CANVAS - Canagliflozin to prevent CV & renal events in T2DM

Neal B, et al. Canagliflozin and cardiovascular and renal events in type 2 diabetes. NEJM 2017

Bottom line:

• In patients with T2DM and ASCVD (or multiple CV risk factors), canagliflozin (at either dose of 100 mg or 300 mg daily) reduced the risk of a composite of CV events over 3.6 years (NNT 200/year);

• But also increased the risk of several important adverse effects, including amputation (NNH ~330/year), fractures (NNH 330/year), genital infections (NNH 42/year for men & 20/year for women), & hypovolemia (NNH 125/year).

Patients (n=10,142)

• Included
  • Adults with T2DM (A1c 7.0-10.5%)
  • Either
    • 30+ y/o + symptomatic ASCVD
    • 50+ y/o + at least 2 risk factors (diabetes duration 10+ y, current smoking, SBP >140 mmHg on at least 1 antihypertensive drug, micro/macroalbuminuria, HDL <1.0 mmol/L)
  • eGFR at least 30
• Baseline characteristics
  • 63 y
  • Diabetes duration 13.5 y
  • ASCVD 66%: Cerebrovascular 19%, CAD 56%, peripheral 21%
    • HF 14%
  • Microvascular disease: Retinopathy 21%, nephropathy 18%, neuropathy 31%
  • Previous amputation 2.3%
  • Labs
    • A1c 8.2%
    • Total cholesterol 4.4
    • triglyceride 2.0
    • HDL 1.2
    • LDL 2.3
    • eGFR 76, macroalbuminuria 8%, microalbuminuria 23%
  • Meds
    • Antihyperglycemics: Metformin 77%, sulfonylurea 43%, DPP-4 inhibitor 12%, insulin 50%

Interventions

• I: Canagliflozin 100 or 300 mg/d
  • CANVAS trial: Randomized to either canagliflozin 100 or 300 mg/d
  • CANVAS-R: Randomized to canagliflozin 100 mg/d, could be uptitrated after 12 weeks to 300 mg/d (~70% uptitrated to 300 mg/d)
• C: Placebo
• A1c ~0.6% lower with canagliflozin (both doses combined) vs placebo
• ~29% in each group discontinued treatment prematurely

Outcomes @ mean 3.6 years (median 2.4 years)

View fullsize

Generalizability & caveats

• P: Enrolled patients had longstanding type 2 diabetes with either symptomatic ASCVD or multiple CV risk factors and A1c >7% despite multiple antihyperglycemic agents
• I/C: Patients were not randomized to A1c goal & non-SGLT2i antihyperglycemic agents were not limited
• O:
  • Consistent reduction in all components of the primary CV endpoint
  • Rates for individual renal outcomes not provided (uncertain whether renal deaths & dialysis rates were decreased, or if driven by '40% reduction in eGFR' component)

Internal validity

• Low risk of allocation, performance, detection & attrition bias
  • Computer-generated randomization with randomly-permuted blocks
  • Randomization by centralized web system
  • Participants & trial staff blinded
  • Intention-to-treat analysis & 4% loss-to-follow-up
• Note: 2-week single-blind placebo run-in phase