Warning message

You need to be logged in to add this content to your library.

When a clinical trial comes up short, typically the data gets parked at the pharmaceutical company and the FDA. No one else uses it; no one learns from the other’s misfortune. Over the course of the past year, the Coalition Against Major Diseases (CAMD), a collaborative initiative of the Critical Path Institute, has developed an open tool to connect those silos of information. Last June, CAMD announced the launch of the C-Path Online Data Repository (CODR). CODR is the first database in which pharmaceutical companies share Alzheimer’s disease placebo and some treatment data for researchers in academia and industry worldwide to use freely.

The database uses a standard format, favored by the Food and Drug Administration, which was developed with the nonprofit standard setting organization CDISC (see Part 2 of this series). CDISC helped CAMD develop the data standard, and scientists at each company then mapped their own dataset to it so the data could be plugged into CODR. In turn, CODR forms the basis for the building of a quantitative drug and disease simulation model of human Alzheimer’s disease. This model—another CAMD project—will give researchers a stronger empirical basis to simulate therapeutic trials. Pharma, academic, and government scientists who gathered at a CAMD conference held 30 November 2010 in Washington, DC, widely agreed that such a simulation model would go a long way toward reducing the risk of failure in those trials.

How far along is this plan? “Last year at this meeting, I said we had interest from six companies in data sharing. The question hung in the room: will they really do the work required to share?” said Steve Broadbent from C-Path. “Now we have that database. More than 200 researchers from all over the world have applied for access and received it; we get some 20 new applications per week.”

What do those researchers find in the database? As of now, placebo and some therapeutic data on 3,600 patients from nine trials by six CAMD member companies, i.e., Abbott, AstraZeneca, GSK, Johnson & Johnson, Pfizer, Sanofi-Aventis, Stephen Kopko of CDISC told the audience. One trial promised by Novartis is yet to come, and a second Pfizer trial is currently being transferred. Eli Lilly and Company has promised to contribute its semagacestat data (see ARF related news story), said Marc Cantillon, who directs CAMD.

At present, Kopko and colleagues are translating their first non-industry trial into the CDISC data standard. It is the homocysteine trial by the Alzheimer’s Disease Cooperative Study group. Next on the list is ADNI, which would add invaluable biomarker and disease progression data to the repository. At this point, the database contains mostly older trials of symptomatic drugs in mild to moderate AD. CAMD wants to expand it with trials that are newer, use biomarkers, and are done at milder/MCI stages. “We were limited so far because companies were not always willing to give us datasets. For NIH-funded trials, we hope that will be easier,” Cantillon said.

Kopko hopes that the January publication of CDISC’s AD data standard will encourage more trial sponsors to align their existing data with this standard and contribute them to CODR. New trials being planned now and in the future can adopt this voluntary standard.

CODR could be a bounty for all, scientists said. To pick a hypothetical example around the latest high-profile setback, Eli Lilly could have theoretically decided not to spend millions of dollars on semagacestat, or design the program differently, had it been able to analyze data by other companies that had been testing γ-secretase inhibitors before. It was common knowledge that Merck and Bristol-Myers Squibb, for example, had been discontinuing compounds with this activity after clinical tests. “As it is, each company has access to only their own, limited AD clinical trial experience,” Cantillon said. Often, poor drugs for various reasons are kept alive too long only to fail in late stages, EMA regulator Cristina Sampaio said at a meeting earlier this year (see ARF Springfield story).

A Pfizer scientist noted that CODR could help companies recruit because it enables them to make a commitment to their trial participants that they will share data equitably and create value from otherwise unsuccessful trials. “We can say to patient groups and patients: The compound we are testing here may die. But your effort is not lost because we give the placebo data back to the community. Enrolling in AD trials is benefiting all AD research, not just this sponsor,” he said. Patients and relatives are frequently dismayed if information gleaned from their research participation is not shared.

A further boon lies in better understanding placebo responses. When a treatment trial falls flat, sponsors sometimes blame insufficient placebo decline, but research has shown that a given trial’s placebo group is often too small to support such claims (see ARF related news story). The CODR shows how placebo groups truly behave.

The database will continue growing for the research community’s use. However, Kopko said, at some point the CAMD disease model group will cut the existing data to build a simulation model of Alzheimer’s disease and ask the FDA and the European Medicines Agency (EMA) to formally qualify that. This model is a central piece of CAMD’s vision.

Why qualify a model? The idea is that a model of the progression of Alzheimer’s would enable all drug developers to optimize a slew of how-to questions about their next planned trial. This is not to replace actual trials with some sort of virtual version, said Klaus Romero of CAMD. Rather, it is to simulate specific questions of trial design—staggered start, when to take samples, to name just two examples—and see how tweaking those would likely affect the outcome. Most importantly, perhaps, a simulation model can help scientists pick the right endpoints to measure.

“The groundbreaking thing here is that we have the regulators along every step of the way,” Romero said. Indeed, Romero and his colleagues presented a draft research plan on how to build the model to the FDA in April and to the EMA earlier this fall. “It was very exciting to have this be a multi-company meeting,” said a scientist at AstraZeneca who helps CAMD with all regulatory aspects. “It was a different feeling. It was collegial.” Working independently, each pharma company would not get this kind of input from the agencies.

“By developing disease models using a broader range of data than any company can generate on its own, research will be able to design more effective treatments and diagnostics,” said Mark McClellan, who co-developed the Critical Path Initiative in 2004 when he headed the FDA. McClellan is now at the Brookings Institution.

Regulators strongly support publicly shared simulation models for trial design, said Jogarao Gobburu of the FDA. He cited one example of a model for non-small cell lung cancer that helps researchers simulate how their drug will behave in trials. Another such model, in pulmonary hypertension, helped FDA scientists to determine what the right endpoint would be in pediatric trials testing drugs originally made for adults. Those sorts of goals are why the FDA qualifies disease simulation models.

The day’s discussion briefly touched on whether such a model could serve as a control in lieu of placebo. In general, regulators prefer placebo controls, but some speakers noted that it can be difficult in late-stage trials to find participants who will honestly stay on placebo and not find a way to get drug. This issue came up at a recent EMA conference on clinical trials in preclinical neurodegenerative disease, and has generated scientific commentary (see Spiegel).

The CAMD disease models group is currently implementing the FDA and EMA’s feedback on their draft research plan for how to build the model. For example, regulators requested that an external review panel add clinical expertise to the model; Lon Schneider of the University of Southern California will lead this panel. Much work remains to execute the plan, and the group will need to consult the agencies again, but CAMD hopes to submit the model for qualification by end of spring 2011, Cantillon said. Pfizer has committed its regulatory staff to write and submit the qualification documents for this model, which will be freely available to all drug sponsors in industry and academia.

The simulation model provides a quantitative description of the disease. It would be a combined drug and progression model, because it draws on CODR placebo data, the published literature of older drug trials in nearly 20,000 patients, and the ADNI database. The last contains biomarker and psychometric data on the natural history of pre-AD and AD, and has led to a staging model of preclinical AD (see ARF Live Discussion). The model should give investigators a range of outcomes depending on how they tweak features of their intended trial, Romero said. Members of the CAMD modeling group published a paper about their collaborative approach this past September in the Journal of Clinical Pharmacology (Romero et al., 2010). Read Part 4 on biomarkers, Parkinson’s, and growing pains.—Gabrielle Strobel.

This is Part 3 of a four-part series. See also Part 1, Part 2, and Part 4. See also a PDF of the entire series.

Comments

No Available Comments

Make a Comment

To make a comment you must login or register.

References

News Citations

  1. DC: Standard Data—Music to Regulators’ Ears
  2. Lilly Halts IDENTITY Trials as Patients Worsen on Secretase Inhibitor
  3. Geneva: At Springfield Meeting, Soul-Searching About Dearth of Drugs
  4. Chicago: Studies Probe Diminishing Placebo Decline, Part 1
  5. London: What Regulators Say About Trials in Familial AD
  6. DC: Biomarkers, Parkinson’s—CAMD Needs All Hands on Deck
  7. DC: CAMD Convenes Stakeholders to Reform Alzheimer’s Trials

Webinar Citations

  1. Together at Last, Top Five Biomarkers Model Stages of AD

Paper Citations

  1. . Pharmacometrics as a discipline is entering the "industrialization" phase: standards, automation, knowledge sharing, and training are critical for future success. J Clin Pharmacol. 2010 Sep;50(9 Suppl):9S-19S. PubMed.

Other Citations

  1. PDF of the entire series

External Citations

  1. Coalition Against Major Diseases
  2. C-Path Online Data Re
  3. CDISC

Further Reading

Papers

  1. . Independent association of an APOE gene promoter polymorphism with increased risk of myocardial infarction and decreased APOE plasma concentrations-the ECTIM study. Hum Mol Genet. 2000 Jan 1;9(1):57-61. PubMed.