Fantastic Voyage

But the results of the initial implants were good enough to raise Cyberonics to the next level. The company got a second round of venture capital, this time for $2 million. Cyberonics brought in an outside CEO and a vice president of clinical studies and regulatory affairs who'd helped Eli Lilly win FDA approval for Prozac.

The company began conducting the clinical studies that are required by the FDA to prove that a new device not only works but is also safe. The VNS device didn't seem to cause problems, and early data looked promising. However, Cyberonics was rebuffed by the agency the first time it applied for approval, since most of its data related to patients who'd had the implant for only three months. It applied a second time in 1994, but was told that its clinical study hadn't included enough patients. This time, the agency's rejection hurt worse: Cyberonics had just gone public, on the assumption that FDA approval was imminent. "We completely misread the FDA," Terry says. "We had 100 patients in our study, and they said we needed 200." The company didn't have the money to give the FDA what it asked for--another yearlong study with 200 participants.

The CEO left, as did the vice president who'd come from Eli Lilly. Cyberonics shrunk from 50 employees to about 35. "We had to cut back advanced development, lay people off, squinch by," Terry says. "It was the lowest point of my career." But the company didn't go out of business. "We had around $20 million left over from the IPO, but that wasn't going to be enough to do the next study," he says. The company arranged private stock sales at low share prices to put more money in the bank. A new CEO was recruited from Johnson & Johnson to help Cyberonics make sure that the third FDA approval attempt was the charm--but he called the board of directors after a week to tell them he'd decided not to take the job after all.

At that point, Cummins, the venture capitalist serving on Cyberonics' board of directors, was drafted to serve as acting chief. He took the job--temporarily, he thought--in part to help Cyberonics' investors cash out. (Sevin Rosen was especially anxious to get the troubled company off its hands.) One part of Cummins's mission as CEO was to scrounge enough money to keep the company moving toward FDA approval, but another was to find a corporate savior that would buy it.

Cummins thought he'd found just that in St. Jude Medical, a Minneapolis device maker. St. Jude invested $12 million in Cyberonics, nearly enough to complete the clinical study the FDA had demanded, and also had the option to acquire the company. Cyberonics began seeing preliminary results from its new study in August 1996. The data looked good but not great. "Patients did better, but the results weren't outstanding," says Shawn Lunney, a longtime Cyberonics employee who is now vice president for market development and engineering. "It wasn't clear-cut enough for a big company."

St. Jude bought two other device makers that fall, but not Cyberonics. "People outside the company thought it meant our device didn't work," Lunney says. "Shareholders were pissed off. The stock went down to $2.50, from about $6.50. But St. Jude's investment had given us not just money but also time and some credibility. Inside, you knew that [the device] worked and that we had enough cash to make our own submission to the FDA."

By January of 1997, Cummins had dropped "acting" from his CEO title. The company went back to the FDA once again. In June, Cyberonics executives and a group of six doctors traveled to Washington, DC, to meet with an FDA advisory panel, which makes the formal recommendation to the agency as to whether a new device should be approved for sale.

The panel unanimously recommended approval of Cyberonics' VNS device for epilepsy. Still ahead were the difficult jobs of convincing often-conservative neurologists that they should consider using the device and persuading Medicare and private insurers to cover the cost. But on July 16, Cyberonics finally received the official approval letter from the FDA in the mail. It had taken 10 years and $50 million.

There are about 80 parts that fit together to make Cyberonics' VNS device. A chip inside sends out precisely calibrated electrical impulses to the nerve along a thin strand of insulated wire; it's paired with a lithium battery that powers the device for seven or eight years. The components are housed in a titanium container that's laser-welded together.

No one knows exactly why the VNS device decreases the frequency and severity of seizures for some epileptics. It may disrupt the patterns of synchronized electrical activity that constitute a seizure, or it may change the way blood flows through the brain in a way that's beneficial to epileptics. Since so little is known about why the device works, it's also hard to tell which patients it will help the most. "The best way to tell," Madsen says, "is to implant someone."