Changing the Paradigm of Recruitment for Rare Disease Trials

RD day 9.1
Deep 6 AI Blog

Deep 6 AI Blog

Get insights about clinical trials, patient identification, health IT, data science, machine learning, and more.

Rare Disease Day is February 28, 2021, which aims to raise awareness of rare diseases and the patients living with them. An astounding 25% of patients with a rare disease can wait 5 – 30 years for an initial diagnosis, and for 40% of patients, the initial diagnosis is wrong. The so-called rare disease paradox is that, while a single rare disease affects relatively few people, there are 300 million people total worldwide (3.5 – 5.9% of the world’s population) living with one of the ~6000 – 7000 rare diseases. 

   

What is a rare disease?

 

The Orphan Drug Act, passed in the United States in 1983 to encourage the development of drugs for rare diseases, defined a rare disease as a disorder affecting <200,000 people in the United States. In Europe, a disease or disorder is considered rare when it affects <1 in 2000 people.

   

Why are rare diseases also known as orphan diseases?

 

Rare diseases or orphan diseases? These terms are often used interchangeably, with the term “orphan” speaking to the fact that treatments for rare diseases have historically been neglected by the pharmaceutical industry. In fact, less than 6% of rare diseases have an approved treatment option, and fewer than 1000 diseases have a minimum set of published scientific knowledge that would benefit disease identification, diagnostics, and therapies.

 

Challenges with research in rare diseases include:

  • Low prevalences
  • Geographically dispersed patient populations
  • Lack of understanding of disease causes and mechanisms of action
  • Patient heterogeneity
  • A large proportion of pediatric patients (onset occurs in childhood for 50% of rare diseases)
  • Lack of standardized vocabulary, making it difficult to identify patients with similar symptoms

     

Patient recruitment using traditional methods is already a lengthy, expensive process. Shrink and disperse the patient population, and it becomes even more challenging to find and retain patients. Often, the symptoms are similar to those of a myriad of other diseases, and the patient’s physician must be aware of a rare disease for it to even be considered as a possible diagnosis. With the smaller sample sizes that typically accompany rare disease trials, reaching statistical power to make confident decisions about drug safety and efficacy can be difficult, and it is typically not cost-effective to develop and market products for these small populations. 

 

To address these challenges, several initiatives and networks were established to support collaborative, coordinated, rare disease research: International Rare Diseases Research Consortium (IRDiRC), European Orphanet, European Reference Networks, Japan’s Initiative on Rare and Undiagnosed Diseases (IRUD), The National Institute of Health’s (NIH) Undiagnosed Diseases Program, the National Rare Diseases Registry System of China (NRDRS), the Deciphering Developmental Disorders project, Rare Diseases Clinical Research Network (RDCRN) Program. The NIH also provides a list of resources for patients to find clinical trials, but what if sponsors/CROs didn’t need to wait for the patient or doctor to find their trials? What if we could proactively find patients for clinical trials?

   

How can we speed up patient recruitment?

 

This is where advanced techniques for data mining of electronic medical records have a role to play—they can increase the potential recruitment pool from the small proportion of people with an accurate diagnosis to all patients within the database that meet the study’s criteria.

 

In a system such as the Deep 6 AI platform, sponsors can build a study, including inclusion criteria such as a list of symptoms, age, imaging findings, and/or genetic test results, and exclusion criteria such as previous treatment. Using this query and the natural language processing (NLP) capabilities of the artificial intelligence (AI) engine, these criteria are precision matched to patients’ records, not only the structured codes and fields but also all the free-form notes and results entered by clinicians. So, study teams can have confidence that the identified patients will be eligible for the trial. Alternatively, physicians could enter their patient’s medical record number into the system and match the patient’s profile with ongoing trials.

 

Imagine also being able to quickly (in mere minutes!) find patients who match a specific clinical profile but have not received a definitive diagnosis yet—purely because the right combination of symptoms, test results, and demographic characteristics was found in their records. The first step in the clinical trial process for these patients could be diagnosis, followed by investigational treatment.

 

With improved recruitment methods such as this, patients could find information about and options to manage their often debilitating diseases. Physicians could offer clinical research as a care option, which might be the only treatment option available for many rare diseases. The scientific community benefits from additional knowledge about diseases to inform future studies. The pharma industry can reduce the cost and time associated with patient recruitment and potentially reach a greater sample to improve the scientific soundness of their trials, which could then encourage more research and potentially reduce the end cost of the drug for the patient. Healthcare costs can be reduced by greater knowledge, improved diagnostics, and better disease management.

   

Have we made progress?

 

 

Financial incentives to develop new drugs for rare diseases, such as those established by the Orphan Drug Act, have been working. Between 1973 and 1983, fewer than 10 treatments for rare diseases were approved. Since the Orphan Drug Act was introduced in 1983, more than 400 drugs and biologic products for rare diseases have been developed and brought to market. However, these address fewer than 300 (of the ~6000-7000) rare diseases.

 

One of the goals for 2017-2027 by the IRDiRC was the approval of 1000 new therapies for rare diseases, the majority for diseases with no currently approved therapeutic options. To do this will require deep collaboration among all clinical trial stakeholders, as well as the embracing of new technologies that harness the immense amount of patient data available. Let’s work together to use those data and decrease the burden of patient recruitment on physicians and sites and proactively identify and recruit the patients who would benefit from rare disease clinical trials and treatments.

 

 

Share this post

Share on linkedin
Share on facebook
Share on twitter
Share on print
Share on email