Is CREDENCE a Game Changer for Diabetic Nephropathy?

F. Perry Wilson, MD, MSCE


April 17, 2019

For nearly 20 years, diabetic nephropathy has been an albatross around nephrologists' collective necks. It's the most common ailment bringing patients across our thresholds, the one for which we have the fewest tools.

Where novel biomarker discoveries breathed new life into glomerular disease and advances in transplantation continue to extend graft function, the principal strategies for the management of diabetic nephropathy have remained unchanged for two decades: control blood pressure, control glucose, and block the angiotensin system.

But nephrologists are optimists. Secondary analyses in studies[1,2] of novel SGLT2 inhibitors, which act to increase urinary excretion of glucose, suggested that they slow progression of kidney disease.

But we've been burned by secondary analyses before. Large clinical trials rarely have significant numbers of patients with chronic kidney disease (CKD), and the CANVAS[1] and EMPA-REG[2] studies (which demonstrated efficacy of canagliflozin and empagliflozin, respectively) were no exception. The nephrology community waited for a definitive trial the way I waited on my prom night: hopeful, but prepared for disappointment.

Enter CREDENCE.[3] In the past 24 hours, the study has been described as "landmark," "breathtaking," and "game changing."

But has the game changed? In my lectures on medical studies, I remind learners that one study should never change practice. In the case of CREDENCE, I may have to eat my words.


CREDENCE was a randomized trial of 4401 patients with diabetic nephropathy, as defined by an eGFR of 30 to < 90 mL/min/1.73 m2 and modest proteinuria of > 300 mg per gram of creatinine. Let me stop here to mention what a feat it is to enroll this many patients with CKD in a clinical trial. An adequately powered study in the CKD population is a landmark already.

Critically, the CREDENCE protocol specified that all patients be on a stable dose of renin-angiotensin system (RAS) blockade. A less savvy or more unscrupulous study team may have tried to avoid the coadministration of angiotensin-converting enzyme (ACE) inhibitors (which have a known benefit in diabetic nephropathy) so as not to "drown out" any signal of benefit from the novel agent. They made the right choice to include a background of standard therapy here.

The primary outcome was the so-called "3 Ds": doubling of creatinine, dialysis, and death. While some may take issue with the composite nature of the outcome (is a doubling of creatinine really as bad as dying?), I don't. Powering for an all-cause mortality outcome alone is daunting, even for an industry-sponsored study like this one, but looking at doubling of creatinine alone risks undercounting those who die or initiate dialysis before the creatinine has a chance to double. In short, the primary outcome is good enough. Let's move on.

That outcome occurred in 11% of those treated with canagliflozin, compared with 15.5% in the placebo group—a relative reduction of 30%, which carried with it an impressive level of statistical significance (P < .0001). Rates of end-stage kidney disease were similarly significantly reduced. And although the difference in overall mortality (17% lower in the canagliflozin group) did not meet our traditional definition of statistical significance, this should hardly discourage use of the medication, given the effect on other outcomes that matter so much to patients.

Prior studies had suggested an increased risk for amputation among individuals treated with canagliflozin, leading the trialists here to mandate examination of patients' feet at each trial visit. Seventy participants in the canagliflozin group and 63 in the control group required amputations of some kind during the study—a nonsignificant difference that should provide some reassurance to both patients and physicians. Rates of acute kidney injury and urinary tract infection were also no higher in the canagliflozin group.

But there were significantly more genital mycotic (ie, yeast) infections in the canagliflozin group, though the rates were quite low (around 2% compared with 0.5% in the placebo group). Diabetic ketoacidosis was also seen more often in the canagliflozin group, with 11 episodes compared with just one in the placebo group. Clearly, canagliflozin has side effects, but they seem to be rare.

More Questions Than Answers

The drug seems to work, but the main unanswered question is why does it work?

By inhibiting the sodium-glucose cotransporter, SGLT2 inhibitors have multiple physiologic effects. They promote glucosuria, which leads to osmotic diuresis, lowering blood pressure. By increasing glucose excretion, they may improve glucose control. They may induce weight loss, reducing glomerular hyperfiltration. Or they may work through as-yet undefined mechanisms.

Trolling through the supplemental data leaves more questions than answers. A1c levels were barely different in the two groups when the study ended, arguing against improved diabetic control as the mechanism of benefit. Systolic blood pressure was lower (by around 2.5 mm Hg) in the canagliflozin group, but if such modest reductions in blood pressure were this protective in terms of renal and cardiovascular outcomes, we'd be awash in useful therapies for diabetic nephropathy.

Those treated with canagliflozin did lose more weight, though—about 1 kg. It's not much, but weight loss is a tough thing to achieve. Finally, canagliflozin significantly reduced proteinuria compared with placebo. It is worth noting that the only other class of medications with proven efficacy at reducing rates of progression in diabetic nephropathy (RAS inhibitors) also significantly reduce protein excretion.

Eating My Words

CREDENCE should change practice. There—I wrote it.

Yes, it is one study. But it is a study that fits firmly into an emerging understanding of the role of SGLT2 inhibitors in diabetes and is bolstered by secondary analyses of previous trials. The sheer magnitude of effect, coupled with the relatively small adverse event rate, suggests that we should reach for SGLT2 inhibitors quickly and fearlessly in diabetic nephropathy.

But not first. All patients in this study had adequate RAS blockade. And cheap, reliable ACE inhibitors and their cousins should still be the first line of defense against progression of this disease. After all, they brought us to this dance two decades ago.

Follow Medscape on Facebook, Twitter, Instagram, and YouTube


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.
Post as: