AI tools find 12 biomarkers that may speed a Sjögren’s diagnosis
New model also IDs 2 drugs that could be repurposed as treatments
Written by |
A computer-based gene activity analysis combined with artificial intelligence (AI) methods has identified a dozen potential biomarkers that may ultimately help clinicians in reaching a Sjögren’s disease diagnosis more quickly, according to researchers in China.
Their study found that activity levels of these new biomarkers were significantly associated with blood levels of self-directed antibodies commonly used in the diagnostic workup of Sjögren’s.
“Our integrative approach identifies 12 robust diagnostic biomarkers for [Sjögren’s], offering new insights into disease mechanisms and highlighting potential therapeutic targets for this challenging autoimmune disorder,” the researchers wrote.
Based on these new findings, the team also identified two medications, one for high blood pressure and the other for cancer, that may be repurposed to treat Sjögren’s.
The study, “Integrative Bioinformatics and Machine Learning Identify Novel Diagnostic Biomarkers and Molecular Mechanisms in Sjögren’s Syndrome,” was published in the International Journal of Genomics.
Sjögren’s is an autoimmune condition that mainly affects the glands that produce tears and saliva, resulting in symptoms that commonly include dry eyes and a dry mouth. Many patients also experience fatigue and joint or muscle pain. In some cases, the disease can affect other organs such as the lungs, kidneys, liver, or nerves.
Because symptoms can be vague and many of them — especially those beyond the tear and saliva gland problems — overlap with other conditions, the disease is often misdiagnosed or diagnosed late.
There’s no simple test to diagnose Sjögren’s
No one test can diagnose Sjögren’s. Doctors typically reach a diagnosis after combining information from symptoms, physical exams, blood tests, and, sometimes, a tissue biopsy.
Blood tests look for certain self-directed antibodies commonly associated with the disease, including anti-ANA, anti-SS-A, and SS-B antibodies. Still, not everyone with Sjögren’s has these antibodies, and when these tests are negative, diagnosis can be even further delayed.
To identify new diagnostic biomarkers, a team of researchers at Panjin Liaoyou Gem Flower Hospital analyzed gene activity data from blood samples collected from 351 patients and 91 healthy individuals, who served as controls.
“The goal was to establish a highly sensitive and specific diagnostic model while uncovering novel molecular targets that illuminate [Sjögren’s underlying mechanisms],” the researchers wrote.
The initial analysis identified 50 genes whose activity differed between Sjögren’s patients and controls. Several rounds of computer refinement narrowed this down to 43 genes.
To screen for precise Sjögren’s diagnostic biomarkers, the team applied a machine learning approach, a type of AI that learns from data and detects patterns. This identified 12 core genes — EPSTI1, IFIH1, CXCL10, TNFSF10, GBP5, PARP9, IFI44, LAP3, IFIT2, IFI44L, PARP12, and OAS1 — whose combined activity differences distinguished patients from controls with an accuracy of 82.5%-99.4%.
Each marker had accuracy of at least 80%
The results showed that all 12 genes were significantly more active in people with Sjögren’s relative to healthy controls. Moreover, each of them was able to discriminate between patients and controls with an accuracy of at least 80%, according to the researchers.
The team also noted that higher activity levels for each of the genes were significantly associated with higher blood levels of anti-ANA, anti-SS-A, and SS-B antibodies.
The strongest links were detected between LAP3 and anti-ANA antibodies, IFI44L and anti-SSA antibodies, and IFI44 and anti-SSB antibodies, “reinforcing their clinical relevance in [Sjögren’s] diagnosis and disease activity assessment,” the team wrote.
Additional analyses confirmed the predominant activity of all 12 genes in blood samples of Sjögren’s patients across various immune cell types. This included monocytes and dendritic cells, as well as antibody-producing B-cells and CD4-positive T-cells.
Further computer analysis implicated these genes in immune regulatory networks. Among them: immune cell trafficking, signaling via interferons (proteins known to fight viral infections), and antigen presentation, or when cells display fragments of microbes to the immune system so it knows what to attack.
The team also found pronounced immune dysregulation in people with Sjögren’s. This was characterized by an imbalance between unactivated B-cells and memory B-cells, which act quickly upon re-exposure to a specific threat, and reduced numbers of cell-killing T-cells and regulatory T-cells, which suppress excessive immune responses.
Finally, the researchers sought to identify approved drugs known to repress the activity of one or more of the 12 core genes, with the potential to be repurposed for Sjögren’s. This analysis detected two drugs: nisoldipine (sold under the brand name Sular), approved to treat high blood pressure by relaxing blood vessels, and exemestane (sold as Aromasin), which is used to treat breast cancer by lowering levels of the estrogen hormone.
The team said their findings set a roadmap for further study “advancing [Sjögren’s] diagnosis and treatment [toward] precision and personalized medicine.”
“Future research should focus on validating this model in [patient groups followed over time], developing clinically suitable detection assays, and further exploring the biological functions of the hub genes and the therapeutic potential of the candidate drugs,” the researchers concluded.
