Even after FDA changes its policies, drug companies would continue
to collect and store scans. The difference is, they wouldn’t have to read
all these scans as part of central review.
At the meeting earlier this week, several ODAC members cautioned
that increased reliance on investigators at trial sites could amount to an
unfunded mandate that may further weaken the already stressed system
of clinical trials.
“We have to be very careful that we simply don’t transfer one
procedure to requiring unnecessary paperwork on investigators who are
already reading these and are already encumbered with a lot,” said ODAC
temporary member Wyndham Wilson, chief of the NCI Center for Cancer
Research Lymphoma Therapeutics Section.
Even before PFS became a widely used endpoint at the FDA oncology
operation, the agency’s default position was to demand that drug companies
use central radiological review.
However, NCI-funded clinical trials cooperative groups were exempt
from the requirement to use central radiology, and instead could rely on
assessment by investigators.
The question of whether central radiology is actually better than
assessment at the sites appears to have come into focus early in the
controversy over Genentech’s application for Avastin (bevacizumab) in
metastatic breast cancer (The Cancer Letter, Dec. 14, 2007).
Avastin received an accelerated approval for breast cancer based
on a cooperative group trial that was not designed to support registration.
The trial—E2100, conducted by Eastern Cooperative Oncology Group—
relied on investigator assessment of progression.
At that time, FDA had a special understanding with cooperative groups:
while drug companies were required to use central radiology, cooperative
groups didn’t have to. However, in September 2006, FDA stunned observers
by demanding that Genentech provide additional documentation with the
Avastin supplemental Biologic License Application.
The sponsor had to locate the scans and reanalyze them using a
central radiology panel. Now, suddenly, the agency was saying that while
the investigator’s determination was good enough in trials that measured
survival, it was no longer sufficient for trials that measured the delay in
disease progression (The Cancer Letter, Sept. 15, 2006).
This controversy could seem obscure six years later, except for having
framed the question of whether assessment of PFS by investigators at the
test sites is indeed less reliable than review by central radiological committees.
In addition to framing the question, the agency made public the data from
both investigator assessments and central radiology assessments of the
same scans.
Is it possible that central radiology was not better than investigator
assessment?
Sitting in the audience that day, Lori Dodd, then an NCI biostatistician,
discerned a scientific question, which led to an influential paper published in
the Journal of Clinical Oncology the following year. (A Q&A with Dodd appears
on page 1.)
Now, turn the clock forward to July 24, 2012.
In addition to Dodd’s paper, a working group of industry scientists
organized by PhRMA industry statisticians has assessed concordance of
investigator assessment and central review in 27 randomized trials. On top
of that, FDA conducted its own, separate meta-analysis of interpretation of
radiographic evidence. Also, the agency and the industry held at least one
public discussion of the question (The Cancer Letter, Nov. 7, 2008).
All of this appears to have prompted FDA to abandon reliance on central
radiology panels and instead rely on investigator assessment, augmented by
audits designed to detect bias.
“Remember, what are we after?” Richard Pazdur, director of the FDA
Office of Hematology and Oncology Drug Products, said at the ODAC meeting
earlier this week. “Going back to the central issue—the presence or absence of bias.
“We are not after some ultimate, absolute truth here of what is the true
value, because that probably doesn’t ever exist.”
