Symlin Approval In Riskier Type 1 Diabetes Prompted By Off-Label Dangers – FDA

Approval of Amylin's Symlin for both type 1 and type 2 diabetes, despite a higher risk of severe hypoglycemia in type 1 patients, was prompted by FDA concerns over potential off-label use for type 1 disease, agency review documents show.

If Symlin (pramlintide) were approved only for type 2 diabetes, off-label use for type 1 "could endanger patients," FDA Medical Officer Dragos Roman, MD, stated in his Feb. 18, 2005 review.

"Such a risk is likely to be reduced by appropriate labeling," Roman said. Risk reduction was a major focus of FDA's review of Symlin, one of the first drugs with serious safety issues approved since the agency announced its drug safety overhaul in November 2004.

Symlin was approved March 16 for type 1 and type 2 diabetes poorly controlled by insulin alone, with a risk management program including a black box warning, patient medication guide and marketing restrictions.

The approval came four years and two "approvable" letters after the NDA was submitted Dec. 7, 2000. Severe hypoglycemia in type 1 diabetics was the primary cause of the lengthy review.

In type 2 patients, "the risk of hypoglycemia…was never deemed an overriding issue in the decision regarding approval," Division of Metabolic & Endocrine Drug Products Director David Orloff, MD, stated in a March 10 memo.

Indeed, FDA was "willing to consider approving Symlin as an adjunct to insulin for use in patients with type 2 diabetes as a single indication" if Amylin re-submitted the NDA, according to minutes of a July 22, 2004 meeting between FDA and Amylin.

Earlier, Office of Drug Evaluation II Director Robert Meyer, MD, suggested in a Dec. 17, 2003 memo that type 2 diabetics not yet requiring insulin could be a population in which Symlin's risk/benefit ratio would be suitable for approval.

Roman acknowledged that limiting pramlintide to type 2 diabetes would be a "desirable course of action" based on its more favorable risk/benefit ratio.

He predicted, however, that approval for type 2 alone "will likely be associated with off-label use."

Off-label use in type 1 patients would be dangerous because "the dose in type 2 diabetes is substantially higher than in type 1 diabetes and the tolerability to the drug is worse in type 1 patients," Roman said.

In the absence of labeling, use in type 1 patients would be informed by the "significant body of literature published on the efficacy of the drug in type 1 diabetes which mentions little of the complexity of a pramlintide/insulin regimen," he cautioned. Dose titration is discussed in patient materials (see related story).

Pramlintide is riskier in type 1 diabetes because such patients "have greater insulin sensitivity, and can more readily achieve daily glycemic excursions in the ranges where hypoglycemia becomes limiting to further control of blood glucose," Orloff explained.

Interaction With Insulin Poses "Catch-22"

The lengthy review of Symlin, the first new drug approved for type 1 diabetes since the advent of insulin, highlights the inherent difficulty of designing development programs in the context of insulin use.

"For insulin, perhaps uniquely in the pharmacopeia, benefit is intimately tied to risk," Orloff wrote.

Like insulin, "the efficacy of pramlintide and the hypoglycemic risk associated with its use cannot be dissociated," Orloff said in a Dec. 2, 2003 memo. "If the pramlintide [dose] is effective, it will significantly add to the known but obviously totally acceptable risk of insulin, a critically necessary drug, hypoglycemia."

"This Catch-22 results, by definition, in an unacceptable risk-benefit profile for pramlintide."

Amylin's NDA originally relied on trials using a fixed dose of insulin to isolate pramlintide's effect and show "improved glycemic control relative to insulin alone," Orloff noted in his March 10 memo.

However, "this method of use could not reasonably be translated to recommendations in labeling as it also engendered a substantial increased risk of severe hypoglycemia with pramlintide/insulin, disproportionate both to the difference from insulin monotherapy in glycemic control and to the absolute level of glycemic control achieved," he said. FDA thus deemed the NDA "approvable" Oct. 10, 2001.

FDA had indicated that Amylin's fixed-dose Phase III program could be inadequate for approval as early as 1997, Medical Reviewer Robert Misbin, MD, reported in his Aug. 30, 2001 efficacy review.

"The sponsor was urged to do an insulin-titration study and to use hypoglycemia as a primary outcomes variable" during an Oct. 28, 1997 meeting with FDA, Misbin noted.

In response to the first "approvable" action, Amylin conducted a trial in type 1 diabetes with flexible insulin dosing and pramlintide titration, in which "rates of hypoglycemia could be compared between treatment groups in the absence of confounding by differences in glycemic control," Orloff said.

The flexible-dose trial (137-150) found a reduced, but still present, increase in severe hypoglycemia rates in pramlintide patients compared with placebo (¹ Pharmaceutical Approvals Monthly July 2003, p. 14).

Nonetheless, FDA issued a second "approvable" letter Dec. 17, 2003. The decision "rested on the conclusion that the achievement of essentially equivalent glycemic control with insulin monotherapy and pramlintide/insulin was evidence that pramlintide offered no particular generalizable advantage," Orloff said. Therefore, FDA determined that "any difference in hypoglycemic risk was unacceptable."

"Neither specific design employed (fixed insulin dosing or insulin dose adjusted to goal) is fully amenable to informing simultaneously both the safe and the effective use of the drug," Orloff concluded. "Neither design translates readily to real-world clinical use. That is, the hypoglycemia risks in the blinded trials are unlikely to be quantitatively reflective of 'actual use' risks, and logically would overestimate risks relative to glucose lowering benefits."

Blinded Trials Seen As Inadequate

"Blinding to therapy sets up a situation in which it is extremely difficult, if not impossible, to address, patient by patient in a clinical trial, both efficacy and safety objectives in clinical management," Orloff said.

"The blinded study approach for pramlintide is a major factor that has made the regulatory development of pramlintide, as neither an insulin nor an oral agent, both difficult and trail-blazing," Roman wrote in his Feb. 18 memo.

"Like novel insulins or insulin pumps, the safe and efficacious use of a novel control variable like pramlintide, when added to insulin, can best be gleaned from a clinical practice setting when neither the patient nor the physician is blind," Roman said.

A postmarketing observational study set to start in September could settle some of the uncertainty over hypoglycemia risk. The "actual use" study will include 2,000 patients over two years. Amylin will try to involve both "targeted" providers who meet the criteria for promotion and "non-targeted" providers.

"It was decided that further blinded trials of pramlintide were, even if ethical based on the risks of insulin-induced hypoglycemia in the blinded setting (which is arguable), not apt to yield further useful results," Orloff said in the March 2005 memo.

Additional data from open-label studies of pramlintide in type 1 and type 2 diabetes "would be adequate for review," he said. Data considered in the third and final review cycle included insulin dose data, unvalidated quality of life questionnaire results, and weight change data, drawn principally from three open-label trials.

In explaining the approval decision, Orloff highlighted Symlin's benefits for type 1 diabetics whose blood sugar goals cannot be reached with insulin alone. The open-label data "suggests that a number [of poorly-controlled type 1 patients] will likely benefit with the addition of pramlintide…some with substantial reductions in weight, reductions and thus simplification of their prandial insulin requirements, and potentially improvements in quality of life," he said.

"Glycemic control in these patients is limited by hypoglycemia and weight gain, itself an undesirable, if often inevitable, consequence of the increasing doses of insulin used, engendering increased caloric intake, and initiating a vicious cycle toward frank obesity and its sequelae," he observed. Labeling notes Symlin's weight loss effect (see related story).

Potential For Class Effects

Symlin's status as the first amylinomimetic drug also raised broader concerns among FDA staffers about the etiology of severe hypoglycemia. Hypoglycemia is linked to gastrointestinal adverse events. "Pramlintide induces nausea, nausea results in smaller meals and an imbalance between the insulin dose and the size of the food ingested," Roman said in his second cycle review.

GI adverse events "may ultimately prove to be a class effect for several new drugs in development which decrease gastric motility and reduce prandial glycemia," Roman's Sept. 6, 2001, review states.

"An overlap between pharmacological and toxic effects may impact on the final tolerability and dose selection for these new drugs," he said. "The potential interplay between two medications (pramlintide and insulin in this submission) and a variable meal content may result in mealtime hypoglycemia."

"Safety signals observed in this review should result in a prospective search of meal-related hypoglycemia and hypoglycemic serious adverse events in other submissions," Roman suggested.

Previous experience with retinopathy in diabetes trials informed FDA's early recommendation for a three-year, 500-patient Phase IV retinopathy study.

Diabetic retinopathy poses a "regulatory dilemma," Misbin said in his 2001 review. While "long-term reduction of HbA1c decreases the risk of diabetic complications, particularly retinopathy…the opposite result, progression of retinopathy, can be expected to occur sometimes during the 6-12 months that is typical of a Phase III trial."

FDA later decided against the Phase IV study because the "dose at which the potential signal of retinopathy was observed (150 mcg) is in excess of the to-be-marketed pramlintide doses of 30 mcg, 60 mcg and 120 mcg," Roman explained in a Feb. 28 memo.