rosiglitazone and the FDA

one of the BMC preceptors asked for my assessment of the shenanigans around rosiglitazone and what we should do (thanks, dan).  so, here is my sense:

1. FDA initially restricted rosiglitazone because of very large meta-anal showing increased cardiac events

2. the RECORD study (Rosiglitazone Evaluated for Cardiovascular Outcomes and Regulation of glycemia in Diabetes study) did not find increased cardiac risk after 3.75 yrs.  re-eval was done at 5.5 years just published in August (see dm RECORD trial amhrtj 2013 in dropbox, or doi.org/10.1016/j.ahj.2013.05.004).  see point 3 below, but the FDA gives lots more weight to this type of trial looking at cardiac outcomes, vs secondary analyses

3. so, they lifted their restriction.

but,...

1. rosiglitazone is not nearly as good as pioglitazone for lipids. for example, a study of 400 people off all other diabetes meds were randomized to pioglitazone vs rosiglitazone. results at 24 weeks:

    --no diff in A1c

    --fasting TG significantly reduced with pioglit but increased with rosiglit

   --HDL increased in both groups but significantly more with pioglit

   --LDL increased with both, more with rosiglit, and (perhaps most important) LDL particle size was larger with pioglit (it is the small, dense LDL particles that are more oxidizable and seem to cause the real problems: inciting inflammation, causing monocytes to differentiate to tissue macrophages, and it is only the oxidized LDLs that bind to the scavenger receptor on the macrophages and lead to uptake of lipids and formation of foam cells)

2. unlike rosiglit which has the rather striking adverse events on the initial meta-anal, pioglit has one of the few studies showing actual cardiovascular benefit. the PROACTIVE study (see dm pioglit PROACTIVE study. lancet 2005 in dropbox, or Lancet 2005; 366: 1279–89) was a prospective RCT of 5238 pts with diabetes with evident macrovascular disease. after 3 years, primary composite endpoint of all-cause mortality, nonfatal MI (including silent MI), stroke, acute coronary syndrome, surgical intervention on heart or legs, amputation) was nonsignificantly better with pioglit.  But, secondary outcome of all-cause mortality, nonfatal MI and stroke was a very significant 16% reduction. [a bit strange that they used these primary/secondary outcomes: their secondary outcome is more objective by not including decisions about surgical intervention, etc.  Each of the secondary endpoints were better with pioglit, though interestingly silent MIs were the same].

3. the RECORD study  specifically looked at cardiac outcomes with rosiglit.  4447 pts with A1C >7 randomly assigned to adding rosiglit, sulfonylurea (if failed metformin), or metformin (if failed sulfonylurea) with 5.5 yr followup, looking at cardiovasc or unknown cause of death, MI, or stroke. No diff in primary endpoints. but small number of events. high dropout rate. and significantly more patients in the rosiglit group on statins (55% vs 46%)!!

 

4. as an aside, the ACCORD trial, which looked at tighter vs less tight control of diabetes, found that there was increased mortality in the tighter control group.  but it turns out that in the tighter control group, there was an increase in the use of TZDs from 58 to 91%, almost all rosiglit....

so, seems pretty clear to me that if you are going to use a TZD, pioglitazone is the only one to use....  it has much better effect on lipids and the PROACTIVE study showed clinical cardiovasc benefit, esp given the dearth of diabetes drugs leading to decrease in macrovascular events (ie, metformin really is the only med with clearcut evidence). both pioglit and rosiglit do lead to more significant heart failure, so should be used cautiously.