Dataset Information

Borrowing Information across Populations in Estimating Positive and Negative Predictive Values.

ABSTRACT: A marker's capacity to predict risk of a disease depends on disease prevalence in the target population and its classification accuracy, i.e. its ability to discriminate diseased subjects from non-diseased subjects. The latter is often considered an intrinsic property of the marker; it is independent of disease prevalence and hence more likely to be similar across populations than risk prediction measures. In this paper, we are interested in evaluating the population-specific performance of a risk prediction marker in terms of positive predictive value (PPV) and negative predictive value (NPV) at given thresholds, when samples are available from the target population as well as from another population. A default strategy is to estimate PPV and NPV using samples from the target population only. However, when the marker's classification accuracy as characterized by a specific point on the receiver operating characteristics (ROC) curve is similar across populations, borrowing information across populations allows increased efficiency in estimating PPV and NPV. We develop estimators that optimally combine information across populations. We apply this methodology to a cross-sectional study where we evaluate PCA3 as a risk prediction marker for prostate cancer among subjects with or without previous negative biopsy.

SUBMITTER: Huang Y

PROVIDER: S-EPMC3196635 | biostudies-literature | 2011 Nov

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Borrowing Information across Populations in Estimating Positive and Negative Predictive Values.

Huang Ying Y Fong Youyi Y Wei John J Feng Ziding Z

Journal of the Royal Statistical Society. Series C, Applied statistics 20111101 5

A marker's capacity to predict risk of a disease depends on disease prevalence in the target population and its classification accuracy, i.e. its ability to discriminate diseased subjects from non-diseased subjects. The latter is often considered an intrinsic property of the marker; it is independent of disease prevalence and hence more likely to be similar across populations than risk prediction measures. In this paper, we are interested in evaluating the population-specific performance of a ri ...[more]

PMID: 22021938

Similar Datasets

Project description:IntroductionOverexpression of the androgen receptor (AR) characterizes a distinct molecular subset of triple negative breast carcinomas (TNBC). The role of AR as a prognostic/predictive biomarker in TNBC is controversial, but increasing evidence suggests that this subset may respond to therapeutic agents targeting AR. Evaluation of AR has not been standardized, and criteria for selection of patients for antiandrogen therapy remain controversial. In this study we determine the appropriate threshold of AR immunoreactivity to define AR positive (AR+) TNBC, describe the clinicopathologic features of AR+ TNBC, and discuss the utility of AR positivity as a prognostic and predictive marker in TNBC.Materials and methods135 invasive TNBC processed in accordance with ASCO/CAP guidelines, were immunostained for AR. Clinicopathologic features of AR+ TNBC were analyzed and compared to AR negative (AR-) TNBC. Patients' age, tumor size, tumor grade, lymph node status, proliferation rate, immunopositivity for EGFR, CK5/6, Ki-67, and disease free survival (DFS) were evaluated statistically.ResultsA 1% cutpoint was confirmed as the appropriate threshold for AR positivity. Using this cutpoint 41% of 135 TNBC were AR+. AR+ TNBC occurred in older women, were larger, had lower mean proliferation rate and increased incidence of axillary metastasis than AR- TNBC. 76% of TNBC with apocrine morphology were AR+. A subset of AR+TNBC expressed basal markers (EGFR and CK5/6). A prognostic model was created.SummaryAR identifies a heterogeneous group of TNBC. Additional evaluation of EGFR expression allowed us to stratify TNBCs into 3 risk groups with significant differences in DFS and therapeutic implications: low-risk (AR+ EGFR-) which represents the LAR molecular subtype with the best prognosis and may benefit the most from anti-androgen therapies; high-risk (AR- EGFR+) which represents the basal molecular subtype with the worst prognosis and may benefit the most from chemotherapy regimens; intermediate-risk (AR+EGFR+ and AR-EGFR-) TNBC with an intermediate prognosis. Prospective trials are required to further validate this prognostic and predictive grouping.

Dataset Information

Borrowing Information across Populations in Estimating Positive and Negative Predictive Values.

Publications

Borrowing Information across Populations in Estimating Positive and Negative Predictive Values.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets