Expert Insight: Urinary NMR diagnostics for inborn errors of metabolism using rule-based interpretation (for research use only)

Inborn errors of metabolism are rare diseases that often provide a typical pattern of elevated metabolites in biological fluids like urine, blood (plasma or serum), or cerebrospinal fluid. More than 1,000 different diseases are known. Because of the different chemical classes of those metabolites, the most used methods are gas chromatography-mass spectrometry (GC/MS), liquid chromatography-mass spectrometry (LC/MS), and other chromatographic methods. However, by using NMR analysis, which is a highly reproducible and quantitative method, it is possible to quantify simultaneously metabolites from a wide range of different substance classes such as organic acids, amino acids, purines and pyrimidines, and other small molecules.

In this on-demand SelectScience webinar, Prof. Dr. med. Friedrich Trefz, Senior Medical Consultant for Bruker BioSpin, and Dr. Georg Frauendienst-Egger, Senior Medical Consultant Neonatology at the Clinical Center Reutlingen, demonstrates the potential of computer-assisted interpretation and linking disease markers identified by NMR to the metagene database.

Read on for highlights from the live Q&A session with Prof. Dr. med. Friedrich Trefz and Claire Cannet from Bruker BioSpin or register to watch the webinar at any time that suits you.

Q: Could you elaborate on the diagnostic utility of urinary NMR?

FT: Yes. It's on a research-level, and you can use it for selective screening of patients suspected of inborn errors of metabolism. You just need a urine specimen of some milliliters and then you can analyze it in a short time. It’s good to have some experience in NMR technology and interpretation, but even without that, you have automatic reports and the support of the Metagene database as shown. The utility is mainly with respect to doctors dealing with patients who have not yet been diagnosed. I would not say it's the best specimen to do neonatal screening but selective screening. That means the more symptoms you have, the better the diagnostic efficiency is.

Q: How are the patient urine samples prepared for analysis?

FT: In urine, we have about 19,000 metabolites. In plasma, you don't have this or many substances. In newborn screening, the Guthrie test is used to measure phenylalanine and tyrosine. Today, you have tandem MS, and this is great. This is only a small selection of diseases, so urine is much more efficient to diagnose rare diseases than blood.

Q: Is it 150 peaks or 150 metabolites? How many peaks are there in total, normally?

CC: It's more than 1000 peaks and we are quantifying automatically 150 metabolites. For that, we are using more than one peak to do the quantification.

FT: Now, we have 151 metabolites where we are sure that not only sensitivity but especially specificity, is guaranteed. Of course, we would like to have more. Every time I meet Claire and the group, I ask about additional metabolites. I'm sure we are working on that and increasing the number of metabolites. However, it's not so easy to guarantee the specificity of the metabolite.

CC: We have 150 metabolites we are quantifying automatically, but as I said before, in urine we have more than 1000 peaks meaning that it's easy to add new metabolites in the automation. For that, we need to make further tests. Every metabolite will come with a certain matching factor and certain correlation factor, meaning that you are able to have an idea if the quantification value is good or not with these two parameters.

FT: If we have 100% as demonstrated in our presentation, then it's fine. Sometimes, we have 70%, and this is good but below that, it would be wise to check the original NMR and see if this is really the metabolite you're looking for.

Q: How does the method compare with mass spec? Perhaps if you could elaborate on some of the pros and cons?

FT: We are a group of about 100 laboratories working together at the Clinical Center Reutlingen. There are not many laboratories using NMR. I remember it's now 50 years ago that I first used GC-MS, which is more sensitive than NMR, but it's not so quantitative except when you're using the stabilized dilution method. GC-MS and NMR can work together, especially if you are using the tool we presented today. For instance, we have a relatively high limit of detections (LoDs) for some substances in NMR, such as succinylacetone. If you are using another method and you have a typical pattern for tyrosinosis type 1 and you see succinylacetone is elevated in other methods, you can add it. I would say it's not a competition, rather a supplementation.

Q: How do you compare the proposed NMR methods with the urine tests that are currently used within the clinic?

FT: This comes back to the answer I gave before. The great advantage here is that you could use NMR in China, then make an analysis in the U.S. and obtain the same result. If you would do that with GC-MS, you would have a big problem because you don't have this quantitative advantage as in NMR and the reproducibility is not so good. There are quite a few advantages with the disadvantage in some respect, being sensitivity.

Q: What field in megahertz would you recommend for such works? Is a cryoprobe required?

CC: The solution which was presented in the presentation is a solution working at 600 megahertz. We are not using a cryoprobe and for this type of sample, it is not required because urine is very salty. We all know, or the people who are doing an NMR know that with salty samples, a cryoprobe is not an advantage, meaning that the work is done at room temperature with 5 millimeters.

Q: What's the impact of different babies' diets on the urine profile? Do you see a noticeable difference between artificial milk and breastfeeding?

FT: Yes, we have so much data, but had no time to publish it. We made a comparison between babies fed with formula milk and breast milk and we could differentiate both groups very nicely with PCA. We have not yet published it because we would like to go a bit deeper and see what makes a difference. I think it may be the oligosaccharides.

CC: It was an oligosaccharide, but the overall profile was different, meaning that it is not only the one marker but there are multiple different markers. This makes the difference between the two.

FT: We also investigated phenylketonuria (PKU) patients and we can clearly see the difference between patients treated with amino acid supplements low in phenylalanine and patients treated with KUVAN. There is a difference between the PKU patients and the normal population in PCA, and this is dependent on the natural betaine the patient is ingesting.

Q: How did you obtain the reference population sample? For example, a pulled urine sample used for profiling against the patient sample for NMR.

FT: We're very happy that we have very good references. These are all children from Germany here. We have investigations documented for healthy children in the first year and up to 4 years of age and later. You must investigate them carefully and if there are any problems, we ask for a urine sample which we can use for NMR. This was approved by the ethical review board.

CC: Yes, we have plenty of samples of healthy patients, and due to the fact that the method is reproducible and standardized means that the reference population sample can be collected everywhere in the world and be measured everywhere in the world. We can then compare to the NMR spectrum, meaning that we have a big, healthy reference cohort for children now. It could be the subject here, in Spain or Australia, it means that we have overall worldwide cohort.

Q: How many metabolites are in the Bruker database? Is the database planned to be expanded?

FT: It's not easy to give the guarantee that the other metabolite is included. Sometimes it's good to have some experience in NMR technology and I must tell you that I'm a pediatrician. I learned GC-MS and I'm sure I also will learn a little bit more about NMR. It's good to know this technology too and not only to focus on the metabolites you automatically have in the quantification tool.

Q: We'd like to know if there is any specific post-program that's required for acquiring the data.

CC: Yes, we are using a standardized post-program. If you want to have the automated reports' analysis of the actual 150 metabolites then you need to follow special sample preparation, and this was one question at the beginning. The sample preparation is easy. It's just adding a buffer solution to the urine sample and putting this sample into the NMR system, and then the experiment is NOESY with gradient experiment. You can do whatever you want but if you want to have these automated reports then you need to follow the special requirements.

Q: Is there a need to further test the system with more confirmed diagnoses?

FT: Yes. Don't forget this is a learning system and the more samples we analyze, the better it is. I come from the University of Heidelberg and this is the main center for metabolic diseases, so we have thousands of samples. The problem is, "What is behind it? What is the diagnosis? Is the diagnosis confirmed? Are the clinical symptoms really clear?" With all these questions, it's very difficult to combine the clinician and the lab people, and there are a lot of barriers because of personal data protection. We really need analyses with confirmed diagnoses so it's absolutely clear. We would like to put in our register that it is increasing and to tune-up the method. We appreciate every colleague sending us a sample with the confirmed diagnosis.

Find out more on this topic by registering to watch the full webinar on demand>>

SelectScience runs 10+ webinars a month across various scientific topics, discover more of our upcoming webinars>>