1. The FDA's 2002 approval letter for Infuse notes the 30-day requirements for reporting adverse events and reactions.
    Read it here
  2. An FDA Advisory Panel of experts met in January 2002 to consider whether to approve Infuse. Trisler's warning appears on page 17.
    Read it here
  3. In 2007 the FDA sent a warning letter to Medtronic's CEO citing failures to submit adverse event report within 30 days, including reports of serious injuries from a device.
    Read it here
  4. The FDA sent a warning letter to Medtronic's CEO following a 2008 inspection citing failures to report adverse events within 30 days and failure to have qualified personnel deciding what events to report.
    Read the letter
  5. Medtronic told the FDA in July 2013 that it assessed the adverse events in the retrospective study for reportability in June 2013, five years after the study was shut down.
    Read the disclosure
  6. The subpoena of documents relating to INFUSE is described on page 37 of Medtronic's 10-K filing to investors in April 2009. The section also notes a later notice requesting Infuse documents for a civil investigation.
    Read the disclosure
  7. A letter from the Senate Finance Committee demands all information on medical complications from rhBMP-2.
    Read the letter
  8. The FDA censored the number of adverse events in Medtronic's summary report from the public, calling it a corporate trade secret.
    See what was censored
  9. After the Star Tribune found the censored summary and complained repeatedly to the FDA, the agency changed the summary report to reveal the number of remaining unreported adverse events. It was 1,039.
    See the unredacted report
  10. The North American Spine Society noted that a lack of reporting may have "led to a skewed understanding of the real risk-benefit ratio."
    See the conclusion on page 21 of the report
  11. A review of clinical research and a list of possible side effects for spinal fusion are included in a package insert for Infuse.
    Read the full label here.
  12. The Yale review was published in 2013. Medtronic said it repeated the known risks for Infuse.
    Read the company's press release about the review findings here
  13. The FDA accepted Medtronic's application to report the events under its Alternative Summary Reporting program.
    Read about the program here

For two years, Medtronic employees pored over the medical records that came flooding in from doctors and hospitals. They documented what happened to 3,600 patients who had received the firm's pioneering bone-fusion product, called Infuse.

The biotech device had won government approval for one specific type of back surgery. Medtronic hoped the records would build a case for a government blessing for a much wider use of Infuse in the multibillion-dollar spinal surgery field.

Medtronic asked doctors to scan patients' files and report any "adverse event" following surgery. Doctors shared more than 1,000 such issues, ranging from minor to serious. Four patients had died.

Federal law requires medical device companies to report possible product-related injuries to the Food and Drug Administration within 30 days of learning about them. Instead, Medtronic employees shut down the study in spring 2008 without telling the government anything.

Scrutiny of Infuse would grow tremendously over the ensuing years. More patients developed problems; thousands filed injury complaints. Congress, the Justice Department and several states opened investigations and demanded information from Medtronic. To this day, neither the company nor the FDA has publicly disclosed full details of the study.

"I, for one, feel duped," said Dr. Charles Rosen, a California surgeon and co-founder of the Association for Medical Ethics. "I used [Infuse] on a lot of people."

Medtronic, which reincorporated in Ireland last year but maintains its operational headquarters in the Twin Cities, said Infuse has been used in more than 1 million patients and has a "well-established" safety profile. The company says the number of complications that turned up among the 3,600 patients with surgeries between 2002 to 2006 wasn't surprising given experiences in earlier clinical trials. And it said reporting doctors concluded that Infuse was not connected to the deaths.

Omar Ishrak, chief executive of Medtronic, the world's largest medical device company, wouldn't comment for this story, but company vice presidents acknowledged that Medtronic should have promptly reported the adverse events when the company first learned of them.

The executives said in a meeting with the Star Tribune that the database of adverse events from the study was misfiled internally, and that it was eventually reported to the FDA after it was rediscovered more than five years later. The company said no patients were hurt by the delay and the company has since improved its procedures.

"There was no analysis to suggest there was a problem … or [that] there was an unusual set of complications and we better keep this quiet," said Richard Kuntz, Medtronic's chief scientific, regulatory and medical officer.

Infuse's packaging has long disclosed potential side effects — bony overgrowth, spinal-nerve inflammation, bowel and bladder problems, male sterility, breathing issues. But at the time Medtronic shut down its study, the FDA's public database of post-approval problems with medical devices contained just 261 reports about Infuse — a quarter of the number of complications the company had gathered.

Medtronic would eventually tell the FDA about more than 1,000 adverse events discovered through the study, more than five years after they were due.

The FDA raised no issues about the late reporting and blacked out the total number of events from the three-sentence summary that became public. That number was revealed just months ago, after the Star Tribune challenged an FDA decision to keep it secret.

Vast medical potential

Infuse Bone Graft is not a traditional medical device.

Its key component is a vial of special powder that gets mixed with sterile water in the operating room. The bioactive solution contains a genetically engineered human protein known as BMP, or bone morphogenetic protein, that works like a drug inside the body to cause bone formation.

In the 1990s, researchers realized that BMPs could transform common surgeries like spinal fusion. That procedure treats back pain caused by degenerative disc disease, but many patients require a second surgery to harvest living bone from the hip to graft vertebrae together.

Lab-grown BMP offered the promise of bone graft without the time and pain of a second surgery.

The commercial potential was clear to Medtronic executives, who were looking to diversify the company's product line beyond its traditional heart devices. In 1999, they paid $3 billion to buy a Memphis company that had exclusive rights to make a specific bone-growth protein, called rhBMP-2.

Three years later, data gathered from clinical trials persuaded the FDA that the product was safe and effective for limited use in humans. The studies, however, looked at just one of the many ways that doctors can perform a spinal fusion with Infuse Bone Graft: Surgeons operated on patients from the front, and inserted the Infuse solution into a specific type of threaded titanium cage that went between vertebrae in the lower back.

Medtronic presented its data for FDA approval in early 2002. Critics like former FDA scientific reviewer Patsy Trisler, then working as a consultant to device companies, questioned whether enough was known about the safety of rhBMP-2, given the significant risk that doctors would use the device in ways not evaluated for safety. Once the FDA approves a device, it is legal for doctors to use it any way they think will help patients.

"After this product type is out in the marketplace," Trisler told the agency's Infuse safety panel on Jan. 10, 2002, "the potential for patient harm is great."

Six months later, the FDA approved the Infuse Bone Graft for fusion of bones in the lower spine using a surgical approach in the gut, but not the back. This narrow approval applied only when doctors combined rhBMP-2 with Medtronic's patented titanium LT Cage, and a specific collagen sponge. More than a million people worldwide have had rhBMP-2 put in their bodies since then.

Doctors try new uses

As Trisler had warned, doctors used Medtronic's new product in all kinds of ways. FDA injury records for early patients show that surgeons used Infuse to fuse bones in the neck, instead of the lower spine. They used it in surgical approaches from the back, instead of the front. They fused more than two bones together.

Medtronic eventually persuaded the FDA to approve use of the Infuse Bone Graft in some dental procedures and others involving tibia leg bones.

Other non-FDA-approved uses continued despite being controversial. Doctors experimented with Medtronic's BMP on children, even though the FDA said from the outset that it was only appropriate for patients whose skeletons were fully formed.

At least 85 percent of the 340,000 BMP surgeries in the United States between 2003 and 2007 were for uses the FDA had not reviewed for safety or effectiveness, according to a study in the journal Spine. BMP use grew fourfold during those years.

Medtronic executives had promised shareholders that Infuse could become the "standard of care in spinal fusion therapy." They wanted more FDA-approved uses. To gather evidence for those new uses, a team reviewed medical records from patients who had received Infuse between 2002 and 2006.

Looking back, current executives say it was a strange decision.

"That's not the mainstream way that we do research," Medtronic's Kuntz said in an interview. "It's not a very effective way to do research, which is probably why we hardly ever do it." Medtronic executives say the company's research and development group settled on the study design after consulting with someone at the FDA, though they declined to reveal who. Kuntz and other executives said the effort had so little statistical significance that they don't even like to call it a real study. The term they use is "retrospective chart review."

Some of the employees who did the work called it something else. They called it "the Mother of All Submissions."

Reviewing patient charts

The project eventually covered the medical records of 3,647 patients at 26 sites — 13 times as many Infuse patients as the studies that led to FDA approval of Infuse a few years before.

Medtronic employees examined every patient record they received, Kuntz said. The charts showed many good outcomes, but more than 1,000 patient complications, some of them serious. Two patients with heart problems died. Another died from cancer, and a fourth from abdominal inflammation. The doctors who reported the deaths concluded Infuse was not the cause. (Medtronic would later tell the FDA that it was unable to determine the definitive causes.)

The company presented a small slice of the retrospective chart data to persuade the FDA in 2008 to approve an untested use of Infuse in a small number of unique cases. But company officials say the FDA would not accept the retrospective chart review as evidence to support broader uses of Infuse. The regulators, whom Medtronic did not name, no longer thought the data would be useful in evaluating additional uses of Infuse, regardless of what they showed.

By May 2008, employees had filed the data in the company's internal archives, without reporting anything to the FDA.

"The question was, of all those things they abstracted from the charts, was that reportable to the FDA? And that's where the clarity wasn't," Kuntz said. "Whether a chart review is covered by our clinical policy is probably something that we were confused about."

At the time, Medtronic was drawing criticism for its handling of adverse events.

A 2007 warning letter targeted Medtronic's neuromodulation device facility for repeatedly failing to meet the 30-day reporting requirement for serious injuries. A subsequent FDA warning letter in 2009 documented more failures to report adverse events. One employee responsible for deciding whether to file such reports "only had a high school diploma with some additional in-house training," FDA inspectors noted.

Medtronic said in 2011 that it had resolved those issues to the FDA's satisfaction.

Kuntz said the company did analyze and record each adverse event in the Infuse retrospective study.

"They characterized them very clearly in the database," Kuntz explained. "When they kind of discovered through discussions that this just wasn't going to be a pathway to lead to an expanded [approval] , they stopped the project and closed out the study."

The company would later tell the FDA in a written report that the study data "were not assessed for reportability until June 2013."

Katie Nyberg said she developed postoperative complications and had to relearn how to walk.Reports of bad outcomes

During the years the data sat unreported in Medtronic's archives, government insurance programs paid hospitals more than $4.5 billion for surgeries that included the Infuse BMP biologic. For Medtronic, sales of Infuse hit $750 million in 2011, according to an analyst estimate.

At the same time, the FDA's adverse-events database began to catalog a growing number of patient injuries.

Pat Beaudry, now 58, went from working as a salesman to living on Social Security disability and taking addictive painkillers because of problems he blames on Infuse.

The Apple Valley man said his first spine surgery in November 2007 worked initially, but his back pain returned worse than ever in September 2008, and his arms and legs began to go numb. As he pondered the revision surgery his doctor recommended, Beaudry said he called Medtronic to ask about problems with Infuse.

"Talk to your doctor," he recalls being told. "We provide him with the information."

Beaudry said he already had asked his doctor about the potential for Infuse to cause nerve-damaging bone overgrowth. The physician, Beaudry said, had answered that it was possible, but he didn't have the numbers to say how often.

"I feel I was a guinea pig for somebody's piece of equipment," Beaudry said.

At the time of the retrospective study, a lack of information about post-market problems created a false risk profile for doctors using Infuse in their practices, said Dr. Eugene Carragee, a Stanford surgeon who led a Spine Journal investigation of Infuse research.

He said, "The company was saying, 'We're just not seeing much problems with this.' "

Katie Nyberg, 45, also had Infuse surgery in 2007. She said she developed bony overgrowth following the operation, which left her unable to stand for long periods or walk without assistance. She relearned how to walk, but said her disabilities caused her to lose the hair salon she co-owned. Without income, she ended up living in a storage garage near St. Cloud for a time.

"I can't figure out why they put something so horrible in me," Nyberg said. "I'm still trying to figure out how this happens in the United States."

Both Beaudry and Nyberg have sued Medtronic.

Requests for information

As questions mounted, the government started looking for more information about Infuse.

In October 2008, the U.S. attorney's office in Massachusetts subpoenaed documents from Medtronic. In June 2011, the Senate Finance Committee demanded all information on "adverse postoperative events" involving rhBMP-2. During the same period, four state attorneys general also asked for Infuse data.

Medtronic handed over thousands of pages — but not the retrospective study, which the company says was misfiled in its archives.

The company said the U.S. attorney's office in Massachusetts, a branch of the Justice Department, closed its criminal and civil investigation of Infuse in 2012. Other state investigations continue, and the company said it shared information about the retrospective study after it was rediscovered in 2013.

Sen. Charles Grassley, R-Iowa, who directed the Senate Finance Committee's Infuse inquiry, told the Star Tribune that the committee should have received the retrospective study data. In 2012, the finance committee concluded that Medtronic had been heavily involved in shaping, editing and writing the research of company-paid physicians who collected $210 million from 1996-2010. Some of that research, the committee said, downplayed Infuse's problems and emphasized the problems of competing treatments.

Amid these inquiries, the North American Spine Society's Spine Journal spent an entire issue reviewing early Medtronic-sponsored Infuse studies. The journal concluded in June 2011 that the risk of adverse events from rhBMP-2 was "perhaps 10 to 50 times the original estimates calculated from industry-sponsored studies."

Later that summer, Medtronic hired Yale University to reanalyze all of its published and unpublished clinical trial data and adverse-event reports filed with the FDA to rebuild public trust and establish a "landmark model" for transparency.

At that point, the researchers working for Yale did not get the retrospective study data either.

James Kirwin is a drug-industry veteran whom Medtronic hired in January 2012 as a vice president of clinical affairs in its spine and biologics division.

He supervised the team that found the unreported study in February 2013. Kirwin says he was forced to resign from Medtronic seven months later; he believes it was related to his advocacy for reporting the study to the FDA.

Company officials say he was not terminated because he pushed to report the study. Further, they say Kirwin was notified about the plan to report the events and raised no concerns.

Kirwin said he worried that the company might have violated federal law in not properly reporting the adverse events — both in 2008, when the study was shut down, and in 2013, when Medtronic waited at least four months to file reports with the FDA.

Medtronic's current spine clinical research director, Kim Michelson, was not at Medtronic in 2008. She said most of the people who ran the study from 2006 to 2008 have left the company.

She said it appears they were not clinical trial experts. Although the FDA says it has not changed adverse event reporting rules since the 1990s, Michelson said employees collecting and cataloging patient complications were not sure what to do with them.

"That's … what I've been able to ascertain, and where I would land," she said. "So, team turnover, unclear guidance, and then lack of expertise in trial management."

Richard Kuntz, Medtronic's chief scientific, regulatory and medical officerYale's assessment

Two independent teams of researchers hired by Yale were nearly done with their work by the time Kuntz told Yale project director Dr. Harlan Krumholz about the retrospective study data, Krumholz said.

Krumholz said he never saw the study database, but he opted not to provide it to his researchers in Oregon and the United Kingdom because they were focused on the results of conventional clinical trials.

The researchers in the United Kingdom found small improvements in fusion rates among Infuse patients, but more frequent short-term pain and "possibly, an increased number of cancer cases."

Using the same data, the researchers in Oregon concluded in June 2013 that in spinal fusion, Infuse "has no proven clinical advantage over bone graft and may be associated with important harms."

The Oregon team's paper, in the Annals of Internal Medicine, said it would have been helpful if more information on Infuse had become public sooner. "Earlier disclosure of all relevant data," the paper said, "would have better informed clinicians and the public than the initial published trial reports did."

Even when Medtronic began to reveal the existence of the study to the FDA, many details remained hidden from the public.

In the summer of 2013, Medtronic filed four death reports and consulted with the FDA about how to report the rest of the events. The company filed another injury report in September 2013.

Four months later, the FDA granted Medtronic's request to file a summary instead of the usual individual detailed narratives for virtually all of the remaining events. "Because this information was dated, and didn't reflect new public health concerns, FDA determined there was not a public health benefit to filing these reports individually," an agency spokesman said in an e-mail.

In February 2014, Medtronic submitted a sixth individual injury report that included a three-sentence summary of the remaining 1,039 adverse events in the "Mother of All Submissions." The events were also presented to the FDA in a 76-page table that identified each event with a medical code.

Only the three-sentence summary became public. The FDA blacked out the number of injuries from public view, calling it a corporate trade secret.

Medtronic and the FDA declined to release the 76-page table. The FDA told the Star Tribune to file a Freedom of Information Act request, which it did in May 2015. In August 2015, the FDA said it would take another 18 months to fulfill the request.

Only in September 2015, after the Star Tribune repeatedly challenged the FDA, did the agency unseal the number of adverse events in the Medtronic study. A spokesman said keeping it confidential had been a mistake.

The fact that the study remained undisclosed for so long raises questions, Minnesota Sen. Al Franken said. "We need to get to the bottom of what happened, both with Medtronic and the FDA," he said in an e-mail. "The public deserves answers."

Medtronic executives argue that the delay in reporting results from the retrospective study didn't hurt anyone, because the work did not show anything new. In addition, they said that when they reported the events to the FDA, they took a conservative approach and disclosed events that didn't fit the standard of reportability.

"We've analyzed the data, and we don't see anything there," Kuntz said. "We've sent it to the FDA … and there has been no action. It's just kind of a no-issue."

A committee of the North American Spine Society said in 2014 that the few adverse events reported in early studies may have encouraged a "laissez-faire" attitude about Infuse among the first doctors to use it.

David Prince of Mitchell Hamline School of Law, who has studied the issues around Infuse, said he believes proper reporting of the retrospective study might have affected treatment decisions.

"In some ways what's curious about this particular case is the FDA's behavior, even more than Medtronic's behavior," Prince said. Infuse "is a widely, routinely used product. When you get that kind of feedback from the marketplace telling you things are going wrong with some frequency, to just shrug and say, 'Well, we won't look at it any further' is the ostrich approach. You're just burying your head in the sand."

Today, Medtronic's lawyers are sorting through more than 6,000 injury claims from Infuse patients. The company is fighting a racketeering lawsuit from the insurer Humana over Infuse claims, and it has paid out $85 million to shareholders who complained that Infuse controversies hurt the stock price.

Medtronic, meanwhile, continues to aggressively market Infuse. Sales, which slumped in recent years, showed double-digit growth in the United States in the most recent quarter. In a March 1 earnings call with stock analysts, Medtronic CEO Ishrak said the company plans to expand its market share for biografts using Infuse.

jim.spencer@startribune.com 202-383-6123joe.carlson@startribune.com 612-673-4779maryjo.webster@startribune.com 612-673-1789UPDATE

On Monday morning, Medtronic issued an official response to this article.