Before the Sugar Spikes: Can Insulin Levels Predict Diabetes?

Biomarker testing has become increasingly popular. Companies like Function Health, Healthspan, Superpower and others offer a wide range of blood tests—many of which a typical primary care physician might not order. One such test is fasting insulin. The rationale is that it may better predict your risk of developing type 2 diabetes than standard tests like fasting glucose or hemoglobin A1c (HbA1c). This makes intuitive sense: type 2 diabetes stems from insulin resistance, a condition in which the body needs to produce more insulin to maintain normal blood glucose levels. Can an elevated fasting insulin level identify diabetes risk even before blood sugar levels rise?

The first challenge is measurement accuracy. There are no standardized assays for measuring insulin, so results can vary widely between laboratories. In fact, a group commissioned by the American Diabetes Association evaluated 12 different insulin assays and found substantial variability—not only between tests but also within the same test (reference here). Several issues contribute to this inconsistency. One is that insulin assays are sensitive to hemolysis (the breakdown of red blood cells), which can occur while the blood sample waits to be processed. Another is cross-reactivity with proinsulin—an insulin precursor—which can cause some tests to overestimate true insulin levels. Even if the measurement were reliable, there is no universally accepted cutoff for diagnosing insulin resistance. Some sources define it as fasting insulin >7.2 mU/L, others use >12.2 mU/L or even >18 mU/L (reference here).

Let’s assume we have a lab that can accurately measure insulin and we agree upon a cut-off value defining an elevated insulin level. Does it add predictive value for the development of type 2 diabetes in the future?

Clinical Studies

Ruijgrok et al conducted a prospective study of about 1350 participants without diabetes at baseline in the Netherlands. Baseline measurements included a fasting glucose, glucose level 2 hours after ingesting 75 grams of glucose (2hPG), HbA1c, fasting insulin, and HOMA-IR. After a mean follow-up of 6 years, about 150 participants developed diabetes. The strongest predictors for diabetes were, in order: fasting glucose, 2hPG, and HbA1c. In their multivariate analysis, fasting insulin was not associated with incidence of diabetes.

Wilson et al conducted an analysis of the Framingham Offspring study, which was a prospective study of about 3000 participants. They also looked at a variety of baseline factors to predict the development of diabetes over an interval of 7 years. Most of the risk could be predicted by family history of diabetes, obesity, fasting blood glucose, hypertension, low HDL, and elevated triglycerides. Adding fasting insulin values did not meaningfully improve risk prediction.

One test that may add some predictive value: glucose level after a glucose challenge. Respected clinician scientist and endocrinologist Ralph Defronzo et al followed 1600 subjects from the San Antonio Heart study without diabetes at baseline to see who developed diabetes. They conducted an oral glucose tolerance test (OGTT) in which participants were given 75 grams of glucose to drink and then had their blood glucose levels tested at various time points. They found that those with a glucose of >155mg/dl one hour later had a 3-fold increased risk for developing diabetes over the next 8 years.

Guidelines

In 2017, Styne et al published the Pediatric Obesity—Assessment, Treatment, and Prevention: An Endocrine Society Clinical Practice Guideline. They wrote:

Many clinicians measure insulin values thinking it adds to the diagnosis of comorbidities. In fact it does not, and such measurements are not recommended. Although obesity is associated with insulin resistance/hyperinsulinemia, attempts to diagnose insulin resistance by measuring plasma insulin concentration or any other surrogate in the clinical setting has no merit because it has no diagnostic value. Fasting insulin concentrations show considerable overlap between insulin-resistant and insulin-sensitive youths…Because of these limitations, measuring plasma insulin concentrations remains a research tool with no clinical value for evaluation of obesity. Measuring fasting insulin concentrations to try to diagnose insulin resistance within general practice should be abandoned.

In 2023, the American Diabetes Association published Guidelines and Recommendations for Laboratory Analysis in the Diagnosis and Management of Diabetes Mellitus. Their recommendation:

“In most people with diabetes or risk for diabetes or cardiovascular disease, routine testing for insulin or proinsulin is not recommended.”

This does not mean insulin resistance is unimportant. On the contrary, it may be one of the most critical drivers of reduced healthspan and lifespan. However, it’s often more valuable to assess the consequences of insulin resistance—such as elevated blood pressure, increased waist circumference, BMI, impaired glucose tolerance, high triglycerides, and low HDL—rather than attempting to quantify insulin levels directly.

While fasting insulin measurement may be useful in select cases, it adds limited value for predicting diabetes in most people. For companies aiming to deliver meaningful insights through expanded biomarker panels, it may be more impactful to focus on alternative markers that are better validated, more consistent, and clinically actionable.