March 20, 2000 – Pharmaceutical testing takes forever and costs a fortune. Swiss giant Roche made a dent in the problem with some assistance from tiny Health Decisions.
Like divorces in Hollywood, clinical drug trials can be long, messy, expensive affairs. So when Dr. Michael Bowden, a pharmaceutical researcher, was put in charge of testing a promising new treatment for Alzheimer’s, he looked for ways to cut time and money.
An intimidating task, since the study involved 1,500 patients from 107 locations in six countries. Bowden was working for a drug-testing arm of the Swiss pharmaceutical titan Roche. On trial: a molecule called lazabemide. Lazabemide impedes an enzyme in the brain in the “monoamine oxidase” family. There is some intriguing evidence that too much MAO-B is connected to Alzheimer’s.
Perhaps slowing the enzyme might slow the dementia. After a decade in development, the drug was ready for a round of testing in patients to determine both its safety and its ability to thwart senility. To get the go-ahead from the Food & Drug Administration Roche had to demonstrate, among other things, that lazabemide could achieve at least a two-point difference in the deterioration of memory and other cognitive skills between patients who took the drug and patients on a placebo. (The test is scaled so that a typical Alzheimer’s patient might experience a four-to five-point change over the course of a year.)
A typical study would take five years and $110 million for the Phase III trial. Could Bowden do better?
In September 1996 Bowden called Dr. Michael Rosenberg, an epidemiologist and the founder and chief executive of Health Decisions, in Chapel Hill, N.C. Health Decisions specializes in developing electronic tools for clinical trials, competing with Quintiles Transnational and others in the trial outsourcing business. With $10.5 million in 1999 sales privately held Health Decisions is dwarfed by Quintiles ($1.8 billion). But it has one important strength: It’s wired. Bowden claims it’s tops in using the Internet to run trials and analyze data.
Instead of relying on doctors’ handwritten notes on each patient, mailed to Roche at the end of the study and entered into a database over the course of a month, Rosenberg adapted an old idea: bubble sheets, like those used in standardized tests. Doctors bubbled in sheets after bimonthly patient visits and overnighted the forms to North Carolina, where they were fed into optical reader machines at a rate of 10,000 pages a day. The system required one-third the number of people needed for traditional data entry, and took a few hours instead of a month.
Rosenberg also developed an Internet-based system to communicate with the doctors at every testing site. If any information appeared to be bubbled incorrectly, a Health Decisions server sent an electronic query to the doctor, asking for clarification. A typical drug study generates four queries per response page, at a cost of $150 per query; Health Decisions’ forms cut the rate to one query in 20 pages. And in the typical drug trial, inconsistencies and data collecting problems get caught at the end, when doctors have forgotten the particulars of a patient’s visit. Working on the Internet, Rosenberg was able to detect immediately that one interviewer was administering the test improperly, and had to have him retrained.
Health Decisions also helped screen for patients who were sick enough to meet the minimum requirements for the study. Four test sites recruited subjects who weren’t as sick as Rosenberg wanted–a fact that was obvious from the database results early in the study. A phone call to the sites fixed it.
All data was submitted by January 1999. Just two weeks later the database was locked up; the delay in a typical study runs into months. The last entry went into the computer on a Friday. The computer crunched away all night. By the next morning Bowden logged on to the Web site to get the final results.
A disappointing conclusion: The drug wasn’t working well enough. Although the course of patients’ deterioration did slow somewhat, the decline was not statistically significant enough for FDA approval, Bowden concluded. Lazabemide delivered only a 1.5-point advantage on the test of cognitive skills, falling short of the 2-point hurdle. That disappointment, and the fact that a few patients had liver problems, perhaps because of the drug, killed lazabemide.
For Rosenberg it was a triumph. His methods spotlighted the drug’s failure in less than three and a half years, ultimately saving Roche $32 million. For its efforts Health Decisions nabbed a $20 million fee over three years, big business for a company that pulled in only $1.5 million the year the project started.
Aftermath: Bowden quit Roche to take a job with Health Decisions, where he heads a European office. Rosenberg is trying to land more big contracts like the one he had with Roche. And Roche is taking the disappointment with monoamine oxidase inhibitors in stride. The market for an Alzheimer’s treatment is so huge that the pharmaceutical industry can afford to spend a few billion dollars in the search for a good one.
Nineteen new Alzheimer’s drugs are in development.
Chances are that few, if any, will be blockbusters.
$24 billion spent on R&D by drug companies in 1999.
33% of R&D spending goes to human clinical trials.
1 the number of drugs approved for every 5,000 compounds tested.
15 years the average length of time for a drug to go from the lab to the market.
20 million Americans may be afflicted with Alzheimer’s by the year 2050.
Reprinted from Forbes.