It’s called the Diasensor 1000. The company that built it, Indiana, Pa.-based Biocontrol Technology Inc., claims it’s the answer to diabetics’ prayers.
The device is said to measure blood sugar levels using an infrared beam of light shot painlessly through the skin. A breakthrough that would liberate millions of diabetics from the often torturous daily ritual of pricking their fingers to test their blood.
The problem is, the Diasensor doesn’t work.
Contrary to the company’s claims of having developed a miracle machine, results of clinical studies on the Diasensor have been horrible, interviews with current and former employees and confidential internal memos show.
Faced with such poor results, Biocontrol ended up doctoring patient test results it submitted to the Food and Drug Administration and staging bogus demonstrations of the Diasensor for potential investors and others in a bid to give the impression the machine was functioning.
The goal was simple: to continue convincing a new flow of investors- Biocontrol’s lifeline – that a salable product was just around the corner.
“The only objective was survival of the company, survival of the project,” said one former employee who witnessed the tampering and said he eventually left the company as a matter of conscience.
Like a number of other former employees, the source asked not to be named. He and the others said they feared reprisals from the company for speaking out.
“We know from a scientific point of view (the Diasensor) doesn’t work,” said one top Biocontrol researcher who just left the company this summer.
Biocontrol Chairman and chief scientist David Purdy and Chief Executive Officer Fred Cooper did not respond to repeated requests for interviews for this story.
In a written response to questions faxed to the company’s offices, Biocontrol called assertions that the Diasensor doesn’t work “preposterous.”
The company said it believed such statements were made by “disgruntled former employees” who “are trying to cause problems for the company but who do not understand the documents and data they are reading.”
Although the FDA has twice rejected Biocontrol’s request to sell the Diasensor in the United States, company officials have for years maintained the machine is ready for market. Most recently, Biocontrol said it was preparing to submit its third application for marketing clearance in the United States and was getting ready to initiate its first sales of the device: in the Far East and the Philippines.
But according to people who have worked closely on the project, the Diasensor shouldn’t be used by anyone.
Rich Wiggins, Biocontrol’s long-time manager of research and development, said in a recent interview that although he believes it’s possible the Diasensor can be made to work someday, he isn’t sure if the device has ever measured glucose.
Wiggins said that when he learned in May that top executives were planning to try to sell he Diasensor in the Far East, he quit in protest.
Unlike the company’s previous promises of overseas sales, this time, Wiggins said, the plans “seemed serious.” That scared him, he said.
Does Wiggins believe that the company, after spending 11 years and more than $100 million raised from shareholders to develop the Diasensor, has proved the machine works on anyone – even one test patient?
“I was never terribly happy with any of the data,” he said.
Over the years, the public has heard a far different version of the Diasensor’s performance, however.
More than a year ago, in a video tour of the company’s research facilities, Biocontrol claimed the Diasensor was performing with a 90 percent accuracy rate. While officials conceded the Diasensor wouldn’t work for everyone, they boasted that two-thirds to three-quarters of diabetics who tried the 40-pound computerized machine would be able to use it. And they described preparations for high-volume production.
The video, hosted by Cooper and Purdy, was sent to shareholders and stock analysts nationwide.
Cooper has contended that the company’s studies show conclusively that the Diasensor works. He has accused the FDA of keeping the machine off the market because of pressure from makers of fingerstick meters that stand to lose millions if the Diasensor is approved.
After the FDA denied approval for the machine following an advisory panel review in February 1996, Cooper called on Congress to intervene, charging the agency had rigged the review process against the company.
After a congressional subcommittee heard his complaints but declined to step in, he hired a private consultant, Jack Nard, to probe the agency for wrongdoing.
But at the same time Cooper was crying foul at the hands of the FDA, the company’s own research scientists were still trying to determine whether the Diasensor was functioning at all.
“Are we getting closer to locking onto glucose?” one company physicist wrote in an internal memorandum several months after the panel review.
Maybe, but who knows.
The Post-Gazette obtained confidential documents detailing the Diasensor’s performance from Biocontrol employees who said they wanted to tell the public, especially diabetics, what they knew about the machine.
Not only did top officials deceive the public about the status of the Diasensor, they said, but the company also sought to mislead the FDA by manipulating and altering clinical test results.
Insiders describe the months leading up to the 1996 panel review as a time of desperation and chaos.
Top Biocontrol researchers, who had been tinkering with the Diasensor for nearly a decade, were still stuck in the early stages of research, but were scrambling to pass off the findings to the FDA as much more.
Readings taken with the Diasensor on test subjects were plotted with dots on charts called error grids. Dots that lined up exactly along a 45-degree line would show perfect correlation between Diasensor readings and the patients’ actual blood glucose levels.
But the dots were landing all over the place.
Frantically, “they kept telling us to remove data,” said one former employee who worked closely on the submission.
He described “cram sessions” in which employees were kept at company labs late into the night reworking the charts, throwing out test results in an attempt to eliminate unfavorable readings.
Often, he said, “they were just making stuff up.”
“They’d say thing like, ‘Since the machine was probably warming up for the first 30 seconds, take that data out.’ Then, ‘Take out the last 30 seconds.'”
Readings that represented very high or very low glucose levels, also were tossed out.
When the grids still came back looking bad, “they’d try another way to extract even more data” by changing their methods for analyzing the findings, the former employee said.
In short, the company was “dredging” the data, a technique used during preliminary research to help form hypotheses for further study. Biocontrol, however, presented its findings to the FDA as results from carefully controlled clinical trials and as proof that the Diasensor worked.
The dredging continued until what had been huge “clouds” of dots on the pages were reduced to a more linear form on a few charts.
Yet, even among the handful of charts considered the best of the lot, readings continued to stray into the “danger” zones. Those areas contained results that were so far off from the test patients’ actual glucose levels that it would be dangerous, possibly fatal, if patients acted on them.
But it wasn’t long before those readings disappeared, too.
All that took was a little white-out.
Patrick Cooper, a PhD in physical chemistry in charge of FDA submissions for Biocontrol, “was very blatantly taking the error grids and applying white-out to dots … to cover up data points that weren’t favorable for presentation to the FDA,” said another former insider who also worked closely on the project.
These were dots in the danger zones, according to two witnesses, on charts that “should have gone directly to the FDA.”
Cooper “would white out the dot, blow on it, and give it to his secretary to photocopy so nobody would know the white-out was used,” one of the sources said.
Employees who expressed concern at the time about the machine’s safety were told not to worry.
“They said once we get past FDA approval, we’ll have time to refine the machine” before any Diasensors land in diabetics’ hands, a former employee said.
In the end, the FDA rejected the Diasensor anyway. The agency doesn’t discuss specifics of research data it receives, citing confidentiality. But the advisory panel cited numerous deficiencies with Biocontrol’s data. Most importantly, panel members said, test results on a handful of patients weren’t enough to prove anything.
Biocontrol said in its faxed statement Friday that the company “has never doctored and manipulated clinical test results.”
“If such an allegation were true, would we not have already received FDA approval with data doctored exactly to their requirement?” the statement said.
As for using white-out, Biocontrol said: “White-out was not used to ‘cover up’ bad results but was used rather to correct computational errors that would have caused the actual correct data to be misrepresented.”
Patrick Cooper, who is not related to CEO Fred Cooper, did not return telephone calls seeking comment.
“We take allegations of alterations of test data very seriously,” Dr. Bruce Burlington, head of the FDA’s Center for Devices and Radiological Health said late last week.
“We’re committed to following up….” he said. “It’s obvious the system relies on the scientific community having confidence in the quality of scientific data.”
When it came to investors, Biocontrol took a different tack to convince them the Diasensor worked.
It rigged demonstrations of the device, programming it to censor results so the readings would appear accurate.
Demonstrations involved taking a glucose reading on a test subject the conventional way using a sample of blood, and comparing that with a reading taken on the Diasensor with an infrared scan on the subject’s forearm.
What spectators weren’t told was the Diasensor had been set up to only display readings within a narrow, predetermined range.
If, for instance, the patient’s actual blood sugar level was expected to be roughly 120 milligrams per deciliter at the time of the demonstration, then the Diasensor’s display range might be narrowed to between 100 and 130 mg/dl.
During the demonstration, if the Diasensor came up with a number outside the preset range, it simply wouldn’t display it.
The machine automatically continued to retest the patient until it hit on a number within the tightly prescribed range.
“It would come up with so many random readings, eventually it would come up with some reading in the desired range” and display that one, a former employee explained.
Voila. The machine appeared to work.
The source said he first heard several managers “laughing” about having rigged the machine. Then he confirmed it.
“We got the data files back and saw the range had been tightly narrowed for the demonstrations,” he said.
The settings had been changed “with prior knowledge of what the patient’s actual glucose was.”
The rigged demos were put on “all the time” at the company’s research center for potential investors, the source said. A demonstration at a news conference at the Indiana, Pa., Holiday Inn in February was performed the same way, several former insiders said.
Biocontrol called those assertions “absurd.”
“Any adjustments to the machine were done merely to update to the most current patient algorithm,” Biocontrol responded in its written statement Friday.
Wiggins, Biocontrol’s former R&D manager who said he rebuilt the Diasensor shortly after joining the company in 1992, said he didn’t believe any demonstrations had been rigged.
“As far as I know, they were legit,” he said. “But I wouldn’t know if they weren’t.”
Wiggins also said he didn’t know if any clinical results had been doctored. He said he regularly reviewed patient test data, but didn’t see formal submissions to the FDA.
“It’s a very closed company,” he said.
But why would the company even request FDA approval for a machine that doesn’t work?
“I don’t know,” Wiggins said. “I don’t understand how the business was run. … I don’t like the way it was run.”
Since it started working on the project in the mid-1980s, Biocontrol and its Diasense Inc. subsidiary have used public and private stock placements to issue roughly 110 million shares – bringing in more than $100 million – to feed operations. Stockholders have included many diabetics and the physicians who treat them.
From the beginning, company officials have attracted those funds with hyperbole.
As far back as 1989, for instance, investors were being told to expect the Diasensor to hit the market within the year.
The company made its strongest official pronouncement the machine was ready in January 1994. That’s when executives first asked the FDA for permission to sell the Diasensor. (That application was rejected shortly after the February 1996 panel review.)
At the time of the filing, not only did the company not have the studies to prove the Diasensor worked, it didn’t have a machine. There was a prototype device, but it wasn’t capable of displaying any readings, or even trying to analyze patient scans.
Still, Cooper, the CEO, announced in a news release that the company was gearing up its manufacturing center in anticipation of FDA approval within 120 days.
In October 1995, Biocontrol announced a multimillion-dollar deal to sell thousands of Diasensors in Germany beginning in January 1996. Shortly afterward, the deal collapsed, with the German distributor saying Biocontrol couldn’t present a working machine.
Cooper, on the other hand, asserted that Biocontrol had canceled the agreement because it didn’t receive the required down payment. Canceling the deal was in the company’s “best interest,” he said, so it could begin selling the Diasensor in England “without a discount” at about $8,000 each. England is still waiting.
All the broken promises haven’t gone unnoticed by shareholders.
A class action lawsuit is pending in U.S. District Court in Pittsburgh charging Biocontrol and its top officers with violating federal securities laws by lying about the progress and effectiveness of the Diasensor.
The case, consolidated from individual suits filed by a handful of shareholders shortly after the 1996 panel review, was certified a class action in March. Depositions have not yet begun.
The company also is facing an inquiry by the Pennsylvania Securities Commission, which is looking into illegal stock sales to Pennsylvania residents. The probe was initiated after a Post-Gazette article last year revealed the company had sold unregistered shares of Diasense Inc. stock in Pennsylvania, telling buyers to use false out-of-state addresses to conceal the sales. Biocontrol officials have denied making such transactions.
Over the years, Biocontrol has racked up more than $100 million in losses, including $22 million last year and $12 million through the first six months this year. Expenses include six-figure salaries and fancy company cars for top officers.
Cooper, the company’s 52-year-old CEO who runs the business side of the operation while Purdy, 69, heads up research, is Biocontrol’s highest-paid executive. He received total compensation of more than $830,000 last year, up from nearly $815,000 in 1995. In 1994, his compensation exceeded $900,000.
Top officers, including Purdy, Cooper and Senior Vice President Anthony Feola, also have awarded themselves warrants to buy hundreds of thousands of shares of company stock. The more the market price of the company’s stock rises, the more valuable the
A big bounce in the company’s stock, which surely would ensue if the Diasensor were approved, would make top officers instant millionaires.
Last October, the company submitted its second application to the FDA to market the Diasensor.
Although Biocontrol had been told to present test results from a statistically valid sample of subjects to show the Diasensor worked, the company again backed its submission with just a handful of patients. The application was quickly turned away.
Mark Arnold, a professor of chemistry at the University of Iowa who has been studying the feasibility of infrared glucose monitoring technology for the last seven years, said he never believed Biocontrol’s breakthrough claims.
The Diasensor essentially is a spectrometer that shoots invisible light into the arm then converts the light coming back out into a spectrum of wavelengths. It then tries to analyze the spectrum for glucose concentrations.
The challenge – some say an insurmountable one – is being able to pick out the glucose molecules, because they are obscured by other body substances such as fats, protein and water.
Arnold studied numerous scientific papers, co-written by Biocontrol researcher Ralf Marbach, that detail experiments and assumptions upon which the Diasensor is based. He said the papers failed to show that researchers measured glucose at all. He said he believes any readings that appeared to relate to glucose happened by chance.
“Researchers cannot distinguish between direct detection of glucose and chance correlations ….” Arnold wrote in a report published last year in the journal Current Opinion in Biotechnology.
Researchers “have made the statement that they can measure glucose, and they’ve made it enough times that some people are starting to believe it. But it’s never been demonstrated,” Arnold said in a recent telephone interview.
Arnold said his work, which is funded by the National Institute of Health and Selfcare Inc., a Waltham, Mass., company that markets conventional glucose meters, is aimed at determining whether infrared glucose monitoring technology is worth trying to develop.
In its statement Friday, Biocontrol contended that Arnold targeted one experiment and overlooked other experiments where “genuine correlation was proven.”
Arnold said that wasn’t so. “They didn’t prove they were measuring glucose even with those other experiments,” he responded.
Opinions vary among ex-Biocontrol employees as to how sincere management has been in its efforts to produce a working Diasensor.
“Dave Purdy’s generally an honest, caring person,” said Wiggins. “I think he’s doing the best he can to get it to work.”
Others complain that employees who speak up about problems with the machine are “blackballed” and pressured to leave. And they say those who speak out too loudly can find themselves out of a job.
One former employee said he began questioning management’s motivations in the months preceding the 1996 panel review. Researchers were conducting preliminary tests on the Diasensor in preparation for a round of clinical trials. But the tests weren’t going well.
“Instead of taking a step back, and realizing the machine needs to be fixed, we went through with the patient trials in the face of common sense,” the source said.
“The scientific thing to do would have been to form a new course of action for what needed to be done to make (the Diasensor) work. That didn’t happen.”
Wiggins said he had the same concerns over how clinical testing was run.
“They never asked me when the research would be done on the machine so we can do clinical trials,” he said.
“They would say, ‘Here’s the date for the next clinical trial,'” turn over a machine.
One ex-management employee described the decision-making process this way:
“We do some tests on a couple patients, but these are not good. We change the instrument, do more tests. It’s still not good. Over and over, the cycle repeats.
“Then the decision is made that it’s time to submit something (to the FDA). So at the current state of development, it’s submitted.”
Insiders say Biocontrol has been able to maintain the loyalty of many of its 100-plus employees and keep them believing the Diasensor works by keeping them in the dark about the project.
Test results on the Diasensor are tightly guarded. And jobs are highly compartmentalized.
Besides a small inner circle of employees, “nobody else is allowed to see what’s going on,” said one former worker.
Some former employees said non-disclosure agreements that they were required to sign when joining the company had made them too fearful to come forward publicly.
When Biocontrol officials became aware that the Post-Gazette was working on this story, the company called former employees asking them if they had been contacted and urging them not to talk.
In recent months, a number of key employees have left the company. Some are jumping ship, others have been laid off.
Biocontrol announced in May it was trying to cut costs and was looking for other ways to pull back. With the market price of its stock at less that 30 cents a share vs. a peak of more than $6 two years ago, the company is under constant pressure to sell increasingly large numbers of new shares to stay afloat.
Some ex-employees say they’d like to forget their years at Biocontrol, but don’t think they ever will. They say they joined the company expecting to be a part of a noble cause, but left feeling bitter.
“You have this Manhattan Project, save the world type-mentality,” said one.
“You’re so proud of what you’re working on. Then, you find out the dark side, that it’s not real … that this has been going on since 1989 and nothing has really changed.
“You realize you wasted part of your life … You feel you’ve been used.
“You feel you’ve been had.”
— 1997, Pittsburgh Post-Gazette.