Fasting and Postprandial Glucose
Fasting glucose reflects overnight hepatic glucose output and basal insulin action, while postprandial glucose reflects meal handling, incretin response and gastric emptying.
Fasting and postprandial glucose are two different physiological questions sharing a unit. A fasting value, taken after at least eight hours without calories, is dominated by hepatic glucose production and the basal insulin needed to restrain it. A postprandial value, conventionally read two hours after a meal or after a 75 gram oral glucose load, is set by how quickly the stomach delivers nutrient, how brisk first-phase insulin is, how well glucagon is suppressed and how much muscle takes up.
Their relative contributions shift predictably. Monnier and colleagues showed postprandial excursions account for most of the excess exposure when HbA1c is near target, while fasting hyperglycaemia dominates once HbA1c climbs into the higher single digits. The thresholds are conventions: 5.6 mmol per litre fasting marks prediabetes and 7.0 diabetes, with the two-hour post-load equivalents at 7.8 and 11.1.
The split maps onto therapeutic choice. Agents that suppress hepatic glucose output or supply basal insulin move the fasting number; agents that slow gastric emptying or amplify the meal-time insulin response, such as short-acting GLP-1 agonists and amylin analogues, flatten the excursion. Knowing which component drives a patient's HbA1c makes the classes non-interchangeable, and why adding a prandial agent to a well-controlled fasting profile still lowers HbA1c.
The measurement is noisier than it looks. Within-person day-to-day variation in fasting glucose is large enough that a single reading near a threshold decides nothing, and glycolysis in an unseparated tube lowers the result. The newer error comes from continuous monitors in people without diabetes, where a post-meal rise to 8 mmol per litre is treated as a pathological spike when it is ordinary physiology, and foods are ranked on differences smaller than the sensor's own error.