Database Exploitation Could Slash Clinical Trial Durations, Cut Costs

A presentation by health care consultancy VitalTransformation shows that a fifth could be cut off clinical trial times and up to $300 million saved if databases were used for trial recruitment.

Using real world information from health care databases to shorten patient-recruitment times to clinical trials could shave up to a fifth of trial duration and rack up significant savings in the process. However, companies understandably are nervous about taking a step in this direction and data-protection laws in the EU could make this difficult to do.

The model has been researched and developed by Duane Schulthess and Daniel Gassull of Brussels, Belgium-based health care consultancy VitalTransformation and was presented at the Royal College of Physicians in London on Oct. 21.

Developing and commercializing medicines for diseases with large unmet medical need takes a long time. The analyses presented suggested that drugs for the central nervous system take just over nine years, anti-cancer drugs about 9.5 years, and drugs for cardiovascular disease closer to 10 years. “I think a fair assessment is that if you do this right, you can probably cut at least six months off of this time without much trouble. A year is a good shot and, if you get all your processes aligned, we think you can do, even potentially, close to two years,” said VitalTransformation’s Managing Director Schulthess in an interview on Oct. 27.

In theory at least, if every process were optimized, manufacturers could be looking at R&D savings for a cancer drug of up to $105 million; for a cardiovascular drug of up to $130 million, and for an Alzheimer’s drug it’s around $300 million. “Let me be very clear: those [figures] are based on industry cost averages and that is just a valuation of the time – it is really just looking at time savings. We did not dig into actual site costs,” Schulthess explained.

Recruitment, according to the VitalTransformation presentation, accounts for roughly 30% of the overall time taken to conduct a clinical trial. It represents about 30 months in total. It is in the interests of trial sponsors to cut this time, which means development would be quicker, as drugs would reach patients faster and savings could be realized that then could be put back into innovation.

The point really is that databases could be used to try and help minimize the failure rate in the development of new drugs, because they can reduce the time needed to identify ideal trial patients. The report suggested that current failures actually increase the average development cost of a marketed drug from $215 million – the clinical trial cost per drug approved – to between $900 million for an anti-cancer drug and almost $2.5 billion for a CNS drug.

“Failure is the norm. That's not a bad thing – it is just the challenge of development and without those failures, you won't have successes. You need those failures in order to find the sub-populations,” said Schulthess.

Using databases to help narrow trial populations as fast as possible also is being explored by those in Europe seeking a more flexible approach to drug licensing and reimbursement, through so-called medicine adaptive pathways to patients (MAPPs) (Also see "Ebola Threat Prompts EU To Focus On Adaptive Licensing" - Pink Sheet, 16 Oct, 2014.).

The key is the ability to harness real-world data. In the case of trial recruitment, this would be done long before licensing, through established databases. For MAPPs, the data is likely to be generated after licensing.

Exploiting Three Database Types

Schulthess initially met with a couple of academics at the annual European Health Forum in Gastein, Austria, some 15 months ago. “We were wondering how you could use patient data better, how you would value it and, if you started using some of these databases that came to mind, how much faster cheaper and better could you do research?” he explained. The main two points of focus were to speed the path of drugs to patients, while cutting the “ridiculously huge” costs of making drugs.

The next step was to examine if there were potential differences between using the various existing types of databases: the U.S. prefers an opt-out model for patients; the EU is looking to build an opt-in model; and there also are patient-led databases.

Schulthess and Gassull decided to look at time as a benchmark, and therefore focused on recruitment for clinical trials. “Also, the amount of failures that are occurring in Phase III simply because sponsors cannot recruit is becoming a really big problem,” Schulthess added.

They therefore looked at existing clinical trials listed in ClinicalTrials.gov, a registry and results database of publicly and privately supported clinical studies in the U.S. The researchers selected three specific trials: one for non-small cell lung cancer, one for Alzheimer’s disease and one for cardiovascular disease.

Key data related to the inclusion and exclusion criteria – the qualifying information analyzed by clinical trial managers, which helps them decide who actually enters a trial.

The normal trial-registration process requires sponsors to use hundreds of hospitals. Once a trial manager and site doctors have been appointed, patients are requested to participate in the trial if they are deemed to be potentially acceptable candidates. “It is all done on what the clinical trial manager or the people running the trial think,” said Schulthess.

Issues taken into account include: disease prevalence in the region; the disease prevalence captured by the hospital in question; the rate of potential candidates entering the hospital; and the normal recruitment rate for the sector, industry and disease area.

Vital Transformation's argument is that, if the inclusion/exclusion criteria were run based on pseudo-anonymous information from patient databases, time and money both could be saved.

The consultancy included two health care professional databases in its research. The first was Oracle Health Sciences Network (HSN) at Penn Medicine, the University of Pennsylvania Health System, which contains information on some 2.25 million hospitalized patients. The purpose of the HSN is to allow Penn Medicine’s industry organization partners to access its data to complete patient-cohort feasibility studies related to potential clinical trials to be performed at Penn Medicine.

In addition to this database, VitalTransformation used NHS England’s Clinical Practice Research Datalink (CPRD), a U.K.-based observational data and interventional research service jointly funded by the NHS National Institute for Health Research (NIHR) and the Medicines and Healthcare products Regulatory Agency (MHRA). It focuses on data from general practitioners. CPRD services aim to maximize the way anonymized National Health Service clinical data can be linked, facilitating observational research and helping to deliver research outputs to improve and safeguard public health. The CPRS database contains some 4.6 million patient records.

VitalTransformation also took the database concept one step further, by examining the potential impact of patient-led databases on its theoretical model, and the results are significant. In fall 2014, Schulthess and Gassull conducted a survey in partnership with two practitioners currently using the Patients Know Best platform. Patients Know Best is an organization that aims to put patients in control of their own medical records and links them to a national database in Britain.

Of the patients who signed up to the survey, 92% said that they would be willing to share their medical records to help researchers find new therapies and cures, and 73% said they would be willing to participate as a research subject by taking new therapies to test their efficacy.

Taking The Plunge

The case made by VitalTransformation appears to be logical and impressive in its simplicity. The problem is that few drug manufacturers/trial sponsors are keen to jump immediately onto this particular bandwagon. Current clinical trial recruitment may be time-consuming, but it is a tried and tested model. While the health care consultancy’s findings represent a realistic conclusion drawn from theoretical modeling, they will require an ambitious and brave trial sponsor to be put into practice.

“If you’re a clinical trial manager there is a tremendous amount of responsibility and a large amount of risk if you try something new,” said Schulthess. He pointed out that, in both the regulatory and reimbursement worlds, change is tough – so it’s not surprising that the same prevailing factors would apply to the clinical trials sector.

It is not that VitalTransformation's research has not piqued the curiosity of drug manufacturers. A spokesperson for the European Federation of Pharmaceutical Industries and Associations said: “We are studying the data with great interest.” Moreover, VitalTransformation confirmed that it continuing with its plans to talk to as many health care stakeholders as possible about its findings.

A significant problem for manufacturers who wish to explore this trial-recruitment pathway is that there just aren’t enough databases around yet.

In Europe, Sweden is extremely advanced in the setting up of registries, but their existence alone may not be enough, said Schulthess. “There are some registries that are up and running, but whether you can use them for this sort of level of work, whether they link with the hospital data and whether they are searchable are serious questions. We tried contacting the Swedes, but they had no desire to join this project,” he added.

An additional complexity in Europe, Schulthess suggested, is that there remains a question over whether such databases even could be used by sponsors to fine-tune their clinical trials processes. “Under the Clinical Trials Regulation that was passed in April 2014, all research has to be done in an ‘opt-in’ fashion and that will radically change the ability to conduct this research,” Schulthess said (Also see "MEPs Pass Clinical Trials Laws But Implementation Will Be Troublesome" - Pink Sheet, 3 Apr, 2014.).

Evidence from the patient-led database survey that VitalTransformation conducted suggests most chronically ill patients would be ready to share their information. “Certainly in those areas of unmet medical need such as melanoma or Alzheimer’s, you bet they are ready,” said Schulthess.

Still, where needs currently are being met – such as in the case of diabetes – not enough patients might opt in to make pursuit of more innovative and effective medicines easier. “Obviously, if you could set up an infinite number of patient-run databases with the European diabetes society and you get 100,000 patients on their hand-held tools supplying data, great,” said Schulthess.

Building a recruitment process around a patient-led database alone would result in something incomplete.

For example, China is trying to build its own population-based database that falls outside the patient-led database parameters and would represent a veritable goldmine. “Imagine being able to do these queries on 1.2 billion people,” Schulthess said. Even most chronic diseases in China seem to have a sizeable population, he noted, when compared with Europe or the U.S.

If trial sponsors could collect information from all three types of databases and recruit on the basis of these data, this essentially would represent something of a gold standard.