Umbilical cord blood collected at birth and routinely discarded may contain measurable biological signals that predict a child’s future risk of developing Type 1 diabetes, according to new research published in the journal Nature Communications. The findings suggest that the biological conditions associated with the disease may begin taking shape as early as pregnancy, well before any immune response becomes detectable and years before a diagnosis would typically occur.
Type 1 diabetes affects the pancreas, specifically the insulin-producing beta cells that regulate blood sugar. While the condition has long been understood primarily as an autoimmune disease, a growing body of evidence points to a more complex picture in which the beta cells themselves become stressed and begin breaking down before the immune system shows signs of attacking them. Infection, increased energy demands, and a smaller pancreatic size have all been identified as potential sources of that early cellular stress. When stressed beta cells stop functioning or die, they release molecular signals that may trigger the immune response that follows, raising the possibility that the immune attack is a consequence of beta cell injury rather than the cause of it.
What the research found
Researchers drew on data from a long-running cohort study in Sweden that has been following mothers and children since the late 1990s. Cord blood samples collected at birth were retrieved decades later and screened for proteins associated with inflammation. A machine learning model was then used to identify which protein patterns were linked to the later development of Type 1 diabetes in children from the general population, not only those with known genetic risk factors.
Several proteins in the cord blood emerged as meaningful predictors of disease risk. They fell into broad categories including proteins involved in moving molecules through the body, proteins associated with environmental exposures such as pollution, proteins involved in maintaining cell structure, and proteins that help regulate immune responses. Two proteins in particular showed the strongest predictive value, one involved in breaking down long sugar molecules that give tissues their structure and flexibility, and another involved in activating the immune system.
Notably, the predictive power of these proteins was not heavily dependent on genetic factors. Including genetic risk information in the model improved accuracy only marginally, suggesting the protein signals themselves were driving the association.
Why this matters for early detection
Current screening for Type 1 diabetes relies primarily on genetic testing and the detection of autoantibodies, proteins that indicate the immune system has already begun targeting insulin-producing cells. By the time autoantibodies appear, the biological changes that set the stage for the disease may already be well advanced. Cord blood biomarkers could provide an earlier window, one that allows clinicians and families to identify risk and begin monitoring or intervention before those changes become irreversible.
Some of the markers identified in the study may also be linked to environmental exposures during pregnancy, including certain synthetic chemicals that pregnant people encounter routinely. That dimension of the findings has implications beyond individual health, pointing toward environmental and public health policy as a potential lever for reducing disease risk at a population level.
Cord blood as a resource worth preserving
Cord blood is already recognized as a source of stem cells used in life-saving treatments for a range of conditions. This research adds to the case for treating it as a diagnostic resource as well, one that carries information about early biological conditions that shape health outcomes across childhood and beyond. The research team is currently extending their work to examine cord blood markers in relation to other conditions including childhood obesity, depression, autism, and inflammatory bowel disease.
The researchers were clear that the findings reflect possibility rather than destiny. The biomarkers identify elevated risk, not a guaranteed outcome, and broader population studies will be needed before the approach can be applied clinically.
