Canagliflozin heart and kidney trials (CANVAS)

Overview

Neal B et al.. N Engl J Med 2017 · PMID 28605608 · DOI 10.1056/NEJMoa1611925

Fewer heart attacks, strokes and heart deaths (-14%), but more toe/foot amputations.

Does the study support the specific claim? Partly · Methodological quality: high

Original publication

Plain-language summary

Scientific in-depth review

1. Full reference

Authors

Neal B, Perkovic V, Mahaffey KW, de Zeeuw D, Fulcher G, Erondu N, Shaw W, Law G, Desai M, Matthews DR, CANVAS Program Collaborative Group

Journal

N Engl J Med

Year

2017

Identifier

PMID 28605608 DOI 10.1056/NEJMoa1611925

Verification

verified – PubMed E-utilities efetch, 2026-10-25

Retracted

no

Corrections

none

Extent checked

full text

Conflicts of interest and funding

Sponsored by Janssen Research and Development (canagliflozin manufacturer), conducted with academic steering committee and George Clinical; analyses carried out independently by sponsor and George Clinical; four authors Janssen employees; sponsor-funded medical writing (MedErgy). Individual disclosures in online forms.

2. Research question

CV, renal and safety outcomes of canagliflozin

3. Study design and population

Study design

Randomised controlled trial – Integrated analysis of two double-blind, placebo-controlled randomised trials (CANVAS NCT01032629, CANVAS-R NCT01989754); primary non-inferiority then superiority test with sequential hierarchical testing; exploratory outcomes with nominal CIs.

Population

Type 2 diabetes (HbA1c 7.0-10.5%) aged >=30 with symptomatic atherosclerotic CVD or >=50 with >=2 CV risk factors; eGFR >30. Mean age 63.3, 35.8% women, diabetes duration 13.5 years, 65.6% prior CVD, 14.4% heart failure, 2.3% prior amputation.

Country

667 centres in 30 countries

Period

CANVAS from December 2009, CANVAS-R from 2014; joint close-out February 2017; mean follow-up 188.2 weeks (median 126.1).

Sample size

10,142 (canagliflozin 5,795; placebo 4,347); 96.0% completed; vital status 99.6%; ~29-30% discontinued study drug prematurely.

Inclusion and exclusion criteria

As above; 2-week single-blind placebo run-in; detailed criteria in supplement (Table S1).

4. Methods and endpoints

Exposure / intervention

Canagliflozin 100 mg or 300 mg daily (CANVAS 1:1:1; CANVAS-R 100 mg with optional increase to 300 mg; 71.4% increased).

Comparator

Placebo

Endpoints

Primary: CV death, non-fatal MI or non-fatal stroke (adjudicated). Secondary (sequential): all-cause death, CV death, albuminuria progression, CV death or HF hospitalisation. Exploratory renal composite; safety incl. amputation, fracture, DKA.

Statistical methods

Cox regression stratified by trial and CVD history (ITT for efficacy); non-inferiority margin 1.3; on-treatment analysis for most safety outcomes, all-time analysis for amputation, fracture, cancer, DKA; multiple imputation sensitivity.

5. Actual results

Results

Primary outcome 26.9 vs 31.5 per 1,000 patient-years, HR 0.86 (95% CI 0.75-0.97; P<0.001 non-inferiority, P=0.02 superiority); with imputation 0.85 (0.75-0.97). All-cause death HR 0.87 (0.74-1.01; not significant, testing stopped); CV death 0.87 (0.72-1.06); non-fatal MI 0.85 (0.69-1.05); non-fatal stroke 0.90 (0.71-1.15); HF hospitalisation 0.67 (0.52-0.87); albuminuria progression 0.73 (0.67-0.79); renal composite 5.5 vs 9.0 per 1,000 PY, HR 0.60 (0.47-0.77) (exploratory). Safety: serious adverse events 104.3 vs 120.0 per 1,000 PY (HR 0.93, 0.87-1.00); amputation 6.3 vs 3.4 per 1,000 PY, HR 1.97 (1.41-2.75), 71% highest at toe/metatarsal; all fractures 15.4 vs 11.9, HR 1.26 (1.04-1.52; heterogeneous between trials); DKA 0.6 vs 0.3, HR 2.33 (0.76-7.17); male genital infection 34.9 vs 10.8; no excess hypoglycaemia, AKI, pancreatitis, cancer or VTE.

Absolute risks

Primary outcome 26.9 vs 31.5 per 1,000 patient-years (4.6 fewer); amputation 6.3 vs 3.4 per 1,000 patient-years (2.9 more); renal composite 5.5 vs 9.0; HF hospitalisation 5.5 vs 8.7; fractures 15.4 vs 11.9 per 1,000 patient-years.

Authors' conclusion

Canagliflozin lowered the risk of CV events vs placebo but increased the risk of amputation, primarily at the toe or metatarsal level; renal benefits possible but not formally significant; care warranted in patients at risk for amputation.

6. Strengths

• Large double-blind placebo-controlled RCTs with ~3.6 years mean follow-up and 99.6% vital status
• Adjudicated CV and renal outcomes and selected safety outcomes
• Pre-specified hierarchical testing; ITT

7. Weaknesses

• Sponsor-designed randomisation schedule, sponsor co-analysis and employee authors
• Integrated analysis of two trials partly necessitated by unmasked interim data used in regulatory filing (CANVAS)
• Moderate event numbers; many exploratory outcomes (false-positive risk acknowledged)
• Amputation excess a new finding with unknown mechanism; fracture finding inconsistent between trials
• ~30% premature discontinuation; more additional glucose-lowering drugs in placebo arm

8. Risk of bias

Orienting tool

RoB 2

Type of assessment

structured assessment based on the full text by a single reviewer (no formal double review)

Full formal assessment

no

Overall risk of bias

some concerns

Rationale

Single-reviewer RoB 2-oriented judgement from the full text; formal assessment needs two independent reviewers and the protocols. Changed from 'low' to 'some_concerns' because the full text documents the post-hoc integration of two trials following unmasked interim data and extensive sponsor involvement; primary efficacy and amputation results themselves are at low risk of bias.

Full-text review (25 October 2026): Population details, follow-up/discontinuation, all secondary/exploratory HRs (HF 0.67, deaths n.s.), full safety table (fractures, DKA, genital infections, no excess cancer) and sponsor role added; integration rationale (unmasked interim data) documented; RoB 2 overall low -> some_concerns; noted amputations also on placebo (relevant to C-ODD-06).

9. Assessment of the statistical interpretation

Benefit (4.6 fewer major CV events per 1,000 PY) and harm (2.9 more amputations per 1,000 PY, mostly toe/metatarsal) of similar absolute magnitude; mortality reductions not statistically significant; renal and HF benefits exploratory. Amputations also occurred under placebo (3.4 per 1,000 PY).

10. Relevance to the specific claim

Supports the safety-signal part of C-ODD-05 (amputation, fracture); contradicts C-ODD-06 (net harm; amputations also occur on placebo).

11. Independent contrary findings

I-ODD-05.

12. Transferability to Germany and Europe

Multinational/US/UK data; transferable to EU practice.

13. Overall methodological judgement

High-quality RCT programme showing both a CV benefit and an amputation (and fracture) signal for canagliflozin; cited one-sidedly by the archive. Contradicts claims that glucose-lowering drugs are net harmful or that amputations occur only with diabetes drugs.

14. Open questions

None beyond those stated.

15. Linked fact checks

• Every major class of glucose-lowering drug (metformin, sulfonylureas, thiazolidinediones, DPP-4 inhibitors, SGLT2 inhibitors, GLP-1 agonists) carries a documented signal of serious harm in the peer-reviewed literature. – Studies supporting the claim
• Glucose-lowering drugs are net harmful and belong among the 'forbidden' items; dapagliflozin (Forxiga) is among the 'dirtiest' drugs and kills the brain; leg amputations happen only to people taking diabetes drugs and insulin. – Studies contradicting or limiting the claim

Use by the Tayebat archive

Note: according to the archive, the studies were compiled by the archive operators. Whether Dr. Elawady himself cited this study is not established.

What the archive says about the study

CANVAS Program (10,142 patients): canagliflozin nearly doubled lower-limb amputations (HR 1.97). (AR adds FDA black-box warning.)

A. What was actually studied?

CV, renal and safety outcomes of canagliflozin

B. What did the authors find?

HR 0.86 primary; HR 1.97 amputation.

C. What conclusions do the authors themselves draw?

Lower risk of CV events but greater risk of amputation, primarily toe or metatarsal.

D. What does the archive derive from it?

Used as proof of SGLT2 harm.

E. Is this derivation scientifically justified?

Amputation numbers correct; the archive omits the primary finding of the same paper (14% fewer major CV events), the lower serious adverse event rate, and that amputations also occurred in the placebo group (3.4 per 1,000 patient-years).

Adequacy of the interpretation (by the archive)

taken out of context