An experimental antidepressant developed by Sage Therapeutics did not meet the main goal of a closely watched study, sending shares in the biotech spiraling downward and undercutting the company’s hopes of proving a clear-cut benefit for the drug. A high dose of the treatment, called SAGE-217, appeared to initially reduce depressive symptoms more than placebo. But, by day 15, the time point at which Sage measured success, the difference in effect between SAGE-217 and placebo did not clear the threshold for statistical validity. Company executives attributed the trial’s failure to confounding factors, like the roughly 9% of study participants who did not appear to actually take the high dose of SAGE-217. Their explanation, however, wasn’t enough to stop shares from sinking by more than half, erasing about $4 billion of the company’s market value.

Sage achieved a major milestone earlier this year in winning Food and Drug Administration approval for its first drug and the first therapy ever cleared to specifically treat postpartum depression.

But, thanks to restrictive labeling and a burdensome dosing schedule, most observers did not anticipate the treatment — called Zulresso — to be a large product for Sage.

Instead, hopes for the biotech’s future as a commercial-stage company centered on SAGE-217, a pill that Sage has pitched as faster acting and safer than currently available antidepressants.

The drug had previously succeeded in Phase 2 studies of patients with major depressive disorder, postpartum depression and bipolar depression, ramping up expectations it could become a major product for Sage.

The Phase 3 trial that just read out results was a major test, and its failure is a significant blow.

Still, Sage CEO Jeff Jonas reiterated the company’s confidence in the drug’s future. “Based on the profile we’ve consistently seen across three pivotal trials, we believe as strongly as ever that SAGE-217 may offer an entirely new approach for the treatment of depression,” said Jonas on a Thursday call with investors.

The trial, dubbed MOUNTAIN, compared two doses of SAGE-217, 30 mg and 20 mg a day, to placebo in adults with MDD. Sage attempted to show its drug reduced depressive symptoms by more than placebo at day 15 on a disease rating scale known as HAM-D-17.

Results showed the 30 mg dose of SAGE-217 delivered a significantly greater benefit at days three, eight and 12. But by day 15, the study’s primary endpoint, the difference between the 30 mg group and placebo group, was no longer statistically clear. Trial participants given the 20 mg dose of SAGE-217 experienced no greater benefit than those on placebo.

Additionally, the drug’s 12.6 reduction from baseline in HAM-D-17 scores at day 15 is noticeably less than the 17.6 point decline Sage reported in Phase 2.

In explaining the failure, Sage executives cited data analysis conducted by the company following the study’s completion, an approach that can be methodologically fraught.

Roughly 9% of patients in the 30 mg arm did not appear to actually take SAGE-217, as determined by tests showing no measurable drug concentration in the blood. If those patients were not considered, the 30 mg dose would have met the study’s primary goal.

It’s not entirely clear, however, why those 9% of patients did not take the drug, although Sage claimed it was not due to side effects.

Additionally, Sage said, the study enrolled more patients with milder depressive symptoms than past trials of SAGE-217. When looking only at those who had more severe depression, the 30 mg dose again would have cleared statistical significance.

SAGE-217 proved relatively well-tolerated, with no loss of consciousness or increased suicidal behavior reported among patients taking the drug.

Jonas said that the safety profile plus SAGE-217’s durable effect  — the nominal declines in depressive symptoms persisted through 182 days — argue for the drug’s potential.

“We view this study — certainly it’s our view and has to be confirmed — as strongly supportive of a generalized MDD filing,” said Jonas, referring to submitting results to the FDA.

“Our intent right now, based on the data from this study, is that the pathway to MDD and PPD combined remains open for us.”

The company is testing SAGE-217 in three other Phase 3 trials, including two maintenance studies in major depressive disorder. Results from those are expected next year and could give Sage further evidence to take to regulators.

For now, though, MOUNTAIN’s failure clouds Sage’s path to market.

While the FDA is well aware of the difficulties in testing drugs for depression, its flexibility has limits. Earlier this year the agency rejected an experimental depression treatment from Alkermes, which was supported by one positive Phase 3 study.

Shares in Sage declined by more than 55% Thursday morning, reducing the company’s market capitalization from nearly $8 billion to approximately $3.4 billion.