Clinical Metabolomics: Will it Deliver?

Numerous research articles have proposed, addressed and promoted metabolomics as one of the key tools for biomarker discovery and personalized medicine. Personally, I am not blessed with a lot of patience, but even those who are might be starting to wonder, “After more than a decade of metabolomics-driven research, can anyone actually name a single resulting biomarker routinely used in the clinic?” I have to admit that, besides trimethylaminoxide and some markers related to gene defects (for example, 7-dehydrocholesterol), nothing comes to mind.

But why? Have we not used the most advanced analytical and computational approaches available? Have we not invested enough money, manpower and dedication? I don’t believe that lack of effort is the problem; I think we just took the wrong path.

In the beginning, when metabolomics was first used in case–control studies, it all seemed pretty straightforward. Many believed that with the right equipment and the right bioinformatics approach, we would easily identify some discriminators between all the molecules we can monitor. But the human body contains more than five liters of blood, we eat more than 500 g of (highly diverse) foods and drinks every day and, to make this picture even more complicated, our molecular fundament depends on genes, sex, weight, race and lifestyle. On top of all these variables, the metabolome is further influenced by circadian rhythm, hormones (mood), menstruation and medication. Say you are looking for a cancer marker – how are we going to find this one molecule, possibly secreted by a few million cancer cells somewhere in your brain or lungs, hidden in a constantly changing five-liter bucket of blood? Frankly, I am not convinced there is a high chance 
of success.

I don’t want to paint too dark a view here, but simply illustrate that metabolomics biomarker discovery is a very complex endeavor. It’s possible that our vision was blurred to the difficulties by the high hopes we had. Nevertheless, I am convinced metabolomics will make its way into the clinic, and hopefully fill the pipelines of clinical chemistry with new molecular tools. In life, you have to fall and get up many times before you learn to walk, and it’s time for clinical metabolomics to take two seminal steps forward.

The first step is to change our mindset – away from traditional biomarker discovery studies and towards understanding the systems effects of metabolites, as outlined in a recent article from Gary Siuzdak’s lab (1). The second step is to define the framework of human metabolism. In other words, what are the actual (true) concentrations of metabolites, what is the range these metabolites are to be expected in vivo, and how are these concentrations affected by circadian rhythm, food intake, tissue distribution and many other factors?

Such steps are increasingly being taken in several recently established phenome centers. In my view, these are exactly the right steps, in the right direction, at the right time (if not a little too late...). Clinical metabolomics has learned from its past failures and too few successes, and is ready to start taking strides into the future.