combining the SIMPO of the selected genes to predict MDD. A validation analysis was then performed using additional DNA methylation data from 194 samples extracted from the GEO database. Furthermore, we applied the mDI to construct a prediction model for the risk of breast cancer using stepwise regression and random forest methods.ResultsThe optimal mDI was derived from 426 genes, which included 245 positive and 181 negative correlations. It was constructed to predict MDD with high predictive power (AUC of 0.88) in the discovery dataset. In addition, we observed moderate power for mDI in the validation dataset with an OR of 1.79. Biological function assessment of the 426 genes showed that they were functionally enriched in Eph Ephrin signaling and beta-catenin Wnt signaling pathways. The mDI was then used to construct a predictive model for breast cancer that had an AUC ranging from 0.70 to 0.67.ConclusionOur results indicated that DNA methylation could help to explain the pathogenesis of MDD and assist with its diagnosis.

Project description:BackgroundWe compared six commonly used logistic regression methods for accommodating missing risk factor data from multiple heterogeneous cohorts, in which some cohorts do not collect some risk factors at all, and developed an online risk prediction tool that accommodates missing risk factors from the end-user.MethodsTen North American and European cohorts from the Prostate Biopsy Collaborative Group (PBCG) were used for fitting a risk prediction tool for clinically significant prostate cancer, defined as Gleason grade group ≥ 2 on standard TRUS prostate biopsy. One large European PBCG cohort was withheld for external validation, where calibration-in-the-large (CIL), calibration curves, and area-underneath-the-receiver-operating characteristic curve (AUC) were evaluated. Ten-fold leave-one-cohort-internal validation further validated the optimal missing data approach.ResultsAmong 12,703 biopsies from 10 training cohorts, 3,597 (28%) had clinically significant prostate cancer, compared to 1,757 of 5,540 (32%) in the external validation cohort. In external validation, the available cases method that pooled individual patient data containing all risk factors input by an end-user had best CIL, under-predicting risks as percentages by 2.9% on average, and obtained an AUC of 75.7%. Imputation had the worst CIL (-13.3%). The available cases method was further validated as optimal in internal cross-validation and thus used for development of an online risk tool. For end-users of the risk tool, two risk factors were mandatory: serum prostate-specific antigen (PSA) and age, and ten were optional: digital rectal exam, prostate volume, prior negative biopsy, 5-alpha-reductase-inhibitor use, prior PSA screen, African ancestry, Hispanic ethnicity, first-degree prostate-, breast-, and second-degree prostate-cancer family history.ConclusionDevelopers of clinical risk prediction tools should optimize use of available data and sources even in the presence of high amounts of missing data and offer options for users with missing risk factors.

Dataset Information

Breast cancer risk prediction with heterogeneous risk profiles according to breast cancer tumor markers.

Publications

Breast cancer risk prediction with heterogeneous risk profiles according to breast cancer tumor markers.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets