Infectious Diseases, is still focused on this research question, and her
publications contributed to the FDA’s decision to move away from its blanket
policy of requiring central review.
Recently, Dodd discussed her exploration of this question with The
Cancer Letter’s editor, Paul Goldberg.
PG: How did this issue reveal itself to you? How did you recognize this
as an important research question?
LD: I was in the audience of the December 2007 ODAC that voted on
bevacizumab, and there were concerns expressed about the discrepancy
rate between the central review radiologists and the local evaluations.
They were reporting values around 30 percent.
There were two things that concerned me about the conversation that
was taking place. Most significantly, the hazard ratios between the two
different methods for reviewing the imaging data were very close. That’s the
bottom line in terms of evaluating whether a treatment is effective.
The other thing that concerned me was that discrepancy rates of 30
percent were not surprising to me, because I had worked with radiologists at
NCI for nearly six years at the time. And that seemed in line with what I knew
about radiographic reading and oncology.
The fact that there were concerns raised really brought the issue to a
point for me where I realized that we needed to look into this further to
determine empirically and analytically what the appropriate approach was to
using progression-free survival from an endpoint evaluation perspective.
PG: So this was an opportunity for a biostatistician?
LD: Right. It was an opportunity for a biostatistician to contribute
something useful, potentially.
