Use Of Pre-Approval CV Data From Ongoing Trial Gets Once-Over At Canagliflozin Panel

There is no statistical reason not to allow a single trial to provide pre-approval and post-approval cardiovascular hazard ratios for a diabetes drug, FDA says, but Janssen’s canagliflozin would be the first. The Endocrinologic and Metabolic Drugs Advisory Committee backed approval 10-5.

Janssen R&D LLC’s NDA for canagliflozin could provide the first opportunity to see if FDA will accept interim data from an ongoing study to show that a type 2 diabetes drug meets the agency’s pre-approval criteria for cardiovascular safety and final results from the trial to meet post-approval requirements.

Several members of the Endocrinologic and Metabolic Drugs Advisory Committee were skeptical a trial could serve both purposes as they backed approval of canagliflozin on a 10-5 vote; FDA convened the Jan. 10 review to seek input on the panel’s comfort level with the drug’s cardiovascular data (Also see "Canagliflozin Should Have Limitations Based On Renal Function, Panel Says" - Pink Sheet, 10 Jan, 2013.).

FDA’s type 2 diabetes-cardiovascular guidance calls for sponsors to show their drugs do not increase the risk for serious cardiovascular events by 30% or more, but because assuring that level of safety requires large trials, the agency allows companies to demonstrate CV safety in two steps (Also see "FDA’s Diabetes Guidance On CV Risk Will Get First Test With Pending NDAs" - Pink Sheet, 22 Dec, 2008.).

Advisory Committee Votes

• 10-5 there is sufficient efficacy and safety data to support marketing of canagliflozin to treat type 2 diabetes mellitus
• 8-7 there are concerns whether the sponsor has excluded a risk margin of 1.8

Pre-approval, the upper bound of the two-sided 95% confidence interval for the estimated risk ratio for cardiovascular events associated with a diabetes drug must be less than 1.8; post-approval, the upper bound must be less than 1.3, the ultimate goal.

The guidance says the 1.8 upper bound can be demonstrated by a meta-analysis of CV events in Phase II and III studies designed with CV events in mind or a single large safety trial, or a combination of both. A CV outcomes study then can be conducted post-approval to demonstrate compliance with the 1.3 upper bound.

Janssen designed the canagliflozin trials to enable the pre-approval meta-analysis, but also began the 4,385-patient CANVAS CV outcomes study to provide interim results to be included in the meta-analysis calculations. The company plans to use final results from the study, which is expected to complete in 2015, to meet the 1.3 upper bound requirements post-approval.

Statistically Possible? FDA Says Yes

This approach could run into statistical problems, according to some on the panel. “FDA will really have to think long and hard about this,” Sanjay Kaul, Cedars-Sinai, said. “How do you allow an interim analysis to impact regulatory decision-making,” he queried. “Say the drug gets approved, how do you ensure that the trial integrity is preserved? How do you ensure that crossover is minimized if, after the drug gets approved, the patients who are randomized to the control arm want to be on active treatment? That will sort of shrink the differences and bias the upper confidence interval to null.”

Division of Metabolism and Endocrinology Products Director Mary Parks said the agency is aware of the difficulties with regard to relying on data from a single ongoing trial to rule out two different risk margins while preserving type 1 error and maintaining data integrity.

But there is no evidence these problems cannot be solved, she indicated. In terms of “the methodology, the technical aspect of whether or not it can be relied on, we could not clearly identify at this point in time that a single trial itself cannot be relied upon to rule out two different risk margins,” she said.

Office of Drug Evaluation II Director Curtis Rosebraugh noted the agency is still at an early stage in applying the cardiovascular criteria. “We have a guidance, but little experience with some of that.” Agency thinking can change as it gains more information, he said, noting that panel comments “would help us as we incorporate knowledge that we gain through development programs.”

Minimizing Post-Approval Trial Crossovers

Addressing the crossover issue, Parks said a lot of options “are being discussed with drug companies, within the agency, with our legal folks.” There can be a conflict between the agency’s need to be transparent about a product’s safety and efficacy and the ability to acquire good data, she said, and there “are active ongoing discussions within the agency [about how to address this], anywhere from limiting the amount of information that can be provided at the time we make a decision, or we’re wholly transparent and share all the information but … [say] one should not make any definitive conclusion on the overall cardiovascular safety of this product,” Park noted. “This is very much uncharted territory.”

Kaul asked for a precedent of accepting interim data for the initial cardiovascular assessment. Parks noted that “things have come in and we have considered this, but nothing [is] in the public domain,” implying that none of those attempts have resulted in an approval.

“Many companies have proposed the single trial to exclude the two risk margins and to meet the requirements under the CV guidance,” she subsequently told “The Pink Sheet,” noting that the agency cannot comment on the stage of development for those programs.

One company with a double-duty trial appears to be Takeda Pharmaceutical Co. Ltd., which reported that it expected interim data from an ongoing trial to allow alogliptin to clear the first CV assessment hurdle when it resubmitted the NDA in 2011 (Also see "Takeda Re-Files DPP-4 Inhibitor Alogliptin, Plays Up Combo With Actos" - Pink Sheet, 1 Aug, 2011.). FDA had issued a “complete response” letter for the drug in 2009, largely because existing studies did not rule out the cardiovascular risk mandated by the 2008 guidance (Also see "Takeda's Alogliptin Suffers Major Setback, Delaying Drug Launch Until 2012 Or Later" - Pink Sheet, 29 Jun, 2009.).

The company launched the EXAMINE cardiovascular outcomes study, which has an estimated completion date of May 2015. Takeda did not get approval for alogliptin the second time around, but said the agency’s concerns are “not specific” to cardiovascular outcomes (Also see "Alogliptin Rejection Dents Takeda’s Diabetes Franchise" - Pink Sheet, 26 Apr, 2012.).

Orexigen Therapeutics Inc. is hoping to make an interim analysis pay off for Contrave (naltrexone SR/bupropion SR) in the obesity space, where FDA wants some assurance of CV safety pre-approval as it does with diabetes (Also see "Recent Round Of Obesity Reviews Suggests CV Safety Standards Have Come Into Focus" - Pink Sheet, 21 May, 2012.).

The agency has agreed to a single CV study for Contrave that rules out a doubling of CV risk at an interim analysis pre-approval and a 40% increase post-approval (Also see "Orexigen Defines A Way To Get The Contrave PDUFA Clock Ticking" - Pink Sheet, 7 Jan, 2013.). The trial was designed under a Special Protocol Assessment with FDA (Also see "With SPA Finalized, Orexigen To Begin CV Outcomes Study For Contrave In Second Quarter" - Pink Sheet, 6 Feb, 2012.).

Mads Rasmussen of Novo Nordisk Inc., who was the industry representative on the panel, noted that an advantage of a single study for diabetes drugs “is that final exclusion of the 1.3 limit will come that much earlier because you don’t have to set up [a new] trial.”

Lingering CV Uncertainty

Aside from the policy considerations about using the interim analysis, many committee members indicated some lingering concerns in that area. The panel split 8 to 7 on the question of “do you have any concern regarding a conclusion that a risk margin of 1.8 has been excluded for canagliflozin,” given the data and points raised by FDA for discussion.

Panel members’ concerns centered not on the 1.8 margin, but on other issues and were shared by those who voted no. Rosebraugh had explained that the purpose of the questions was “to help us get a sense of your comfort with the certainty of CV data that’s provided.”

The overall estimated hazard ratio for MACE-plus (a composite of adjudicated cardiovascular death, non-fatal myocardial infarction, non-fatal stroke and hospitalization for unstable angina) in nine clinical trials plus the CANVAS study interim analysis was 0.91 (0.68, 1.22).

FDA had questioned whether this was accurate because during the first 30 days of the CANVAS trial the estimated hazard ratio was 6.5 (0.85, 49.66), although after day 30 it was 0.89 (0.64, 1.25). The imbalance of events in the early timeframe could have undermined the assumption of proportional hazards necessary to interpret the Cox model used in the pre-specified analysis, agency statisticians suggested (Also see "Cardiovascular Uncertainty Accompanies Canagliflozin To Advisory Panel" - Pink Sheet, 8 Jan, 2013.).

Statisticians on the committee found the overall estimated hazard ratio acceptable. If there is some effect from the disproportionality, it is not very large, noted Erica Brittain, National Institute of Allergy and Infectious Diseases. And even in CANVAS, once past the 30-day point, “survival curves turn around and not only converge, there seems to be some suggestion that there may be a possible advantage [for canagliflozin] … at six months, just a suggestion of that. When you put all that together, I’m pretty comfortable that the hazard ratio estimate is a fairly reasonable measure of a treatment difference.”

If the early versus large imbalance is real, “it’s reasonable to sort of combine those within the overall hazard ratio and conclude that things are pretty good,” Michael Proschan, also of NIAID, said.

Other points of concern raised by FDA included a hazard ratio exceeding 1.0 for non-fatal stroke, compared to less than 1.0 for the other components of MACE-Plus, and an increase in LDL cholesterol levels.

William Hiatt, University of Colorado, voted that he had concerns about the CV data but agreed that the 1.8 criterion had been met. However, he suggested that the longer-term data might have an upper bound of 1.3 or more if the upward trend in LDL-C levels continued long term.

More data is needed to determine the effect on stroke, noted Acting Chair Abraham Thomas, Henry Ford Hospital in Detroit, who voted that he did not have concerns about the CV data. “You need probably several years to see the impact on LDL cholesterol,” he added.

If approved, canagliflozin, with a proposed brand name of Invokana, could be the first sodium glucose co-transporter 2 (SGLT2) inhibitor to receive approval to treat diabetes. AstraZeneca PLC’s andBristol-Myers Squibb Co.’s dapagliflozin failed to gain approval in its first review cycle; a resubmission is expected in mid-2013 (Also see "Bristol Signals Dapaglifozin’s Back On Track, Plans FDA Resubmission" - Pink Sheet, 1 Nov, 2012.).

The SGLT2 inhibitors cause renal elimination of glucose, a new mechanism of action for treating type 2 diabetes. Janssen is seeking an indication as an adjunct to diet and exercise with dosing at 100 mg and 300 mg once daily.