Clinical Chemistry - Podcast

Clinical Protein Analysis by Mass Spectrometry: A New Higher Order

Brian A. Rappold



Listen to the Clinical Chemistry Podcast


Article

Brian A. Rappold. Clinical Protein Analysis by Mass Spectrometry: A New Higher Order Clin Chem 2021; 67:3 461–62.

Guest

Brian Rappold is the research and development director for Biochemical and Integrated Genetics for the Laboratory Corporation of America in Research Triangle Park, North Carolina. Dr. Andy Hoofnagle is a professor in the University of Washington in Seattle.


Transcript

[Download pdf]

Bob Barrett:
This is a podcast from Clinical Chemistry, sponsored by the Department of Laboratory Medicine at Boston Children’s Hospital. I’m Bob Barrett. If you have an interest in clinical proteomics – and frankly, who wouldn’t? – the March 2021 issue of Clinical Chemistry deserves your special attention. In addition to scientific articles and a special report featured in other podcasts from this issue, these works are the subject of an editorial in that issue by Brian Rappold who is the research and development director for Biochemical and Integrated Genetics for the Laboratory Corporation of America in Research Triangle Park, North Carolina. Also appearing in that issue is a Q&A feature that asks five experts the question, “Should LC Tandem MS be the reference
measurement procedure to determine protein concentrations in human samples?” The moderator for that Q&A feature is Dr. Andy Hoofnagle, who is a professor in the University of Washington in Seattle, where he helps run the clinical laboratories and serves as the director of clinical mass spectrometry. We are delighted to have both Brian Rappold and Andy Hoofnagle as our guest in this podcast. So, first of all, why are the measurements of proteins important in medicine? And, Andy, we’ll let you start.

Andy Hoofnagle:
The measurement of proteins and taking care of patients is really one of the most important things that we can do. We do a lot of measurements in the chemistry lab, whether it’s for ions, like sodium, or chloride, or calcium, or small molecules like glucose and creatinine, which we use to try to figure out what to do with patients on the inpatient or the outpatient basis. But proteins, there are really different kind of biomarker that can be used for the diagnosis, the prognosis, and the therapeutic management of patients. A great example is for cancer biomarkers, small molecules and ion, they really aren’t very important. But proteins, specific proteins that come from the tumor, it can be used to figure out who has recurrence. Other examples before looking at systemic inflammation or look at something like CRP or IL-6, which has been really important in some of the recent immunological therapies that we use in cancer. So, proteins really are our centerpiece of us taking care of our patients and without them, we wouldn’t be able to take care of patients as well as we do.

Bob Barrett:
So, Brian, how does mass spectrometry fit into those measurements in proteins for patients?

Brian Rappold:
The protein measurement by mass spectrometry piece is really -- it’s few and far between. We’ve got a few niche components here and there. So, Andy’s done a lot of work on thyroglobulin. There are some interesting things around plasma renin activity which are protein and peptide-based measurements. But, by and large, protein measurement by mass spec -- mass spec is oftentimes too specific to try to measure things. It also has a lot of upfront analytical challenges, so that when we try to measure protein by mass spectrometry, the upfront processes of aliquoting and digestion typically for doing bottom-up proteomics. It has fundamental flaws in it. These are difficult experiments to control, and perhaps the most important piece to doing mass spectrometry is an internal standard and isotopically labeled material that acts as a normalizing surrogate for the actual compound and we don’t have very good protein-based internal standards for mass spectrometry assays as of yet. They’re just difficult to manufacture. Andy, do you want to add anything on to that?

Andy Hoofnagle:
Yeah, I agree that internal standards are important, and I think that field has made great advances in understanding exactly what internal standards need to be for proteomic assays that use mass spec really to measure proteins. I guess I should be more specific, but I think that while it’s hard and there’s lots of sample prep going into it, and yes, we need to be thoughtful about the internal standards, but I think the benefits of how mass spec could fit into the care of patients are just ginormous. You mentioned thyroglobulin before, but that is such a canonical example of how bad immunoassays can be for the measurement of proteins. We’re trying to use these proteins to take care of patients and then we get the wrong answer, and we don’t even know it, right? The immunoassays can’t tell us that they’re wrong.

So, I agree, we’ve got to be really thoughtful about the workflows that we invent to measure proteins by mass spec. But I think that they really could hold a very special place in the care of patients if we do it right. So, I love that you mentioned internal standards, because that really is one of the bugaboos. You’ve mentioned the workflows; that’s going to be hard, but the ability to actually get the right answer is ginormous.

Now, you mentioned specificity. That’s one of my favorite things, because it begs the question like you said: “What is the protein that we’re measuring?” because proteins are really this amorphous class of things that are related to each other that have lots of different post-translational modifications, whether it’s splicing variants in different isoforms, or even possible relation different like oscillation events. And so, we don’t even really know what we’re measuring in an immunoassay, and now we’re asking the mass spec to do the same measurement and maybe it never will be able to. So, I think there’s a lot of potential for mass spec to fit under the care of patients in a very special way that rivals immunoassays for sure. But I agree with you, we are -- we’ve got our work cut out for us for sure.

Brian Rappold:
Well, I think bouncing off that, Andy, I don’t want people to assume that a mass spectrometer is really just a hammer looking for a nail, because we don’t need to measure every single version of a protein for any particular disease state. I think that there’s a number of good immunoassays out there that do excellent jobs in (at least the cost perspective) of being able to capture important information. So, liver enzyme screening is a -- I don’t know that we’re going to have a need to really replace those sorts of assays with mass spectrometry, because there might not be underlying value. But I think that also sort of brings up the question of whether or not we have explored that space efficiently and mass spectrometry I think is where we really might be able to slide a lot more of these more elegant studies to try to get a better lens or deeper lens into the biology of certain diseases.

Andy Hoofnagle:
I think what you’re saying is like liver enzymes because we can do them by entomology and things like CRP where the nephelometric assay is just fine. Why would we ever use mass spec in those situations? I agree with that. But, when you’re talking about biology, are you saying that we need to figure out different biomarkers for disease using mass spec. or are you saying we can apply mass spec to the current biomarkers?

Brian Rappold:
I think it’s probably going to be both with a lean towards the ladder. So, the biomarkers that we have, we know they’re affiliated with disease, but I think the question that we have around some of these protein-based biomarkers is, “How well can we stratify a particular disease state?” and I don’t believe that the current technology allows us to do that. Whereas mass spectrometry that it really has a fantastic opportunity to perhaps dive a bit deeper into what the -- as you mentioned, the post-translation modifications by oscillation, phosphorylation components – what those mean as a function of disease, which is where there is no technology like mass spectrometry when it comes to that.

Bob Barrett:
Well, from the sound of this, is measuring proteins by mass spectrometry difficult? And if so, then why bother doing it that way?

Brian Rappold:
Well, yes, it is incredibly difficult to do really, really well. I think that Andy’s recent example of the lipoprotein(a) reference method and the thought and the detail that goes into the underlying experiments to prove that what we say we’re measuring we are actually measuring, is really fundamental to try to measure proteins by mass spectrometry.

Andy Hoofnagle:
I agree. There are a lot of steps as you mentioned before in trying to measure a protein. And let’s use bottom-up proteomics where we use a protease to digest the protein into peptides before measuring a peptide as a surrogate for the protein. If we think about that process, we have to take the three-dimensional structure of the protein and try to make it one dimensional, if at all possible. It’s pretty difficult to do as you know, we might even produce the disulfide bonds and then alkylate them to prevent the reformation of the disulfide bonds. And then we’re going to bring in an enzyme, a protease, but it’s going to ideally digest things up into the analytes that we’re actually going to measure.

That’s asking a lot. We know that enzymes have lots of -- they’re going to respond to their environment. They’re going to have things like product inhibition. There are many reasons why an enzymatic proteolysis is problematic. And so, we need to be really thoughtful about the robustness of the digestion and the robustness of the preparation of the proteins for digestion. It’s complicated. What I like though is that the peptides that result from that digestion are just the small molecules that you and Russell had been teaching the field for more than a decade, how to do it well. We now know that we can measure peptides as well as we can measure small molecules, which is pretty exciting for me. I think that’s the one part that makes it easier. The hard part as you said, there’s a lot to do before those proteins become measurable analytes.

Brian Rappold:
Absolutely. The supporting cast of what it takes to measure a protein, the problems in digestion, but, furthermore, as a laboratory instrument, an LC-MS/MS system is not terribly helpful at times. They do have particular issues. They’re not always the most rugged systems in terms of operation. So, there’s a lot I think that goes into this, the entire process of measuring proteins, and specifically to bottom up, because that’s where we are going to have to live for quite a while. While the future looking view of top-down proteomics is really enticing, there’s just far too many experimental variables and questions outstanding in that particular procedure. But for the bottom-up approach, there’s so much that needs to be addressed from a first principles and granular level as to how we make this assay really work in the clinical setting.

Bob Barrett:
What exactly is a reference method procedure, and why should laboratorians, or even patients for that matter, care about them? Andy you’re a co-author on the paper on the Proposed Candidate Reference Method for Lipoprotein(a) so you want to take this question?

Andy Hoofnagle:
Happy to. So, reference methods are the way that we’re actually going to get laboratories to get the same answer. And that was actually a hallmark of what President Obama wanted from clinical laboratories as part of his health care reform and precision medicine. He asked us, “Can you guys please get the same answer from lab to lab?” and that was a very reasonable request.

But it’s hard. It’s hard because manufacturers have different antibodies that they put into their assays. As a result, we have different potential interferences we have and because of the post-translational modifications that Brian and I have been talking about, the protein you think you’re measuring might not be exactly the same in person-to-person. So, it’s hard. And we would like everybody to get the same results.

So, on average, we can do that by having something to point to so the reference measurement procedure is that which if we take a sample and measure it with the reference measurement procedure and get a number, every assay should get that number. And the reason they call it the reference measurement procedure is because we have done the analytical chemistry as well as we possibly can to have as much faith in that method as we possibly can. In other words, we don’t know of a way to do it any better.

Now because of the nuances of what we just talked about with proteomics and doing it from bottom-up, the LPA assay is a good example of how complicated this can because LPA exists in many different isoforms and blood different from person-to-person. So, we need to find a part of the protein that we can really understand on a mole basis. What are the number of moles of the protein that are present in the sample? Not the molecular weight, because the molecular weight is different from person-to-person. So, in the candidate reference measurement procedure that we published, we did as much as we possibly could to think that we got the right answer. And that’s great; there are other efforts now in the IFCC to generate other candidate reference measurement procedures that will use a slightly different approach, but ideally, it will get the same answer for both reference methods (that would be ideal). But the goal is to give data for specific samples so that people can use them to harmonize their results and since there’s a reference measurement procedure, we can use the word “standardize.” We can standardize everybody to get the same accurate answer. And that’s the goal of the reference measurement procedure.

But we really do, we have to drill down and figure out, at every step of the way, are we doing it the best we can? Are we getting as much recovery as we possibly can? So that we can refer back to that assay as the gold standard.

Bob Barrett:
Brian, how do you think that reference method procedures for proteins by mass spectrometry will change the way medical testing is performed?

Brian Rappold:
It’s going to be massive; I think on two particular fronts. The first is that what mass spectrometry will bring is a definitive answer. When we think about immunoassays and Andy mentioned this earlier, we don’t really know exactly what we’re measuring when it comes to a protein. There’s a couple of different forms that may or may not interact with the antibodies or work properly in the technology that we’re trying to apply. Mass spectrometry is an absolute definition of truth, and having reference method procedures for any particular assay, we will use one of the lipoprotein(a), Andy’s method that he recently published, that approach is going to make every single laboratory in the world look at their results and say, “How far off are we, and what does this mean in regards to a disease state?” because by and large, most laboratories will not always follow the process of making their own reference intervals or determining what is normal as a function of biomarkers concentration to disease, or there’s generally published guidelines that are accepted but not all assays achieve the same measurement as what you have that’s actually set the reference interval for this international guideline.

The second part that I think is going to be really cool is that having opportunity to look at proteins by mass spectrometry is going to just open up the amount of available data. So, when we start to look at why we have certain patient samples that have a very high concentration when we measure them by non-mass spectrometric technology, what was making that flag? We have an opportunity to look at the isoforms in the PTMs that may well be incredibly meaningful as a function of therapy or disease that we just cannot get from anything else. So, when we think about reference method procedures, it’s not going to be just setting the benchmark for everybody to have the same result across the world, which is I think one of the more stellar components of a reference method procedure, but I think it’s also going to give us a path to look at how well we do tests and what more information can we derive from a particular test, and how can we change the current testing paradigm to get to I think to Andy’s referral earlier, to a more personalized medicine approach.

The assays that we use right now are generally built from historic technology that has – some of these assays for proteins have not changed in decades and decades and if we want to bring up to a personalized medicine approach, obviously, using historic technology that never afforded us that is not the way to go forward. We want to be able to get to a definitive answer of a known concentration of a molecule, but also tease out those components that might be very, very important in terms of therapy or diagnosis.

Andy Hoofnagle:
Brian, I love what you’re talking about with respect to understanding what it is we’re measuring in the personalized medicine approach, but I want to go back to the first part of your answer, which was that maybe one day instrument manufacturers will try to get the right answer. I will express my frustration with manufacturers. As an example testosterone has now, with the help of CDC and lots of great academic investigators shown that there is a potentially a generalizable cut off, a number that we should use as a normal testosterone in men. And there’s even a CDC standardization program to try to get immunoassay manufacturers to get the right answer.

I had to actually stop measuring testosterone using immunoassay and switch to mass spec. Not because, necessarily, is mass spec needed in adult males where they have plenty of testosterone around them. The Immunoassay usually gets it right. The calibration was so wrong, that even though there is a reference measurement procedure and a standardization program. So, all of my endocrinologists came to me and said, “We can’t do this anymore. Your reference range is wrong, getting the wrong answers, we have to change.” So, we changed.

There’s other examples: c-peptide and insulin. None of the manufacturers are excited because they have to resubmit something to the FDA to change their calibration strategy to get the right answer. So, I’m not as optimistic that these reference measurement procedures are going to change medicine quickly, as I think you might be, but I will reiterate, I’m really excited that we’ll understand what we are measuring back to that personalized medicine approach like you talked about. We’re going to understand what’s really important, but without investment from the manufacturers we’re going to be stagnant in our ability to harmonize and standardize even with good awesome reference measurement procedures available.

Brian Rappold:
Well, Andy, that’s a great point, because it’s the dirty underbelly of laboratory medicine at times where the things that we would hope would always be in agreement are not, but to your point, that CDC procedure and testosterone, that is a benchmark that much like your endocrinologist said, “You have to switch.” I suspect that when we have more and more reference method procedures for proteins, we’re going to see many, many more people saying, “Look, I just don’t get as much good information out of your test as this alternative that has linked itself metrologically to an RMP or a reference method procedure. So, it’s not going to be any degree of a flipping the switch. It’s going to be a very, very slow dimming down of certain assays that we don’t trust and a slow elevation of assays that we actually have metrological and scientific faith in that’s going to bring about this this change in protein analysis for the clinical testing field. At least, that’s my opinion.

Bob Barrett:
Just 10 or so years ago, apart from papers on toxicology, it was uncommon to find mass spectrometric methods published in the pages of Clinical Chemistry. Obviously, that has changed dramatically. So, I’ll throw this out to both of you. What do you think the future of mass spectrometry holds for medical testing?

Andy Hoofnagle:
I’m a black box kind of guy, Bob. I think that automated mass spectrometers are going to change the way we take care of patients hospital to hospital. I think, as we’ve talked about here, really understanding what the analyte is and being able to specifically quantify it is really an exciting paradigm shift rather than looking in at that indirect signal that we get in spectrophotometry or in immunoassays, even chemiluminescent, the flash of light. That flash of light is an indirect measurement, and indirect understanding of what’s actually in the sample. So, ideally, to spread the good analytical chemistry of mass spectrometry to every hospital lab, we’re going to need an automated system so they can fit it into their lab. But, I think as Brian most elegantly has pointed out, we have the opportunity to really get and understand what it is you need to measure and we get to prove to ourselves that we’re measuring it well.

Brian Rappold:
Yeah, I think that’s absolutely right, Andy, and the dream of more hospitals having access to mass spectrometric technology is really a beautiful one because it does offer so much in terms of higher quality test results. I think that it’s going to be a price proposition for many of these facilities; the black box is certainly going to help with that. It’s going to really normalize what we have as an expectation for how to bill for a mass spectrometric test, to how we actually afford that technology. That’s going to be a fantastic I think 10-year plan is to see more mass specs and not just in the toxicology component where it’s in pain management clinics, but to actually see mass spectrometers perhaps on the desk of your GP when they get small enough and can run things properly.

The future for mass spectrometry is massive. If you really think about what the technology can do, definitive measurements of compounds with a linkable metrological source all the way back to the electron, fundamentally. There’s no way to actually frame out where it couldn’t go. And that’s probably the more important question, is what mass spec can’t do right now or what it won’t be able to do in the future. It’s certainly not going to be a tricorder, but it is going to offer enough detailed information that when physicians have pertinent questions about a patient’s status, they can go and get the correct answer. And that’s where I think mass spectrometry is going to be a big player and I would suspect that’s going to be a lot more reference method procedures showing up in the journals of Clinical Chemistry and we’re going to have a lot more candidate methods proposed in probably the next four or five years, for a whole slew of analytes that we actually really need to care about a lot more in the clinic.

Andy Hoofnagle:
Yeah, I agree and I think that’s what’s so exciting about this issue from Clinical Chemistry – is really putting a lot of thought and I will say that the editorial that you wrote was incredibly insightful. It’s one of the best editorials that’s ever been written in Clinical Chemistry as far as I’m concerned. I think that the brains behind the operation, people like Christa, people like Tomáš, who worked with me to make this method and Mark Lowenthal there’s just so many great metrological minds that are thinking about these things and I’m super excited to watch and take part as much as I can. But mass spec is going to change the way we take care of patients. And as you said, there’s really no limit to what it can do.

Bob Barrett:
Lastly, in your editorial, Brian, you paraphrased a recent Nobel Laureate in Literature, would you mind singing that stanza for our listeners? We don’t nearly have enough music on this podcast.

(Musical Interlude)

Bob Barrett:
I have my Bic lighter up, and I’m putting it from side-to-side

Andy Hoofnagle:
That was great, Brian, inspirational as always.

Brian Rappold:
It’s a real pleasure always talking with you, Andy and Bob, I really appreciate your time.

Bob Barrett:
All right. Thanks guys.

Andy Hoofnagle:
Thanks, Bob. Thanks, Brian.

Bob Barrett:
That musical interlude was from Brian Rappold, research and development director for Biochemical and Integrated Genetics for the Laboratory Corporation of America whose editorial on Clinical Protein Analysis by Mass Spectrometry appears in the March 2021 issue of Clinical Chemistry. He was joined by Andy Hoofnagle, professor of Laboratory Medicine and director of Clinical Mass Spectrometry at the University of Washington in Seattle. He served as moderator for a Q&A feature on that topic, appearing in the same issue. I’m Bob Barrett. Thanks for listening.