What researchers call “the largest proteomic study in the world” could lead to new clinical laboratory assays for determining genetic risk for multiple cancers
Examining blood proteins may be superior to clinical information in determining an individual’s risk for developing multiple diseases. That’s according to a new study conducted by researchers from the UK, America, and Germany who determined that measuring thousands of proteins from a single drop of blood can predict the onset of several illnesses.
The findings may provide clinical laboratories and physicians with new assays to more accurately predict an individual’s risk for more than 60 diseases.
“With data on genetic, imaging, lifestyle factors and health outcomes over many years, this will be the largest proteomic study in the world to be shared as a global scientific resource,” said Naomi Allen, MSc, DPhil, chief scientist at UK Biobank and professor of epidemiology, University of Oxford, in a UK Biobank news release. “These combined data could enable researchers to make novel scientific discoveries about how circulating proteins influence our health, and to better understand the link between genetics and human disease.”
“Measuring protein levels in the blood is crucial to understanding the link between genetic factors and the development of common life-threatening diseases,” said Naomi Allen, MSc, DPhil (above), chief scientist at UK Biobank and professor of epidemiology, University of Oxford, in a news release. With further study, this research could lead to new clinical laboratory assays that help physicians predict an individual’s risk for certain diseases including many forms of cancer. (Photo copyright: UK Biobank.)
Protein Signatures Outperform PSA Testing
To conduct their research, the team collected data from the UK Biobank Pharma Proteomics Project (UKB-PPP). This initiative is “one of the world’s largest studies of blood protein concentrations” and “aims to significantly enhance the field of ‘proteomics,’ enabling better understanding of disease processes and supporting innovative drug development,” according to the Biobank’s website.
The scientists analyzed the values of approximately 3,000 plasma proteins among 41,931 participants in the UKB-PPP. They examined the 10-year potential of developing certain diseases by measuring the plasma proteome and linking those observations to incident cases noted in electronic health records (EHRs).
The team specifically looked at the pathology types for several illnesses and utilized advanced techniques to identify a signature of proteins associated with those various diseases. They found their protein-based model exceeded traditional prediction methods when comparing the models with polygenic risk scores.
“We are therefore extremely excited about the opportunities that our protein signatures may have for earlier detection and ultimately improved prognosis for many diseases, including severe conditions such as Multiple myeloma and idiopathic pulmonary fibrosis,” she added. “We identified so many promising examples; the next step is to select high priority diseases and evaluate their proteomic prediction in a clinical setting.”
Identifying Individuals at High Risk for Certain Diseases
Of the thousands of known proteins in humans, the team focused on about 20 proteins found in blood. With as few as five proteins and as many as 20, they were able to do a risk assessment on 67 diseases, including:
The model could prove to be beneficial in the development of new therapies for certain diseases.
“A key challenge in drug development is the identification of patients most likely to benefit from new medicines. This work demonstrates the promise in the use of large-scale proteomic technologies to identify individuals at high risk across a wide range of diseases, and aligns with our approach to use tech to deepen our understanding of human biology and disease,” said Robert Scott, vice president and head of human genetics and genomics, GSK, and co-lead author of the study in the UCL news release.
“Further work will extend these insights and improve our understanding of how they are best applied to support improved success rates and increased efficiency in drug discovery and development,” he added.
“We are extremely excited about the opportunity to identify new markers for screening and diagnosis from the thousands of proteins circulating and now measurable in human blood,” said Claudia Langenberg, PhD, director of the Precision Healthcare University Research Institute (PHURI) at Queen Mary University of London and professor of computational medicine at the Berlin Institute of Health, in the UCL news release. “What we urgently need are proteomic studies of different populations to validate our findings, and effective tests that can measure disease relevant proteins according to clinical standards with affordable methods.”
More research and studies are needed before the protein-based model can be used to predict disease in clinical settings. However, the model could someday provide clinical laboratories, pathologists, and physicians with new assays that more accurately forecast an individual’s risk for certain illnesses.
Biomarkers used by clinical laboratories and anatomic pathologists are usually biological substances or states that can be measured during testing either in vivo or in vitro. However, clinical laboratories may soon be working with biomarkers based on measurable aspects of external human anatomy. One such biomarker employs facial analysis and facial recognition technology to produced phenotypic data that could help pathologists diagnose rare genetic disorders. A human phenotype is data comprised of a person’s “observable characteristics or traits.”
Phenotypic Data from Photographs
Three genomics companies: FDNA, GeneDx, and Blueprint Genetics, are collaborating on a unique project, dubbed Face2Gene Labs. They are using a facial recognition application called Face2Gene developed by FDNA. The application uses artificial intelligence (AI) and phenotyping technology to extract data from facial photographs of patients. The data is then examined and compared to a database of hundreds of thousands of patterns that were generated from photos of patients with known rare genetic disorders. The algorithm then compiles a list of possible diagnoses. The goal is to produce phenotypic data that clinicians can transmit in real-time directly to medical laboratories for analysis.
“Trying to diagnose patients with genetic sequencing is like searching for a pin in a 22,000-needle haystack,” stated Dekel Gelbman, CEO, FDNA, in a news release. “By providing accurate phenotypic and clinical data to the lab directly at the point of genetic interpretation, we are truly realizing the promise of precision medicine. And, with the power of artificial intelligence behind it, clinicians will be pointed toward potential diagnoses that they may have never otherwise considered.”
The Face2Gene application developed by FDNA uses artificial intelligence to compare digital photographs of patients’ faces against hundreds of thousands of stored patterns to help clinicians identify genetic disorders in children. (Photo copyright: FDNA.)
Solomon goes on to praise GeneDx and Blueprint Genetics as examples of innovative and renowned labs adopting technology that will lead the way in pinpointing rare disease and promote further medical advancements.
“This is an important collaboration for several reasons,” states Ben Solomon, MD, a Clinical Geneticist and Managing Director of GeneDx, in the news release. “It’s a great way to leverage clinical and genetic information and machine learning approaches to find answers for the clinicians, patients, and families GeneDx serves. Aside from providing answers, this integration will make the diagnostic testing process easier, smoother, and more enjoyable for clinicians.”
85% Increase in Diagnostic Yield with Addition of Phenotypic Data
A recent multi-center study called PEDIA (short for Prioritization of Exome Data by Image Analysis) looked into the accuracy of genetic testing when using FDNA’s Face2Gene tool. The study, conducted by researchers at the Berlin Institute of Health and Charité University of Medicine in Berlin, showed promising results of the collaboration.
“We estimate that the addition of phenotypic features [encoded in HPO terms] increases the diagnostic yield to about 60% [from 25% without],” stated Peter Krawitz, MD, PhD, and Principal Investigator for PEDIA. “When adding facial analysis, FDNA’s technology, to that process, the diagnostic yield increases to more than 85%,” he explained in the FDNA news release.
The Rarity Paradox and Diagnosing Genetic Disorders in Children
According to Global Genes, a rare disease patient advocacy non-profit organization, one in 10 Americans (approximately 30 million people) suffer from a rare genetic disorder. These disorders also affect the same percentage of people worldwide, or about 350 million people. There are more than 7,000 distinct rare diseases known to exist and approximately 80% of those illnesses are caused by faulty genes. In addition, about half of the people affected by rare genetic illnesses are children.
“We call it the rarity paradox,” stated Gelbman in an article published in Wired. “Each rare disease in itself affects very few people, but on aggregate the effect is pretty staggering.”
The three companies hope their collaboration will help clinicians determine faster, more accurate diagnoses, while diminishing anxiety among patients and their families regarding the unknowns of rare genetic disorders.
“Since 2012, Blueprint Genetics has been developing technological innovations in sequencing and clinical interpretation to improve the quality and performance of rare disease diagnostics,” noted Tero-Pekka Alastalo, MD, PhD, President, Chief Medical Officer of Blueprint Genetics, in the FDNA news release. “It’s great to see how these innovations are now helping the genetics community and patients suffering from inherited disorders. Combining these technological innovations with our transparent approach to diagnostics and next generation phenotyping tools like Face2Gene represents the next steps forward in molecular genetic diagnostics.”
Pathology groups and clinical laboratories are advised to monitor this exciting development in genomic research. It illustrates how unrelated technologies, such as facial analysis software, could soon be used for diagnostic purposes to detect the presence of genetic disorders, and to determine the best therapies for patients. Labs will want to be prepared to engage with clinicians who adopt this technology and to answer patients’ questions about it.
