Access to full, appropriately de-identified datasets from clinical trials can benefit the biopharmaceutical industry by improving the efficiency of drug development, enhancing comparative-effectiveness analyses and reducing duplication of effort among trial sponsors, officials from the European Medicines Agency (EMA) insist.

The officials, including senior medical officer Hans-Georg Eichler and executive director Guido Rasi, have sprung to the defence of EMA’s draft policy for access to, and publication of, clinical-trial data following claims by industry that the proposals threaten patient privacy, trial sponsors’ proprietary interests and the integrity of the regulatory system.

The three-month public consultation period for the draft policy ended on 30 September and the EMA intends to implement the new transparency procedures from 1 January 2014.

“A managed-release environment that allows sharing of patient-level data while ensuring patient privacy would create a level playing field for all stakeholders,” Eichler and colleagues argue in the New England Journal of Medicine.

“What is sometimes labelled as ‘free riding’ may ultimately pay dividends for innovative companies and for public health.”

Not sustainable

Industry is right to be concerned about the sustainability of the existing drug-development and -business model, the EMA officials write.

The timelines and costs of clinical drug development “are increasing relentlessly, and the attrition rate of assets in development remains high”, they note.

At the same time, “growing cost pressures in all healthcare environments are forcing restrictions on drug use, aiming to limit coverage only to patients who can be expected to benefit from a given intervention and for whom that intervention is clearly cost-effective”.

Design and analysis

Access to the full datasets of completed clinical studies would, in the first place, lead to improvements in the design and analysis of subsequent trials, Eichler et al maintain.

“For example, available information about numerous variables can be used to identify and validate prognostic factors. Relevant validated prognostic factors can then be selected for use in the stratification of subsequent trials to reduce unwanted variability, minimise type I and type II error rates, and inform pre-specification of statistical modelling and subgroup analyses.”

The identification and validation of factors that predict treatment response also enable active sampling or population enrichment in subsequent clinical trials “to avoid having a treatment appear ineffective because the trial has been conducted in a diluted population”, the authors point out.

Enrichment can reduce sample sizes as “it makes larger treatment effects easier to detect”, they explain.

The inclusion of patient-level data can also generate comprehensive, quality-controlled databases with potential to inform future research projects and questions, Eichler and colleagues add.

Meta-analyses of patient-level data “can suggest that a trial is not needed because of the weight of existing evidence”.

Analyses such as these “have been essential in validating surrogate endpoints and speeding up the clinical development of subsequent drugs for HIV and colorectal cancer, have reduced the need for blinded independent central review in cancer trials, and may lead to shorter efficacy trials of drugs to treat schizophrenia”, they note.

Moreover, the availability of these data would allow well-characterised historical controls to be used in drug development where randomised controlled trials (RCTs) are not feasible due to the rarity of a disease, the authors say.   

Heterogeneity of effects

Secondly, they contend, lessons from past clinical trials about the heterogeneity of treatment effects will not only streamline drug development but may also “enhance a drug’s value in the marketplace”.

Identifying a population with high unmet need in which a new treatment may be more cost-effective than other available options can help sponsors during reimbursement negotiations, Eichler et al note.

Comparative effectiveness

They also emphasise the growing importance of comparative-effectiveness insights to patients, prescribers and positioning of new medicines.

While head-to-head RCTs are considered the gold standard for assessing comparative effectiveness, a dearth of these “has led to increased use of indirect comparison methods that rely on data from placebo-controlled regulatory trials”, the authors explain.

Data from individual patients on both outcomes and co-variates “can alleviate some of the weaknesses of this approach, such as the need to make assumptions about heterogeneity and consistency of effect on the basis of the summary data that are currently in the public domain”, they say.

Wider access to patient-level data from clinical trials will allow sponsors to present more robust comparative-effectiveness information about their product “soon after licensing and at a very limited cost” compared with head-to-head trials.

Doomed from the outset

Finally, Eichler et al write, one of the “inherent inefficiencies” of data secrecy is the repetition of trials and projects “that are doomed from the outset”.

Drug developers may continue to pursue a given target “even though clinical trials conducted by others have demonstrated the effort’s futility”, they maintain. With patient health at risk and limited resources for research, “the high opportunity cost of clinical-data firewalls is difficult to justify”.

Addressing concerns

Given the array of potential uses for patient-level data in facilitating research and development, it is surprising that few drug developers have been sharing data voluntarily, the authors comment.

Commonly voiced concerns, they note, include the risk of jeopardising patient privacy, of clinical trials being misinterpreted due to “inappropriate” analyses, and of commercially confidential information being disclosed to competitors.

In the EMA officials’ view, standards for de-identifying personal data “are available and continue to evolve to ensure adequate protection”. Legally binding data-sharing agreements can provide an additional level of protection, they point out.

Secondary analyses

While a “truly open” approach to clinical-trial data does carry a risk of inappropriate secondary data analysis and conclusions, this risk “exists for any type of secondary analysis, regardless of the nature of the data”, Eichler et al argue.

Two-way transparency is crucial to address this risk, “since it allows critical review of any secondary analysis by the public and the EMA”, they add. Moreover, there must be strong safeguards in place “to ensure that the clinical investments and intellectual property of innovators are not jeopardised by ‘free riders’”.

Clearly, though, legitimate interests in intellectual property and the protection of private investments “must be weighed against other legitimate interests, such as transparency regarding the outcomes of clinical trials and the protection of public health”, the authors contend.

Striking the right balance of these interests, they insist, “is a duty for all responsible stakeholders involved, not just for regulators”.