Using Adaptive Trial Design to Get Ahead of Tougher Regulations

Although the European Commission has not passed the proposed update to its medical device regulations, it likely will do so in the next few years. Under the proposed revision to European Union device directives, the review process for medical devices could get much more stringent. Despite this delay, device developers depending upon revenue from the European market will need to get smarter about product development now.

Enhanced regulation will have a filtering effect on device developers based on the decisiveness of their reactions. The most substantial compensatory strategies will require significant changes and organization-wide adoption. Two currently underused strategies—adaptive trial design and risk-based monitoring—have the potential to mitigate increased costs and time delays while simultaneously meeting the spirit of regulatory reform to improve patient safety and the utility of clinical trial data for physicians, regulators and patients. As the regulatory path for these strategies is already clear, companies that adopt these approaches ahead of regulatory reforms will benefit from earlier cost savings as well as a streamlined clinical development process fit for the impending regulatory environment.

First, What Do These Regulations Mean?
Enhanced regulation, in whatever final form, has the potential to drive up costs, increase the need for more clinical data, and slow down the entire development and approval process. As the European Parliament’s ENVI committee (which is in charge of the environment, public health and food safety) advised lawmakers to place patient safety above innovation, execution of trials will have to include more safety checks along the course of the trial, such as product assessments to detect non-compliance, assessment of quality systems, and unannounced audits to verify legal compliance. Additionally, new regulations will force sponsors to replace the basic statistics used in current device trials with standardized statistical methods to validate therapeutic claims.

Strategy 1: Adaptive Design
Rising costs and regulatory burden will force the device industry to become more efficient and strategic in advancing product lines. The ability to anticipate and react to the regulations will be a major differentiator. Companies that do not react to the changes will operate less efficiently and possibly run into regulatory issues, both of which will delay approvals. Moving forward, a solution is for companies to implement adaptive designs for their device trials. Although such designs are a relative rarity in the device space, they represent one in five clinical trials for drugs. Regulatory agencies are aware of and support adaptive designs—e.g., guidance on the use of adaptive design is available on the U.S. Food and Drug Administration (FDA) website and the agency’s Center for Devices and Radiological Health indicated that it is very receptive to adaptive design, having reviewed 120 adaptive trials in the last five years.

Adaptive trial designs allow pre-planned adaptations to a trial based on interim analysis of data without compromising trial integrity and improve trials along a paradigm called “cheaper, faster, better.” As the phrase suggests, adaptive designs can reduce trial costs, shorten the duration of the trial and, importantly, enable companies to make better decisions during and after the trial—minimizing risk as development proceeds.

There are three main types of adaptive designs suitable for device trials:

1. Sample Size Reassessment or “Right Sizing Your Trial” 
This is the most commonly used type of adaptive design for device trials. This type of adaptive design allows for interim review of data and an assessment of the optimal sample size needed to achieve statistical success of the trial. There is a direct link between the trial objective and the required sample size, which can be illustrated simply: To show that a device provides a 20 percent improvement in the relevant endpoint a trial may require 100 patients. If the device only can achieve a 10 percent improvement, the trial will need 200 patients. The ability to adjust sample size during the course of the trial could save the trial from failure if more patients are needed for statistical success. It also can ensure that the sample size is not too large as to needlessly expose patients to the device under study and lengthen the recruitment period. Validated software packages allow these types of trials to be conducted easily.
2. Combination of Pilot and Pivotal/Pre-Market Approval (PMA) Trials
This type of design saves time by combining the activities performed separately in the pilot and PMA stages into a
single trial. This type of design also may reduce the number of enrolled patients if the appropriate statistical methodology is used to combine the analysis of patients from both stages for the final analysis.
3. Bayesian Designs
Bayesian designs have been in use in medical device trials for the last 10 years. In this type of adaptive design, some “prior knowledge” of trial parameters reduces the number of patients needed and therefore the duration of the trial. Upon interim analysis, if this prior knowledge is correct, a company already is a step ahead of the game in terms of accelerating the trial. If trial data prove the prior knowledge to be incorrect, the current trial data take over and provide the correct results.

Strategy 2: Risk-Based Monitoring
The standard practice of 100 percent source data verification (SDV), which is the process of manually verifying that data in clinical record forms have not been transcribed inaccurately from the original source notes, consumes an estimated 25 to 30 percent of the total cost of a Phase III trial.^1,2 This resource-intensive quality assurance approach may seem straightforward and robust, but can provide a false sense of data integrity. It has been reported that 15 percent of errors remain in clinical trial databases after human review.³ Furthermore, the most serious problems with clinical data do not result from a miscopied number, but rather whether, for example, clinicians are properly trained and follow the trial protocol as they should.

Risk-based monitoring is an alternative data quality assurance strategy that can maintain data integrity while reducing the number of onsite monitoring visits and related costs. Rather than treat all clinical trial sites as equals, risk-based monitoring efficiently allocates monitoring resources to the sites determined to be higher-risk and that produce data with more errors (Figure 1). The result is conservation of data integrity and marked reduction in resource use. For example, in simulations by the Aptiv Solutions Risk-Based Monitoring Team, a risk-based monitoring approach for a trial deploying 25 sites (340 patients) reduced the number of site visits by 46 percent, resulting in cost savings of $450,000.⁴ This savings then can allow monitors to focus attention toward vital training and adherence issues that can lead to more serious data quality issues. Simply focusing on the output (the data originally entered) and not the input process per se means that data quality never improves.

Regulatory Acceptance
In the last few years, regulatory authorities have clarified their position on acceptance of justified alternative risk management and monitoring approaches. In January 2011, the European Medicines Agency issued a document titled “Quality Risk Management” (ICH Q9). In October 2011, the MDC/DH/MHRA Joint Project released a document titled “Risk-adapted Approaches to the Management of Clinical Trials of Investigational Medicinal Products.” In August 2013, the FDA released industry guidance titled “Oversight of Clinical Investigations—A Risk-Based Approach to Monitoring.” From the regulatory perspective, risk-based monitoring does not lead to any less vigilance in the oversight of the trial. It actually focuses activities on preventing/mitigating important and probable risks to data quality such as protocol eligibility criteria and processes critical to subject protection and trial integrity.

Path to Adoption
While initiatives that reduce source data verification, including risk-based monitoring, have been proposed for years, they have not been widely adopted in the medical device industry. With a clear regulatory path, risk-based monitoring is a validated opportunity to significantly reduce clinical trial costs. Just as manufacturing companies employ statistical techniques to examine random batches of products to assess quality and refine the manufacturing process, medical device companies have no choice but to modernize and more intelligently and efficiently conduct clinical trials.

Adopting this novel approach to quality management unsurprisingly requires new thinking, starting early in the trial planning phase. There are three important considerations:

1. One size does not fit all. A risk-based monitoring approach defines the extent and nature of monitoring required after consideration of the study objective, protocol design, complexity, geography, size, critical data points, risks, primary and secondary endpoints, and individual site personnel and characteristics that each contribute to the understanding of risk for the trial.
2. Real-time data entry and analysis is required for risk-based monitoring. Monitors, sites and investigators need to assess trial data as the trial progresses to guarantee proper control of data quality. For example, if a predicted lower-risk site produces data with errors that exceed the acceptable threshold, monitoring for that site can revert to 100 percent SDV, depending on the monitoring plan, to ensure data quality. Real-time data entry and analysis also allow faster data scrubbing and early prevention of repeated errors.
3. A common misconception is that risk-based monitoring can be performed by undertaking 100 percent SDV in only the first, third and fifth visits. This approach is not grounded in sound statistical theory. The FDA guidance document on risk-based monitoring supports the use of a statistical sampling approach.⁵ Statistical sampling allows control of variability, reduction in costs, and improvements in quality and has been used in a myriad of industries for many years. Manufacturing companies employ advanced statistical techniques to examine random samples (or batches) of products to assess their quality. Based on the results of these quality assessments, they refine their manufacturing processes to continually improve the quality of their product.

Filter Effect
Companies of all sizes can benefit from adopting adaptive designs and risk-based monitoring. Large companies with several product lines will increase their competitiveness. Small companies will see an increase in funding opportunities, as adoption demonstrates to the currently more cautious investor community that they are efficient, innovative, and control the risks associated with any clinical development.

The new reality is that costs and regulatory demands are increasing and will continue to do so. It is imperative to view regulation not as a burden but as a solution to improve clinical development. The focus must now be on strategies such as adaptive designs and risk-based monitoring that enable companies to make better product development decisions and help increase clinical efficiency. 

References

1. Funning S, Grahnen A, Eriksson K, Kettis-Linblad A. “Quality assurance within the scope of Good Clinical Practice (GCP): what is the cost of GCP-related activities? A survey with the Swedish Association of the Pharmaceutical Industry (LIF)’s members.” Qual Assur J. 2009;12:3-7.
2. Eisenstein EL, Lemons PW II, Tardiff BE, Schulman KA, Jolly MK, Califf RM. “Reducing the costs of phase III cardiovascular clinical trials.” American Heart Journal. 2005;149:482–488.
3. Society for Clinical Data Management. “Good clinical data management practices.” Washington DC; 2005.
4. Grieve AP, Fardipour P, Zippel E. “AptivInSite and Verification by Statistical Sampling: A Novel Approach to Risk-Based Monitoring.” www.aptivsolutions.com/aptivinsite/. 2013.
5. Grieve AP. “Source Data Verification by Statistical Sampling: Issues in Implementation.” Drug Information Journal. 2012; 46(3):368-377.

---

Laurent Spiess, Ph.D., is vice president of business development for adaptive clinical trials at Aptiv Solutions. He assists companies designing and implementing adaptive trial methodologies that improve device and drug development programs by reducing clinical trial cycle times and optimizing late stage trial success rates. Spiess received his Ph.D. in quantum physics from Northwestern University.

Erdmann Zippel is director, clinical operations, at Aptiv Solutions. For the past 13 years, he has lead teams at medical device firms, contract research organizations and biotech companies that conducted medical device clinical studies throughout Europe in the fields of cardiology, angiology, anesthesia/intensive care, orthopedics and neurology. Prior to joining the industry in 2001, he spent 12 years as a registered nurse focused on hospital intensive care and anesthesia.