A meeting of the FDA’s Peripheral and Central Nervous System Drugs Advisory Committee voted unanimously that Vanda Pharmaceuticals Inc.’s candidate for non-24-hour sleep-wake disorder, Hetlioz (tasimelteon), was safe and voted 10-0, with one abstention, that the company provided substantial evidence of efficacy for the drug.

Although the vote was consistent with the FDA’s generally positive review, it shocked some observers who criticized the company sharply for abandoning its original planned primary endpoint and substituting another a month before the study was unblinded.

That new endpoint, entrainment of the circadian rhythm, was created by Vanda and had never been used in a clinical trial for a sleep drug before. Further, the FDA disagreed with Vanda about its use of that endpoint, and dinged the company in its briefing documents for not having pre-specified primary clinical endpoints.

Non-24-hour sleep-wake disorder (non-24) affects blind people, whose bodies are unable to use cycles of day and night to calibrate their internal circadian rhythms. Non-24 causes irregular sleep patterns, leading to social isolation and difficulties with school and work.

Tasimelteon is a dual melatonin receptor agonist with selective activity at the MT1 and MT2 receptors. It is designed to reset the master body clock in the suprachiasmatic nucleus, entraining the melatonin and cortisol rhythms to align to a 24-hour day-night cycle. The FDA granted Vanda priority review for the drug. (See BioWorld Today, July 31, 2013.)

The Washington-based company’s original Phase III trial design, as reported by the company in 2010, was improvement in total sleep time during the night (nTST), with secondary endpoints evaluating parameters of daytime sleep and laboratory measures of synchronization between the internal body clock and the 24-hour environmental light-dark cycle.

The FDA’s reviewer did not find statistical significance for nTST, however, meaning that according to its original design, the trial was a failure. Vanda reported success with its new primary endpoint for the trial: entrainment of the circadian melatonin rhythm as measured by the urinary metabolite of melatonin, aMT6s.

The agency did not accept that biomarker-based endpoint, on the basis that benefit for disordered circadian rhythms should occur in a brief period of time, and be readily measurable in terms of sleep quality. That meant that secondary endpoints of lower quartile of nighttime total sleep time (LQ-nTST) and upper quartile of daytime total sleep duration (UQ-nTST) became important for decisions about efficacy of the drug.

Panelists exhibited some confusion on why the FDA and Vanda were unable to agree on endpoints, and some members noted that a melatonin-based biomarker would have relevance in the development program and potentially in diagnosis of patients. The FDA responded that clinical benefit is of primary importance.

According to Vanda, 20 percent (eight out of 40) of tasimelteon patients were entrained by week two, compared to 2.6 percent (one out of 28) in the placebo group. It said a subsequent analysis of entrainment for some patients at month seven showed an entrainment rate of about 59 percent.

Vanda’s second primary endpoint was clinical response based on entrained melatonin rhythm and a score of at least 3 on the N24CRS scale, calculated from “the totality of 2.5 circadian cycles up to 180 days of observation.”

The FDA, however, focused on the secondary endpoints of LQ-nTST and UQ-dTSD, which resembled the total night time sleep measure that Vanda began with, and which according to the agency reflected the clinical benefit one would expect of a drug designed to address the disordered circadian rhythms of a person with non-24.

Vanda reported that tasimelteon 20 mg improved LQ-nTST by 56.8 minutes – essentially that the patients’ worst nights of sleep were improved by nearly an hour. At the same time, patients experienced a decrease of unwanted naps on the worst days of daytime sleep by 46.48 minutes. Those numbers were in comparison to 17.08 minutes LQ-nTST and 17.87 minutes for UQ-dTSD in the placebo group.

The FDA’s review accepted those endpoints as clinical evidence of efficacy, but warned that defining the endpoints posthoc risks a Type I error in either study. A Type I error is when a null hypothesis is incorrectly rejected.

The risk that the trial had actually failed to show efficacy of tasimelteon was of little concern to the committee, however, which overwhelmingly agreed that efficacy had been established.

Still, some dissent was aired.

Oregon Health and Science University professor of psychiatry Robert L. Sack voted yes, but with “mixed feelings.”

“There is quite strong evidence that an already available agent, namely melatonin, is about as effective,” Sack said. “I understand that there’s no commercial support for going through expensive trials for melatonin to be approved by the usual FDA criteria. Consequently, it has remained underutilized.”

Sack called for more public awareness of non-24, noting that the drug seemed to be on the path to approval, but that it would be an expensive drug that a typical patient would take for life.

Justin Zivin, a professor emeritus at the University of California San Diego, abstained from the voting, explaining, “It failed on its primary endpoint. It succeeded on subsequent compound endpoints which I have problems with, and therefore I can’t decide whether this drug is effective or not.”

A broader question was raised during the open public hearing portion of the meeting, by a representative of Public Citizen, about the precedent set for the regulatory process for other drugs, with changes having been made to the protocol for tasimelteon very late in the clinical trial process, and essentially in contradiction with instruction given by the FDA.

“This would set an ominous precedent for future development programs. How would the FDA ensure that such an arbitrary and late-in-the-game change in clinical trials is not done in the future?” the representative asked.

The panel also voted on the validity of non-24 as a disorder, and the members unanimously agreed that the evidence was sufficient to support it. The panel also agreed that the drug was safe.

Shares of Vanda (NASDAQ:VNDA), which were suspended during the advisory meeting, were up 21 percent in after-hours trading Thursday.