An FDA advisory panel is currently meeting to analyst a re-analysis of a clinical trial of the GlaxoSmithKline drug Avandia. For a summary of why this is important, read my piece from last night. This is a summary of what happened, culled from my tweets and those of others.
#FDA's Parks says RECORD like verb, such as "Remember to recórd the show, rather than noun: I'll keep a récord of that request #Avandia $GSK
— Donna Young (@ScripDonnaDC) June 5, 2013
Really weird vote at end of this Avandia panel. C will be modify the restrictions on sale -- could include relaxation and tightening.
— Matthew Herper (@matthewherper) June 5, 2013
Ok Marciniak's on. His opinions are not the opinions of the FDA. "IT is not a personal view as is erroneously stated" in the program.
— Matthew Herper (@matthewherper) June 5, 2013
Marciniak was denied access to the readjudication.
— Matthew Herper (@matthewherper) June 5, 2013
On 60 minutes DCRI's founderexplained how they missed Anil Potti: they reviewed a data set provided by Potti.
— Matthew Herper (@matthewherper) June 5, 2013
Core argument here: $GSK funded the analysis, provided the data. Therefore, it is not trustworthy.
— Matthew Herper (@matthewherper) June 5, 2013
At first blush, this Marciniak discussion is stronger than I expected.
— Matthew Herper (@matthewherper) June 5, 2013
Sanjay Kaul agrees the study design is biased. This is a key exchange. Sanjay, generally a critic, saw role for Avandia.
— Matthew Herper (@matthewherper) June 5, 2013
FDA really should have gotten more of the previous panel here. I don't care if they had to pester them.
— Matthew Herper (@matthewherper) June 5, 2013
FDA's Bob Temple providing useful summary of current FDA standards for ruling out CV risk of different drug classes.
— Larry Husten (@cardiobrief) June 5, 2013
Are FDA managers bad at managing difficult but smart people? (Graham, Marciniak).
— Matthew Herper (@matthewherper) June 5, 2013
@john_lamattina @matthewherper If no one has authority, who's accountable? FDA often acts like 25 cats with their tails tied together.
— John Alan Tucker (@JohnTuckerPhD) June 5, 2013
Murray Stewart from $GSK has a great voice. Relaxing. Makes me sleepy.
— Matthew Herper (@matthewherper) June 5, 2013
GSK's Murray: RECORD had to be open-label because in Europe #Avandia use with insulin was contraindicated. #FDA #diabetes
— Miriam E. Tucker (@MiriamETucker) June 5, 2013
On criticisms of RECORD, GSK's Stewart omits the inexplicably low event rate. See Psaty, Furberg: nejm.org/doi/full/10.10…
— David Juurlink (@DavidJuurlink) June 5, 2013
Number of Avandia patients dropped from 117,000 to 3400 after the REMS.
— Matthew Herper (@matthewherper) June 5, 2013
Wow. Duke's K M. shows slide of 30 companies he's consulted for. He speaks at CME events too. Published full disclosures on web.
— Matthew Herper (@matthewherper) June 5, 2013
DCRI COI disclosure website: dcri.org/about-us/confl…
— Larry Husten (@cardiobrief) June 5, 2013
Mahaffey didn't really respond to Marciniak's criticisms. Unfortunate.
— Larry Husten (@cardiobrief) June 5, 2013
Really brief attempt to quiet Marciniak's allegations. Moving onto prepared presentation on RECORD readjudication.
— ACC Advocacy (@Cardiology) June 5, 2013
Mahaffey's 2010-2012 total consulting pay in Dollars for Docs database: $116k. Does not include many conflicts listed by Duke.
— Matthew Herper (@matthewherper) June 5, 2013
@matthewherper Try again with this link: projects.propublica.org/docdollars/sea… (You can remove the tick term with no problems)
— Adam Dunn (@AdamGDunn) June 5, 2013
RECORD re-analyzation being shown. It does still support MI, but not an increase in deaths.
— Matthew Herper (@matthewherper) June 5, 2013
That seems driven by the paucity of strokes. Is the composite endpoint the right one here?
— Matthew Herper (@matthewherper) June 5, 2013
These results really do stand up to massaging, don't they? They look pretty similar to what was reported.
— Matthew Herper (@matthewherper) June 5, 2013
@matthewherper Numerically lower deaths is pretty compelling for safety even in an unblinded trial.
— C. Michael Gibson MD (@CMichaelGibson) June 5, 2013
@matthewherper There were 20 additional deaths found by DCRI.I think that adds strength to the analysis.
— C. Michael Gibson MD (@CMichaelGibson) June 5, 2013
@ivsin @matthewherper Yes it would.
— C. Michael Gibson MD (@CMichaelGibson) June 5, 2013
@matthewherper Missing pt data was clearly at random, = in both arms, not influenced by tx assignment. No informative censoring present.
— C. Michael Gibson MD (@CMichaelGibson) June 5, 2013
Kaul: Did DCRI approach GSK or other way around?
— C. Michael Gibson MD (@CMichaelGibson) June 5, 2013
Kaul: does the cause of death tell you anything about the adjudication?
— Matthew Herper (@matthewherper) June 5, 2013
Kaul: does the cause of death tell you anything about the adjudication?
— Matthew Herper (@matthewherper) June 5, 2013
Kaul picks up double-counting in composite. MI+CV death is one patient, counted in components too.
— Matthew Herper (@matthewherper) June 5, 2013
Kaul picks up double-counting in composite. MI+CV death is one patient, counted in components too.
— Matthew Herper (@matthewherper) June 5, 2013
Kaul asks it: do you believe the RECORD study allows a cardiovascular assessment of safety of rosiglitazone?
— Matthew Herper (@matthewherper) June 5, 2013
Duke answers: we have identified all of the events based on the data that was investigated you can see here...
— Matthew Herper (@matthewherper) June 5, 2013
Duke answers: we have identified all of the events based on the data that was investigated you can see here...
— Matthew Herper (@matthewherper) June 5, 2013
Kaul: you did not answer my questions. Strictly speaking all you can conclude is that Avandia is no worse than the comparator.
— Matthew Herper (@matthewherper) June 5, 2013
Kaul: I am not aware of any evidence of any evidence establishing the safety of the comparator.
— Matthew Herper (@matthewherper) June 5, 2013
Kaul: If you were to do an indirect comparison, how would metformin compare to placebo?
— Matthew Herper (@matthewherper) June 5, 2013
Mahaffey repeats that the data of RECORD is reliable but finally concedes that the design is open to question
— Larry Husten (@cardiobrief) June 5, 2013
Kaul questioning appropriateness of study design, comparators, etc. DCRI: Not our task. We were tasked with re-examining RECORD data.
— ACC Advocacy (@Cardiology) June 5, 2013
@matthewherper @ivsin I agree.
— C. Michael Gibson MD (@CMichaelGibson) June 5, 2013
@matthewherper @cmichaelgibson You know what would be nice right about now? Data from TIDE.
— Paul Ivsin (@ivsin) June 5, 2013
@matthewherper Very similar to what was originally reported.
— C. Michael Gibson MD (@CMichaelGibson) June 5, 2013
Marv Konstam: 811 rosi events referred for adjudication of MI, 663 in control.
— C. Michael Gibson MD (@CMichaelGibson) June 5, 2013
Konstam: Were there differences in post randomization mgt that impacted events?
— C. Michael Gibson MD (@CMichaelGibson) June 5, 2013
Panelist: Increase in CHF with Rose but no increase in mortality? Why was CHF not adjudicated? Why the disconnect?
— C. Michael Gibson MD (@CMichaelGibson) June 5, 2013
The discussions will resume this afternoon.