Verdict on clinical trials registries? Good, but must do better

from Nature Reviews Drug Discovery

Meredith Wadman

Critics say users find it hard to navigate even the best available option

The number of trials submitted to ClinicalTrials.gov surged as researchers tried to meet a 13 September 2005 deadline set by medical journal editors. Graph adapted from Zarin, D. A. et al. NEJM 353, 2779–2787; 2005.

Late in 2004, the upper management at GlaxoSmithKline had a decision to make. Within months, they needed to determine where to post details of all their actively enrolling clinical trials, in order to meet a deadline set by medical journal editors. With more than 350 clinical trials registries to choose from it looked like a tricky decision. But it wasn't. The answer was obvious.

“ClinicalTrials.gov is the largest clinical trial register in the world, accepts studies conducted outside the US and has a global reputation and reach,” says Rick Koenig, a GSK spokesman. Besides, he adds, it made practical sense to register all the company's trials on a database where, as required by US law, the company had already begun to register trials aimed at serious and life-threatening diseases.

Other companies, from the smallest biotechs to the biggest drug firms, have reached the same conclusion, with the result that ClinicalTrials.gov (www.clinicaltrials.gov) is rapidly becoming the Google of the clinical trials registry world. “I don't see anything else that's even close in size and scope,” says Barbara Godlew, Clinical Trial Registry Manager at Novartis.

As of 14 February 2006, ClinicalTrials.gov, which is run by the National Library of Medicine at the US National Institutes of Health, contained 26,972 clinical trials posted by 2,400 organizations in 130 countries. The number of trials registered has more than doubled in the past year, due mainly to a 13 September 2005 deadline for registration set by the International Committee of Medical Journal Editors (see GRAPH). After GSK settled a lawsuit from the New York attorney-general Eliot Spitzer, alleging that GSK had deliberately suppressed negative results from four clinical trials of its antidepressant Paxil (paroxetine), the medical editors' committee decided it would no longer publish the results of trials that had not been registered in advance in an independent database (De Angelis, C. D. et al. JAMA 292, 1363; 2004)

The journal editors said that two registries satisfied their criteria, though last month they admitted several more to this exclusive club (see Box). Before then, trial sponsors had been obligated to post to either the UK-based ISRCTN Register (http://www.controlled-trials.com/), which currently contains data on 4,164 trials, or the much larger ClinicalTrials.gov. The US database was launched in 2000 as a result of a new law that required sponsors to upload their open clinical trials for serious and life-threatening illnesses to an NIH-maintained registry.

But critics say that although ClinicalTrials.gov might be the registry of choice, it remains an incomplete solution to a complex problem. The registry is lacking in certain fundamentals such as comprehensiveness and lay-user-friendliness, they say.

For instance, cancer trials posted at ClinicalTrials.gov often list 'prior cytotoxic regimen' under 'exclusion criteria,' notes Paul Parry of the Arlington, Virginia-based Abigail Alliance for Better Access to Developmental Drugs. “The patient needs it to say: 'you can't participate if you've had chemotherapy'.”

Those who run the registry say that it is not intended to provide an exhaustive, lay-friendly description of any given trial, but rather a starting point where patients can learn that trials exist, then pursue more information with help from their doctors. Says Annice Bergeris, who oversees trials registration at ClinicalTrials.gov: “We strongly suggest, because we feel that ClinicalTrials.gov should not stand alone, that a patient should go to their doctor [with information from the registry]”.

Of equal concern, says Frank Burroughs, the Abigail Alliance President, is that despite the surge in trials registered in ClinicalTrials.gov, information still remains patchy. Although “it's far more adequate for all trials than any other database than I'm aware of, it's terribly incomplete,” he says.

When the database staff analysed registrations entered over several months last summer, they found that only three-quarters specified what primary outcome the trial was measuring — one of the twenty data points required by the editors' group — and that these entries varied markedly in their specificity (Zarin, D. A. et al. NEJM 353, 2779–2787; 2005).

Some companies have resisted completing all 20 data fields required for ClinicalTrials.gov registration for competitive reasons. “There may be a proprietary need to keep it confidential for a period — not forever,” says Alan Goldhammer, the Associate Vice President for Regulatory Affairs at the Pharmaceutical Research and Manufacturers' Association.

But ClinicalTrials.gov staff contend that the registry's growing momentum is going to drive the database to near-completion. “ClinicalTrials.gov is becoming the worldwide standard. So I think those gaps will close,” says Bergeris.

Patient advocates are far from convinced. And they argue that there are better approaches to corralling a large proportion of all trials than the Field of Dreams strategy of 'if you build it they will come'. They point to the Lorenzen Cancer Foundation, based in Monterey, California, which has used data-mining techniques and algorithms to produce a database of pancreatic cancer trials (http://www.pancreatica.org/Search.cfm). It lists 435 currently recruiting trials that accept patients with pancreatic cancer. Entering 'pancreatic cancer' in a focused search at ClinicalTrials.gov turns up 98 open trials.

“It happens all the time,” says Dale O'Brien, Director of Information Services for pancreatica.org, when asked whether cancer patients find trials on his registry after failing to find them on ClinicalTrials.gov. O'Brien has even jokingly coined a term — 'The Fresno Conjecture' — to describe this disparity, after two patients from Fresno, California, learned from the registry about clinical trials taking place in Fresno that he says hadn't been listed on ClinicalTrials.gov.

In part, the gap is accounted for by a difference in search terms. Searching under 'solid tumours' at ClinicalTrials.gov, for instance, turns up an additional nine trials for which pancreatic cancer patients are eligible, and which are also listed at pancreatica.org. Yet it's not clear how a patient coming to the government database could know this.

The discrepancy could also reflect the fact that 63% of the pancreatica.org database is populated with Phase I trials. These early trials do not need to be registered at ClinicalTrials.gov under either US law or the journal editors' mandate — and as a result just 15% of the trials in the government database are Phase I trials.

O'Brien, a physician whose group has joined an alliance aiming to produce similar databases for other cancers, contends that comprehensive databases for five leading cancers could be produced in one year for the equivalent of roughly three oncologists' salaries. “We think that this kind of database should be available for all cancers. It's just ridiculous that it's not,” he says.

Still it's clear that the ClinicalTrials.gov database, while imperfect, remains the highest profile option in a world in which a mind-boggling array of trials databases run by hospitals, patient groups, public health organizations, companies and lobbying interests are available at the click of a mouse.

So the hope is that addressing critics' concerns will help ClinicalTrials.gov to evolve into a universally helpful resource. In industry, public health and academia, “we're all looking for a common website with common standards,” says Alan Breier, Chief Medical Officer at Eli Lilly. “And ClinicalTrials.gov is being looked to now as a really good site that could assume this broader responsibility.”