Thursday, December 22, 2016

omics whole genome data can improve the accuracy of breast cancer survival predictions

Tags

omics whole genome data can improve the accuracy of breast cancer survival predictions -

accurate predictions of whether a tumor is likely to spread would help clinicians and patients choose the best treatment. But current methods fall short of the required accuracy. A new study reveals that the samples of primary profiling of tumors using genomic technologies can improve the accuracy of predictions of breast cancer survival compared to single clinical information. The study was published in the journal GENETIC , a publication of the American Genetics Society.

Although this method is not ready for use in the clinic, shows evidence of principle study that survival predictions increase when they incorporate detailed data on which genes are active in samples of tumor relative to non-cancerous tissue from the same patient. This is also true for the methylation data of the whole genome, which maps the parts of DNA that carry "tags" that molecular activation influences gene. If developed for use in the clinic, the approach could save some patients from unnecessary chemotherapy.

After surgery, about 80% of breast cancer patients are treated with adjuvant therapies, including chemotherapy and radiotherapy. These treatments often have serious long-term side effects, including heart damage, infertility, memory problems, and a higher risk of developing a new independent cancer. But not all patients necessarily need adjuvant therapy; Breast cancer is estimated to recur or metastasize in 40% of patients, which suggests that a significant number of patients suffer side effects unnecessarily. For now, the widespread use of adjuvant therapy remains inevitable because we can not predict which primary tumors may metastasize and become fatal, and that will stay put.

Today, doctors use a variety of clinical information to help choose the best treatment for an individual patient, including the patient's age and ethnicity, size of the tumor, the type of cell it arises, how it evolved (stage), and the presence of different types of receptors and other molecular signatures of tumor cells (subtype of cancer). To help narrow down the choices of treatment, several commercial tests estimate that the risk of cancer recurrence by measuring the activity (expression) of a set of genes that influence cancer progression. For example, the panel Oncotype DX widely available expression analysis of 21 genes in tumor samples and is recommended for patients with certain types of breast cancer.

But cancer is a complex disease and its behavior is likely affected by thousands of genes. Advances in genomic technology mean it is now possible to measure the gene expression of the tumor through the entire genome. Samples may also be contoured for a variety of other measures of genome-wide, including the variation in the DNA (e.g., deletions or mutations) and methylation. The authors of the new study examined whether these genomic data, either alone or in combination, could actually improve forecasts of breast cancer survival.

"Instead of pre-selection that handful of genes could better predict survival, we used data of all genes in the cancer cell --approximately 17,000 in our study - and let our model calculation select those information, "says study leader Ana I. Vazquez Michigan State University

in. test their approach, Vazquez and his colleagues used data from the Cancer Genome Atlas, a national Institutes of Health project that the profiles of several types of genome-wide data in thousands of cancer samples. the samples are matched normal tissue from the same individual, as well as clinical database for the patient. the polished team in primary breast cancer samples from 285 patients who had adequate clinical monitoring information to allow the team to analyze the survival rate.

the authors used this data set to build calculation models that predict the outcome of a patient (eg survival) using different types of data. They compared the performance of these models using cross-validation. In this method, data is randomly divided into two: one part is used to build and edit a predictive model, and the other part is used to test the accuracy of the model runs. This procedure is repeated hundreds of times to new divisions random data, and the results are marked to reveal the model is more reliable predictions.

This showed that the gene expression of the entire genome data were better predictors of survival than one source of information currently used by doctors, including cancer stage (how advanced the cancer east) and molecular subtype (eg, hormone receptor status). By combining gene expression data with clinical data provided better predictions that all clinical predictors together. The data of gene expression of the entire genome also outperformed the forecasts made with genes in DX Oncotype panel in the subset of patients who met the criteria for the panel. Oncotype DX is a well validated test used in the clinic since 04.

Data only methylation was also more predictive than any standard clinical information, and have also improved most predictions when combined with clinical data . Finally, by combining clinical information, gene expression of the entire genome, and Methylation data provided the most predictive models examined in the study.

"Overall, we can conclude that the forecasts continue to improve as you add data 'omics" said Vazquez. "This gives us promising leads in genomics for the future implementation of the clinic."

Not all types of information across the genome were as predictive as gene expression data or methylation, the team found. the prediction accuracy from clinical data was not affected by the addition of genomic profiles of microRNAs, small molecules that can influence gene expression. And although the accuracy was improved by combining clinical data with genomic profiles of a particular type of DNA change (known as a variation of the number of copies), the improvement was much smaller than the gains provided by gene expression or methylation data.

None of the models can not accurately predict survival. Vazquez said that although the method is promising, a major limitation of the study was the small number of samples available for developing the models. To be applied by clinicians, the method should be validated using the thousands of patient data, rather than hundreds. His team is also studying how to incorporate other factors into their models, including treatment regimens. Ultimately, this can help physicians and patients responding to the best course of treatment with the individual characteristics of each tumor.


EmoticonEmoticon