PART 1 | Clinical metabolomics is just like basic research metabolomics – right?

We’re kicking off a 3-part series of blog posts to help explain what clinical metabolomics is, why it’s important and how it’s being applied in rare diseases, especially inherited metabolic disorders.

Kirk Pappan, PhD – Associate Director, Scientific Discovery and Application, Precision Medicine, Metabolon

To get started, let’s tackle a very important question. Isn’t clinical metabolomics just like basic research metabolomics?

Umm, well…
Maybe I’ll answer that with an analogy.

Are you familiar with the Piano Guys? They just play the piano – right? Well, I guess, if you consider airlifting a grand piano to the top of a tall desert mesa to record their cover of Coldplay’s Paradise, “just playing the piano.” Go ahead, invest five minutes to watch the video, then you can read the rest of this post followed by a really, really productive day of work without any goofing off. I won’t tell anyone.

When I joined the Precision Medicine group of Metabolon a little over a year ago, I learned first-hand about the features that distinguish metabolomics for clinical applications – or clinical metabolomics – from metabolomics used for basic research. Having analyzed and interpreted hundreds of datasets as a study director on the basic research services side of the company, I am well aware of the comprehensive UHPLC-tandem mass spectrometry profiling technology and iterative improvements that have expanded its coverage across all areas of metabolism while reducing variance (reported as median standard deviation) to less than 5%. I also know that a large reference standard library, which includes information about retention time on a chromatography column, accurate mass of the parent ion and preferred adducts, as well as accurate mass fragmentation data from tandem mass spectrometry, enables the unambiguous identification of approximately 1,000 named metabolites in plasma.

It is important to point out that metabolomics studies involve groups of subjects, while the clinical use is to evaluate an “n of 1.” The question has been “what are the biomarkers that separate the groups?” In the clinic, however, the question is “what are the molecules that are outside the expected range?” Then, more importantly, “what does all this tell us about disease or the health status of an individual?”

While a robust biochemical profiling platform is a necessary element, it is not sufficient to meet all the criteria needed to define clinical metabolomics. Clinical significance is about more than analytical precision. From my perspective, there are two additional elements, which when combined with the core platform technology, complete the definition of clinical metabolomics.

Specifically, clinical metabolomics involves the application of clinical laboratory standards, protocols and oversight to a global biochemical profiling technology whose results are interpreted relative to a reference cohort. While there are many other aspects of a clinical test, I’m going to focus on those mentioned above.

All clinical laboratory testing adheres to standard practices, protocols, training, oversight and record-keeping standards. You’d find no less in our clinical metabolomics laboratory which follows SOPs, personnel training, an alphabet soup of standards such as CAP, CLIA, CAPA and HIPAA, analytical validation of the technology, clinical validation of observed disorders, continuous quality and performance analysis, and project management to meet delivery timelines. In other words, clinical metabolomics adheres to the same long-established practices that define all clinical tests that have the potential to lead to life-altering decisions by families, doctors and other medical professionals.

In retrospect, the comparison of biochemical levels to expected ranges defined by a reference cohort is obvious for clinical laboratory results. Nonetheless, this final distinguishing element of clinical laboratory testing tripped me up initially.

On the basic metabolomics research side of the company, we use sample powering to drive the statistical results at the end of a study. For example, a common study design for analysis of samples collected from humans might involve 50 healthy controls and 50 disease cases to look for biomarkers or mechanisms of disease. For basic research studies, the most common data analysis approach involves comparing the means of two groups by a method such as the Welch’s t-test to calculate a p-value.

However, clinical laboratory testing doesn’t look at group averages – it focuses on how the individual  differs compared to a reference population. This is a bedrock principle of clinical laboratory test result reporting and is likewise a required standard needed to adapt metabolomics to precision medicine.

Thus, the final step of moving metabolomics into the clinical testing laboratory involves the acquisition of biochemical profiles from hundreds of individuals to comprise a reference population. From this population, the mean and standard deviation can be determined for each biochemical detected and can be used to define cut-offs for the expected reference range. Calculating a z-score for every biochemical detected in a patient sample then allows the biochemical to be understood in terms of where it lies in relation to the reference population.

Metabolomics is only beginning to penetrate the world of clinical laboratory testing, but its future in the space is bright. While the technology will continue to advance, it will also be interesting to watch how the application of clinical laboratory standards, protocols and oversight, as well as the use of reference populations, advances and adapts as clinical metabolomics gains its footing. A good first step in this evolution will be to remember that it takes more than a robust platform technology to conduct clinical testing.

To learn more about our clinical products and services or to order a Meta IMD test, please click here.