Questions Raised About Invokana Label Expansion For CV Risk Reduction

It hasn’t received a lot of media attention but on Tuesday the FDA approved an expansion of the canagliflozin (Invokana, Janssen) label to include a reduction in the risk of heart attack, stroke, or CV death in adults with type 2 diabetes and established CV disease.

Canagliflozin, an oral SGLT2 inhibitor, thus becomes the third diabetes drug approved for CV risk reduction in this population. Another SGLT2 inhibitor, empagliflozin (Jardiance, Lilly and Boehringer Ingelhiem) is approved for CV death reduction while liraglutide (Victoza, Novo Nordisk), an injectable GLP-1 RA, is approved for the reduction of major adverse cardiovascular events.

I’m not now regularly updating CardioBrief (I may resume next year) but I received an anonymous message raising questions about the approval that prompted me to ask Sanjay Kaul (Cedars Sinai) for additional perspective. I thought my readers would be interested in these messages.

Here is the original message I received:

“Off the record:  Canagliflozin gets FDA approval to claim MACE risk reduction according to company press release.  The basis for approval was a meta-analysis of two negative trials that differed somewhat in design, one of which (CANVAS) was unblinded.   Interim results for this trial were presented to the public in the 2012 FDA Endocrinologic and Metabolic Drugs Advisory Committee.  Is this a new bar for substantial evidence for a CV risk reduction claim at FDA?  How was bias excluded for anything other than mortality?  Is this the first time FDA grants a CV risk reduction claim based on a meta-analysis?”

Here is Sanjay Kaul’s response:

Long story!

Because of the potential adverse impact of public disclosure of the interim analysis results of CANVAS Cohort A study during the AdCom panel in January 2013, FDA considered two options, either reject the CANVAS Cohort-A results and ask the sponsor to conduct a new large and adequately powered trial to address CV safety (rule out 1.3 risk margin) or to borrow ‘valid’ unbiased data from CANVAS Cohort-A and combine with a medium-sized CANVAS Cohort A-like new CVOT. The pooling was to be done in a blinded fashion and all trial conduct issues were to be ensured to be high quality and standardized. Thus, because of the extenuating circumstances related to the public release of results from interim analysis of CANVAS Cohort-A, the FDA accepted an integrated analysis plan whereby data from extended follow-up of CANVAS Cohort-A post-Nov 20, 2012 (date of last public disclosure to EMA) would be combined with new data from the CANVAS-R trial (originally designed to evaluate treatment impact on albuminuria) to address CV safety as assessed by exclusion of post-marketing HR of 1.30. One of the essential requirements for data integration was that the 2 trials be very similar in design and conduct; have similar primary and secondary endpoints and event adjudication (both trials were to be adjudicated by a common CEC). This was done to minimize bias and maximize statistical power to allow reliable estimation of CV, kidney and safety outcomes. The prespecified statistical hierarchical testing strategy, jointly agreed by the FDA and the investigators, included a stepwise evaluation of the endpoints. Of note, superiority for MACE was not prespecified in the testing sequence, even though the investigators stated in the Statistical Analysis Plan (SAP) that if the null hypothesis was rejected and the upper bound of the HR is <1.0, it would be concluded that canagliflozin was superior to placebo. Clearly this was an accommodation (exception to the rule) by the FDA because of the extenuating circumstances.

Personally, I can argue against approval for CV efficacy indication, because superiority for MACE was not prespecified in the hierarchical testing plan; effect size was not c/w substantial effectiveness criterion; and in sensitivity analyses, 4 out of 6 analyses failed to be significant. I am generally supportive of FDA decisions, but there are legitimate reasons for my disagreement in this case!  Regardless of the FDA-approved indication, I would argue that the amputation hazard renders the benefit-risk balance undesirable relative to empagliflozin. I don’t see the expanded label having a material impact on the downward trajectory on canagliflozin sales.

Hope this helps!




  1. Is that the one that makes your genitals rot and drop off and increases lower limb amputations? Not sure I wouldn’t prefer the heart attack.

    Oh wait, I don’t eat many carbs so maybe I can have the best of both worlds . . .

Speak Your Mind