By Allison Proffitt
November 11, 2021 | Yesterday, the U.K. 100,000 Genomes Project published the results of a pilot study seeking diagnoses for 4,660 participants from 2,183 families with rare diseases. They found that whole genome sequencing measurably improved diagnoses for patients with a wide range of genetic illnesses and saved money for the healthcare system. Results were published in the New England Journal of Medicine (DOI: 10.1056/NEJMoa2035790)
Whole genome sequencing has been used for diagnoses of rare disease cases for years, but it has not yet been incorporated into routine care. That was a main goal of the pilot project, said Professor Sir Mark Caulfield of Queen Mary University of London, lead author on the paper and former Chief Scientist at Genomics England, in a briefing earlier this week.
With the data uncovered during this pilot study, Caulfield said they’ve “transformed the NHS”. The team built a combined NHS-research infrastructure, created an NHS Genomic Test Directory comprising everything from single-gene tests to whole-genome sequencing, and launched a national Genomic Medicine Service to provide consistent and equitable access to testing.
Some of these steps have been in place for several years, built, Caulfield said, on the earliest results of the pilot study. The full study results published in NEJM yesterday are meant to serve as a model for other countries to emulate. “Today we’re sharing our findings so that other healthcare systems can consider adoption to address unmet need in rare disease worldwide,” he said.
The study itself looked at 4,660 undiagnosed participants from 2,183 families with suspected rare diseases. Patients were included from nine English hospitals with a broad spectrum of disease categories. Family members were enrolled as well, when possible. Patients were recruited in 2014 and 2016; results were returned between May 2016 and April 2019.
Participants were chosen representing 161 disease categories across a broad range of disorders: from cardiovascular disorders, endocrine disorders, growth disorders, to hematologist disorders, intellectual disability, renal or urinary tract disorder, respiratory disorder, tumor syndrome and more.
“A really distinctive feature of this study is that this is the first to apply WGS to a wide range of diseases with unmet diagnostic need, selected by clinicians in the NHS who are involved in direct healthcare of those people,” Caulfield said.
Whole genome sequencing was performed using the TruSeq DNA polymerase-chain-reaction (PCR)–free sample preparation kit (Illumina) on a HiSeq 2500 sequencer. Reads were aligned to the Genome Reference Consortium human genome build 37 (GRCh37) with the use of Isaac Genome Alignment Software and variant calling was done with the Platypus variant caller. In addition, the researchers collected data on clinical features using Human Phenotype Ontology terms.
Diagnostic yield across the cohort was 25%—itself a significant improvement for a population of persistently undiagnosed patients—but for hearing and ear disorders, metabolic disorders, intellectual disability, neurological or neurodevelopment disorders, and ophthalmologist disorders, the diagnostic yield was higher. “Some were up to 40-45% diagnostic yield,” Caulfield said. “This is beginning to be really successful.”
“Of these diagnoses,” the authors write, “60% were made on the basis of coding single nucleotide variants (SNVs) or insertions/deletions (indels) in the applied panels; 26% were made on the basis of coding SNVs or indels affecting well-established disease genes not included in the applied panels (diagnoses were made through phenotype-based prioritization or expert review by the study clinicians or the clinical genetics teams from Congenica or Fabric Genomics); and 14% were made on the basis of genome-wide, phenotype-agnostic research analysis that investigated beyond SNVs and indels, coding regions, and disease genes in the applied panels”
The authors estimate that 14% of the diagnoses they found would not have been made by other tests because 4% were single nucleotide variants that were not in coding regions, 2% were SNVs in coding regions poorly read by other tests, and 8% were structural changes, many in non-coding regions.
Cost of Care
A diagnosis is not just a relief for patients and families, the authors point out can represent significant cost savings to the medical system.
To estimate the cost of the delayed diagnosis, the authors compared the cost to the healthcare system of the undiagnosed participant and compared it to the costs incurred by their unaffected siblings. They found that affected participants used 183,273 episodes of care at a cost of £87 million, while their siblings used 53,706 episodes of care at a cost of £21 million.
“What we see is that once we make a diagnosis, the NHS can make a much more focused clinical care, and we see fewer focused episodes over the next 18 months,” Caulfield said.
The authors shared a few specific examples:
- A 10-year-old girl whose previous seven-year search for a diagnosis had multiple intensive care admissions over 307 hospital visits at a cost of £356,571. Genomic diagnosis enabled her to receive a curative bone marrow transplant (at a cost of £70,000). In addition, predictive testing of her siblings showed no further family members were at risk.
- A man in his 60s who had endured years of treatment for a serious kidney disease, including two kidney transplants. Already knowing his daughter had inherited the same condition, a genomic diagnosis made by looking at the whole genome for him and his daughter enabled his 15-year-old granddaughter to be tested. This revealed she had not inherited the disease and could cease regular costly check-ups.
- A baby who became severely ill immediately after birth and sadly died at four months but with no diagnosis and healthcare costs of £80,000. Analysis of his whole genome uncovered a severe metabolic disorder due to inability to take vitamin B12 inside cells explaining his illness. This enabled a predictive test to be offered to his younger brother within one week of his birth. The younger child was diagnosed with the same disorder but was able to be treated with weekly vitamin B12 injections to prevent progression of the illness.
The authors report that 25% of the diagnoses had immediate clinical utility.
“What that means is [the diagnosis] changed a medication, led to additional surveillance for the [patient] or their relatives, led to eligibility for clinical trials they wouldn’t have got into, or the diagnosis could inform future reproductive choices,” Caulfield said. “And there were other benefits—simple benefits—that would be really easy to do if we knew what the cause of these diseases were, for example: dietary change or vitamin supplements.”
While the benefit to families was the authors’ primary concern, said Richard Scott, Chief Medical Officer at Genomics England, diagnosis for just these rare disease families is not the ultimate goal.
“There’s a real potential there in the future to grow what, at the moment, is quite a small proportion of people,” Scott said. “Let’s turn this from a conversation about diagnosis to one about real impact.”
Caulfield agreed. “You need the complex research-NHS ecosystem that Richard outlined to get the full value for the patients and the participants. That’s an important innovation that’s now live in the NHS as part of the Genomic Medicine Service.”
“What’s usual about the UK is not just the scale of the investment, but paired investment from healthcare and research and recognizing the two need to work really closely together to deliver the best for the patient,” Scott said. Elsewhere, he observed, “it’s very hard… to get that paired investment that’s coordinated, that recognizes that the two benefit each other enormously.”
The National Health Service has set a long-term goal of 300,000 whole genome sequences. That’s a lot of data, and Caulfield said an international coalition of 3,600 researchers from 33 countries have offered to work on these data to drive up the value.
Genomics England and its partners are also planning a new pan-rare disease re-analysis to add other RNA and protein data.
“We hope to involve a lot more functional characterization of the variants so we don’t end up with 500 signals that we can’t resolve,” Caulfield said. “We want to go from the patient to the genetic change to the function. Because if you have the function, you can describe the biology of the disease and that allows you, potentially, to find therapies.”