Lugano-CH, Brussels- BE, 2 May 2013
An important new study has revealed the clearest picture yet of precisely how much measurement variation influences gene expression profiles of breast cancer.
The results show, for the first time, which gene expression measurements may benefit from pooling of biopsies from a single tumour, researchers said at the 5th IMPAKT Breast Cancer Conference in Brussels, Belgium.
These findings represent an important step toward allowing doctors to more precisely tailor an individual’s treatment to a detailed analysis of their tumour’s gene expression.
Over recent years, scientists have identified many genes, and groups of genes, that can provide crucial information about how an individual patient’s cancer will respond to treatment with different drugs.
But a number of hurdles need to be cleared before tests to measure the expression of these genes can be used in clinical situations.
One important challenge is the fact that many different cell types can be present within a single tumour (known as intratumoural heterogeneity), each with different patterns of gene expression and potentially different sensitivity to drugs.
“Performing these tests with a single biopsy may or may not accurately represent that cancer, depending on intratumoural heterogeneity,” explains lead author Dr Rosanna Lau from the University of Texas MD Anderson Cancer Center, US.
A further complication is that some of the variation between test results can arise from technical variations in the testing process, and not by real differences between samples (analytical variance).
To differentiate between these sources of variation in breast cancer, Dr Lau and colleagues performed DNA microarray analysis on three biopsies each from 51 breast cancers.
“Our results indicate that analytical variance, resulting from technical aspects of the assay, can be dramatically reduced by standard data processing methods such as normalizing and scaling,” Dr Lau says. “Pre-analytical sources of variance, such as tissue preservation method and ischemia mostly did not affect gene measurements.”
The dominant source of variance between biopsies from the same tumour was due to intratumoural heterogeneity, Dr Lau’s group found. However, the extent of that variation depended on the particular gene or groups of genes being studied.
“Some genes, such as ESR1 and HER2, are very consistently expressed across the tissue, thus gene expression measurements display little variation between biopsies. However, other genes such as MKI67, which is known to be highly variable, is expressed less consistently, and therefore can produce vastly different results depending on the area of the tumour that is sampled,” Dr Lau says.
For the first time, Dr Lau’s group also showed how combining samples of two or three biopsies from a single breast tumour could effectively overcome this variation for selected genes.
“Differences between tumours are much greater than variability within a tumour or a test. Our current study shows that we can get a comprehensive picture of the genes being expressed in the tumour by sampling multiple areas of the tumour and pooling the samples together. This increases the precision of the assay and allows us to make more reliable predictions related to the disease. The trade-off is that intratumoural heterogeneity is also averaged to a single, more consistent measurement.”
This study is an excellent example of how researchers are rising to the challenges of tumour heterogeneity, comments Prof Charles Swanton, Chair in Personalised Cancer Medicine at the UCL Cancer Institute in London and from the Cancer Research UK London Research Institute, member of the ESMO Translational Research Working Group, who was not involved in the study.
“Developing accurate biomarkers that are not subject to real tumour sampling bias is of critical importance. This intricate study will likely be a gold standard by which other studies in this area are measured. Such in-depth analyses will ultimately be essential in the biomarker qualification process,” he said.
“The study also emphasises the need to limit the potential for tissue processing or assay technologies to lead to spurious measurements through well-defined standardised operating procedures,” Prof Swanton said.
Notes to Editors
Session info: Heterogeneity of breast cancer: How much is there at the start and how much happens over time?
Thursday, 2 May 2013, 3:30 PM – 5:10 PM (CEST). Place: Gold Hall
Please contact the IMPAKT Press Office at email@example.com to schedule remote interviews or for any inquiry.
About the Breast International Group (BIG)
The Breast International Group (BIG) is a non-profit organisation for academic breast cancer research groups from around the world, based in Brussels, Belgium.
Founded by leading European opinion leaders in 1996, BIG now constitutes a network of 50 groups based in Europe, Canada, Latin America, Asia and Australasia. These research entities are tied to several thousand specialised hospitals and research centres worldwide. More than 30 clinical trials are run or are under development under the BIG umbrella. BIG also works closely with the US National Cancer Institute (NCI) and the North American Breast Cancer Groups (NABCG), so that together they act as a strong integrating force in the breast cancer research arena.
To make significant scientific advances in breast cancer research, reduce unnecessary duplication of effort, and optimally serve those affected by the disease, large-scale cooperation is crucial. Therefore BIG facilitates breast cancer research at international level, by stimulating cooperation between its members and other academic networks, and collaborating with, but working independently from, the pharmaceutical industry. To find out more about BIG, please visit: www.breastinternationalgroup.org
About the European Society for Medical Oncology (ESMO)
The European Society for Medical Oncology (ESMO) is the leading European professional organization committed to advancing the specialty of medical oncology and promoting a multidisciplinary approach to cancer treatment and care.
ESMO’s mission is to advance cancer care and cure through fostering and disseminating good science that leads to better medicine and determines best practice.
As a trusted organization with 35 years of experience, ESMO serves its members and the oncology community through: a brand of excellence in post-graduate oncology education and training; leadership in transforming evidence-based research into standards of cancer care in Europe; dedicated efforts to foster a more favorable environment for scientific research; innovative international platforms to share expertise, best practices and disseminate the most up-to-date scientific research to as wide an audience as possible.
ESMO’s scientific journal,Annals of Oncology, ranks among the top clinical oncology journals worldwide. ESMO events are the meeting place in Europe for medical oncologists to update their knowledge, to network and to exchange ideas. To find out more about ESMO, please visit: www.esmo.org
Contribution of analytical, pre-analytical and introa-tumoural heterogeneity to variance in gene expression measurements from human breast cancers.
R. Lau1, H. Sun1, R. Gould1, C. Hatzis2, W..F. Symmans1
1Houston, TX/US, 2Saugus, MA/US
Purpose: To estimate the contributions of intra-tumoural heterogeneity (3 biopsies), pre-analytical sample integrity (ex vivo ischemic time), and analytical assay error (technical replicates) to the overall variance of microarray-based measurement of gene expression or multi-gene signatures in a clinical laboratory.
Methodology: Intra-tumoural heterogeneity (IH) was estimated from 51 breast cancers, each sampled with 3 biopsies and profiled on Affymetrix U133A arrays. Analytical variance (AV) was estimated from technical replicates of each key step of the standard microarray profiling procedure (repeated RNA extraction, cDNA synthesis, in vitro translation [IVT], and hybridization) in a single biopsy from 20 of the 51 tumours. Pre-analytical variance (PAV) from 6 levels of ex vivo ischemia in 11 breast cancers was estimated from a previously published cohort (Hatzis et al, JNCI 2011). Within each study, we estimated the total within sample variance using a linear mixed effects model. The variance associated with IH and PAV was adjusted by subtracting the AV, and all components were reported relative to the robust standard deviation (SD, i.e. interquartile range/1.349) in the combined population.
Results: RNA extraction and array hybridization were the predominant contributors to analytical variance of measurements of single genes (ESRI, MKI67) and multi-gene indices (SET index and GGI), whereas cDNA synthesis and IVT were highly reproducible. The SD associated with MKI67 measurements was approximately 70% of the biological SD, limiting the clinical utility of this marker. However, for ESRI, SET index and GGI, the total measurement SD accounted for 18-33% of the biological SD. Generally, the IH dominated the other contributions, but to a different extent for each gene or signature.
Conclusions: In general, intra-tumoural heterogeneity, analytical and pre-analytical variance contributed to approximately 30% of the total standard deviation in single- and multi-gene measurements. Quality improvement in the clinic might include pooling 2 or more biopsies, optimizing sample preservation and laboratory attention to RNA extraction and array hybridization.
Disclaimer: Information contained in this press release was provided by the abstracts’ authors and reflects the content of the studies. It does not necessarily express ESMO's or BIG’s point of view.