CANVAS: Straight Talk Needed on Amputation Risk

Harpreet S. Bajaj, MD, MPH


October 16, 2017

Since the US Food and Drug Administration announced its black-box warning earlier this year for canagliflozin, many of us have been concerned about the amputation risk observed with this drug in the CANVAS trial (hazard ratio [HR], 1.97 compared with placebo).[1] Despite the overall positive cardiovascular, heart failure, and kidney disease-related benefits seen with canagliflozin, a key clinical concern among prescribing clinicians such as me relates to predictive factors for amputations, and whether this risk correlates with the other treatment-related adverse effects observed (eg, volume depletion and hemoconcentration).

Harpreet S. Bajaj, MD, MPH

I expected further data clarifications surrounding this issue at the 1-hour CANVAS session[2] at the European Association for the Study of Diabetes (EASD) 2017 meeting, held in September in Lisbon. But I was disappointed; no new data related to amputations from the CANVAS program were shared.

During the presentation, the discussants did include a prepublication meta-analysis of the CANVAS and EMPA-REG OUTCOMES trial programs. As a reminder, a previously published post hoc analysis from the latter cardiovascular outcomes trial had suggested no increase in amputations with empagliflozin.[3] The new meta-analysis slides[2] show an increased risk for amputation with combined data from the two sodium/glucose cotransporter 2 (SGLT2) inhibitors (canagliflozin and empagliflozin), with a significant HR of 1.44.

So, how are we supposed to interpret this discrepancy?

Given the neutrality of amputations with empagliflozin in EMPA-REG OUTCOMES (HR, 1.02), it is intuitive to conclude that the harm in this new combined meta-analysis is entirely derived from the CANVAS program results. This is also supported by the large I 2 statistic (with a significant P value of 0.01), suggesting that the variation across the two study programs is probably due to their true heterogeneity rather than to chance.

In my humble opinion, this unambiguous proof of heterogeneity raises a conflict of interest issue: What was the motive behind performing such a meta-analysis in the first place, with subsequent inclusion in the EASD presentation of CANVAS? I won't even bother going into details of the observational data presented in Lisbon, or elsewhere, which obscures canagliflozin's link to amputation, because such "real-world" data—with numerous biases and confounding—do not stand up to the rigor and design of the randomized CANVAS trial.

To reassure us of the benefit/harm balance with canagliflozin, we clinicians need more data-mining from CANVAS and additional long-term randomized controlled trials. One such trial is the ongoing CREDENCE trial with canagliflozin, in which a more careful and comprehensive prospective assessment of foot complications is mandated by a recent protocol amendment.

Until those results become available (expected in 2019-2020), it may be prudent for the CANVAS publication committee to share a post hoc mediation analysis that explains their findings. In particular, such an investigation should look at whether there is an association of amputations with the reported volume depletion and/or hemoconcentration observed with canagliflozin during CANVAS.

Moreover, to clarify whether the amputation harm could be a class effect of SGLT2 inhibitors, a thorough collection of prospective data on amputations and other foot-related complications should be required of all ongoing and future long-term trials within the class (eg, DAPA-CKD, DAPA-HF, and DECLARE with dapagliflozin; EMPEROR-Preserved and EMPEROR-Reduced with empagliflozin; and VERTIS CV with ertugliflozin).


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.