Clinical Lab Manager - News, Editorial and Products for the Clinical Laboratory
Innovations in Newborn Screening for Metabolic Disease

Innovations in Newborn Screening for Metabolic Disease

Untargeted clinical metabolomics could provide a new and inexpensive way to screen for metabolic conditions

Sarah H. Elsa, PhD, FACMG

Dr. Sarah H. Elsea is a professor of molecular and human genetics at Baylor College of Medicine and senior director of biochemical genetics at Baylor Genetics. Dr. Elsea earned a BSc in chemistry with a minor in biology from Missouri State University and a PhD in biochemistry from Vanderbilt University. She completed postdoctoral fellowships in molecular and biochemical genetics at the Baylor College of Medicine and is a board-certified geneticist by the American Board of Medical Genetics and Genomics. She started her research lab as an assistant professor at Michigan State University where she also ran a clinical diagnostic laboratory, continued her research at the Medical College of Virginia at Virginia Commonwealth University, and then returned to Baylor College of Medicine and the Medical Genetics Laboratories, now Baylor Genetics. Her research is focused on the discovery, diagnosis, pathomechanisms, and treatment of rare disease, particularly neurodevelopmental and neurometabolic disorders. She is passionate about her work, is a member of several professional societies, and has authored more than 150 scientific and lay articles.


Q: What is included in newborn screening? Have there been any recent advances? 

A: Technology for newborn screening of metabolic conditions has remained fairly constant for the last decade, with the exception of the addition of microfluidics for screening of some lysosomal enzyme deficiencies. State laboratories use targeted tandem mass spectrometry for screening the majority of the conditions on the Recommended Universal Screening Panel (RUSP). Newborn screening primarily includes metabolic disorders that are due to enzyme deficiencies; however, other conditions are also screened, such as hypothyroidism and cystic fibrosis.

Newborn screening is better now than it has ever been—but it can continue to improve. Not all states screen for the same conditions, so newborn screening is dependent upon where you are born in the US. My firm view is that all newborns in the US should have the same screening available to them. 

Q: Why is it important to screen for inborn errors of metabolism?

A: Many inborn errors of metabolism (IEMs) are treatable. But without screening and prompt treatment, most of these conditions will lead to neurodevelopmental problems, including intellectual disability, movement disorders, and even death. Treatments may be as simple as a vitamin supplement, as in biotinidase deficiency, or thyroid supplementation for hypothyroidism.  Others may require dietary changes or certain medications. Not every newborn screening condition has a fully effective treatment; however, new approaches to treatment are always being developed that prevent negative patient outcomes and lead to an improved quality of life. 

Q: How is untargeted metabolomics different from traditional metabolomics approaches?

A: Using metabolomics is not new, but using untargeted metabolomics is new to clinical laboratories. Up to this point, metabolomics has really focused on case-cohort studies looking at differences or changes across groups, where outliers are thrown out. Because metabolic disease is rare, we always have to consider that we’re not going to have big numbers. For our purposes, we’re looking for outliers—rare outlying cases that are different from all of the others that probably warrant a diagnosis. The key piece for developing clinical untargeted metabolomics has been developing a process that allows us to do an n=1 analysis, in that we need to be able to clearly analyze a patient sample without bias or other samples.

Q: What are the advantages of using untargeted metabolomics for newborn screening?

A: Traditional screening approaches that include plasma amino acids, plasma acylcarnitines, and urine organic acids are effective for a subset of metabolic conditions—and most of those conditions are now covered by current newborn screening. While targeted panels are clinically very sensitive, they can lead to serial testing, where if the first test results come back negative, another test is ordered, and several tests may be needed to get a result. The difference in our approach using clinical metabolomics is that we include most of the metabolic newborn screening disorders, plus many more conditions beyond the newborn screen. Thus, untargeted clinical metabolomics provides an inexpensive and timely method to screen for many conditions with one small blood sample, reducing health care costs by eliminating serial testing and multiple clinic visits, and improving the sampling process and shortening the time to diagnosis. 

Q: Tell me about your metabolomics pipeline for diagnosis and management of IEMs.

A: We do this work in collaboration with Metabolon, Inc., a metabolomics company in North Carolina. We’ve been working with them since 2013 to develop a clinical pipeline. At this time, we collect data from the patient and require a clinical note to inform the analysis. Then, we perform the analysis using plasma, urine, and/or cerebrospinal fluid. Because the analysis is untargeted, we don’t have absolute values for every molecule. Using a statistical approach, we develop Z-scores for each molecule based on a reference population. For some diseases, the detected molecules are rare and would not normally be detected in a person with a normal metabolism. So, instead of being assigned a Z-score, those molecules are scored as “rare.”

We then integrate the metabolomic, clinical, and sometimes genomic data to make diagnoses or to recommend additional testing based on a detected abnormality. Right now, we have a turnaround time of about 21 days, but with a more focused approach, we could get to a five-to-seven-day turnaround, which would be ideal.

Q: What are some of the challenges involved in untargeted metabolomics screening?

A: A challenge inherent in untargeted mass spectrometry analyses is that you may not see every molecule in every sample because you’re only looking at about 1,000 molecules. For me, that says we’re not quite ready for an untargeted newborn screening approach because you don’t want to miss an important molecule in the wrong patient. That potential situation worries me, and I wouldn’t want to take that approach, but I do think we could define a very clear panel of molecules that we know we could identify every time.

Of course, another one of the challenges is state laboratories being able to manage this kind of testing because the instrumentation is state of the art, and not all state laboratories are able to keep up with that need.

Q: What are you working on to advance this technique?

A: This methodology does have limitations. It only detects small molecules—which means that it doesn’t detect large lipids or carbohydrates, so some metabolic pathways are not well covered. A next step would be to address larger molecules.

We’re also trying to use artificial intelligence (AI) to identify new associations in the data, because sometimes we don’t see what we don’t know. When we use AI to analyze the data, we’ve identified other pathways affected by a known metabolic defect that we were not aware of before. We would never have made that connection without using a computational or AI approach. 

Q: What do you see as the future of IEM diagnostics?

A: It’s clear to us that doing small panels of analytes is not the way forward, especially with the use of exome sequencing for diagnosis, which has become the preferred approach. The recommendation that just came out a few weeks ago from the American College of Medical Genetics is to screen any child with developmental delay via exome sequencing. Coupling the exome sequencing with metabolomics testing will be the most efficient way to quickly rule out a metabolic disorder, which in many cases may be treatable.

Right now, we are the only group that’s doing this type of testing clinically, as the n=1 has been difficult for a number of groups to manage. It would be nice to make this more broadly applicable, which is absolutely possible, but groups need to come together to make it happen.