Marshall Summar is Director of the Rare Disease Institute at the Children’s National Hospital. He runs the genetics and metabolism programme to build new clinical models to take care of patients with rare diseases. Marshall talks about how the definitions of a “rare” disease may vary, the opportunities and challenges of emerging therapies and his work in improving clinically actionable protocols.
Please note the transcript has been edited for brevity and clarity.
FLG: Hello, and welcome to the latest “A Spotlight On” interview. Today, I’m joined by Marshall Summar, and we’re going to be talking about rare disease. Marshall, if you could please introduce yourself and tell everyone a little about what you do?
Marshall Summar: Thank you so much. My name is Marshall Summar and I work in Washington DC at the Children’s National Hospital, where I run the Rare Disease Institute in the genetics and metabolism programme. I’ve been in the field since about 1985. My career has been split between Vanderbilt University and the Children’s National Hospital. The goal of what we’re trying to do here is to build out new clinical models to care for patients with rare disease. It’s a very exciting field and a very exciting time. I think we’ll have lots to talk about.
FLG: That’s so exciting. Thank you. Could you give us your definition of a rare disease?
Marshall Summar: That’s actually a really good question. It depends on where you are. If you’re in the European Union, they use a 1 in 2,000 definition. In the United States, we use the definition of 200,000 patients or less, which works out to about 1 in 1,600. Different countries do it in different ways. For example, Taiwan has a list of rare disease. If it’s not on the list, it doesn’t occur, so theirs might be more like one in 100,000, or one in 50,000 from the population.
A good working definition I’ve used over the years to define a rare disease: It’s uncommon enough that a general practitioner might have heard the name but has probably not encountered many patients or only a single patient with that condition. They wouldn’t be expected to know the complex management details for those patients. That’s my working definition.
FLG: That’s an excellent working definition. What are some of the factors affecting the prevalence of rare disease? And are we seeing an upward trend due to better diagnosing?
Marshall Summar: We are absolutely seeing an upward trend and three things are going on. One is that molecular genetic diagnoses are taking common conditions and breaking them down into smaller genetic subgroups. For example, in breast cancer, there are now 30 molecular subtypes based on DNA and genetic changes. Those groups start to fall into a rare disease level of incidence. So, you’ve got your incidence (how many per 1,000), then your prevalence (how many are living and how long in the population at a given time).
Another example is Down’s syndrome, which is a borderline rare disease. In the 1980s, life expectancy was about 27. Now it’s approaching 60, and prevalence has gone way up. Cystic Fibrosis used to be lethal. Typically in the teenage years, it adds one year to life expectancy every year, because we don’t know what the back end of it is. Sickle cell anaemia is another great example, prevalence has also gone way up. It used to be one of those diseases where we would see early death; now, we’re seeing patients live longer and longer. Again, we don’t know what the back end of that is. So, the incidence of these conditions has not changed that much. What molecular genetic diagnosis has allowed us to do is identify more and more cases.
I worked as part of the Human Genome Project in the 1990s. Before that, we could accurately diagnose a few dozen things. Then, when we got the draft sequence of the human genome, we were suddenly able to start applying that. As the cost of sequence analysis has come down, we’re identifying new conditions where we’re linking a molecular genetic change to a clinical phenotype, in other words, a disease, at the rate of 10 to 12 a week. This is unlike any other field of medicine. These patients always existed; we just did not know what they had, and we didn’t have a way to test them.
FLG: That’s fascinating. You talked about a unique model of medicine. What are other factors that are making rare disease such a unique space?
Marshall Summar: Rare disease is coming out as a field of medicine where the rules are a little different. One is our incidence is very small. In fact, we don’t even know the incidence for most of these diseases because there are so few patients. And these are genetic in origin, which makes them lifetime conditions. You’re looking at lifespan diseases which can manifest differently at different stages of life. The next part that adds a layer of complexity is genetic heterogeneity. That means that the genetic change that causes that disease can be very mild or very severe, giving you a clinical phenotype that can vary across the age span. A great example is urea cycle disorders, defects in ammonia and nitrogen processing. Patients can present in the first day of life with severe hyperammonaemia. Or the patient may not show symptoms until their 80s when a series of environmental exposures bring out that underlying genetic defect. Same genes, same biochemistry, but just different degrees. That’s another layer of complexity.
Another is that we know very little about most of these conditions. Our classic models for consensus treatment for these patients really fall short when you start looking at the rapid discovery rates and everything else. The classic model is your Delphi criteria, where you want a large meta-analysis of the population like hypertension, where you have lots of data that you can analyse. In rare disease, we may have a dozen patients. We had one disease where we were doing a clinical trial, and we only had five patients. I remember the statistician saying, “That’s not a representative sample.” My reply to him was, “It’s not a sample, that’s the entire population”. You have to think of these things differently.
How do you do evidence-based medicine when your evidence tends to be very thin? There’s an interesting thing we’ve observed too. When you look at rare disease, you don’t get the big meta-studies. Published literature on rare disease is a little peculiar because you’ll often have a report of the first few patients. But the only thing people can really get published after that are the exceptions. If you base your treatment guidelines on that, you end up teaching to the margins. And how do we actually capture expert opinion? What I would say is a really unique feature in the field is that it’s one of the closest interactions between the patients, families, and physicians. Our families are often the world’s experts on their version of their rare disease. Many times physicians will not have seen this before. The family often has to be the bearer of the information about the disease. The physician can listen to them and actually learn about these things. The partnerships between the care providers in rare disease and the families, I think, is closer than in any field of medicine. I think there are some real distinct features here.
FLG: Absolutely. And for a bit of a timeline, can you give us an overview of the types of therapies we’ve seen over time and which are the growing and emerging options?
Marshall Summar: Let’s go back in time to 1983 and the passing of the Orphan Drug Act. The European Union has something very similar and Britain has something very similar. Before that, there was almost no incentive to develop therapies for rare disease. Large pharmaceutical companies would not be incentivised to do that. It’s arguably the most effective incentive programme we’ve seen to get industry to participate in the field. Looking at the data for last year, 50% of the new chemical entities approved by the FDA were for orphan designation. Now, about half of those are for cancer. But that means 25% of the drugs coming out of the FDA for approval every year are for rare disease, which is a startling increase.
What type of therapies are we looking at? There are drug and pharmaceutical therapies; that’s where a lot of the development is. But we have to remember that, for many of our patients, there’s probably never going to be a drug. But we’ve really started working on maximising the quality of life for our patients. Again, that’s where that partnership with parents and families works, but also the educational system, all of these different things. Down’s syndrome is a good example. It used to be that folks would say a child with Down’s syndrome will never read or be able to socially interact. And we never gave those kids a chance. What we’re finding now is we train to failure, so to speak. We try to allow these patients to do as much as possible by providing positive learning and social situations where they can grow and develop and become their own people. We’re finding the results really do improve that way.
Now, let’s talk about our classic drug model. We start with your small molecules; that’s kind of your classic drug. We try to find the genes or the things we know about that disease, where we can put a lever in and improve some outcomes. That’s still a very viable development pathway because tissue penetration with small molecules is pretty good. Then we come to the age of the enzyme therapies that started back in the 90s when we were able to make recombinant enzymes that would carry out the activity of a missing enzyme. This has typically been in lysosomal storage diseases that require infusions of enzymes on a regular basis. It’s been a mainstay of therapy for many, many years. Now, we have some new generation things coming along, such as substrate reduction. So, if you’re building up a toxin and can’t quite get rid of that, can you go upstream to that? By building a metabolic block higher and a pathway at a non-toxic level, you prevent the formation of a toxic substance later on down the pathway. Dietary therapy can be highly effective therapeutically for patients with PKU (phenylketonuria), something we’ve been doing since the 1960s.
Now, let’s get into our nucleic acid therapies; let’s start with messenger RNA. That’s something not many people were thinking about. And we all got a big dose of it last year in our arms. Messenger RNA therapies are growing pretty quickly, and we have found they’re pretty safe. I think we’ve just had the largest clinical trial experiment in history. While there’s been a few reactions, not many actually. Messenger RNA therapy is coming along and for metabolic diseases is probably where it’s going to be very impactful.
Then we get into the genome and the productions of RNA. We have things that will affect the splicing of messenger RNAs; let’s say there’s a defect we can get to splice out or splice in. That’s one strategy. Sometimes, you don’t want to make the protein causing the problem. So we’ll use antisense oligonucleotides, things that will go in, bind to the RNA, or bind to the gene and prevent the production of the messenger RNA.
Now, let’s come to gene therapy. So there are two kinds of branches. One is vehicles that will deliver a normal copy of the gene into the tissue, it may or may not integrate into the DNA, but it will sit there and make things. The most common one that’s being studied these days is the adeno-associated viruses, and they’re different viruses that will infect different tissues, so you can get some specificity there. What we’re now figuring out with these types of viral gene therapies is, is one dose enough, and is that going to wear out with time? For example, in a liver-based therapy, if we dosed once, will the effect be diluted when liver tissue divides? But, gene therapy in tissues that don’t divide a lot, like muscle and brain, may be more effective. Or it may be that we simply buy some time for our patients and get them through their developmental years.
Now, CRISPR is evolving quickly. It’s not quite ready for prime time. I keep hearing the term five years from now; that was the term we started using with gene therapy back in the 80s. I don’t know what the timeframe is on CRISPR. Obviously, if you can change the genome to get rid of the deleterious change, that’s a good thing. What everyone worries about are off-target effects. Did you get an additional gene that might be on an oncogene or something else that could cause problems later on? Most of the worries are around cell cycle regulators. If you interfered with the cell cycle regulator, might you turn that tissue into a tumour? While risks may be low, if you’re doing that to billions and billions of cells, the risk may actually become appreciable.
I’m cautiously optimistic about the gene editors. But I think it’s one of those things that you have to keep an eye on. There are efficiency issues; how many of the cells can you revert to a normal phenotype? Stay tuned. There’s a lot of effort to make those systems more efficient, specific and portable. And bacterial systems can afford to be sloppy. If the bacteria gets a mutation, so what, there’s 10 billion right behind me that got it right. When you’re doing eukaryotic or human cells, you want to be a little more careful than that. I think we’ll probably see some refinements in the technology before that moves forward.
FLG: That’s an excellent review. Thank you so much for diving into the opportunities as well as the limitations there. And on limitations, what are the biggest obstacles you face when trying to approach treating rare disease?
Marshall Summar: I work in a lot of different places and talk to a lot of different folks. First, large healthcare systems are not always well designed to deal with rare disease. If you get below a certain level of factors, such as infant mortality, the disease may not be that noticeable. So in countries with a lot of other challenges, rare disease are not something that’s popping up much. Rare disease always slots in last, it seems. And I’m seeing this in a number of systems. They base a lot of those systems on big data from lots and lots of patients, so they can make rational choices about allocating resources. The individual therapies for rare disease are quite expensive. They’re caught between this, ‘We would like to take care of you, but you don’t really fit our model. This therapy is very expensive, so we’re not sure how we’re going to slot it in. Oh, by the way, we’ve now spent our annual budget on the things we already knew about, and we’re not sure what to do with you. That’s one model.
The next are the kind of hybrid models where you have public and you have private insurance. Those systems might actually be doing a little bit better with rare disease. Let’s take gene therapy, for instance. We’re finding a pretty good acceptance of employing gene therapy for these diseases. While it’s very expensive, it improves your patient’s quality of life. But, from the payer standpoint, you may recognise cost savings by spending a lot of money at one point but not having to spend as much down the line. I see a little bit more flexibility in those systems.
Now talking about the workforce as a challenge. The fields that care for patients with rare disease are typically non-procedural fields. Now, let me explain why that’s important. In most healthcare systems and models, a lot of the financial emphasis is on surgeries, ICU beds, procedural base, laboratory base, and things like that. Thinking about a patient doesn’t pay in almost any medical system. It’s rewarding beyond belief; I love doing what I do. But it doesn’t have a great financial model. That’s one reason we tend to have fewer people in the field than I would like to see because the financial incentives are not as great. And in a national system, we have a dearth of providers. One thing we’re finding is that students graduating from medical colleges, a lot of them are coming out with a lot of debt. Particularly in the United States and some other countries, and they have to look at fields that may have better pay. We try to get folks interested in the field; it is a fascinating field. The patients are wonderful. It’s intellectually stimulating. So any of you out there who are trainees, please, goodness gracious, it’s a great field to be in! But we do have workforce issues.
With genetic counselling, which are our professionals who go through and work with patients explaining the testing, we’re actually competing with industry, the large sequencing houses, and companies that provide commercial sequencing. They hire a lot of genetic counsellors, so the counsellors we would normally have in our clinics helping diagnose, many of them are actually on the industry side right now. Part of it is our own fault; we have not done a great job of providing tools and systems for primary care providers and front-line practitioners to be able to diagnose rare disease well.
One thing we’re working on is building out protocols for when a primary care provider is presented with a patient; they actually have some recipes they can follow to help that patient get a diagnosis. It shouldn’t mean that you have to wait a year to see the person who can do it. And that’s the typical wait time for rare disease physicians and medical geneticists. The digital environment can also help push things forward. That leads into what I call the diagnostic odyssey for patients. There are some diseases that can often go a decade or more of not being diagnosed, even though the patient has symptoms. And there are a lot of approaches trying to get those diagnostic tools pushed closer out to the front line. Everyone talks a lot about using sequence analysis and newborn screening. In my opinion, I would say that we’re not ready for that yet because we don’t know how to interpret so much of what we find in the human genome. So our false positive rate would be enormously more than the true positive rate of things we would catch. Think about that for a parent when they’ve got a new baby. Someone calls and says, ‘We found a change that might kill your child, but we’re not really sure, but call us if anything funny happens.’, That’s not going to be particularly reassuring.
FLG: Definitely not. What can other areas of medicine learn from the rare disease sector as we shift our systems to patient-led more widely? With sometimes complex diagnoses, we’re really bad at understanding risk. Empowering the patient often comes with more responsibility on them as well. So what challenges do you see and what opportunities?
Marshall Summar: I tend to see opportunities more than challenges. What I would say is, in rare disease, the patient has to be a partner because they know more often about their own version of it than anyone else. How I think that translates into the rest of medicine is going against a trend we’ve had, where we want to lump everyone into an enormous group, and figure out the best way to treat that enormous group. But what we find is that an enormous group is made up of individuals. As molecular diagnoses continue to evolve and improve, we’re going to find changes that will impact even routine health care. How do you do this surgery for this patient with a joint problem? Or do you wait because they may have a better regenerative capacity than someone else and let them go forward? I think we’re going to start to see those kinds of changes work their way in.
Some of it’s just getting the system used to the fact that you’ve got 25,000 plus different genes interacting together. So, it really does speak to the fact that these patients each have an individual presentation, but there’ll be some commonalities. That’s one of the ways we can spot what’s going on. But you have to treat that patient as an individual and be flexible about your therapy. One size fits all does not work well in rare disease because of the variability in genetic heterogeneity, expression, phenotype, all those things. You may start with a generic here’s how we’re going to address this, but you’re going to end up having to tailor it for that specific patient to meet their needs, whether they’re biochemical or other needs, so that you can optimise what you’re trying to do. You have to start with a cookbook and start with the recipe. But you’ve also got to monitor carefully and adjust and adapt. I would say that’s one of the ways it’s very different, and I think we will be teaching other fields. As more common conditions split into small molecular subgroups, you’re going to have to start tailoring therapies for those breakdowns in those smaller groups. And like I said, they look a lot more like rare disease when you start doing that.
FLG: Thank you. And to link to your earlier point about data. This year, we had a biotech company announce their $100 genome. That’s making data more easily available than before. So what challenges and benefits do you see of reduced costs in genetic sequencing?
Marshall Summar: I am all for reducing costs. The thing to remember is the sequence itself is the least expensive part of sequence analysis. If we do a whole genome, we’re sequencing several billion bases of DNA. It was only a couple of months ago that we considered we had a complete human genome sequence, we had most of it, but there were some really hard-to-sequence parts we didn’t get until about three months ago. If I sequenced both of us, I would find about 10 million variations between us. Let’s say one of us has a disease and the other does not; which of those 10 million variations caused the disease? There’s a bit of an interpretive dance involved. If it’s something we’ve seen before, that’s a lot easier. You know, the delta f508 mutation in the CFTR gene is known to cause cystic fibrosis. That one’s pretty easy. But what if you find a change in that gene that no one has seen before?
We’re going to hear the term variants of unknown significance thrown around a lot. If I sequence anyone, I’m going to find some serious changes in some genes. But are they relevant to the clinical picture in front of us? So, $100 genomes are fine. But you still have to analyse the data and see if there’s a link between the changes you find. I guess one way of putting it is, we’ve got a giant monster-size puzzle with lots of pieces, and we’re starting to fill in some of the edge pieces. We’ve got to fill in a whole lot more before you can really see the impact of some of the cheaper sequencing. It’s the sequence analysis because we still require a knowledgeable human intermediary to say, ‘Okay, this makes sense that this variation might link to this.’ But the most common answer in human sequence analysis right now is ‘maybe’. What you get is a ‘maybe this variation may have caused this, there’s a high probability’. It’s one of the reasons we like to sequence families as much as possible, as opposed to an individual. Could this come from the parents? Are there any siblings with the same change who don’t have anything or may have the same thing? So we’re still kind of feeling our way through. I think that’s going to take a few years because it’s a complex problem. But every day, we fill in a little bit more of the puzzle.
FLG: That’s fantastic. As we’re almost out of time, I’d love to finish off by revisiting your fun project on expert opinion. And we can dive into that.
Marshall Summar: You’ll see this coming out sometime in the next year. We got together a lot of folks in the field said, ‘Okay, how do we solve this workforce problem and get some standard care to the patients?’ Anytime in any rare disease field, when have pulled together our best treatment practices, we improve outcomes. Cystic fibrosis is a beautiful example. Before there were any new drugs, simply pulling the best practices together almost doubled the life expectancy of the patients.
We have a programme we’re working on with the same folks that built the REDCap programme at Vanderbilt University. This will be called RARECap. What it is, is clinically actionable protocols. They are based around scenario presentation, confirming diagnosis, chronic management, impact on routine medical care and special situations like pregnancy and chemotherapy. What do you do with that disease in those situations? That’s what I call Friday night guidelines. So, it’s Friday night, the specialists have all gone home, do you have something in front of you that you can use to do that?
But we’re taking a little bit of a different approach. Sort of Wikipedia meets Reddit. You’ll have a core protocol that addresses what’s known about the disease, but the people using it, who are registered users, have to have some knowledge in the field. They comment on the protocol, what they might do differently or what they think might improve it and the community that’s using that can vote those up and down, make comments, and comments are very carefully moderated. But then you can actually use that to improve the protocol with time. You’re pulling best practices. Is it a 100% solution? Absolutely not. It capitalises on that expert opinion with people who are working with those patients on the front line. And the nice thing is, we can ramp this up to cover 1000’s of diseases, as opposed to waiting 20 years until you’ve got published literature. Then you find the published literature is self-contradictory, so you don’t really know anyway!
FLG: Thank you so much, Marshall. I’ve learned so much. I can’t wait to see what that puzzle ends up looking like, whether it’s in the next five years or the next five years after that. And I’m excited to see where patient-led medicine will take us in the learnings we’ll take from rare disease. So thank you for joining me.
Marshall Summar: It’s been my pleasure. Thank you very much.
Watch more interviews in our series here.
