Landmark CREDENCE full results: J&J’s Invokana confers 30% risk reduction on primary renal endpoint without amputation signal – April 15, 2019

Executive Highlights

  • Just-released results for J&J’s CREDENCE renal outcomes trial (n=4,401) of Invokana in type 2 diabetes and CKD show that the SGLT-2 inhibitor drove a 30% relative risk reduction (95% CI: 0.59-0.82, p=0.00001!) on the primary renal endpoint of ESKD, doubling of serum creatinine, and renal or CV death (NEJM paper and editorial, press release, WCN presentation). Results were also very positive for secondary endpoints of three-point MACE (21% RRR, 95% CI: 0.67-0.95, p=0.01) and the composite of CV death and hospitalization for heart failure (31% RRR, 95% CI: 0.57-0.83, p<0.001). See Table for summary of all efficacy and safety data.  

    • We highly recommend watching the full presentation at the World Congress of Nephrology in Melbourne, Australia by Drs. George Bakris, Meg Jardine, Vlado Perkovic, Kenneth Mahaffey, David Wheeler.
    • Based on these results, investigators estimated that, if used in 1,000 people similar to the population of CREDENCE for 2.5 years, canagliflozin would prevent 47 occurrences of the study’s primary outcome (see above), with a number needed to treat (NNT) of 22 (95% CI: 15-38). Blood pressure therapy losartan, which was the last drug to be approved for CKD in 2000 had a very similar estimated NNT for the prevention of ESKD – 17 over 3.4 years.

    • The NNT drops to 16 when only considering the population with an eGFR between 30 and 45 ml/min/1.73 m2the group which Dr. Wheeler suggested nephrologists are understandably most concerned with. As Drs. Ingelfinger and Rosen put it in the accompanying editorial, “such [NNTs] are certain to be welcomed by patients with diabetes and chronic kidney disease and by the clinicians who treat them.”

    • At baseline, mean eGFR was 56 ml (all had Stage 2 or 3 CKD), UACR 927 mg/g, BP 140/78 mmHg, and A1c 8.3%; 50% had established CVD. All were on background antihypertensives (ACEi or ARB) when tolerated, meaning the renal benefit is on top of standard-of-care for CKD. 

  • Importantly, there was no significant imbalance in amputations (HR=1.11, 95% CI: 0.79-1.56) – a very reassuring result in a high-risk population, particularly following the amputation signal in CANVAS (HR= 1.97, 95% CI: 1.41-2.75, p<0.001). A total of 133 lower-extremity amputations were collected for CREDENCE (70 on canagliflozin, 63 on placebo), with rates just over 1% of patients per year in both arms.

  • A nonsignificant 0.11% reduction in A1c was found with canagliflozin. Importantly, this suggests that the mechanism of canagliflozin’s renoprotection is likely to be independent of A1c, with authors proposing that it may come instead from a reduction in intraglomerular pressure. Both systolic and diastolic blood pressure also decreased relative to placebo, 2.38 mmHg and 1.44 mmHg, respectively – a seriously impressive feat on top of maximum-tolerated ACE inhibitors or ARBs.

  • Thought leader reaction to CREDENCE results has been highly positive, reflecting enthusiasm for the first therapeutic advance in treating CKD in type 2 diabetes in nearly two decades. Dr. Sanjay Kaul deemed these results to be a “major advance” while Dr. Dan Drucker characterized these findings are “transformative for the field.” While still expressing enthusiasm, Dr. John Buse reminds the community that issues of access are still critical. See below for a full section of key opinion leader commentary.

  • Last month, J&J submitted an sNDA to FDA requesting a novel renal indication for Invokana. This indication stands to broaden Invokana’s recommended use, from eGFR’s ≥45 down to ≥30 ml/min/1.73 m2, for the 100 mg dose. We’re impressed by the speed of this submission and would certainly expect it to be positive – this should open the door up to a broader patient group.

J&J just published full CREDENCE renal outcomes trial results for SGLT-2 inhibitor Invokana (canagliflozin) in NEJM (editorial and J&J press release), along with full presentation at the World Congress of Nephrology in Melbourne, Australia by Drs. George Bakris, Meg Jardine, Vlado Perkovic, Kenneth Mahaffey, David Wheeler. In our interpretation, these results are highly positive, confirming Invokana’s renal-protective capacity and impact on both heart failure and MACE outcomes, while also supporting safety – including amputation risk – in a high-risk population.

In our interpretation, these results are highly positive, confirming Invokana’s renal-protective capacity and impact on both heart failure and MACE outcomes, while also supporting safety – including amputation risk – in a high-risk population.

Here are the topline results of the trial in one table:

Key Thought Leader Quotes

  • “The data shared so far indicates compelling evidence that SGLT2 inhibitors for patients with type 2 diabetes and chronic kidney disease provides robust reno-protection in addition to cardiovascular benefits. Given the seriousness of chronic kidney disease for patients with type 2 diabetes, evolving new strategies such as SGLT 2Is to interdict renal compromise or failure address an unmet need in improving outcomes and quality of life.” – Dr. Will Cefalu (Chief Science, Medical and Mission Officer, ADA)
  • "The CREDENCE results are impressive and impactful, providing direct support for previously reported secondary observations of the cardiovascular outcomes trials completed to date across the class, moving the needle substantially toward supporting the use of SGLT2i’s in patients with type 2 diabetes with compromised kidney function. As has been published in a meta-analysis of 3 prior trials, and as was confirmed in CREDENCE, diabetic kidney disease development and progression is potently impacted by these therapies – virtually halting progression, and this observation seems pretty consistent across the spectrum of baseline eGFR from normal to moderately impaired kidney function that predominated the cohorts studied in the cardiovascular outcomes trials, and here in CREDENCE with even more advanced CKD. Importantly, in all of these trials, this is on the backdrop of highly prevalent use of ACEIs and ARBs and with overall very good blood pressure control – state of the art management for diabetic kidney disease.

    But similarly importantly beyond the kidney effects is the observation that the CV benefits in general, and heart failure benefits specifically, are exaggerated in those with more advanced CKD, despite the consistently observed waning of glycemic efficacy in that context. This is important for 2 reasons: a) underpins the likelihood that little to none of these benefits are driven by effects on glycemic control; and b) suggests that the product label contraindications for use in patients with moderate or worse kidney disease should be reconsidered. To this latter point, unlike almost every other drug with cautions and contraindications in the product label based on eGFR or creatinine clearance that are based on safety concerns due to lack of clearance of a medication, for the SGLT2i’s, the contraindications are based completely on the observed waning efficacy of glycemic effects and not on any safety concerns. So if cardiovascular and kidney outcomes are favorably affected across the spectrum of kidney function, then likely clinical use should follow.” – Dr. Darren McGuire (UT Southwestern Medical Center, Dallas, TX)

  • "I agree this study is big news since, although we strongly suspected SGLT2 inhibitors would have a reno-protective effect, we now have definitive evidence. Renal physicians are excited to see this given the need for drugs in this space. Hopefully, the criteria to enable use of these drugs in care will be rapidly expanded. The safety findings are also reassuring but in the context of smaller amputation event numbers, more data from forthcoming trials are needed to add further reassurance. Even so, the rationale to use these drugs is increasing. Many more guidelines will no doubt change as a result." – Dr. Naveed Sattar (University of Glasgow, UK)

  • I agree with the editorial that the importance of CREDENCE "cannot be overstated" for all of the obvious reasons. I congratulate the investigators and the sponsor for moving forward so quickly with a label addition, particularly given our lack of recent progress in DKD for type 2 diabetes. However, after just returning from the ACP Annual Meeting in Philadelphia, I am reminded by the vast majority of internists that few of their patients have access to the SGLT2 inhibitors and the GLP1 Receptor Agonists. Even though ADA and AACE guidelines put them (appropriately) high in the treatment algorithm, particularly for those patients with ASCVD and DKD, they simply are not accessible. The most common comment: ‘even when covered at a tier 2 level, the co-pays are not affordable'. This isn't like insulin in type 1 diabetes as appropriate patients for canagliflozin won't die without, but the stakes are still high. I am anxious to see the economic analysis from CREDENCE as too many who could benefit won't at this point in time. The other issue: is the renal benefit a class effect? At this time, we don't know.” – Dr. Irl Hirsch (University of Washington School of Medicine)

  • The findings represent a major advance in that for the first time we have a therapeutic intervention which can reduce renal and cardiovascular outcomes in patients with diabetic CKD on the background of RAAS inhibitors. The results also underscore the point that patient enrichment strategy is more likely to pay off rather than softening the criterion for success by including softer outcomes such as eGFR decline >40% which is not a validated outcome for nephroprotection. The placebo arm had the highest incidence of renal or cardiovascular outcomes and mortality among all SGLT2 inhibitor trials, which might explain the impressive and clinically important and statistically robust effect size of treatment intervention. It is somewhat reassuring to note that the incidence of fracture and specifically lower limb amputation is not higher with canagliflozin in CREDENCE despite 24% with h/o PAD (c/w 21% in CANVAS). However, I would have liked to see greater than 130 amputations events accrue in CREDENCE to get a more reliable estimate of risk. Remember, there were a total of 187 major or minor amputation events in the CANVAS program. In this regard, I question the justification for stopping the trial prematurely as it precludes an opportunity for amputation risk assessment. I am not a big fan of stopping trials prematurely (even with prespecified stopping rules) except for an overwhelming mortality benefit or emergence of an unacceptable safety signal or evidence of futility. Overall, this trial extends the portfolio of evidence in favor of SGLT2 inhibitors from CV outcomes to renal outcomes, which is a big deal!” – Dr. Sanjay Kaul (Cedars-Sinai Medical Center, Los Angeles, CA)

  • "Regarding the justification for the IDMC recommendation to stop CREDENCE early and to rebut some public comments of criticism that (expectedly) have arisen: As the chair of the IDMC, I can attest that the issue of amputation event accrual weighed very heavily in our considerations, not just at the time of the interim analysis but for the entire duration of trial data monitoring. At the interim analysis, alongside this and other myriad considerations of benefits and risk typical of IDMC deliberations, and beyond any pre-specified single-analysis stopping guidance, we recommended to stop the trial. At days’ end, we determined the pros of stopping the trial outweighed the cons, with the ethics of maintaining placebo control  with a robust superiority finding on the primary outcome and in the context of accumulating data from other trials about the kidney benefits of SGLT2 ihibitiors being key factors. I would make the same decision today and I am confident each member of this IDMC would say the same (and probably most if not all of the critics, faced with the totality of the data at the time).” – Dr. Darren McGuire (UT Southwestern Medical Center, Dallas, TX)
  • My view is that the CREDENCE study simply adds to the already strong and consistent data showing that SGLT2 inhibitors improve outcomes for those with diabetes and associated complications. I view the new data important for five particular reasons:

    • The previously reported increased risk for amputation conferred by canagliflozin was not seen in this study;

    • The agent improved outcomes for those with pre-existing kidney disease with GFRs as low as 30 mL/min/1.73 m2;

    • The early decline in glomerular filtration is more than made up over time (akin to the worsening retinopathy that can be observed when glycemia control is implemented);

    • The overall profound improvement in the most relevant endpoints- including all-cause mortality; and

    • While it is risky to say with certainty, the profound benefit for those with T2D included in the CREDENCE study (since individuals with T1D were excluded) may well apply to those with T1D and I believe the question must be squarely addressed considering the profound effect on the most important endpoints.” – Dr. David Harlan (UMass Memorial Health Care, Worcester, MA)
  • CREDENCE provides the first direct evidence of a nephro-protective effect of SGLT-2 inhibitors. This effect was already appreciated within cardiovascular outcome trials, but CREDENCE is the first one having renal outcomes as primary endpoint. Though the study shows a protective effect for type 2 diabetic patients with low eGFR (<60 ml/min x 1.73m2) already on ACE-inhibitors or angiotensin receptor blockers, an effect also was apparent in those with milder renal impairment (60-90 ml/min x 1.73m2). The implication is that the drug may not just reduce the progression of diabetic kidney disease but also delay in a substantial manner loss of GFR in patients with normal renal function. Of interest safety profile was satisfactory with no sign of imbalance for amputation and with only a small but apparent increase in DKA. If these data will be confirmed by other ongoing studies with SGLT2-inhibitors, the introduction of these compounds early in the natural history of the disease may become compelling. In the early stage of the disease SGT2-inhibition can provide full glucose lowering efficacy with no risk of hypoglycemia, favor body weight and blood pressure control while exerting a direct renal protection. In these patients safety should be of an even less concern as DKA may be less likely to occur in patients who retain better pancreatic function as it occurs in the early stage of the disease.” – Dr. Stefano Del Prato (University of Pisa, Pisa, Italy)
  • “These results are transformative for the field, and further enable personalized diabetes medicine in the clinic to prevent kidney complications, improve cardiovascular outcomes, and save lives, for people with type 2 diabetes.” – Dr. Daniel Drucker (University of Toronto, Toronto, Canada)

  • I eagerly awaited this data for two reasons: there is a huge unmet need for treatments to prevent progression of diabetic kidney disease. I also was relieved to see that the safety signals failed to demonstrate increase in amputations or fractures. The question I would have:  what specifically were the instructions given for foot care and when drug was discontinued in the presence of a concern about risk?” – Dr. Carol Wysham (University of Washington School of Medicine)

  • "CREDENCE is the first major clinical trial, that was statistically powered, to demonstrate further slowing of kidney disease progression in almost 20 years. This major advance for slowing time to renal replacement therapy with the SGLT 2 inhibitor canagliflozin is long overdue." – Dr. George Bakris (University of Chicago, Chicago, IL)
  • "I think this is a great study, showing for the first time a benefit of an SGLT2i on renal outcomes assessed as the primary endpoint. The results apply to a population with a wide range of eGFR, spanning to 90 to 30, and this is clinically important, even though all patients were macroalbuminuric, somewhat limiting the generalizabililty of the results. Of note, amputations were less of a concern than in CANVAS, and the fracture risk was not an issue at all. I think this study clearly shows that the renal benefit of canagliflozin is independent of glucose control since the major effects were on hard renal endpoint – proteinuria not even being part of the composite primary or secondary endpoints – and occurred in the presence of very small reductions in hyperglycemia (due to reduced eGFR). I would also note that patients were actually treated with a target HbA1c level around 8.0%!" – Dr. Francesco Giorgino (University of Bari Aldo Moro, Bari, Italy)

Renal Outcomes

Canagliflozin conferred a significant 30% relative risk reduction (HR=0.70, 95%: 0.59-0.82, p<0.0001) on the primary endpoint of time-to-first-occurrence of end-stage kidney disease (ESKD), doubling of serum creatinine, and renal or CV death. ESKD was defined as the need for renal replacement therapy, including chronic dialysis or renal transplant, and this finding was consistent across endpoint components and 15 prespecified subgroups. Breaking this down, canagliflozin demonstrated a 32% RRR for ESKD (HR=0.68, 95% CI: 0.54-0.85, p=0.0015), 28% RRR for dialysis, kidney transplantation, or renal death (HR=0.72, 95% CI: 0.54-97, p=NA), 22% RRR for CV death (HR=0.78, 95% CI: 0.61-1.00, p=0.05), and NS 17% RRR for death from any cause – wow! Nearly every single Kaplan-Meier curve for these endpoints showed early and consistent separation (see below).

It’s now well-established that SGLT-2 inhibitors can protect against eGFR decline. The eGFR curves over time display this trend perfectly (see below), with the estimated GFR slope decline lessened by 2.74 ml/min/1.73 m2 per year (95%CI 2.37-3.11) – a 60% reduction (-1.85 ml/min/1.73 m2 per year with canagliflozin vs. -4.59 with placebo).

Subgroup Analyses

There were no significant between-group differences for any renal event in CREDENCE, suggesting similar benefit regardless of baseline eGFR or UACR. While there did seem to be a trend toward greater benefit in those with a baseline UACR ≤1,000 (a lower risk population) compared to those >1,000 (see below), the same trend did not exist with eGFR (see below). Most importantly, the renal benefit remained in those with an eGFR between 30 and 45 ml, a population for which Invokana 100 mg (the dose investigated) is not yet recommended. This makes a compelling case for the indicated population of Invokana to be broadened to this population as part of any potential renal indication (see more below).

Patient Population

As a reminder, all 4,401 CREDENCE participants had type 2 diabetes, Stage 2 or 3 CKD (eGFR ≥30 to <90 ml/min/1.73 m2), and macroalbuminuria (UACR >300 to ≤5,000 mg/g). All were receiving maximum-tolerated ACE inhibitors or ARBs – meaning the risk reduction was on top of standard-of-care for CKD. In a conversation with our team, Dr. James List (Global Therapeutic Area Head, CVM, Janssen R&D) emphasized the renal effect with Invokana is “as large if not larger” than that seen with these antihypertensives. At baseline, mean eGFR was 56 ml, UACR 927 mg/g, BP 140/78, and A1c 8.3%; 50% had established CV disease, 15% known heart failure, and 5% prior amputations. This compares to ~66% with CVD (compared to 50% with CVD in CREDENCE) and a mean eGFR of 77 ml in CANVAS – so, overall a sicker population in CREDENCE. See the study design and baseline characteristics paper for more, and check out the figure below used to develop this point by several speakers at the presentation.

It’s difficult to understate the importance of these findings for people with diabetes, particularly those with diabetes that has progressed beyond what current interventions can provide. As thought leaders have pointed out, SGLT-2 inhibitors represent the first innovation in CKD in ~20 years. Slowing the progression of kidney disease is a tremendous win for patients and providers, particularly as data show the burden of CKD is increasing, with diabetes as the primary cause of renal failure. Per the CDC, CKD prevalence among adults with diabetes rose from 4.1% in 2006 to 7.2% in 2015 – we are uncertain if the 2006 rate was not properly assessed or if the growth is actually hypergrowth (nearly doubling in nine years is not normal growth).

CV and Metabolic Outcomes

  • In line with CANVAS results and canagliflozin’s cardioprotective indication for the 3-point MACE, the therapy conferred a 20% RRR (HR=0.80, 95% CI: 0.67-0.95, p=0.0121) in MI, stroke, and CV death. Perhaps this finding supports the thinking that MACE benefit with SGLT-2s is more pronounced in higher-risk cohorts – though only 50% of CREDENCE participants had known CVD (compared with 69% in CANVAS), CKD is considered a CVD equivalent. The RRR for CV death and hospitalization for heart failure was 31% (HR=0.69, 95% CI: 0.57-0.83, p=0.0001), with 39% RRR (HR=0.61, 95% CI: 0.47-0.80, p=0.0003) in heart failure alone. (This is particularly notable from our view given potential systems savings – see the clinical implications section for the number needed to treat.)

  • A nonsignificant 0.11% reduction in A1c vs. placebo was found with canagliflozin after 42 weeks, suggesting that the mechanism of canagliflozin’s renoprotection is likely to be independent of A1c. Authors propose that it may come instead from a reduction in intraglomerular pressure. This is crucial for the renally impaired population, such as that studied in CREDENCE, since the glycemic and weight benefits of SGLT-2 inhibitors are attenuated at lower eGFRs. Of course, A1c equipoise between the arms was intended in the study in order to isolate the renal and CV effects of canagliflozin.

  • Those treated with canagliflozin lost, on average, 0.88 kg (~2 lbs) more than those treated with placebo over the course of 42 weeks (-1.99 kg vs. -1.11 kg, respectively). Both systolic and diastolic blood pressure also decreased relative to placebo, 2.38 mmHg and 1.44 mmHg, respectively – a seriously impressive feat on top of maximum-tolerated ACE inhibitors or ARBs.

Safety Outcomes

  • Importantly, CREDENCE demonstrated no significant imbalance in amputations (HR=1.11, 95% CI: 0.79-1.56) or bone fractures (HR=0.98, 95% CI: 0.70-1.37). Diving deeper, Dr. List stated that there were slightly more major amputations in the placebo arm and numerically more minor amputations with canagliflozin. While no amputation is “minor” to a patient, we see this result as very encouraging; even if the point estimate could have been more neutral, this is undoubtedly a positive result, particularly given the CREDENCE population is very high-risk. After the near-doubling of amputation risk with canagliflozin vs. placebo in CANVAS (HR= 1.97, 95% CI: 1.41-2.75, p<0.001) resulted in a black box warning on the product label, these results should assuage concerns over risk in the renally-impaired population, particularly given that the magnitude of the cardiovascular benefit is so much higher in magnitude than the amputation risk. Researchers acknowledged that this difference in the two trials could be due to differing trial populations, protocols, or chance. Very notably, Dr. Wheeler pointed out in his clinical implications perspective on the results that the CREDENCE population was at a much higher risk for amputations compared to CANVAS – the rates of amputation in the former were 12.3% on canagliflozin (70/2200) and 11.2% (63/2197) on placebo vs. 6.3% (140/5790) on canagliflozin and 3.4% (47/4344) on placebo in CANVAS. Methodologically, Dr. List assured our team that special efforts were made to collect amputations in CREDENCE, as they can easily be masked in datasets as procedures rather than adverse events.

  • As to be expected with an SGLT inhibitor, diabetic ketoacidosis was more common in the canagliflozin arm (2.2 events per 1,000 patient years) than the placebo arm (0.2 events per 1,000 patient years), despite these rates overall being low (actually, nonsignificant).

Clinical Implications

  • Dr. List characterized CREDENCE findings as a breakthrough in CKD treatment. He explained how SGLT-2 Invokana (approved in early 2013) changed the trajectory of kidney disease in CREDENCE: “After you reached steady state with GFR, the slope of change was dramatically different. People on placebo had a decrease of a little over 4.5 ml/min every year, which is very typical of DKD. At steady state, the slope for those on Invokana decreased to <2 ml/min every year. For context, people without kidney disease typically have a decrease of about 1 ml/min each year as they age.” Clearly, it will be critical to decide what patients are at highest risk and then getting them on this SGLT-2 – it will obviously be important to work out who should get this therapy when. This will require global healthcare systems to invest in prevention – long-term savings (fewer people on dialysis and with kidney disease) that require shorter-term investment (actually paying for drugs that are not SFUs and metformin only).


Competitive Implications and Landscape

  • As a reminder, J&J just submitted an sNDA to FDA requesting a new indication for Invokana for reducing risk of the study’s primary endpoint. Currently, Invokana 100 mg – the dose investigated in CREDENCE – is contraindicated for those with eGFR <30 ml and not recommended below 45 ml; the 300 mg dose is not recommended below 60 ml. To our understanding, any approved indication would only be applied to the 100 mg dose of canagliflozin. However, CREDENCE enrolled down to an eGFR of 30 ml, and the results were concordant across subgroups of kidney disease based on eGFR and albuminuria. To this end, canagliflozin will certainly at least be considered for CKD across the entire investigated eGFR range, potentially expanding canagliflozin’s indicated population. 

  • As of October 2018, Invokana has an FDA-approved indication for reducing risk of three-point MACE in people with type 2 diabetes and established cardiovascular disease. Although this is speculation on our part, we believe that these results provide J&J with incredible reason to invest again in Invokana – we believe there should also be tremendous patient education on this front.

  • These results have us seriously excited for future renal outcomes trials with SGLT inhibitors and novel glucose lowering agents in general. Lilly/BI are investigating Jardiance (empagliflozin) in the EMPA-KIDNEY renal outcomes trial (n=5,000, with or without type 2, and including type 1) and AZ is investigating Farxiga (dapagliflozin) in the Dapa-CKD trial (n=4,000, with or without type 2). Expected completion dates are June 2022 and November 2020, respectively. Additionally, Novo Nordisk will conduct the FLOW renal and CV outcomes trial of Ozempic (semaglutide) in type 2 and CKD (n=3,120 to complete August 2024).

Close Concerns’ Questions

  • What are the main learnings from this trial from a health economics perspective?

  • What were the differences in A1c, weight loss, or amputations for those in CREDENCE?

  • Was CGM used in this trial and, if so, what did it show? (Obviously the point of the trial was to assess risk changes on the kidney front, but we are also interested in other learnings gleaned by CGM.)

  • What are expectations for FDA and other global regulatory agencies?

  • Do these results indicate class effects or is there something specific about the canagliflozin molecule?

  • From a clinician perspective, how should Invokana be prescribed in practice? For example, at what eGFR should an HCP start thinking about renal protection (and how does trajectory play into this)?

  • How often are clinicians already prescribing SGLT-2s for renal protection, and how is J&J going to promote that use?

  • Could more detailed treatment standards around SGLT-2s for CKD improve uptake of the class?

  • How meaningfully is J&J investing in Invokana (e.g., through patient discounts or promotion)?

  • Will these results assuage amputation concern from patients and providers?


-- by Ann Carracher, Peter Rentzepis, Martin Kurian, and Kelly Close