The latest British study on Intratumor Heterogeneity and Branched Evolution Revealed by Multiregion Sequencing may hinder personalized medicine strategies that depend on results from single tumor biopsy samples.
A reply to a Dr. Robert Nagourney blog posting about this study, brought up the Oncotype DX assay. The woman's breast tumor was early stage 1/grade 1 NO ER/PR positive Her2 negative and her Oncotype DX showed 8% recurrence rate.
After reading her pathology report, three years ago, she raised the question as to why the whole tumor was not looked at after the lumpectomy to make sure the diangosis was the same throughout the tumor. She was told it would be to lengthy, costly and she had nothing to worry about. She was told that they only test a slice of the tumor and throw the rest away. She was taken back by the oncologist's comment. Indeed, so would I.
After she read an article on the British study, is she to assume she may have been right? Or was the pathologist correct in telling her with his assessment of her tumor coupled with the Oncotype DX that things looked good for her and she was not on the Titanic. She was very depressed for someone who was three years out and wondering whether or not the treatment she received was really what she needed.
Indeed, intratumor heterogeneity can lead to underestimation of the tumor genomic landscape portrayed from single tumor-biopsy samples and may present major challenges to personalized medicine and genetic biomarker development. Intratumor heterogeneity, associated with heterogeneous protein function, may foster tumor adaptation and therapeutic failure.
The study is significant because it suggests relying on one sample could overlook important genetic biomarkers that help make tailored treatments effective, explaining perhaps why personalized cancer therapy has been less successful than expected.
It's a theoretical but overrated problem. The same problem applies to ER, Her2, EGFR mutations, KRAS, OncotypeDx. Even worse for trying to do studies on individual cells, e.g. as from circulating tumor cells. Less of a problem for cell function analysis, since they are sampling a much bigger mass of cells and are homogenizing the mass (actually homogenizing the distribution of microclusters).
It's analogous to the Gallup poll. You are projecting the behavior of a national electorate, based on a sample of 1,500 voters, who may or may not be representative of the whole. Rasmussen and Gallup have the same sized sample, but select different people for their polling ("likely voters" vs "all voters"), so their projections often disagree.
It is one of the reasons why (1) "resistance" predictions tend to be more accurate than "sensitive" predictions (of the cancer is resistant anywhere, it pretty much doesn't matter), if you use the "resistant" drug, the patient will have progressive disease and (2) the tests are more analogous to using the barometric pressure to predict for rain than they are analogous to a serum sodium level; i.e. the predictions are useful (assay "sensitive" drugs being seven times more likely to work than assay "resistant" drugs), but they aren't perfect (i.e. 100%), no diagnostic test in medicine is.
Intratumor Heterogeneity and Branched Evolution Revealed by Multiregion Sequencing
http://cancerfocus.org/forum/showthread.php?t=3640