Just Buzzwords or Can Innovative Technologies Truly Transform Clinical Trials?

Real world data, real world evidence, artificial intelligence, and blockchain are four of the most hyped terms in today’s healthcare regulatory ecosystem. The underlying data, techniques, and technology hold real promise for improving regulatory and clinical decision making. Here, we look beyond the hype to see how these innovations can be brought together to improve the lives of patients.

With the prevalence of health information being collected in digital formats, the amount of real world data (RWD) available for analysis has grown explosively, leading to similar growth in real world evidence (RWE) derived from this RWD. FDA defines RWD as data relating to patient health status and/or the delivery of healthcare routinely collected from a variety of sources (electronic health records, claims and billing activities, product and disease registries, health-monitoring devices, etc.); RWE is the clinical evidence regarding the usage and potential benefits or risks of a medical product derived from analysis of RWD.

Sponsors, clinicians, regulators, and patients are re-evaluating the hierarchy of evidence, in which randomized controlled trials (RCT) have traditionally served as the gold standard for clinical evidence. The use of RWE in regulatory decision making was emphasized in 2016 following the 21st Century Cures Act that requires the FDA to evaluate the use of RWE in the approval of new indications for previously approved drugs. This shift in thinking was further demonstrated by the statement from FDA Commissioner Dr. Scott Gottlieb at the September 2017 National Academy of Sciences workshop on the Impact of Real World Evidence on Medical Product Development. In his remarks, Dr. Gottlieb stated, “For those who’d challenge the suitability of our effort to incorporate real world evidence into our regulatory model, I’d challenge you with the opposite intention: Should a product be marketed based on a data set that speaks to a limited and rigidly constructed circumstance, when the clinical use, and in turn the evidence we might have to evaluate the product, could have been far richer, far more diverse, and more informative?” The answer to this question is a clear “No.” Appropriate data sources beyond information collected in controlled trials should be considered when evaluating evidence to support medical products. RWE can be leveraged throughout the drug development lifecycle, from helping inform the clinical development strategy to supporting post-market safety assessment.

But, how do you adequately and efficiently analyze the abundant amount of RWD? And more specifically, how can RWD be harnessed to support clinical trial outcomes? Enter innovative technologies including artificial intelligence (AI) and blockchain.

AI is the process by which computer systems perform tasks that normally require human intelligence. From AI driven chat-bots, which can provide patients with medical information to machine learning- enabled pharmacovigilance systems, which can monitor post-market drug safety, AI can be used to analyze unstructured RWD to provide clinical insights.

Below are some examples of how AI could be used to harness RWD, and ultimately support clinical trials:

1) Patient Recruitment: AI and natural language processing can be used to mine data from medical records to identify patients and cohorts for clinical trials, accelerating patient recruitment.

2) Clinical Trial Site Selection: Data subscriptions for external clinical research resources exist, and companies are able to use AI, machine learning, and natural language processing to mine the data for key predictors of site performance. The data collected from global clinical trial sites can help predict which sites may have the highest probability of investigator performance.

3) Precision Medicine Approaches: AI can be used to analyze research datasets, including -omics datasets to help uncover new mechanisms, indicate potential biomarkers, and support personalized medicine efforts.

However, AI analysis will only be as good as the algorithms used to program the systems as datasets may have inherent limitations or biases built into them by humans. Also determining under what context AI-analyzed RWD can provide substantial RWE to drive regulatory and clinical decision-making is still under discussion, and requires input from stakeholders, such as clinical experts, regulators, and data scientists. There are two parts to this discussion-1) general AI considerations for managing and analyzing information, regardless of source (RCT vs. real world), and 2) considerations surrounding the validity of data not collected under a controlled setting, but in the ‘real world’. Although these can be viewed as independent topics, the main question for stakeholders is whether the application (an AI algorithm or a RWE study) is fit for purpose. Examining the question at hand, the available evidence, and other contextual factors should be part of the ‘fit for purpose’ discussion.

For RWE to be able to support clinical trials and inform healthcare decisions, RWD must be of high quality. Data integrity, reproducibility, and sharing, together with regulatory approval, patient recruitment, and personal data privacy concerns are some of the biggest challenges facing clinical trials today. Utilization of RWE also faces similar challenges of data sharing, data integrity, personal data privacy concerns, and the misuse/misinterpretation of data. Blockchain can help allay these challenges. Often thought of in relation to bitcoin and other cryptocurrency, blockchain is a technology with uses and applications that far exceed financial services. Blockchain is a distributed, peer-to-peer network database or ledger that maintains time-stamped, immutable records of all transactions. Each of these time-stamped records is a ‘block’ and is linked to a preceding block, forming a blockchain. Cryptography guarantees data security, allowing users to access only the parts of the blockchain they “own”, via private keys. A recent report from the IBM Institute of Business Value stated that 56% of healthcare executives surveyed expect to have a commercial blockchain solution in place by 2020, with clinical trial records listed as an area with greatest potential benefits. Here’s how blockchain can help in the utilization of RWD/RWE, and ultimately support clinical trials:

1) Patient Recruitment: Patient enrollment constitutes up to 30% of the drug development lifecycle. It has been reported that up to 50% of clinical trial sites recruit one or no patients, and 80% of clinical trials fail to meet their enrollment timeline. Identification of patients eligible to participate in a particular clinical trial remains a challenge. Blockchain can help by giving investigators access to decentralized, trusted data without compromising patient privacy via smart, or self-executing, contracts. The terms of agreement of a smart contract are written directly into the program code and the contracts themselves are executed automatically within the blockchain. In this scenario, a smart contract could specify which aspects of healthcare, genetic, and/or demographic data can be shared with investigators; all other data in a patient’s record would remain secure and inaccessible. This allows both, patients to be connected with trials anonymously and investigators to search for potential clinical trial participants solely based on genetic, geographic, and demographic criteria.

2) Data Sharing: Roughly 50% of global clinical trials are left unreported, with investigators failing to share trial results. Additionally, only 10% of trials on ClinicalTrials.gov list results. Without access to all data generated in a clinical trial, policy makers and other stakeholders in the healthcare product development continuum are unable to make fully informed decisions. Blockchain can maintain tracked, chronological, immutable records of clinical trials, protocols, and results that allow for secure sharing of data without errors. Once again, smart contracts can assist by acting as ‘trusted administrators’ and automating record keeping.

3) Data Integrity: Data tampering techniques including endpoint switching, data dredging, and selective reporting of results tarnish the credibility of clinical trial findings, resulting in inaccurate conclusions around the benefits and risks of treatments. Data in blockchain are time-stamped and unchangeable and, thus, tamper resistant. The immutability of data stored in blockchain makes the data extremely trustworthy.

A recent consideration around blockchain and the sharing of data and personally identifiable information (PII) is EU’s General Data Protection Regulation (GDPR) that went into effect on May 25, 2018. GDPR gives individuals the right to restrict use or request deletion of their PII. Data stored in blockchain is immutable and cannot be deleted, potentially going against GDPR stipulations. However, blockchain can continue to be utilized provided it is executed correctly: information can be stored “off-chain” and linked to a blockchain via public and private cryptographic keys. In this instance, PII may be stored in a modifiable database with simply a one-way hash of that information stored on the blockchain.

Thus, the current constraints around the utilization of RWE can be mitigated via innovative technologies including AI and blockchain. However, clear communication and conversations among stakeholders on the efficient generation and use of RWD/RWE are critical to advancing its application in regulatory and clinical decision-making. If the field explores the potential use cases above, and others, heeding this call for clear communication, we can reach beyond the hype and buzz to bring these innovations together in service of better regulatory and clinical decision making-and ultimately, improved patient outcomes.

References

1. https://www.fda.gov/ScienceResearch/SpecialTopics/RealWorldEvidence/default.htm. Accessed July 16, 2018

2. Real-World Data and Evidence in Drug Development. Accessed July 16, 2018. https://www.fda.gov/downloads/drugs/developmentapprovalprocess/smallbusinessassistance/ucm572939.pdf

3. https://www.fda.gov/newsevents/speeches/ucm576519.htm. Accessed July 16, 2018

4. Miksad and Abernethy. Clin Pharmacol Ther. 2018 Feb;103(2):202-205

5. Jadhav, Sujay. Blockchain Technologies Poised To Disrupt Stagnant Trial Timelines. Accessed July 16, 2018. https://www.forbes.com/sites/forbestechcouncil/2018/02/02/blockchain-technologies-poised-to-disrupt-stagnant-trial-timelines/#4408cba67390

6. Marr, Bernard. A Complete Beginner's Guide To Blockchain. Accessed July 16, 2018. https://www.forbes.com/sites/bernardmarr/2017/01/24/a-complete-beginners-guide-to-blockchain/#34caeec66e60

7. Healthcare rallies for blockchains: Keeping patients at the center. Accessed July 16, 2018. https://www-01.ibm.com/common/ssi/cgi-bin/ssialias?htmlfid=GBE03790USEN

8. Miseta Ed. Bring Down The Cost Of Clinical Trials With Improved Site Selection. Accessed July 16, 2018. https://www.clinicalleader.com/doc/bring-down-the-cost-of-clinical-trials-with-improved-site-selection-0001

9. Recruitment Roles, Sponsors, CROs, and investigator sites must all work together for effective patient recruitment. Accessed July 16, 2018. http://www.appliedclinicaltrialsonline.com/recruitment-roles

10. Gaudette, Andy. Blockchain’s Emerging Healthcare Applications. Accessed July 16, 2018. http://hms.com/blockchains-emerging-healthcare-applications/

11. Leveraging Blockchain for Clinical Trials/Research. Accessed July 16, 2018. https://beyondstandards.ieee.org/blockchain/leveraging-blockchain-clinical-trials-research/

12. Nugent, Timothy; Upton, David; Cimpoesu, Mihai. Improving Data Transparency in Clinical Trials Using Blockchain Smart Contracts. F1000Research 2016, 5:2541 (doi:10.12688/f1000research.9756.1)

13. Mearian, Lucas. Will blockchain run afoul of GDPR? (Yes and no). Accessed July 16, 2018. https://www.computerworld.com/article/3269750/blockchain/will-blockchain-run-afoul-of-gdpr-yes-and-no.html

Ranjini Prithviraj, PhD, PMP, Sr. Managing Editor & Associate Director, DIA Publications