This "intratumor heterogeneity" issue is not a new revelation to cell function assaysts. As you can see, searching for these genetic predispositions, it is like searching for a needle in a haystack. One can chase all the mutations they want, because if you miss just one, it may be the one that gets through. Or you can look for the drugs that are "sensitive" to killing all of your cancer cells, not theoretical candidates.
Contrary to anlayte-based genomic and proteomic methodologies that yield static measures of gene or protein expression, functional profiling provides a window on the complexity of cellular biology in real-time, gauging tumor cell response to chemotherapies in a laboratory platform. By examining drug induced cell death, functional analyses measure the cumulative result of all of a cell's mechanisms of resistance and response acting in concert. Thus, functional profiling most closely approximates the cancer phenotype.
Testing of one sample of the tumor may well not render an accurate environment, unless you are recognizing the interplay between cells, stroma, vascular elements, cytokines, macrophages, lymphocytes and other environmental factors. The human tumor primary culture microspheroid contains all of these elements. Studying cancer response to drugs within this microenvironment would provide clinically relevant predictions to cancer patients. It is the capacity to study human tumor microenvironments that distinguishes it from other platforms in the field.
They have observed some degree of "genetic drift" where mets tend to be somewhat more resistant to drugs than primaries. Over the years, they have often encouraged physicians to provide nodal, pleural or distant site biopsies to give the "best shot" at the "most defended" of the tumor elements when metastatic disease is found.
The tumor of origin (as in the NEJM study as well) and the associated mets tend to retain consanguinity. That is, the carcinogenic processes that underlie the two populations are related. This is the reason they do not see "mixed responses" (one place in the body getting better and another place in the body getting worse), but instead, generally see response or non-responses.
Heterogeneity likely underlies the recurrences that are seen in almost all patients. This is why they try to re-biopsy and re-evaluate when recurrences are observed. Heterogeneity remains a theoretical issue no matter what platform one uses. Why complicate this fact by using a less biologically relevant method like genomics that only scratches the surface of the tumor biology?
http://robertanagourney.wordpress.co...omic-analysis/