Jardiance's CV Benefit: Can A Single Study Carry The Day At FDA?

Agency finds support for cardiovascular mortality benefit with Boehringer/Lilly's diabetes drug, but advisory committee will review weaknesses in outcomes trial's component data.

FDA's analysis of the EMPA-REG OUTCOME data for Boehringer Ingelheim GMBH and Eli Lilly & Co.'s typed 2 diabetes drug Jardiance (empagliflozin) reflects the limits of how far a single study designed for safety purposes can be stretched to support efficacy claims.

In advisory committee briefing documents released June 24, FDA reviewers generally agreed that empagliflozin, an SGLT2 inhibitor, reduced the risk of CV death and all-cause mortality in the 7,000-patient outcomes trial. However, they raise concerns about a possible adverse impact on non-fatal stroke, one of three components of the composite primary endpoint of major adverse cardiovascular events (MACE), which also includes CV death and non-fatal myocardial infarction.

"It does remain an issue on whether the results from a single study initially designed for safety will be sufficient to obtain efficacy claims," FDA's statistical review states.

Given the seemingly contradictory findings on CV death and stroke and concerns around how MIs were assessed in the study, the agency questions whether EMPA-REG, which was designed as a safety trial to exclude an increased risk for MACE, is sufficient to support a new efficacy claim.

The agency's review also suggests Boehringer's requested claim for reduction in heart failure-related hospitalization is a non-starter. Design factors with EMPA-REG limit the conclusions that can be drawn from the positive heart failure findings, agency reviewers said, citing the need for a dedicated trial on this issue.

First Panel To Weigh CV Benefit Claims

FDA's Endocrinologic and Metabolic Drugs Advisory Committee will meet June 28 to discuss the sNDAs for Jardiance and Synjardy (empagliflozin/metformin) for CV benefit claims (see box). The efficacy supplements were received by FDA Nov. 4, which suggests the user fee dates are in early September.

Jardiance, Synjardy Proposed Indications

To reduce the risk of all-cause mortality by reducing the incidence of cardiovascular death and to reduce the risk of CV death or hospitalization for heart failure in adult patients with type 2 diabetes mellitus and high CV risk.

The meeting will mark the first advisory committee review of an outcomes trial conducted pursuant to FDA's December 2008 guidance on CV safety assessments for type 2 diabetes drugs that purported to demonstrate a CV benefit (Also see "Empagliflozin's CV Benefit Impresses At EASD, But What Will FDA Say?" - Pink Sheet, 17 Sep, 2015.).

The committee's view on the robustness of the data and the ability of a single trial to support a CV benefit claim will have important implications for other diabetes drug sponsors, most notably Novo Nordisk AS, whose LEADER trial has demonstrated a CV benefit with the GLP-1 agonist Victoza (liraglutide) (Also see "Will Novo’s LEADER Trial Move GLP-1 To The Front Line In High-Risk Diabetes?" - Pink Sheet, 4 Mar, 2016.).

FDA's draft questions indicate the committee will be asked to vote on two issues: whether the EMPA-REG results demonstrate that use of empagliflozin for glycemic control would not result in an unacceptable increase in CV risk, and whether the trial provides substantial evidence to establish that the drug reduces CV mortality in the population studied.

FDA also seeks discussion around the various components of the MACE endpoint results and on the heart failure data. Notably, the agency does not seek a vote specific to a heart failure claim.

Demonstrating Efficacy Through A Safety Study

In choosing to seek approval only for claims specific to all-cause mortality, CV death and heart failure hospitalization, Boehringer and Lilly opted against pursuing a broader indication based on the less impressive results of the MACE primary composite endpoint (Also see "Empagliflozin CV Outcomes Data: Is Sum Of The Parts Greater Than Whole?" - Pink Sheet, 17 May, 2016.).

In contrast, Novo has said it intends to pursue a broad CV benefit claim for liraglutide based on the composite endpoint (Also see "Novo Will Go After MACE Indication For Victoza Based On LEADER" - Pink Sheet, 14 Jun, 2016.).

An interim analysis from EMPA-REG was used to support the drug's CV safety at the time of its August 2014 approval, with the outcomes study continuing post-approval. FDA's briefing document executive summary contrasts EMPA-REG with some other CV outcomes trials for diabetes drugs in that it "was not prospectively designed (i.e., sized) with the expressed intent of demonstrating a cardiovascular benefit of the new antidiabetic therapy."

"CV safety trials conducted to meet the FDA guidance generally specify that non-inferiority and superiority hypotheses will be tested in a sequential manner in their analysis plans (i.e., regardless of what the trial is initially powered to show)," the executive summary states.

"There is no precedent on using these types of safety trials for efficacy since this is the first of the diabetes safety trials to be considered for an efficacy claim," the briefing document's statistical review states. "It does remain an issue on whether the results from a single study initially designed for safety will be sufficient to obtain efficacy claims."

Silent MIs And Not Assessable Deaths

Empagliflozin was found to be both non-inferior and superior to placebo on the primary MACE endpoint, with a 14% reduced risk. On the MACE+ endpoint, which includes hospitalization for unstable angina, empagliflozin was non-inferior, but not superior, to placebo. The benefit on the primary endpoint was driven primarily by the effect on CV death, and there was a numerically unfavorable imbalance on stroke (see table).

EMPA-REG OUTCOME Results
Endpoint	Hazard Ratio (95% CI)
MACE	0.86 (0.74, 0.99) p=0.04 for superiority
MACE+	0.89 (0.78, 1.01) p=0.08 for superiority
CV Death	0.62 (0.49, 0.77)
Nonfatal MI	0.87 (0.70, 1.09)
Nonfatal Stroke	1.24 (0.92, 1.67)
All-Cause Mortality	0.68 (0.57, 0.82)
Heart Failure Hospitalization	0.65 (0.50, 0.85)
Source: FDA advisory committee briefing documents

However, FDA review staff raised questions about the exclusion of certain types of MIs from the analysis and about the number of CV deaths that were deemed not assessable.

So-called "silent" MIs based solely on ECG parameters were not included in the primary MACE analysis. The statistical review notes that "this is an event that it is difficult to detect" with only about half the patients in the trial screened for it. Of the 53 patients who experienced a silent MI, 38 were in the empagliflozin group.

EMPA-REG's safety objective "may be why certain aspects of the trial are different from a trial that is directly targeting efficacy, such as the non-inclusion of silent MI in the non-fatal MI endpoint," the review states.

Under various exploratory analyses that take the silent MIs into consideration, empagliflozin still demonstrated non-inferiority on the MACE primary endpoint but not superiority, the statistical review states. "The issue of silent MIs, however, only affects the non-fatal MI component of MACE and does not affect the CV death component, which is what is driving the differences between the two arms."

The agency's clinical summary notes that "while occurrence of a true incident silent myocardial infarction event represents a clinically significant event, analyses using silent MI events … are limited due to lack of complete event ascertainment and adjudication. The division is therefore uncertain about the meaningfulness of retrospective analyses results which include this event in the 3-point MACE endpoint. Whether and how these events should be considered in the overall interpretation of the trial results is a point for discussion."

On the CV mortality endpoint, FDA's clinical summary notes that 40.1% of all CV deaths in the trial were labeled as "not assessable," which was defined as all deaths not attributed to specified categories and not attributed to a non-cardiovascular cause.

"It is not clear whether these events are truly CV deaths," the review states. If all 124 of these not assessable cases were excluded from the CV death analysis, there is still a statistically significant reduction in the risk of CV death for empagliflozin compared to placebo, but superiority is lost on the primary MACE endpoint.

Stroke Concern

The agency's biggest concern appears to be the unfavorable imbalance in stroke.

"There appears to be an imbalance in strokes early after initiation of treatment with empagliflozin," the clinical summary states. "By day 150, differences between arms disappear and the incidence is similar between the treatment arms by one year. A separation is again observed starting at approximately day 600 which persists for the remainder of the observation period."

"The clinical relevance of this numerical imbalance is unclear," the clinical summary states. "While the finding is not statistically significant, it remains of concern for this product and drug class."

EMPA-REG was not designed solely for exploring stroke events, FDA said, adding that standard of care for stroke work-up may have varied across regions. "Additionally, the applicant collected no information on disability related to stroke. Thus it is not clear if treatment with empagliflozin, while perhaps reducing the risk of death, leads to more stroke-related disability."

However, a consult from the Division of Neurology Products suggests less concern about the potential stroke signal.

Substantial evidence supports reduced risk of CV death but not a lower MACE risk, FDA's Cardio-Renal reviewers said.

"The increase in the non-fatal stroke rate observed in those who received either dose of empagliflozin compared to those who received placebo is likely due to chance and not to treatment with empagliflozin," medical officer Jody Green said in a May 5 memo. Rather, the imbalance might have resulted, in part, from regional differences in disease management or because of unblinding due to side effects, the review states.

The Division of Cardiovascular and Renal Products also weighed in with its view of the data.

"Considering the totality of the data including the stroke findings, the silent MI findings, and the missing data, we do not believe that the trial provides substantial evidence that empagliflozin lowers the risk of MACE, and specifically, strokes and MIs," medical officer Karen Hicks and biostatistician John Lawrence said in a May 5 review. "We do, however, believe the trial provides substantial evidence that empagliflozin reduces the risk of CV death."

Heart Failure Claim A Non-Starter?

FDA views the requested claim for reduced risk of heart failure-related hospitalization as premature and in need of confirmation.

"This endpoint was not included in the plan to control for type 1 error," the executive summary states. "The trial was not designed to assess heart failure outcomes, and data that may be important to the interpretation of this finding are missing or if present may not be as reliable as in a dedicated trial (e.g., New York Heart Association Functional Classification or ejection fraction, baseline therapies for the treatment of heart failure to assess adequacy of treatment at baseline). It is unclear whether the patient population was appropriate for assessing this or whether patients were receiving optimal therapy for heart failure."

The cardio-renal review states that because of its diuretic effect, "it is certainly plausible that empagliflozin could reduce the risk of HF hospitalization (in patients with a preserved or reduced [ejection fraction]); however, we believe this hypothesis should be confirmed in a well-designed and well-conducted trial in patients with HF."

Boehringer and Lilly are unlikely to have been surprised by FDA's view on the heart failure claim. In April, the companies announced plans for two new outcomes studies of empagliflozin and treatment of chronic heart failure in people with and without type 2 diabetes.