Diabetes and the HbA1c biomarker: from average glucose to glycemic control

Start here: what diabetes is

Glucose is a sugar carried in the blood and used as fuel. Insulin is a hormone that helps the body move and store glucose and restrains the liver from releasing too much. Diabetes is a group of diseases in which blood glucose stays too high because the body makes too little insulin, responds to it poorly, or both ^[1].

The two main forms reach that problem by different routes. In type 1 diabetes, an autoimmune process destroys the pancreatic beta cells that make insulin, so insulin replacement is essential. In type 2 diabetes, body tissues become less responsive to insulin and beta cells can no longer make enough insulin to compensate; the balance between those defects varies from person to person and changes over time ^[1].

Control matters because chronic high glucose is linked to eye, kidney, nerve, and cardiovascular complications, while treatment that pushes glucose too low can itself be dangerous. The aim is therefore not the lowest possible number, but a safe, individualized range that reduces long-term risk without creating an unacceptable burden or hypoglycemia risk ^[2].

This review follows that care loop through three pillars: measurements show what is happening, medicines change the biology, and progress asks what newer tools can do better. The simple A1c model between the first and second pillars connects day-to-day glucose to the slower laboratory number used for long-term control.

Pillar 1: measurements

HbA1c is glycated hemoglobin: glucose reacts nonenzymatically with hemoglobin inside circulating red cells, so the measured fraction integrates glucose exposure over the red-cell age distribution. The clinical shorthand is "about three months," because red-cell lifespan is commonly around 90-120 days, but the weighting is not a perfect flat window and changes when red-cell turnover changes. That is why anemia, hemoglobin variants, kidney disease, transfusion, erythropoietin, pregnancy, and other conditions can make HbA1c disagree with directly measured glucose ^{[3] [4] [5]}.

HbA1c became the control metric because it is standardized, convenient, prognostic, and stable enough to guide population-level targets. Current Standards of Care still use HbA1c for diagnosis and treatment goals, while emphasizing individualized targets and hypoglycemia avoidance; a common nonpregnant-adult target is below 7%, but goals are relaxed or tightened according to age, comorbidity, pregnancy, frailty, treatment risk, and patient preference ^{[1] [2]}.

Continuous glucose monitoring (CGM) measures interstitial glucose every few minutes, so it captures variability, overnight patterns, hypoglycemia, post-prandial excursions, and treatment responses that a 3-month average hides. The consensus CGM language is time in range, commonly the percent of readings in 70-180 mg/dL for many nonpregnant adults, plus time below range and time above range. CGM therefore complements HbA1c: HbA1c estimates chronic exposure; CGM shows the path by which that exposure happens ^{[6] [7]}.

Fasting plasma glucose (FPG) and the oral glucose tolerance test (OGTT) are diagnostic stress tests on shorter time-scales. FPG is easy and reproducible but emphasizes basal hepatic glucose output and fasting physiology. OGTT challenges the system with a glucose load, making it useful when post-load dysglycemia is the concern, though it is slower and more burdensome. Standards continue to define diabetes using HbA1c, FPG, 2-hour OGTT glucose, or unequivocal hyperglycemia with symptoms ^{[1] [8]}.

Fructosamine and glycated albumin are shorter-window glycation markers, roughly reflecting glycemia over the lifespan of serum proteins rather than red cells. They are useful when HbA1c is biologically misleading, but they have their own albumin/protein-turnover caveats and are less universally standardized than HbA1c ^{[9] [10]}.

Centerpiece: a simple A1c model

The ADAG study linked laboratory HbA1c to contemporaneous glucose profiles and gave the widely used linear conversion ^[11]:

eAG_mg_dL = 28.7 * A1c_percent - 46.7

The inverse form is:

A1c_percent = (eAG_mg_dL + 46.7) / 28.7

This regression is not a mechanistic law. It is a population calibration that lets clinicians translate HbA1c into estimated average glucose units. A simple simulatable kinetics model starts by treating HbA1c as a weighted average of glucose over the red-cell lifespan. Let G(t) be plasma or CGM glucose in mg/dL, E(t) be the eAG-equivalent hemoglobin-glycation state in mg/dL, and tau be a red-cell turnover time constant in days. A first-order integrator is:

dE/dt = (G(t) - E(t)) / tau

A1c(t) = (E(t) + 46.7) / 28.7

Parameters: tau = 90-120 days for a simple red-cell-lifespan scale; initial E(0) can be set to the starting mean glucose or inferred from a starting HbA1c by E(0) = 28.7 * A1c(0) - 46.7. Inputs: a daily or subdaily glucose profile G(t). Outputs: E(t) and A1c(t). Numerically, with time step dt in days:

E_next = E + dt * (G - E) / tau

Assumptions: glucose drives glycation approximately linearly over the clinical range; red-cell turnover can be represented by one time constant; the ADAG regression maps the simulated eAG-equivalent state to A1c. Limitations: real red-cell age distributions are not a single exponential, individual glycation gaps exist, and altered red-cell lifespan or hemoglobin biology can shift HbA1c independent of glucose ^{[4] [5] [12]}.

This model is useful because it separates three ideas that are often conflated: the instantaneous glucose trajectory G(t), the red-cell integrator E(t), and the ADAG reporting conversion into A1c. It predicts why a dramatic glucose improvement is visible immediately on CGM but only gradually in HbA1c, and why a single A1c can hide two very different glucose profiles.

Pillar 2: medicines

Insulin replaces or supplements insulin action and is essential in type 1 diabetes and often required later in type 2 diabetes. The field is moving from multiple daily injections toward lower-burden formulations and devices; once-weekly basal insulin icodec showed noninferior glycemic control versus daily basal insulin in type 2 diabetes trials, while safety and patient-selection questions remain central ^{[13] [14]}.

Metformin remains a common first-line type 2 diabetes drug because it lowers hepatic glucose output, improves insulin sensitivity, is inexpensive, and has long clinical experience. Sulfonylureas stimulate insulin secretion by closing beta-cell KATP channels, making them effective but prone to hypoglycemia and weight gain. DPP-4 inhibitors prolong endogenous incretin signaling, giving modest glucose lowering with generally low hypoglycemia risk. Current pharmacologic standards position these older classes inside an individualized framework that also weighs cardiorenal disease, weight, cost, and hypoglycemia risk ^{[15] [16]}.

SGLT2 inhibitors lower glucose by blocking renal glucose reabsorption, causing glycosuria. Their importance expanded because cardiovascular and kidney outcome trials showed benefits that go beyond HbA1c lowering, especially in heart failure and chronic kidney disease contexts ^{[17] [18]}.

GLP-1 receptor agonists increase glucose-dependent insulin secretion, suppress glucagon, slow gastric emptying, and reduce appetite. Semaglutide demonstrated cardiovascular benefit in type 2 diabetes and later made the weight-loss dimension of incretin therapy impossible to treat as secondary ^{[19] [20]}.

Dual GLP-1/GIP agonism pushed this trend further. Tirzepatide produced large HbA1c and weight reductions in type 2 diabetes trials, including head-to-head superiority over semaglutide 1 mg in SURPASS-2 and benefit when added to basal insulin in SURPASS-5 ^{[21] [22]}. The therapy frontier is therefore no longer just "lower glucose"; it is glycemia plus weight, cardiovascular risk, kidney risk, tolerability, access, and durability.

Pillar 3: progress

Automated insulin delivery closes the loop between CGM and insulin pumps. Algorithms adjust basal insulin and correction dosing from real-time glucose data, shifting care from manual reaction to continuous control. Randomized trials of closed-loop systems and bionic-pancreas approaches improved time in range and HbA1c-related outcomes, with usability and safety now as important as algorithmic performance ^{[23] [24] [7]}.

Incretin therapies are changing the remission conversation. Weight loss from GLP-1 receptor agonists and GLP-1/GIP dual agonists can reduce insulin resistance enough that some people with type 2 diabetes move toward medication de-escalation or remission-like glycemic states, though durability, muscle preservation, discontinuation effects, access, and long-term safety still matter ^{[25] [21] [26]}.

CGM is becoming standard infrastructure. It is already central in type 1 diabetes and increasingly used in insulin-treated type 2 diabetes, high-risk dysglycemia, and therapy adjustment. Its value is not only a better metric; it creates the data stream needed for automated delivery, behavioral feedback, and earlier recognition of treatment failure ^{[7] [6]}.

Beta-cell and islet replacement are moving from concept to clinical signal. Stem-cell-derived islet programs and encapsulated beta-cell products aim to replace the missing secretory tissue in type 1 diabetes. Recent reports show glucose-responsive insulin production and, in some settings, insulin independence or improved glucose control, but immune protection, durability, manufacturing, and safety remain unresolved ^{[27] [28] [29]}.