Editor

Chelsea Therapeutics Inc.'s earnings-call news last week that the firm is asking the FDA to allow an endpoint change in the still-blinded, second Phase III test with droxidopa for low blood pressure caused some Wall Street ripples, and sparked speculation on the wisdom of such a move.

"I can't recall any program where someone convincingly changed the efficacy endpoint after the fact," said David Apelian, chief medical officer for GlobeImmune Inc., though often trials will include two co-primary endpoints (as Apelian did with the hepatitis B virus drug Baraclude, when he was with New York-based Bristol-Myers Squibb Co.).

BMS tested the efficacy of Baraclude (entecavir) by measuring viral load and doing biopsies. GlobeImmune, of Louisville, Colo., has an ongoing, adaptively designed Phase II trial with its prostate cancer drug GI-4000. Adaptive trial design seems to be catching on fast. (See BioWorld Today, Oct. 30, 2009.)

Chelsea's strategy looks like some maximum version of the approach, but it's not, since the endpoint change is being sought based on the outcome of an entirely separate, finished study.

Apelian, a trial-design expert who also helped Madison, N.J.-based Schering-Plough Corp. with studies that won approval of Rebetron (ribavirin) for HCV in pediatric patients, called the Chelsea move "a risky proposition. Generally, what people have done in the past is deploy the co-primary type structures."

Droxidopa, designed to treat a condition called neurogenic orthostatic hypotension - a drop in blood pressure upon standing from a lying or sitting position - blew up in the first Phase III study, called 302, measuring improvement in dizziness and lightheadedness.

In 302, droxidopa did better than placebo but missed statistical significance, sending Charlotte, N.C.-based Chelsea reeling and knocking 61 percent from the company's shares on the day the outcome was disclosed. A later analysis showed better news in various clinically relevant assessment criteria, but Chelsea didn't recover much. (See BioWorld Today, Sept. 25, 2009.)

The drug scored 3.5 on a symptom assessment while the placebo scored 4.0 in study 302, with the higher score indicating worse symptoms. Chelsea hopes a more sensitive instrument in the new endpoint for 301 will let droxidopa prove itself.

Meanwhile, though, the company plans to keep 301 blinded pending a powwow with the FDA - which could ask for a bigger enrollment, if 301 is going to be the basis for an NDA. Regulators have said they could approve the compound on the basis of satisfying results from a single, robust trial, but the unveiling of 301's results is delayed until at least the first part of next year.

Simon Pedder, Chelsea's president and CEO, said during the conference call on third-quarter earnings that an ongoing analysis of the failed trial, "which we plan to share in detail with the FDA," suggest that part of the questionnaire used to evaluate results with droxidopa "may not be the most reliable" in determining benefit.

Benefit there seems to be, at least in Japan, where the compound has been sold since 1989. Marketed by Dainippon Sumitomo Pharma Co. Ltd., droxidopa first won approval for frozen gait or dizziness associated with Parkinson's disease, orthostatic (postural) hypotension and syncope (loss of awareness of one's surroundings), and dizziness associated with Shy-Drager syndrome, as well as amyloidotic polyneuropathy.

In 2000, the Japanese expanded the label to include vertigo, dizziness and weakness associated with orthostatic hypotension in hemodialysis patients.

If approved, droxidopa would compete with midodrine, the only FDA-blessed therapy for orthostatic hypotension, which has limited use because of side effects and a black-box warning for supine hypertension.

Interestingly, the further analysis of droxidopa's failed first trial turned up cause for optimism in the second study - based on the very design that Chelsea now is asking regulators to allow tweaking. And that optimism, on the part of an expert consulted by Leerink Swann, is because the second trial's design is more like a midodrine study in a younger and less pre-treated population.

In 302, the trial that fell short, patients were screened for response and an effective dose of droxidopa, stayed on their doses of active drug for one week, and then randomized to get either that same droxidopa dose or placebo for two weeks.

In 301, by contrast, patients were screened and then underwent a seven-day washout period, then randomized to drug or placebo, and evaluated after just one week.

To change endpoints in the course of a Phase III trial doubtless makes investors nervous, but it's not unprecedented, albeit for differing reasons. Earlier this month, London-based Antisoma plc said the FDA gave its nod to an endpoint change in the Phase III study with AS1413 called ACCEDE. The trial is comparing Antisoma's DNA intercalator plus cytarabine against daunorubicin plus cytarabine in newly diagnosed secondary acute myeloid leukemia.

By agreement with regulators, the primary endpoint of ACCEDE will continue to be based on remission rate, defined as the proportion of patients whose AML backs off completely, or does so with an incomplete blood-count recovery. What's different in the setup now is that remission won't need to be confirmed before further intervention - initial remission is the revised endpoint.

Antisoma asked for the change because investigators didn't think it was good practice to wait for bone-marrow sampling results before starting therapy after the disease goes away. The switch means ACCEDE is no longer covered by the FDA's special protocol assessment, but the agency promised no impact on whether the trial's results will be accepted as the basis for an NDA filing.

Endpoint shuffling has led to difficulties for some. In July, the FDA's Oncology Drugs Advisory Committee turned thumbs down on Centocor Ortho Biotech LP's Yondelis (trabectedin), an investigational new molecular entity, in combination with Doxil (doxorubicin HCl liposome injection) as a safe and effective treatment for patients with relapsed ovarian cancer.

In a resounding 14-to-1 vote, the advisory panel said the risks of adding Yondelis, a tris tetrahydroisoquinolone alkaloid, to Doxil outweighed the benefits.

A major sticking point was the jump from a primary endpoint of overall survival to progression-free survival during the Yondelis study, despite the FDA's stern cautioning against doing so. Richard Pazdur, director of the agency's Office of Oncology Drug Products, called the action by Ortho a "buyer beware" move. In September, the FDA followed up with a hardly unexpected complete-response letter for Yondelis. (See BioWorld Today, July 16, 2009.)

Apelian noted that some firms with cancer drugs have tried to use overall survival data as market-worthy evidence of efficacy, after narrowly missing in progression-free survival, but "even in cancer, it's not a gimme" - and changing the endpoint in an still-blinded, low blood pressure study, as Chelsea hopes to do, could prove a particularly hard sell.