OTraces Inc., a biomedical diagnostics company with offices in Deerfield Beach, Florida and Sykesville, Maryland, has published a retrospective analysis claiming its proteomic biomarker method can identify which children will develop Stage 3 (symptomatic) type 1 diabetes with “overall predictive power above 92%” up to five years before diagnosis. It’s a notable claim if it holds up, but it’s important to be clear about what this actually is right now: a company’s own retrospective look at existing data, not a peer-reviewed publication or a prospective clinical validation.
What OTraces Actually Analyzed
The analysis draws on real, NIH-funded data from the TEDDY study (The Environmental Determinants of Diabetes in the Young), a multicenter prospective cohort study that has followed roughly 8,676 genetically at-risk children from birth since 2004 specifically to study how type 1 diabetes develops. OTraces looked at data from 590 children who went on to develop T1D and 568 who didn’t, applying what it calls Proteomic Noise Suppression technology combined with spatial proximity classification to four established biomarkers (ZnT8 autoantibodies, mIAA, IA2A, and GADA). The company reports negative and positive predictive values of 92.9% and 96.0% at one year before diagnosis, 94.9% and 97.1% at three years, and 89.2% and 83.3% at five years. For context, type 1 diabetes is staged clinically in three phases under a framework endorsed by JDRF, the Endocrine Society, and the American Diabetes Association: Stage 1 is autoimmunity with normal blood sugar, Stage 2 is autoimmunity with abnormal but not yet diabetic blood sugar, and Stage 3 is the onset of symptomatic disease, the stage OTraces says its method predicts years in advance.
The Comparison OTraces Is Drawing
The company frames its numbers against two benchmarks: a conventional logistic regression model it ran on the same data, which it says only achieved 29.1% to 54.4% sensitivity depending on the time horizon, and previously published TEDDY-based prediction models, which it characterizes as achieving 35-50% specificity at a fixed 90% sensitivity, compared to the 84-94% specificity range OTraces reports for its own method. Alain Cappeluti, OTraces’ Co-founder and President, framed the clinical stakes: “The ability to identify children at risk for T1D years before onset represents a transformative opportunity for early intervention and clinical trial recruitment.”
Why the Caveats Matter Here
OTraces itself is upfront about the limits of what’s been shown so far, and those limits are worth stating plainly rather than burying. In the company’s own words: “These findings come from a retrospective analysis of existing TEDDY cohort data and have not been evaluated in a prospective clinical study.” The release also states plainly that “OTraces’ analytical methods described here are for research use only and have not been cleared or approved by the U.S. Food and Drug Administration for the prediction or diagnosis of type 1 diabetes.” In practice, that means these numbers come from OTraces analyzing historical data it already knew the outcomes for, a useful first step, but a different and much lower bar than a prospective study that tracks new patients forward in time without knowing in advance who will develop diabetes, or than an independent, peer-reviewed publication. Readers should treat this as an early, company-reported result worth watching rather than a validated diagnostic breakthrough.
OTraces has previously worked on proteomic diagnostic approaches for ovarian, pancreatic, and prostate cancer, and lists both a Florida and a Maryland address, though the specific role of its Sykesville, Maryland location isn’t detailed in this release. If OTraces pursues independent peer review or a prospective study on this T1D method, that would be the next real test of whether these numbers hold up outside the company’s own analysis.
