It is utterly impossible to pigeonhole Dr. Jerald Schindler. We’ve tried.

Schindler confounds the three easiest caricatures that any self-respecting journalist would reach for. Schindler is a biostatistician, but speaks fluidly, lucidly. He’s a vendor, but leaves his Cytel sales pitch in the overhead luggage bin on the plane to the meeting. President of the research division at Cytel, he’s a regular on the conference circuit. But he’s always got something fresh.

The etrials people got Schindler down to their recent 2007 user meeting, and he did not disappoint. After so much talk about the possibilities of adaptive trials, it would seem there is little left to say. Schindler came up with something. He recognized the people in the room already understood the value of electronic data, and gave them another reason to nudge their organizations to gather more of it.

Great Explainer

Despite its usual mixed signals, even after an invitation-only conference late in 2006, the FDA basically likes adaptive trials. As part of the FDA’s Critical Path wish list, regulators hope to spend a little money researching the technique. There is only the question of how fast they’ll push for such studies.

The industry, for its part, sees adaptive designs as a bone saw to cut money and time from clinical development schedules. Statisticians are on board. And Schindler can explain all of that more cogently than anyone.

The basic idea, Schindler told the crowd in Orlando, Florida, is that an adaptive trial uses data not available at the start of a trial to modify the trial. As an example, dosage arms of a trial can be dropped or added without any sacrifice of statistical power. In a few years, he speculates, entire phases of development may be eliminated or merged into each other. Considerable savings are possible.

Best of all, sponsors will not guess so much. “You wind up reducing your risk of clinical development,” Schindler told the etrials meeting. “You have this opportunity to focus on better treatments.”

Quicker Decisions

In some trials, sponsors will learn a drug works sooner. In others, the decision to kill a compound will be made earlier. “If the drug isn’t working, you just stop,” Schindler says. “You don’t have to finish the entire program. From an ethical perspective, why enroll patients in a trial where the product doesn’t work?”

All too often, Schindler went on to say, drug companies realize they chose the wrong dose only at the end of a trial. Further development of the compound is killed rather than resuming the search for the best dose. The great allure of adaptive designs is that they might allow sponsors to explore multiple dosages in parallel, early in the process.

Get Data. Use Data.

Here’s how he explains the idea: “You can pick multiple doses and have a competition. You allow them to naturally select themselves and let the winner show up.” Schindler often invokes this Darwinian, natural selection metaphor for finding the right dose using technology. It’s spot on.

For most of his presentation, Schindler focused on clinical data and continuous electronic access to it.

Cultural Obstacles

But the first obstacle to using adaptive design is cultural and attitudinal. It’s new. The industry is conservative. Schindler, having spent time at Wyeth, is diplomatic about this. The adaptive approach, he says, “impacts everyone across the process of clinical development. It is not just a statistical methodology.”

“This is new,” he says. “We’ve never done it before. There is this inertia thing. Companies have to accept that there is process change.”

Real-time Data

Beyond that, adaptive designs will take technology. It’s not just a question of investing in electronic data capture (EDC). That’s almost non-negotiable. Why? Because paper case report forms just take too long to work their way through any company in the industry.

“It’s not just enough to get the data in the database,” says Schindler. “You have to have tools to look at the data. Then you have to have a process that people actually look at the results that the statisticians come up with.”

Another must-have, at least for the savviest organizations, is a system that links different stakeholders, all with different roles and domain expertise. It sounds simple. It isn’t. “The way to make this work effectively, I’ve found, is to have a nice e-clinical system that everybody can access 24/7, where ever they are.”

No Silos

The goal is to have one source of data, available via the web, so that everyone in the organization has the same level of visibility into a newly stopped (or started) arm of a trial. Ideally, Schindler said, “everybody goes through the same source. So you don’t have people in data management having one answer and someone in clinical research having a different answer.”

Yes, there are lingering regulatory worries. There are a few arcane statistical nuances. But Schindler doesn’t think those are the real issue limiting the usage of adaptive or flexible designs. “In some ways, the statistical and regulatory issues were the smokescreen for data availability and data access,” Schindler says.

In the blogosphere, they call that the money quote. In church, they say “Amen.”

As he related to ClinPage, Schindler believes the merits of interactive voice response (IVR) systems are not sufficient to support adaptive decisions on their own. That’s because the clinical data is not in IVR systems. The data does live in EDC systems, and in clinical data warehouses and repositories.

Further Thoughts

Half a year ago, we did a cover story on adaptive trials for our former employer. More recently, we wrote a much shorter article about Genzyme.

Do you or someone in your company have insights into good strategies for deploying adaptive designs? Please contact the . He is keen to learn whether anyone out there is a) actually doing an adaptive trial, b) doing one that includes modeling and simulation, and c) interested in a public discussion of noncompetitive, operational ideas to help the industry explore the approach.

d9A2t49mkex