Have the Blockbuster Diabetes Drug Trials Been Biased?

Miriam E. Tucker

January 15, 2020

Imbalances in glycemic control, blood pressure, and diuretic use between treatment and placebo arms could have biased the cardiovascular and renal outcomes of recent large trials in favor of the study drugs for treating type 2 diabetes, some experts assert.

The cardiovascular outcomes trials (CVOTs) were mandated by the US Food and Drug Administration in 2008 to ensure the safety of newer agents being developed for type 2 diabetes following the debacle of rosiglitazone.

Results from some of the CVOTs and other subsequent dedicated trials showing cardiovascular and renal benefits have influenced clinical guidelines for type 2 diabetes and led to FDA approval of additional indications for some of the drugs beyond glucose lowering.

In nearly all these manufacturer-funded trials of a number of drug classes — dipeptidyl peptidase-4 (DPP-4) inhibitors, glucagon-like peptide 1 (GLP-1) agonists, and sodium-glucose cotransporter 2 (SGLT2) inhibitors — glycemia and blood pressure were not strictly controlled but were left to the discretion of the treating physician, although some studies did include "rescue criteria."

Now in an article recently published online in the Journal of Pharmaceutical Policy and Practice, Japanese researchers Rumiko Shimazawa, PhD, of the Department of Clinical Pharmacology at Tokai University School of Medicine, and Masayuki Ikeda, MD, point out that A1c levels were significantly higher in placebo groups than in treatment groups in all of the CVOTs.

Those imbalances, they argue, placed patients in the placebo groups at potentially higher cardiovascular risk and thereby biased the results in favor of the study drug.

"Reanalysis with adjustment for the [A1c] imbalance is absolutely indispensable for the correct evaluation of the CVOTs," Ikeda, of the Department of Medical Informatics at Kagawa University Hospital, Japan — who had no disclosures — told Medscape Medical News.

Similar views were expressed in 2018 by former Bristol-Myers Squibb investigators Simeon I. Taylor, MD, PhD, an endocrinologist now at the University of Maryland, Baltimore, and nephrologist Bruce R. Leslie, MD, now of Seventh Doctor Consulting, Princeton, New Jersey.

Taylor and Leslie additionally point out that blood pressures were also imbalanced between the CVOT study arms.

And, Leslie told Medscape Medical News in an interview, those same imbalances — as well as in diuretic use — also occurred in more recent dedicated trials of the effect of SGLT2 inhibitors on kidney function and heart failure, including CREDENCE and DAPA-HF.

"The imbalance is baked into how these studies are done. Whether intentional or inadvertent, there's an imbalance," asserts Leslie, who owns stock in Bristol-Myers Squibb, Pfizer, and Lilly.

Any Difference Is "Subtle" and Unlikely to Account for Benefits

Asked for comment, Silvio Inzucchi, MD, director of the Yale Medicine Diabetes Center, New Haven, Connecticut, and a senior investigator for several of these trials, told Medscape Medical News: "It is extremely difficult to conduct a trial with absolutely equal A1c levels between the treatment groups when you allow an extra drug in one arm."

"So, all of the CVOTs have shown about a 0.4% to 0.7% difference [in A1c] between the groups, sometimes even more depending on the potency of the drug. To have equal A1cs in both groups, the study sites would have to assume complete responsibility for glucose management. That would be a much more complex and much more expensive study...It's also no longer a reflection of 'real-world' practice," he explained.  

And in response to similar arguments about the imbalances made in a letter to the New England Journal of Medicine following publication of the renal results of the EMPA-REG Outcome trial with the SGLT2 inhibitor empagliflozin (Jardiance, Lilly/Boehringer Ingelheim), Inzucchi and two other EMPA-REG coauthors called the differences in glycemic and blood pressure control "subtle."

They write, "Treatment with SGLT2 inhibitors results in a reduction in hyperglycemia and blood pressure, and these effects may indeed have contributed to the improved outcome with empagliflozin."

"However, the magnitude and duration of the observed reductions are unlikely to fully account for the positive renal effects...it is more likely that the effects of empagliflozin on reducing intraglomerular hypertension played a more fundamental role than glycemic or hypertension control in mediating the renal effects," they state.

It Shouldn't Be Difficult to Conduct Post-Hoc Adjustments of A1c

In their article, Shimazawa and Ikeda analyzed results from 12 CVOTs published through December 2018 that followed the FDA's 2008 guidance.

These included three studies of SGLT2 inhibitors (EMPA-REG OUTCOME, CANVAS, and DECLARE-TIMI 58), four of DPP-4 inhibitors (CARMELINA, EXAMINE, SAVOR-TIMI 53, and TECOS), and five of GLP-1 agonists (LEADER, SUSTAIN-6, HARMONY, EXCEL, and ELIXA).  

In most of the trials, patients had a high risk of atherosclerotic cardiovascular disease (CVD) or established CVD with baseline A1c levels ranging from 7.2% to 8.7%.

All received active drug or placebo, but they weren't truly "placebo-controlled" trials, as additional glucose-lowering medications were allowed, Shimazawa and Ikeda point out. 

There was significantly greater use of additional glucose-lowering drugs in the placebo groups of the 10 trials that reported such data.

But regardless of use of such additional medications, A1c levels were significantly higher in the placebo groups in all the trials, ranging in percentage point difference from 0.27 (in ELIXA) to 1.00 (in SUSTAIN-6).

And despite better glycemic control in the treatment groups, heart failure rates were higher in the treatment groups in EXAMINE and SAVOR-TIMI, leading to warnings regarding this on the labels of two DPP-4 inhibitors.

Ikeda told Medscape Medical News that it shouldn't be difficult to resolve the imbalance problem by adjusting for A1c, as the CVOT investigators "have the critical data of their own, and the post-hoc analyses with adjustment for the imbalance are elementary statistics."

In fact, he noted that this was actually done in one of the CVOTs, ELIXA, resulting in a loss of a significant advantage for lixisenatide in percent change in urinary albumin-to-creatinine ratio (from P = .004 to P = .07). 

Are SGLT2 Inhibitors Merely More Expensive Diuretics?

Leslie is less convinced that the imbalance in glycemic control would have made a major difference in cardiovascular outcomes, at least in the short-term.

"The duration of studies is relatively short. For 3- to 5-year follow-up it seems unlikely that differences in glycemic control can explain the cardiovascular benefit," said Leslie.

However, regarding the CVOTs and other major trials of SGLT2 inhibitors, Leslie said, "My belief is that the difference in outcomes is mostly due to blood pressure difference and diuretic use imbalance, which are intimately related."

He points to evidence — including some of his own work — that SGLT2 inhibitors have diuretic properties and that they enhance the renoprotective effects of renin–angiotensin–aldosterone system (RAAS) inhibitors by potentiating their antihypertensive and antiproteinuric actions.

Indeed, in a letter to the New England Journal of Medicine following publication of CREDENCE, which showed renal benefit for the SGLT2 inhibitor, Leslie and coauthor Leslie E. Gerwin, JD, of Princeton University, New Jersey, write: "In this trial, canagliflozin — a drug with diuretic properties — was administered to patients with diabetic kidney disease, nearly all of whom were receiving a [RAAS] inhibitor."

"In the placebo group, however, fewer than half the patients were taking diuretics," they point out.

There was also a blood pressure imbalance in CREDENCE of 3.30 mmHg (systolic) and 0.95 mmHg (diastolic).

Leslie told Medscape Medical News that the same is true of the CVOTs of SGLT2 inhibitors, including CANVAS, EMPA-REG, and DECLARE TIMI 58, potentially influencing the heart failure outcomes.

"It was the same structure. Less than half of the placebo group was being treated with a diuretic at baseline, but all the treatment group patients got a diuretic [as well as] an SGLT2 inhibitor along with RAAS inhibitors," noted Leslie.

Thus, he said, "all the SGLT2 inhibitor CVOTs, as well as CREDENCE, contain an unbalanced therapeutic design...that leaves unanswered the question of whether the cardiovascular and renal benefits they describe can be reproduced by inexpensive generic thiazide diuretics."

SGLT2 Inhibitors Are Unique

In response to Leslie and Gerwin's letter, CREDENCE lead investigator Meg J. Jardine, MB, PhD, of The George Institute for Global Health, Sydney, Australia, and two coauthors replied: "Diuretics have not been shown to prevent kidney failure."

"The benefits observed in the CREDENCE trial were also consistent, regardless of baseline diuretic use, so we think it is unlikely that the diuretic effect explains the benefits of canagliflozin," they note.

Leslie commented, "Diuretics don't prevent kidney failure, but neither do SGLT2 inhibitors. They just slow it down, same as diuretics."

Inzucchi, who has multiple disclosures relating to the companies conducting these trials, told Medscape Medical News he disagrees with Leslie's assertion that the diuretic effects of SGLT2 inhibitors are the same as those of thiazide diuretics.

"I don't agree that SGLT2 inhibitors are 'just like thiazides.' They work in a totally different part of the nephron, and although they are relatively weak natriuretics, their effect on sodium excretion may be more sustained than with other diuretics," he said.

"This is perhaps because they inhibit sodium reabsorption proximal to the macula densa, so the resultant loss of urinary sodium — and subsequent volume contraction — does not appear to simulate the normal neurohormonal compensatory mechanisms like conventional diuretics that serve to attenuate efficacy over time. These hormonal changes — increases in catecholamines, renin, aldosterone, and antidiuretic hormone — may also have deleterious effects on the heart," he explained.

In addition, Inzucchi said, "Thiazides have never been shown to reduce heart failure hospitalizations or mortality as do the SGLT2 inhibitors. So the gliflozins may be unique diuretics."

And in response to another letter expressing concern about the glycemia and blood pressure differences in CREDENCE, Jardine and colleagues write: "Pooled analyses of intensive blood pressure and glucose lowering have not shown clear renal benefits, so these are also unlikely explanations, particularly given the modest differences between the two groups."

"The trial protocol encouraged investigators to deliver the best guideline-based care to patients according to blood pressure and glucose and lipid levels. None of these interventions (ie, the use of diuretics and intensive blood pressure and glucose lowering) has been shown to have benefits of the magnitude observed in the CREDENCE trial, despite multiple trials," they state.

In their letter regarding CREDENCE to the New England Journal of Medicine, Leslie and Gerwin suggest a clinical trial could be conducted comparing canagliflozin added to RAAS inhibition with a generic thiazide diuretic added to RAAS inhibition in patients with diabetic kidney disease and otherwise controlled hyperglycemia.

This, they argue, "could help to determine whether the renoprotective qualities of canagliflozin are anything more than those of an expensive diuretic."

And, as for the CVOTs, Leslie told Medscape Medical News he agrees with the Japanese researchers that post-hoc analyses could provide some answers.

With regard to the diuretic question, the sponsors could address the concern with the data they already have by performing a subanalysis comparing cardiovascular or renal outcomes for patients taking study drug without a concomitant diuretic to the outcomes for patients taking a diuretic and placebo.

"This sort of post-hoc analysis is not as pure as a prespecified one, but at least the data are readily available," said Leslie.

But of course, Leslie and Gerwin also note, the fact that the companies have no incentive to conduct such analyses "exemplifies a deficiency in the pharmaceutical regulatory system."

"Sponsors are not required to ascertain whether the results of [SGLT2 inhibitor] therapy and those of more cost-effective diuretic therapy might be similar," they conclude.

Ikeda has reported no relevant financial relationships. Leslie has reported owning stock in Bristol-Myers Squibb, Pfizer, and Lilly. Inzucchi has reported serving on clinical trial executive/steering/publications committees for AstraZeneca, Novo Nordisk, Boehringer Ingelheim, and Sanofi-Lexicon; advisory boards for AstraZeneca, Novo Nordisk, vTv Therapeutics, and Abbott/Alere; and has accepted lecture fees from Boehringer Ingelheim and Merck.

J Pharm Policy Pract. Published online November 18, 2019. Full text

For more diabetes and endocrinology news, follow us on Twitter and Facebook.


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.