A Rare Disease Revolution
Genomic medicine is on the upswing – and rapid, reliable, accessible sequencing could significantly shorten the diagnostic odyssey for many rare disease patients
Stephen Kingsmore | | 15 min read | Review
Genomics started off as a purely research exercise – but, over the last decade, it has increasingly come to the fore in medicine. Scientific advances have led to significant improvements in the diagnosis and management of rare diseases with genetic causes. Most of the burden of such illness is pediatric and many of these diseases are fatal in childhood. In fact, they are the leading cause of infant mortality in the first world – and therefore the biggest determinant of childhood health. People don’t realize that because genetic disease is like an iceberg; we see the patients who are diagnosed early or survive long enough to receive treatment, but not those who don’t.
The (pediatric) rare disease landscape
Currently, there are over 7,200 known genetic diseases – and that number increases by about one a day. We have had some spectacular success in identifying some relatively common diseases, such as sickle cell anemia, thalassemia, or cystic fibrosis. However, the vast majority of rare diseases affect fewer than one in 10,000 individuals and are massively underdiagnosed. Why? Because until recently, we had no way to diagnose these diseases other than by family history.
That all changed when we gained the ability to sequence and decode the entire genome. That’s when we realized just how much we had been operating in the dark. We hadn’t even known that many of these diseases were genetic; we had theorized that they might be environmental or simply labeled them as “unknown etiology.” For example, there are well over 1,000 different genetic diseases that cause seizures in children. The same is true of deafness or intellectual disability. We had always believed that patients with these phenotypes had complex disease affected by environmental factors – when, in fact, most of the conditions were rooted in genetics all along. This meant that many of our patients had been receiving ineffective treatments because they were being treated for non-genetic causes.
Nowadays, we can diagnose a genetic disease in under a day from delivery of a blood sample to result. What proportion of children benefit from this technology? I would estimate about 0.01 percent. The tragedy is that it will take years – hopefully not decades – for children around the world to have access to these diagnostics. There are so many barriers: infrastructure, reimbursement, and training. The vast majority of physicians were never trained in genomics and therefore don’t know how to practice genomic medicine. So it’s a complex issue – not just a matter of legislation or evidence, but also one of upskilling the medical workforce.
The diagnostic odyssey
Rady Children’s Hospital is unique in our approach to whole genome sequencing for all patients. If your child is seen in our autism clinic, for example, genome sequencing is available on the first visit because there are so many genetic causes underlying autism spectrum conditions. If your child is admitted to our neonatal intensive care unit, genome sequencing is immediately available to diagnose any potential genetic conditions. Even at Rady, doctors still underrecognize genetic disease, but we’re getting there. We also have 81 partner children’s hospitals across North America who send us medical records and DNA samples from children who may benefit from whole genome sequencing. Genomics is becoming increasingly recognized and democratized, but the need still vastly outweighs the provision – or the reimbursement.
For patients who don’t have access to whole genome sequencing, the diagnostic journey can be disheartening, painful, or even dangerous. It can take decades of going from physician to physician, specialist to specialist, subspecialist to subspecialist, until finally the penny drops, genomic testing is ordered, and we get an answer. Families who travel this road go through years of suffering, being misunderstood, and having their children dismissed as having emotional or psychological problems. Many experience organ failure before diagnosis – and that cannot be remedied, especially in neurological conditions. The tragedy today is that we have treatable illnesses that are under- or misdiagnosed, leading to outcomes that are needlessly bad. My patients and their families often ask, “How can this happen in the 21st century?” – and I sympathize with their frustration.
A story of success
We had a five-week-old patient whose mother brought him into our emergency department late on a Sunday night. She explained that her baby – who had no previous health issues – was irritable and couldn’t be consoled. Fortunately, the pediatrician on call noticed that his eye movement was not quite normal and ordered a stat CT scan that showed white lesions all over the patient’s brain.
He was immediately admitted to our neonatal intensive care unit and, by lunchtime the next day, a decision had been made to perform whole genome sequencing. The patient’s blood sample arrived at our institute at 5:00 pm. By 7:30 am, we had a diagnosis – thiamine metabolism dysfunction syndrome 2 (1). Without treatment, the disease is rapidly fatal; in fact, in the short time between admission and diagnosis, the patient had already experienced a seizure, was lethargic, and couldn’t feed. We immediately administered treatment – large doses of biotin and thiamine – and, within six hours, he was back to normal. He was alert, happy, peaceful, and feeding from a bottle. A couple of days later, he went home. One year later, he’s an affectionate, loving little toddler hitting many of his developmental milestones.
A decade earlier, the patient’s sister had presented with similar symptoms. Undiagnosed, she was treated with every antiepileptic drug available – but none controlled her seizures and she died in infancy of profound neurological devastation. Whole genome sequencing was the difference between a premature death and a couple of days in the hospital. Our patient will be on high-dose biotin and thiamine supplementation for life, but that life will probably last 80 to 90 healthy, happy years. Everyone wins.
The future of rare disease
Our hospital has committed to decoding the genomes of 10,000 children in families of need over the next three years. I think of us as an “icebreaker” – somebody has to go through that ice first and create the knowledge base for the entire medical establishment so that other centers follow suit. We began with our most critically ill patients – those in intensive care – because they have the highest incidence of genetic disease. Now, though, we’re broadening our focus to our outpatient clinics.
At the same time, we’re working on more affordable testing options, because diagnostic whole genome sequencing – especially if it’s done in one day – is very expensive. We’re planning to introduce newborn screening by whole genome sequencing so that, over the next decade, every newborn baby is fully screened for genetic diseases. The ideal solution is to diagnose every child at birth, before onset of symptoms, and start them on effective treatment immediately (where available). That, in turn, will spur massive pipeline development by pharmaceutical companies – because a diagnosis without a treatment is better than no diagnosis at all, but what we really want are effective therapies for our patients so that they can live full, healthy, normal lives.
That’s the journey that we’re on – breaking the ice, focusing applications of whole genome sequencing, developing affordable testing, and partnering with the pharmaceutical industry to accelerate drug development pipelines. It sounds like a lot, but it’s not just us; there’s an ever-growing community of people who are pushing to make these dreams a reality.
Genomic science has evolved very quickly, so traditional pathology is now playing catch-up. Genomics was originally part of clinical pathology, but it’s fast becoming a standalone specialty. At Rady, we have conventional pathologists, molecular pathologists, and laboratory directors who have PhDs rather than MDs. The division of duties and expertise is in flux – it feels a little bit like, “Who’s on first?” I would say that traditional pathology hasn’t yet fully entered the molecular era for rare, inherited genetic disease, so we need to democratize knowledge and make it mainstream for every pediatric subspecialty. If you’re a pediatric nephrologist, you need to know the genetic disorders of the kidney – but not necessarily the liver or the brain. The idea that a singular person will cover this space is a fallacy; there are far too many of these diseases and they’re highly variable in terms of the organ systems that they affect. The key is to work collaboratively.
Children with rare diseases tend to be medically complex patients who require care from a coordinated team. Those whose diagnoses don’t come with effective treatments will need a team-based care approach throughout their lives. Historically, these patients often died in infancy or early childhood – so now, for the first time, adult medicine is encountering the need for this kind of care. Our patients can’t continue seeing pediatricians well into their 40s; we need to train adult practitioners and establish adult clinics for rare diseases whose treatment and survival rates are improving.
The first time we set the Guinness world record for fastest whole genome sequence was in 2011 and it took us 56 hours. In 2017, we improved it to 26 hours, then 19 hours in 2019 and 13.5 hours in 2021. Unfortunately, I think we’ve hit the limit on recertifying the official record – but we’re still getting faster!
Why do we do this? Because we’re trying to change the world. Researchers don’t usually do this kind of publicly visible “fun” – but genomics still has a lot of resistance to overcome. To say, “We’re going to decode your baby’s genome” – that’s a big concept for a parent to take in, so we need things like the Guinness world record to get people excited about genomics.
There are still many misunderstandings and misconceptions around genomic medicine, which is why we consider education and engagement one of our most important responsibilities. We need the public to hear about the benefits of genomics and we need scientific and medical professionals to understand the research so that they can communicate it to their patients. We don’t want to wait 20 years for a new generation of doctors to emerge; we want the current medical establishment to realize how exciting and fulfilling genomics can be.
If I could say one thing to my colleagues in pathology and laboratory medicine, it would be, “Genomes are good.” Always be thinking about the genome. Anytime something is atypical, think genome. Even if whole genome sequencing turns up no results, you’ve gained valuable information about your patient – we even published studies showing that 72 percent of clinicians found negative results useful (2) and 97 percent of parents agreed that it was useful despite only 23 percent of children receiving genomic diagnoses (3). Diagnoses are valuable, but ruling out genetic causes of disease can be equally important to help focus diagnostic attention elsewhere – and to reassure parents that their child’s genome is healthy. So – think genome!
The Rare Disease Revolution: Awareness
To simplify the diagnostic odyssey, we must collect and use all of the information available to us – from genetic sequences to literature review
The biggest challenge facing rare disease diagnosis is the awareness and recognition of each of these diseases – each of which individual clinicians very rarely encounter – and the challenge associated with interpreting genetic testing results.
The widespread adoption of next-generation sequencing has led to rapid whole genome and whole exome sequencing, in turn increasing the rate at which rare diseases – often not present on limited gene panels – are diagnosed. In addition, result interpretation has been accelerated by the increased use of automated pipelines for secondary and tertiary analytics, meaning that the time from sequencing to diagnosis can be as little as a day. In the future, I anticipate that these advances will result in whole genome and whole exome sequencing as a standard of care – and that next-generation sequencing will be universally applied to newborn screening to increase early diagnosis of genetic disease.
The medical literature, which contains millions of articles and is continually expanding, is extremely difficult to navigate – especially for rare diseases – but holds valuable clinical and functional evidence that can impact patients’ diagnostic journeys. To ensure accurate and efficient diagnoses, we need a way to automatically aggregate and annotate evidence from the literature.
In addition, patient advocacy groups often collect clinical and genetic information from their patients that could be instrumental in better understanding individual diseases and their causes – but organizing and widely sharing that information with the clinicians, researchers, and pharmaceutical companies that could benefit from it presents a challenge. These groups need a more efficient way to gather, standardize, and share information to maximize its value for continued research.
Collectively, rare diseases are not all that rare, and so must always be considered in the diagnostic process. Genetic testing will be crucial in ensuring that this process is both efficient and sensitive; this is especially true when using whole genome or whole exome sequencing, because these approaches offer the highest likelihood of identifying a causative mutation. In addition, establishing more efficient ways to aggregate and use clinical and genetic information from the literature, patient advocacy groups, and other sources will be necessary to ensure that all patients are diagnosed rapidly, efficiently, and appropriately.
Mark Kiel is Founder and Chief Scientific Officer at Genomenon, Ann Arbor, Michigan, USA.
The Rare Disease Revolution: AI
The biggest challenge in rare disease diagnosis – and how AI can help overcome it
The phenotypes clinicians describe are largely encoded in the form of text-based clinical notes, often for months or years before the patient receives a definitive rare disease diagnosis. This means that the greatest impediment to advancing lifesaving rare disease research lies in synthesizing each patient’s longitudinal journey to understand their condition’s natural history – and, once diagnosed, the outcome of each treatment option they explore.
The ongoing artificial intelligence (AI) revolution is enabling machine-augmented curation of tens of millions of patients’ electronic health records (EHRs) – a total of billions of clinical notes from over a century of care. This synthesized institutional biomedical knowledge is revealing, for the first time, the natural history of each rare disease, from the first observable phenotypes through many years of missed or incorrect diagnoses, eventual correct diagnoses, and treatments. In many rare diseases, AI is also making connections to more common conditions whose FDA-approved treatments may offer a glimmer of hope in treating related rare diseases.
Bringing pathology into the digital era at scale by digitizing billions of pathology slides and creating the world’s largest digital knowledge base of pathological and normal tissues will be a game-changer in the AI revolution – but realizing this promise to transform biomedical research will not materialize unless these digital images are tied meaningfully to curated health records, including disease phenotypes, attempted treatments, and patient outcomes.
Venky Soundararajan is Co-Founder and Chief Scientific Officer at nference, Cambridge, Massachusetts, USA.
The Rare Disease Revolution: Genomics
Bringing whole genome sequencing into rare disease diagnosis, prognosis, and treatment
Rare disease diagnostics using whole genome sequencing (WGS) are essential for many people suffering from unknown illnesses. The greatest challenge with WGS is the myth that sequencing and analysis are too expensive – or too complicated – to implement in a diagnostic setting. There’s a perception that WGS requires massive capital outlays and expertise, but this is no longer the case. The cost of sequencing has come down so significantly that many labs can now purchase their own instruments or send out samples for affordable contract sequencing. In addition, customers can now also access cost-effective turnkey solutions – or even full-service support – for help with analysis and interpretation while still delivering turnaround times of 48 hours or less. Even if you don’t have genomic experience or technology in-house, you can still provide rare disease patients with rapid whole genome analysis.
Access and equity
With many rare diseases, particularly childhood diseases and time-sensitive medical conditions, reducing the time to diagnosis can significantly improve outcomes and quality of life. Today, we can go from sample to result in just a few days and prioritize high-value diagnostic candidates for clinicians so that they can focus their attention on a smaller number of high-value, potentially causative candidate genes identified by artificial intelligence (AI). We’ve come a long way in the last couple of years; now, our task is to increase awareness of – and access to – techniques and technologies that can alleviate the burden of the diagnostic odyssey.
Many of the challenges with access and equity are tied to overall awareness. There has been so much discussion of things like the “race to the $1,000 genome” that interpretation – making sense of those three billion base pairs – has been overshadowed by hardware discussions. Today, services that provide ready access to whole genome data and AI-powered clinical decision support platforms are making it possible for any hospital anywhere to offer state-of-the-art whole genome diagnosis without requiring genomic infrastructure. And, as reimbursement for WGS continues to expand, that equity will also expand.
AI is really what makes genomic analysis at scale possible. There is no way one human can study millions of variants in a whole genome sequence and turn around an answer in a day or two – and, even if there were, without artificial intelligence, the cost becomes unmanageable at scale.
Training AI on large, population-level WGS datasets has paved the way for better interpretation tools that can automate and accelerate what has been a time-consuming – and, in some cases, impossible – task. With AI, instead of looking at the entire genome, pathologists can focus on the variants with the highest likelihood of causing disease at a cost that is increasingly reasonable and at a speed that is feasible for time-sensitive, critical care situations.
And, if that seems like too much, clinics can start with a simpler custom panel based on WGS and consistently test exactly what they’re looking for now – with the added ability to go back to the samples and data later if they find other validated genes they want to study. These panels can start small and be expanded as needed because the data are there – and they can be used for research across populations, too.
Ultimately, AI analysis of whole genomes means you can get high accuracy and rapid clinical reporting, allowing laboratories to provide greater access to the lifesaving answers a genome may hold. And, with expanding reimbursement for rare disease sequencing and treatment, we can look forward to a time when all rare diseases are identified early and treated fast – potentially saving many patients’ lives.
Jeanette McCarthy is Vice President of Precision Medicine, Fabric Genomics
- MJ Owen et al., “Rapid sequencing-based diagnosis of thiamine metabolism dysfunction syndrome,” N Engl J Med, 384, 2159 (2021). PMID: 34077649.
- DP Dimmock et al., “An RCT of rapid genomic sequencing among seriously ill infants results in high clinical utility, changes in management, and low perceived harm,” Am J Hum Genet, 107, 942 (2020). PMID: 33157007.
- JA Cakici et al., “A prospective study of parental perceptions of rapid whole-genome and -exome sequencing among seriously ill infants,” Am J Hum Genet, 107, 953 (2020). PMID: 33157008.