March 31, 2006
failure: Why we need cancer cell profiling to match treatment
to patient and why we are dropping the ball (continued):
Meanwhile, back on the DNA and RNA and protein ranch, what is new?
The next great things in cancer treatment are so-called "targeted"
therapies, the most prominent of which now are tyrosine kinase
inhibitors (e.g. gefitinib, erlotinib, sunitinib, sorafenib).
These drugs are very expensive and only work in a minority of cases
(occasionally dramatically; more often modestly). Investigators
at the Dana-Farber Cancer Center (an internationally-renowned,
Harvard-affiliated institution) developed a predictive test, based on
"We are eager for this technology to be widely available to physicians
and their lung cancer patients, as it can help identify those who are
likely to dramatically respond and survive for extended periods of time
with a relatively benign treatment," said [one Dana-Farber oncologist;
from a press release].
"We are entering an exciting new era in cancer therapy," [another
Dana-Farber oncologist, from the same press release] said. "We are now
beginning to be able to use molecular profiling to provide personalized
treatment that offers hope of improved survival and less toxicity from
therapy. And the ability to provide this molecular diagnostic test on a
worldwide basis with rapid turnaround will be critical for further
clinical research and application in the clinic."
Dana-Farber licensed the test to Genzyme, a large US cancer
laboratory. Genzyme is actively
marketing the test as a non-investigational service to cancer
patients, at a cost of about $1,000 (to predict for the activity of a
The Director of Thoracic Oncology at Dana Farber has been quoted as
saying that 50% of all non-small cell lung cancer patients at Dana
Farber are now
receiving the Genzyme test to obtain information to guide their
Now, what were the data supporting the use of this test at the time of
the above quotations?
Prospective, randomized trials showing improved treatment outcomes in
patients so tested? Nope. Prospective trials, showing survival
advantages in patients with "positive" test results (as exist to
validate cell culture drug resistance testing)? Nope.
trials, showing response advantages
in patients with "positive" test results (as exist to validate cell
culture drug resistance testing)? Nope. Retrospective
trials, showing both response and survival advantages
in patients with "positive" test results, in thousands of patients,
from multiple laboratories (as exist to validate cell
culture drug resistance testing)? Nope. Two
entirely retrospective studies, from two Harvard-affiliated hospitals,
response, but not survival, advantages with a grand total of 26
assay/treatment correlations? Yes.
A subsequent study from another laboratory did not
show correlations between gene mutations and patient survival.
Now, in January, 2006, I submitted an abstract to ASCO describing
correlations between cell culture assay results (cell death in response
to gefitinib exposure) and survival of 31 patients with non-small cell
lung cancer who had received extensive prior chemotherapy. These
correlations were based on the actual assay results which had been
reported, in real time, prospectively to the doctors who had ordered
our laboratory tests. There were striking
correlations between test
results and patient survival, yet the ASCO reviewers did not even
select this work for poster presentation. (They will publish the
abstract in this year's ASCO Proceedings, which is why I'm not posting
it at this time).
Now, I can't begin to understand the thinking which goes into reviews
like this (and like the other similar reviews detailed elsewhere on
this website). Can the reader begin to appreciate that I may be
correct in my charges that there is a certain closed-mindedness which
has been rampant now for two decades relating to cell culture
What do the EGFR gene mutations code for? Anti-apoptotic
pathways. So we inhibit anti-apoptosis with gefitinib or erlotinib, and
the cells undergo apoptosis and die. And we detect that at the
whole cell level with our cell culture assays and report out --
prospectively -- that this correlates strikingly with patient survival,
and ASCO determines that it's not even worthy of a look? If
I'm right, then not only will we have a very important predictive test,
but we'll have a unique tool for identifying newer, better drugs,
testing drug combinations, serving as a "gold standard" to develop new
DNA, RNA, and protein-based tests of drug activity, and so on.
And these same close-minded ASCO reviewers condemn the use of the whole
cell function assays until they've been shown to improve patient
treatment outcomes in prospective, randomized trials. While
enthusiastically endorsing the clinical application of DNA content
assays which have been shown to correlate only with response and not
survival, and only in a handful of patients, and only in entirely
What is going on here? It's the same mentality which wouldn't
even agree to look at data indicating a near doubling in the survival
of patients with platinum resistant ovarian cancer (see below).
And wouldn't agree to consider data showing striking correlations
between platinum activity and patient survival in previously-untreated
ovarian cancer (see below). And wouldn't agree to publish a
comprehensive meta-analysis of scores of studies reporting response and
survival correlations in thousands of patients (see below).
While turning a blind eye to the corrosive
practice of selecting
drug regimens based on the profit motive.
In the coming weeks, I'll take a closer look at "molecular" assays vis
a vis cell culture assays.