An Experimental 2-HB Assay Flagged Post-Meal Dysglycemia

A study in the August issue of Diabetes Care tested an experimental blood marker called 2-hydroxybutyrate, or 2-HB, as a way to identify post-meal dysglycemia that conventional fasting measures may not capture. The assay showed useful rule-out performance in one cohort. It did not establish a new screening standard.

The distinction matters because the strongest-looking statistic in a diagnostic study can depend heavily on who was tested, how common the target condition was, and whether the assay is being used to rule a condition out or rule it in.

What the researchers measured

The analysis included 772 people from the Biomarkers of Personalized Medicine cohort. The researchers measured fasting 2-HB with a named enzymatic assay called XpressGT and compared it with dysglycemia defined by a two-hour glucose result above 140 mg/dL during an oral glucose tolerance test.

They also examined a prespecified subgroup of 534 people whose fasting plasma glucose was below 110 mg/dL and whose A1c was below 5.7 percent. In that subgroup, 2-HB produced an area under the receiver operating characteristic curve of 0.79 for post-meal dysglycemia. Each one-standard-deviation increase in 2-HB was associated with more than twice the odds of dysglycemia after adjustment, with an odds ratio of 2.23.

An AUC of 0.79 means the marker separated the study's dysglycemia and non-dysglycemia groups better than chance. It does not mean the test was 79 percent accurate, and it does not tell us how it would perform in a general screening population.

The rule-out result was stronger than the rule-in result

At a 2-HB cutoff near 39 micromoles per liter, the study reported 90 percent sensitivity. Supplementary Table S1 reports 53 percent specificity, a positive predictive value of 10 percent, and a negative predictive value of 98.1 percent at 39.15 micromoles per liter.

That pattern is the heart of the paper. A result below the cutoff may be useful for identifying a low-probability group in a population like the one studied. A result above the cutoff was not close to a diagnosis: at that threshold, only about one in ten positive results represented the study's oral-glucose-tolerance outcome. Many people would still need the reference test.

Predictive values also change when prevalence changes. A high negative predictive value from this cohort cannot simply be carried into every clinic, age group, or risk population.

The paper's fasting-glucose label needs context

The paper describes the 534-person subgroup as having normal fasting glucose and A1c, but its fasting-glucose cutoff was below 110 mg/dL. The American Diabetes Association's 2026 Standards of Care define impaired fasting glucose as 100 to 125 mg/dL. That means the study subgroup could include people whose fasting glucose already fell within the ADA prediabetes range.

This does not invalidate the analysis. It does make the phrase beyond normal fasting measures less clean than it sounds. The study is best read as an assay-performance result under the authors' prespecified thresholds, not proof that 2-HB routinely discovers disease after standard screening has declared everything normal.

The commercial connection is direct

Two authors list affiliations with DirectSens GmbH. DirectSens markets XpressGT, describes a research-use-only kit as available, and says it is seeking licensing, laboratory-developed-test, and co-development partners.

Company involvement does not make a peer-reviewed result wrong. It raises the value of independent replication, transparent regulatory status, and real-world studies that compare the full workflow with established screening strategies. The paper did not show that using 2-HB improves health outcomes, prevents diabetes, or reduces unnecessary testing in routine U.S. care.

Where current blood testing fits

The ADA continues to list fasting plasma glucose, the two-hour oral-glucose-tolerance result, and A1c as appropriate tests for screening and diagnosis. They measure different parts of glucose metabolism and do not always identify the same people.

An A1c test estimates average blood glucose over roughly two to three months. A fasting glucose test is a point-in-time measurement after fasting. Neither is interchangeable with the experimental 2-HB assay, and LTS does not offer 2-HB testing.

The useful conclusion is narrow. This study gives 2-HB enough signal to justify more validation as a possible rule-out tool for post-meal dysglycemia. Its low positive predictive value, threshold definitions, single-cohort design, and commercial development stage are reasons not to treat it as a new consumer screening test.

This article is editorial commentary and is not medical advice. It has not been reviewed by a physician and should not be used to make decisions about diabetes screening, diagnosis, testing frequency, medication, or treatment.