Optimizing diagnostic development through applied statistical techniques
Beckman Coulter uses JMP® software from SAS to rapidly adapt to evolving regulatory standards while building in-house statistical capabilities
|Challenge||Pressure to improve data analysis to adhere to statistical guidelines from regulatory bodies and to analyze data more deeply for more efficient diagnostic development.|
|Solution||User-friendly yet comprehensive, JMP software helps the diagnostic team respond to regulatory requirements and implement a design of experiments (DOE) approach without a dedicated on-site statistical team.|
|Results||Enhanced study designs that allow for in-depth analysis of complex molecular interactions and growing interest in JMP implementation companywide.|
Finding clinical applications for basic research is a key priority in both the public and private sectors, driving the development of new diagnostic tools for clinical use. Many factors affect the ability of researchers to get novel diagnostic products to market; in the US, the Food and Drug Administration (FDA) and Centers for Medicare and Medicaid Services (CMS) regulate diagnostic tests, based on the intended use. Researchers seeking approval for their diagnostic tests must comply with evolving requirements for statistical analysis from these regulatory bodies, and diagnostic teams must evolve alongside these regulations to integrate robust statistical analysis into their research. Thus, diagnostic companies need comprehensive yet simple statistical tools that can be quickly adopted by researchers who may lack statistical training.
Emmanuel Romeu is a research and development manager in the Diagnostics branch of Beckman Coulter, a medical diagnostics company, where he specializes in creating novel diagnostic assays for commercial laboratories. In response to the growing demand for rigorous statistical analysis from regulatory bodies, Romeu began to incorporate statistics into his routine diagnostic work. Originally, he sought merely to align his team with the guidelines set by the FDA. Now he serves as a local statistician for both the Life Sciences and Diagnostics branches in France, where he integrates design of experiments (DOE) and other statistical approaches to streamline the development process and identify key areas for testing.
Understanding the need for a comprehensive, integrated statistical approach to diagnostic development, Beckman Coulter selected JMP as a companywide statistical software solution several years ago. Like Romeu, many Beckman Coulter staff members lacked statistical training or the initial interest in developing these skills. JMP provided an accessible software solution that gave users an opportunity to immediately integrate statistical analyses into their work, while letting users grow their statistical understanding over time. “At the beginning I was not really interested,” confesses Romeu. But through his required JMP training, he learned how to start using simple statistical analyses without understanding the mathematical background. In particular, the myriad data visualizations available through JMP brought the statistics to life and helped him grow his statistical understanding and enthusiasm. “After, as I was really interested, I tried to learn more and go into the mathematical aspects.”
Using DOE to optimize diagnostic development
After beginning to use JMP in his research, Romeu has drastically changed his approach to diagnostic development. “I’m collecting more data than I used to do in the past because I’m using a lot of design of experiments,” he says.
The importance of DOE for diagnostic development cannot be overstated. Using this statistical technique, researchers can assess multiple factors simultaneously as opposed to a single factor at a time. While some molecules are difficult to diagnose, others are simple but are highly reactive; DOE allows researchers to tailor their studies to the unique properties of each molecule and refine their formulations. Despite the increased data output, this process is straightforward, due to JMP software’s comprehensive visualization capabilities, which are key to both the DOE analysis itself and to engaging new JMP users.
Prior to DOE implementation, iterative testing pushed Romeu’s team to continue to add elements to their buffers, making them more chemically complex without the ability to evaluate all of the intricate molecular interactions. Now the Diagnostic team can develop simpler, more sensitive buffers capable of diagnosing complex products.
While using these new techniques requires additional effort in the short term, “in the long term it’s really less work,” says Romeu. “We understand better the design space and its function in quality by design … So we know in advance where are our strengths, our weaknesses, and we can focus on that.”
Creating a new assay
Testing for vitamin D is becoming increasingly common, as physicians realize the prevalence of vitamin D deficiency in both pediatric and adult patients. Many people experience no symptoms associated with vitamin D deficiency, but for some, levels that are too low can lead to brittle or misshapen bones, muscle weakness, pain or fatigue. Furthermore, recent research suggests that low levels of vitamin D can increase the risk of several chronic illnesses, including hypertension, Type 1 and Type 2 diabetes, and multiple sclerosis. Fortunately, this condition can be remedied through a combination of supplements and increased sun exposure. Unfortunately, developing a diagnostic test for it is less simple.
“I worked on the vitamin D diagnostic, which is one of the most difficult to develop,” recalls Romeu. Initial expectations were that this project would require at least four years with a team of 20 people to succeed. “The use of statistics, DOE really helped us … and advanced some of the issues,” allowing the team to meet the four-year deadline with only half the allotted staff.
Although not the first diagnostic test for vitamin D, this new assay improves on existing tests in several ways. “As one of the first assays standardized to the NIST-Ghent ID-LC-MS/MS reference method, the new assay offers greater diagnostic confidence in patient results, as well as enhanced stability and ease of use as a result of new innovative packaging,” said Arnd Kaldowski, President of Beckman Coulter Diagnostics. This test is now incorporated into Beckman Coulter’s diagnostic instruments and sold to commercial laboratories.
Engaging new JMP® users to transform diagnostic development
The proven utility of in-depth statistical analysis to the Diagnostic team has garnered the attention of others in the company. Although JMP is already the company-approved statistical software, plans to broaden the use of JMP are already underway. Romeu has shared his enthusiasm for the program widely, communicating the solution’s ease of use and potential for further streamlining operations and elevating the quality of ongoing research.
Colleagues previously reluctant to implement JMP software and new analytical methods now actively promote JMP to other employees, showing the appeal and simplicity of the software even to older, less tech-savvy personnel. The simple functionality of JMP has also been a boon for new employees. As they largely come from a biology or engineering background, new employees usually have only a basic understanding of statistics; training on DOE using JMP builds on their limited knowledge and allows them to apply relevant statistical aspects to their work quickly.
Romeu’s enthusiasm for JMP extends to improving the very regulations that spurred him to implement JMP in the first place. “I’m personally involved in one of the guidelines for the moment, and I realize that it truly depends on the influence of the different people [involved],” Romeu says. “Some push for something very good in statistics, but it’s too complex, so people will not use it. So it’s better to propose something simpler, even if it’s not as evolved, to be sure that people really use it.” As JMP increases the accessibility of DOE and other sophisticated statistical techniques to both the Diagnostics and Life Sciences teams, Beckman Coulter will continue to prove to be an innovative leader in the field.