Insulin is the principal anabolic hormone of the body, produced exclusively by the beta cells of the pancreatic islets of Langerhans. Its physiological roles extend well beyond glucose lowering: insulin suppresses hepatic glucose output, promotes glycogen and lipid synthesis, inhibits lipolysis and proteolysis, and drives cellular uptake of glucose, amino acids, and potassium. An understanding of endogenous insulin biology is the essential foundation for rational therapeutic use of exogenous insulin preparations, because the pharmacological goal of insulin therapy is to replicate the time course and magnitude of normal insulin secretion as closely as the available preparations and delivery systems permit.
Insulin biosynthesis begins in the rough endoplasmic reticulum with the translation of preproinsulin from the INS (insulin gene) located on chromosome 11p15.5. A 24-amino-acid signal peptide is cleaved co-translationally as the nascent chain enters the ER (endoplasmic reticulum) lumen, generating proinsulin, a single-chain 86-amino-acid precursor in which the B chain (30 amino acids) and A chain (21 amino acids) of mature insulin are connected by the C-peptide (connecting peptide, 31 amino acids). Within the ER, proinsulin folds and forms three disulfide bonds (two inter-chain bonds between A and B chains, one intra-chain A chain bond) that are essential for biological activity. Proinsulin is then packaged into clathrin-coated vesicles at the trans-Golgi network and delivered to secretory granules, where the conversion enzyme PC1/3 (prohormone convertase 1/3) cleaves at the B-C junction and PC2 (prohormone convertase 2) cleaves at the C-A junction, liberating mature insulin and the equimolar byproduct C-peptide. Both insulin and C-peptide are co-secreted in equimolar amounts, a fact exploited clinically: C-peptide measurement distinguishes endogenous insulin secretion from exogenous insulin administration, since commercial insulin preparations do not contain C-peptide.2
Glucose-stimulated insulin secretion (GSIS) from beta cells is mediated through a tightly regulated membrane depolarization mechanism. Glucose enters the beta cell via GLUT2 (glucose transporter 2), the low-affinity, high-capacity transporter whose transport rate is proportional to extracellular glucose concentration across the physiological range. Intracellular glucose metabolism by glucokinase (hexokinase IV), the rate-limiting step acting as the beta cell glucose sensor, generates glucose-6-phosphate that enters glycolysis and ultimately increases the cytosolic ATP (adenosine triphosphate) to ADP (adenosine diphosphate) ratio. The resulting rise in the ATP-to-ADP ratio closes the ATP-sensitive potassium channel (KATP channel), a hetero-octameric complex of four Kir6.2 pore-forming subunits and four SUR1 (sulfonylurea receptor 1) regulatory subunits. KATP channel closure prevents outward potassium current, depolarizing the beta cell membrane from approximately -70 mV toward the threshold for voltage-gated calcium channel activation. L-type and P/Q-type voltage-gated calcium channels open, calcium influx triggers exocytosis of insulin-containing secretory granules, and insulin is released into the portal circulation.2 This KATP channel-mediated pathway is the primary target of the sulfonylurea class of oral hypoglycemics, which close KATP channels by binding SUR1 independently of glucose metabolism, explaining both their glucose-lowering efficacy and their hypoglycemia risk.
Physiological insulin secretion exhibits a characteristic biphasic pattern in response to a glucose challenge. The first phase (first-phase insulin response, FPIR) occurs within 1 to 3 minutes of glucose exposure and represents exocytosis of a readily releasable pool of pre-docked secretory granules positioned adjacent to the plasma membrane; this phase lasts approximately 10 minutes and produces a sharp insulin spike. The second phase begins 10 to 20 minutes after glucose stimulation, is more sustained, and reflects recruitment of reserve granules from deeper within the cell, new granule biosynthesis stimulated by glucose, and amplifying signals from incretin hormones including GLP-1 (glucagon-like peptide-1) and GIP (glucose-dependent insulinotropic polypeptide) released from intestinal L and K cells respectively in response to nutrient absorption. The incretin effect accounts for approximately 50 to 70 percent of postprandial insulin secretion in healthy individuals, explaining why oral glucose provokes substantially more insulin release than an equivalent intravenous glucose load (the incretin effect). Loss of first-phase insulin secretion is among the earliest detectable defects in the progression from normal glucose tolerance to type 2 diabetes mellitus (T2DM), predating the onset of overt hyperglycemia by years.3
Insulin exerts its biological effects by binding to the IR (insulin receptor), a heterotetrameric transmembrane glycoprotein of the receptor tyrosine kinase (RTK) family consisting of two extracellular alpha subunits and two transmembrane beta subunits linked by disulfide bonds. Insulin binding to the alpha subunits induces conformational change that activates the intrinsic tyrosine kinase activity of the beta subunit intracellular domains, leading to autophosphorylation of multiple tyrosine residues within the beta subunit activation loop. This activates the receptor and enables it to phosphorylate cytoplasmic docking proteins, chief among them IRS-1 (insulin receptor substrate-1) and IRS-2 (insulin receptor substrate-2), on tyrosine residues. Phosphorylated IRS-1 and IRS-2 recruit and activate PI3K (phosphatidylinositol 3-kinase), which generates PIP3 (phosphatidylinositol 3,4,5-trisphosphate) at the inner leaflet of the plasma membrane. PIP3 recruits PDK1 (phosphoinositide-dependent kinase-1) and its substrate Akt (protein kinase B), and PDK1-mediated phosphorylation of Akt activates this serine-threonine kinase, which is the central node of downstream insulin signaling.1
Activated Akt drives the principal metabolic effects of insulin through multiple downstream targets. In skeletal muscle and adipose tissue, Akt phosphorylates AS160 (Akt substrate of 160 kDa), a GAP (GTPase-activating protein) that normally maintains the small GTPase Rab in its inactive GDP (guanosine diphosphate)-bound state. AS160 phosphorylation inactivates the GAP, allowing Rab to adopt its active GTP (guanosine triphosphate)-bound state, which triggers translocation of intracellular GLUT4 (glucose transporter type 4)-containing storage vesicles to the plasma membrane. GLUT4 insertion into the membrane dramatically increases glucose uptake capacity; skeletal muscle alone accounts for approximately 75 to 80 percent of postprandial glucose disposal in the insulin-stimulated state. In the liver, Akt activates glycogen synthase kinase 3 (GSK-3) phosphorylation and inactivation, promoting glycogen synthase activity and hepatic glycogen synthesis, while simultaneously suppressing expression of gluconeogenic enzymes through Foxo1 (forkhead box protein O1) phosphorylation and nuclear exclusion. In adipose tissue, Akt-mediated phosphorylation and inactivation of PDE3B (phosphodiesterase 3B) raises intracellular cAMP (cyclic adenosine monophosphate), and Akt also directly phosphorylates and inhibits HSL (hormone-sensitive lipase), suppressing lipolysis and free fatty acid release.1 The parallel MAPK (mitogen-activated protein kinase) arm of insulin signaling, activated downstream of IRS-1 through the Grb2 (growth factor receptor-bound protein 2)/SOS (son of sevenless, a guanine nucleotide exchange factor)/Ras pathway, mediates insulin mitogenic and growth effects but does not contribute to glucose lowering.
C-peptide is secreted equimolarly with insulin and is absent from all commercially available insulin preparations. Measurement of fasting C-peptide (normal 0.5 to 2.0 ng/mL) distinguishes residual endogenous insulin secretion from exogenous insulin in a treated patient, guides the distinction between type 1 and type 2 diabetes when phenotype is ambiguous (e.g., latent autoimmune diabetes in adults, LADA), and helps identify factitious hypoglycemia from surreptitious insulin injection (high insulin, suppressed C-peptide). C-peptide also has modest biological activity: it appears to promote sodium-potassium ATPase activity in the kidney and nerve, and low C-peptide in type 1 diabetes may contribute to microvascular complications. These activities remain an active research area and do not yet affect clinical management.
Hepatic insulin extraction is a quantitatively important aspect of insulin pharmacokinetics that distinguishes the physiological route of delivery from subcutaneous insulin administration. Insulin secreted from pancreatic beta cells enters the portal circulation and is delivered to the liver, where approximately 50 to 60 percent of secreted insulin is extracted during first pass through the hepatic sinusoids. This portal delivery creates a hepatic-to-peripheral insulin concentration gradient in which hepatic insulin concentrations are two- to fourfold higher than peripheral concentrations during the postprandial state. This gradient is physiologically critical: it ensures that hepatic glucose production suppression (a highly insulin-sensitive response) occurs at insulin concentrations that do not produce peripheral hypoglycemia. Subcutaneous insulin administration bypasses portal delivery entirely, presenting insulin to peripheral tissues (muscle, fat) before the liver. To achieve adequate hepatic effects (glycogen synthesis, suppression of gluconeogenesis), peripheral insulin concentrations must be higher than those produced by equivalent portal delivery, predisposing to peripheral hypoglycemia at doses required for hepatic glucose output suppression. This fundamental pharmacokinetic difference from physiological insulin delivery is an inherent limitation of all current subcutaneous insulin regimens.4
Commercially available insulin preparations span a spectrum from rapid-acting analogs designed to mimic the sharp first-phase insulin spike of physiological meal-time secretion to ultra-long-acting analogs engineered to provide a near-peakless basal insulin background over 24 to 42 hours. Each preparation reflects specific structural modifications to the human insulin molecule that alter its self-association behavior, receptor binding kinetics, or subcutaneous depot pharmacology to achieve the desired time-action profile.
Human insulin (regular insulin, crystalline zinc insulin) consists of the native 51-amino-acid sequence and at pharmaceutical concentrations (U-100, 100 units/mL) exists predominantly as zinc-stabilized hexamers in solution. Following subcutaneous injection, hexamers must dissociate through dimers to biologically active monomers before absorption across the capillary endothelium into the circulation, a process that takes 30 to 60 minutes and produces a peak effect at 2 to 3 hours with duration of 5 to 8 hours. This delayed onset necessitates injection 30 minutes before meals to match peak insulin effect with postprandial glucose excursion, a requirement that is practically challenging and contributes to both pre-meal hypoglycemia and post-meal hyperglycemia when timing is imperfect. Regular insulin retains clinical relevance for continuous intravenous infusion (only insulin formulation suitable for IV use given the risk of particulate formation with other preparations), for mixed insulin regimens, and in the U-500 concentrated form for highly insulin-resistant patients requiring very large doses.5
Rapid-acting insulin analogs were developed by introducing amino acid substitutions that destabilize hexamer formation and promote the monomer-dominant state at pharmaceutical concentrations, allowing near-immediate absorption after subcutaneous injection. Insulin lispro (Humalog) reverses the ProB28-LysB29 sequence to LysB28-ProB29, disrupting the beta-sheet contacts between B chain C-termini that stabilize dimer formation without altering the receptor-binding surface. Insulin aspart (NovoLog/NovoRapid) substitutes ProB28 with aspartic acid, introducing charge repulsion at the dimer interface. Insulin glulisine (Apidra) substitutes AsnB3 with lysine and LysB29 with glutamic acid, also destabilizing self-association. All three analogs have onset of action within 10 to 15 minutes of subcutaneous injection, reach peak effect at 1 to 2 hours, and have duration of 3 to 5 hours. This pharmacokinetic profile allows injection immediately before or even briefly after meal initiation (useful in patients with unpredictable appetite or gastroparesis), more closely replicating physiological first-phase secretion and producing superior postprandial glucose control compared with regular insulin in most clinical studies.5 Ultra-rapid formulations of insulin aspart (Faster Aspart, Fiasp) and insulin lispro (URLi, Lyumjev) incorporate additional excipients (niacinamide with faster aspart; citrate and treprostinil with URLi) that further accelerate subcutaneous absorption, with onset within 2 to 5 minutes and superiority over standard rapid-acting analogs in limiting postprandial glucose excursions.
NPH (neutral protamine Hagedorn) insulin is an intermediate-acting preparation produced by complexing human insulin with protamine and zinc at neutral pH, forming a suspension that dissolves slowly after subcutaneous injection. Its onset of 1 to 2 hours, peak at 4 to 8 hours, and duration of 12 to 18 hours make it suitable for twice-daily basal dosing but produce clinically significant nocturnal hypoglycemia (peak effect between 2 and 4 in the morning when injected at bedtime) and variable day-to-day glycemic control due to inconsistent suspension resuspension before injection (a recognized source of dosing error). NPH remains in widespread use in lower-resource settings due to its low cost and compatibility with regular insulin for premixed formulations, but it has been largely supplanted in higher-resource settings by basal analogs with superior pharmacokinetic profiles.11
Long-acting basal insulin analogs were designed to eliminate the pronounced peak of NPH (neutral protamine Hagedorn) insulin and provide a sustained, near-flat insulin background that more closely approximates physiological basal insulin secretion. Insulin glargine (Lantus, Basaglar, Toujeo) is produced by substituting AsnA21 with glycine (stabilizing the molecule at neutral pH) and adding two arginine residues to the B chain C-terminus (shifting the isoelectric point from pH 5.4 to 6.7). The formulation is acidic (pH 4), fully soluble in the vial, but precipitates upon injection into the neutral subcutaneous environment, forming a depot of microprecipitates that dissolve slowly. The standard U-100 formulation (Lantus, Basaglar) provides 20 to 24 hours of duration; the U-300 concentrated formulation (Toujeo) creates a smaller, more compact subcutaneous depot with slower dissolution, extending duration to approximately 36 hours and reducing intra-patient variability.7
Insulin detemir (Levemir) achieves prolonged action through a distinct mechanism: removal of ThrB30 and fatty acid acylation of LysB29 with a C14 (fourteen-carbon) myristic acid chain allows detemir to self-associate and reversibly bind albumin in the subcutaneous interstitium and bloodstream, creating a prolonged-release depot with duration of 18 to 22 hours at typical doses, often requiring twice-daily dosing.6 Insulin degludec (Tresiba) extends this principle further: a C18 (eighteen-carbon) fatty acid attached to LysB29 allows degludec to form large multi-hexamer complexes in the subcutaneous depot that slowly dissociate, producing duration exceeding 42 hours, a near-peakless pharmacodynamic profile with substantially less intra-patient variability than glargine U-100, and a half-life of approximately 25 hours that permits flexible once-daily dosing within an 8-hour window without loss of glycemic control.14
Premixed insulins combine a rapid- or short-acting component with an intermediate-acting protamine-complexed component in fixed ratios (e.g., 70/30 = 70% NPH/30% regular; 75/25 = 75% protamine lispro/25% lispro; 50/50 formulations). They offer injection convenience (two injections daily) but sacrifice the ability to independently titrate basal and bolus components, making them poorly suited for patients with variable meal schedules, active titration, or hypoglycemia risk. They are most appropriate for patients with stable meal patterns, limited dexterity, or cost constraints. Premixed analogs (BIAsp 70/30, Humalog Mix 75/25) provide faster onset than regular insulin-containing premixes, better matching postprandial glucose excursions.
Concentrated insulin formulations address the large-volume injection problem in highly insulin-resistant patients. U-500 regular insulin (Humulin R U-500) contains 500 units/mL (five-fold concentration), reducing injection volume by 80 percent for patients requiring more than 200 units per day, a clinically significant advantage given subcutaneous absorption limitations (absorption becomes erratic when volumes exceed approximately 50 units per injection site). U-500 regular also has a prolonged duration of action (up to 24 hours at high doses) due to depot pharmacokinetics at high local concentrations, allowing twice-daily or three-times-daily dosing as a combined basal-bolus regimen in some patients. U-200 insulin degludec (Tresiba U-200) and U-300 insulin glargine (Toujeo) provide concentrated basal analogs with reduced injection volumes and, for U-300 glargine, extended duration and reduced variability compared with U-100 glargine, without pharmacodynamic differences significant enough to require recalibration of doses beyond accounting for the concentration difference.5
The ADME (absorption, distribution, metabolism, and excretion) profile of subcutaneously administered insulin is profoundly different from that of small-molecule drugs, reflecting the physicochemical complexity of the protein molecule, the biology of the subcutaneous depot, and the dual-organ (hepatic and renal) degradation pathway. Variability in subcutaneous absorption is the single largest source of day-to-day glycemic variability in insulin-treated patients and the primary practical limitation of current insulin pharmacotherapy.
Absorption of subcutaneously injected insulin proceeds through a sequence of physicochemical and physiological steps. Insulin is injected into the subcutaneous adipose tissue, where it initially exists as a depot of hexamers (for regular insulin and the basal analogs) or primarily monomers/dimers (for rapid-acting analogs and ultra-rapid formulations with added excipients). Hexamers and dimers cannot cross the capillary endothelium; absorption requires dissociation to monomers at the injection site. For regular insulin, this dissociation is rate-limited by zinc release and dilution by interstitial fluid, taking 30 to 60 minutes; for rapid-acting analogs engineered to disfavor self-association, dissociation is nearly immediate. Monomers then traverse the subendothelial basement membrane and enter the capillary circulation, primarily by transcapillary diffusion rather than lymphatic uptake (lymphatic absorption of insulin is quantitatively minor, less than 20 percent). Local blood flow at the injection site is a major determinant of absorption rate: increased blood flow (heat application, exercise of the injected limb, massage) accelerates absorption and can precipitate hypoglycemia; reduced blood flow (cold, lipohypertrophy, peripheral vascular disease) delays absorption and blunts the insulin effect.8
Injection site location systematically affects insulin absorption rate. Absorption is fastest from the abdomen (due to greater subcutaneous blood flow), intermediate from the arm, and slowest from the thigh and buttock. This translates into clinically meaningful differences: abdominal injection of rapid-acting insulin produces a sharper, earlier peak better suited to bolus dosing; thigh or buttock injection of basal insulin may prolong depot residence and reduce peak concentration. Standard clinical guidance recommends using a consistent anatomic region for each injection type (abdomen for bolus, thigh or buttock for basal) and rotating within that region to prevent lipohypertrophy. Injection depth also matters: intramuscular injection accelerates absorption (faster onset, higher peak, shorter duration) and is a recognized cause of unpredictable hypoglycemia, particularly in lean patients using longer needles. Needle length recommendations (4 to 6 mm for most adults) are designed to reliably achieve subcutaneous placement across a range of body habitus while minimizing intramuscular injection risk.8
Distribution of absorbed insulin is effectively limited to the vascular compartment and insulin-responsive tissues; the volume of distribution is approximately 0.1 to 0.2 L/kg, consistent with distribution in the extracellular fluid. Insulin does not cross the blood-brain barrier under normal conditions. Approximately 5 percent of circulating insulin is loosely bound to plasma proteins; the remainder circulates free. Insulin detemir is the exception: its C14 (fourteen-carbon) fatty acid chain produces approximately 98 percent albumin binding in plasma, substantially reducing the free fraction and contributing to its pharmacokinetic buffering effect (the reservoir of albumin-bound detemir smooths absorption fluctuations). The volume of distribution for detemir is lower than for other insulins at approximately 0.1 L/kg, consistent with its predominantly vascular and extravascular distribution limited by high albumin binding.5
Insulin metabolism occurs primarily in the liver and kidney via insulin-degrading enzyme (IDE), a zinc metalloprotease that cleaves insulin after receptor-mediated internalization and within endosomes. The liver clears approximately 50 percent of portal insulin during first pass, as described above. In the peripheral circulation, the kidney is the dominant site of insulin clearance, accounting for approximately 30 to 40 percent of peripheral insulin degradation. Renal insulin clearance occurs through two mechanisms: glomerular filtration of free insulin (insulin is below the molecular weight threshold for filtration at approximately 6,000 Da for monomers) followed by tubular reabsorption and degradation, and peritubular uptake from the post-glomerular capillaries. The clinical consequence of renal impairment is reduced insulin clearance, prolonging insulin action and predisposing to hypoglycemia. This effect is seen across the entire insulin class but is most clinically relevant for regular insulin and NPH (neutral protamine Hagedorn) insulin; the long-acting analogs have relatively flat pharmacodynamic profiles that blunt hypoglycemia risk from prolonged action somewhat, though dose reduction is still required as eGFR (estimated glomerular filtration rate) falls below 30 mL/min/1.73m2. The half-life of circulating insulin monomers in the peripheral circulation is approximately 4 to 6 minutes; apparent pharmacodynamic half-lives are much longer due to ongoing absorption from the subcutaneous depot.9
Lipohypertrophy (subcutaneous fatty tissue enlargement at injection sites) affects an estimated 30 to 50 percent of insulin-injecting patients and is the most common cause of unexplained glycemic variability in clinical practice. Repeated injection at the same site causes local mitogenic effects of insulin on adipocytes and a reduced local inflammatory response, creating a fibrofatty tissue with markedly reduced vascularization and delayed, erratic insulin absorption. Insulin injected into lipohypertrophic tissue may have an onset delayed by 60 to 120 minutes compared with healthy tissue. Diagnosis requires systematic palpation of all injection sites; patients often prefer hypertrophic sites because injections are less painful there. Management is rotation to healthy tissue with temporary dose reduction, since absorption will dramatically increase on switching sites. Prevention requires structured rotation protocols.
Hepatic insulin clearance is altered in states of hepatic dysfunction. Cirrhosis reduces first-pass hepatic insulin extraction, increasing peripheral insulin exposure and predisposing to hypoglycemia; cirrhotic patients often require substantially reduced insulin doses despite peripheral insulin resistance driven by elevated glucagon and counter-regulatory hormone excess. NAFLD (non-alcoholic fatty liver disease) and its progressive form NASH (non-alcoholic steatohepatitis) is associated with hepatic insulin resistance (reduced hepatic insulin signaling and impaired glycogen synthesis) but preserved or enhanced hepatic insulin extraction, producing the common clinical scenario of hyperinsulinemia with hyperglycemia. Understanding which component of hepatic insulin physiology is predominantly impaired helps explain the paradox of elevated fasting insulin levels coexisting with fasting hyperglycemia in T2DM (type 2 diabetes mellitus) patients with significant hepatic steatosis.4
Effective insulin therapy requires matching insulin pharmacokinetics to the physiology it is designed to replace. In type 1 diabetes mellitus (T1DM), where endogenous insulin secretion is absent, the goal is complete physiological replacement: a basal insulin providing continuous suppression of hepatic glucose output between meals and overnight, combined with bolus insulin at each meal to cover postprandial glucose excursions. In T2DM (type 2 diabetes mellitus), insulin is typically added to failing oral therapy, often beginning with basal-only regimens before advancing to basal-plus or full basal-bolus as beta cell function further declines.
The starting total daily dose (TDD) for insulin-naive patients is typically estimated at 0.4 to 0.5 units per kilogram of body weight per day in T1DM, or 0.1 to 0.2 units per kilogram for basal-only initiation in T2DM patients beginning insulin therapy. In T1DM basal-bolus regimens, TDD is divided approximately 50 percent basal and 50 percent bolus (distributed across meals). The initial split can be adjusted based on fasting versus postprandial glucose patterns: persistently elevated fasting glucose suggests insufficient basal insulin; elevated postprandial glucose with satisfactory fasting levels suggests insufficient bolus coverage. Patients with significant insulin resistance (T2DM with obesity, concurrent glucocorticoid therapy) often require substantially higher doses, sometimes exceeding 1 to 2 units per kilogram per day, before transitioning to concentrated formulations.10
The basal insulin titration algorithm most supported by clinical evidence uses structured self-monitored fasting glucose to drive stepwise dose adjustment. The commonly used treat-to-target approach titrates the basal insulin dose by 2 units every 3 days when the mean fasting glucose of the preceding 3 days exceeds the target (typically 80 to 130 mg/dL per ADA guidelines), holding or reducing the dose if any fasting glucose is below 80 mg/dL. More aggressive titration algorithms (e.g., the Insight titration algorithm: increase by 4 units if mean fasting glucose exceeds 140 mg/dL, by 2 units if 110 to 139 mg/dL) reach target faster at the cost of slightly higher hypoglycemia rates. The evidence base for aggressive titration in T2DM is strong: the landmark treat-to-target trials (TTT) demonstrated that most T2DM patients could achieve HbA1c (glycated hemoglobin) below 7 percent with basal insulin alone using algorithmic titration, with hypoglycemia rates lower than historically observed with less systematic approaches.10
Bolus insulin dosing in basal-bolus regimens is guided by three parameters: the CIR (carbohydrate-to-insulin ratio), the CF (correction factor, also called insulin sensitivity factor), and the pre-meal glucose target. The CIR defines how many grams of carbohydrate one unit of rapid-acting insulin covers; a typical starting CIR in T1DM is 1 unit per 15 grams of carbohydrate (the 500 Rule: 500 divided by TDD approximates the CIR in many patients). The CF defines how much one unit of rapid-acting insulin is expected to lower blood glucose (in mg/dL); a common starting estimate is the 1800 Rule: 1800 divided by TDD approximates the CF in mg/dL per unit. A typical mealtime bolus dose is calculated as: (grams of carbohydrate / CIR) + (current glucose minus target glucose / CF). These formulas are starting estimates requiring individualized titration; patients and providers should refine them based on systematic post-meal glucose checking (2-hour post-meal target 140 to 180 mg/dL). The advent of CGM (continuous glucose monitoring) has substantially improved bolus titration by providing real-time post-meal glucose curves rather than isolated fingerstick values.11
Continuous subcutaneous insulin infusion (CSII), commonly called insulin pump therapy, uses a programmable pump to deliver rapid-acting insulin (typically aspart or lispro) continuously as a basal infusion rate and as user-triggered bolus doses at meals. CSII replaces the need for separate basal insulin injections and allows programming of multiple basal rates across the 24-hour cycle (e.g., higher rates in the early morning to counter the dawn phenomenon, lower rates during exercise). The dawn phenomenon reflects the early morning rise in cortisol and growth hormone that increases hepatic glucose production and insulin resistance between approximately 3 and 8 in the morning, producing pre-breakfast hyperglycemia; this is best managed with CSII by programming an increased basal rate during this interval, which is not achievable with once-daily basal analog injection. Advanced HCL (hybrid closed-loop) systems integrate CSII with CGM (continuous glucose monitoring) and a control algorithm to automatically adjust basal delivery in response to real-time glucose trends, approximating an artificial pancreas. Multiple randomized trials have demonstrated HCL superiority over standard CSII in TIR (time-in-range, percentage of time with glucose 70 to 180 mg/dL) and hypoglycemia reduction in T1DM (type 1 diabetes mellitus).11
The target blood glucose during surgery and in the postoperative period is 140 to 180 mg/dL per major society guidelines, balancing the risks of hyperglycemia (impaired wound healing, infection, osmotic complications) against hypoglycemia (which is more dangerous in the anesthetized patient who cannot report or respond to symptoms). Key principles: (1) Long-acting basal insulin is continued at 75 to 80 percent of the usual dose the night before or morning of surgery; (2) All rapid-acting bolus insulin is held while the patient is NPO (nil per os); (3) Correction scale insulin (sliding scale) is appropriate as a supplement to basal insulin but not as the sole regimen; (4) Patients on insulin pumps can often continue pump therapy intraoperatively with anesthesiologist oversight; (5) For major surgery or critical illness with anticipated prolonged NPO status, intravenous regular insulin infusion with glucose monitoring every 1 to 2 hours provides the most precise glycemic control. Glucocorticoid-treated patients require anticipatory dose increases of 20 to 50 percent or more, with particular attention to afternoon and evening glucose elevations that mirror the pharmacodynamic peak of the steroid.
Intravenous insulin infusion using regular insulin is the standard approach to glycemic management in the intensive care unit (ICU) and for acute hyperglycemic emergencies. Regular insulin is the only preparation suitable for intravenous use; other formulations (analogs, NPH, glargine) must not be administered intravenously due to concerns about altered pharmacokinetics and, for NPH and premixed formulations, particulate matter. Insulin stability in IV solutions is generally adequate for clinical use: regular insulin adsorbs to PVC (polyvinyl chloride) tubing but this effect is clinically managed by flushing the tubing with insulin-containing solution before beginning the infusion. Standard ICU protocols use structured algorithms that adjust infusion rate based on current glucose and the rate of change (delta glucose), targeting glucose 140 to 180 mg/dL as recommended by the NICE-SUGAR (Normoglycemia in Intensive Care Evaluation and Survival Using Glucose Algorithm Regulation) trial, which demonstrated that intensive glucose control targeting 81 to 108 mg/dL increased 90-day mortality compared with conventional control in severely ill patients, largely attributable to severe hypoglycemia events.12
Hypoglycemia is the most clinically consequential adverse effect of insulin therapy and the primary barrier to achieving optimal glycemic control. Beyond hypoglycemia, the adverse effect profile of insulin includes weight gain, injection site complications, and immunogenicity, each of which affects adherence and long-term outcomes. Special populations including pregnant patients and those with advanced CKD (chronic kidney disease) require specific approaches reflecting altered insulin physiology and pharmacokinetics in these states.
Hypoglycemia is defined by the ADA (American Diabetes Association) Whipple triad: symptoms consistent with hypoglycemia, a measured low plasma glucose (below 70 mg/dL for level 1, below 54 mg/dL for clinically significant level 2, any glucose with severe cognitive impairment requiring external assistance for level 3), and resolution of symptoms with glucose administration. Neurogenic (autonomic) symptoms (sweating, tremor, palpitations, hunger) are mediated by the sympathoadrenal response triggered at a glucose threshold of approximately 65 to 70 mg/dL. Neuroglycopenic symptoms (confusion, difficulty concentrating, behavioral change, visual disturbance, seizure) appear at approximately 55 to 60 mg/dL and reflect direct neuronal glucose deprivation. The brain has essentially no glycogen reserve and depends on continuous glucose delivery; sustained neuroglycopenia below 50 mg/dL produces irreversible neuronal injury and can be fatal without prompt treatment. Common precipitants of insulin-induced hypoglycemia include excess insulin dose, delayed or missed meals, increased physical activity (exercise increases GLUT4 (glucose transporter type 4) translocation by an insulin-independent, AMPK (AMP (adenosine monophosphate)-activated protein kinase)-mediated mechanism, lowering glucose requirement for physical work), alcohol ingestion (which suppresses hepatic gluconeogenesis by competing for NAD (nicotinamide adenine dinucleotide, the oxidized form NAD+) in the conversion of ethanol to acetaldehyde), and inadvertent intramuscular injection accelerating absorption.13
Hypoglycemia unawareness (impaired awareness of hypoglycemia, IAH) develops in approximately 25 percent of T1DM (type 1 diabetes mellitus) patients and represents a dangerous complication of recurrent hypoglycemia. Repeated hypoglycemic episodes reduce the glycemic threshold at which counter-regulatory responses (epinephrine release, glucagon secretion) and neurogenic symptoms are triggered, a phenomenon termed hypoglycemia-associated autonomic failure (HAAF). In patients with IAH, the first indication of hypoglycemia may be neuroglycopenic impairment (confusion or loss of consciousness) without the preceding adrenergic warning symptoms. The counter-regulatory response is also impaired in long-standing T1DM due to progressive loss of the glucagon secretory response from alpha cells, making epinephrine the primary (and then also impaired) counter-regulatory defense. Structured patient education programs emphasizing blood glucose awareness training (BGAT) and temporary upward revision of glucose targets can partially restore hypoglycemia awareness by avoiding recurrent hypoglycemia for several weeks; real-time CGM (continuous glucose monitoring) with threshold alerts is recommended for all patients with IAH as it provides an alarm system independent of symptomatic awareness.13
The treatment of hypoglycemia follows the rule of 15: 15 grams of fast-acting glucose (glucose tablets, 4 oz juice or regular soda, glucose gel) orally, recheck glucose in 15 minutes, repeat if still below 70 mg/dL. Pure glucose is preferred over sucrose-containing foods because sucrose requires intestinal hydrolysis before absorption. For severe hypoglycemia with inability to take oral glucose, glucagon (1 mg IM [intramuscular] or SC [subcutaneous] for adults) is the standard out-of-hospital treatment; intranasal glucagon powder (Baqsimi, 3 mg) has been approved as an equivalent alternative that does not require reconstitution, substantially improving usability by caregivers. Intravenous dextrose (25 to 50 mL of 50% dextrose, D50W) is the definitive in-hospital treatment. Glucagon works by activating hepatic glucagon receptors (Gs-coupled, cAMP-mediated), activating glycogen phosphorylase, and mobilizing hepatic glycogen stores; it is therefore ineffective in patients with depleted glycogen (prolonged fasting, alcohol intoxication, hepatic failure).13
Multiple drug classes predictably alter insulin requirements and should prompt anticipatory dose adjustment and glucose monitoring. Agents that increase insulin requirements: glucocorticoids (dose- and timing-dependent; afternoon/evening glucose most affected), atypical antipsychotics (olanzapine, clozapine; promote insulin resistance and weight gain), thiazide diuretics at high doses (reduce insulin secretion and increase insulin resistance), protease inhibitors (HIV therapy; direct beta cell toxicity and insulin resistance), calcineurin inhibitors (tacrolimus more than cyclosporine; directly toxic to beta cells). Agents that decrease insulin requirements or mask hypoglycemia symptoms: non-selective beta-blockers (propranolol, nadolol) blunt tachycardia and tremor but not sweating as a hypoglycemia warning, and may prolong hypoglycemia by blocking hepatic glycogenolysis. ACE inhibitors and ARBs may modestly improve insulin sensitivity. Alcohol: inhibits gluconeogenesis and glucagon release, profoundly potentiating insulin-induced hypoglycemia, particularly with delayed onset 4 to 8 hours after ingestion.
Weight gain is a predictable and clinically significant adverse effect of insulin intensification, averaging 2 to 4 kg in T1DM initiation trials and 3 to 6 kg over 6 to 12 months in T2DM (type 2 diabetes mellitus) intensification studies. The mechanisms are multiple: reversal of glycosuria (caloric loss via urine is eliminated when hyperglycemia is corrected), enhanced lipogenesis and suppressed lipolysis from insulin anabolic effects, increased appetite driven by hypoglycemia episodes and defensive carbohydrate consumption, and reduction in resting energy expenditure as glucose metabolism is normalized. Weight gain worsens insulin resistance, driving a cycle of escalating insulin requirements in T2DM. Adding a GLP-1 (glucagon-like peptide-1) receptor agonist or SGLT-2 (sodium-glucose cotransporter-2) inhibitor to insulin therapy are evidence-based strategies for attenuating insulin-associated weight gain while maintaining glycemic control, and both combinations have cardiovascular outcome trial support in T2DM.11
Insulin immunogenicity, while substantially reduced with modern human and analog insulins compared with older animal-derived preparations, remains clinically relevant. Anti-insulin antibodies (AIAs) develop in virtually all insulin-treated patients, though their clinical significance varies widely. High-titer AIAs create an insulin-antibody complex that acts as an insulin reservoir, releasing insulin unpredictably and contributing to erratic glucose control; this mechanism explains the rare syndrome of immune-mediated insulin resistance (requiring doses exceeding 200 units per day). Local injection site reactions (erythema, induration, pruritus) within minutes to hours of injection reflect IgE-mediated hypersensitivity and are most common with newer patients and with pump infusion set changes. Systemic allergic reactions are rare with modern recombinant human insulins but can occur; desensitization protocols are available for patients with confirmed allergy who require insulin therapy. Lipodystrophy at injection sites may reflect immune reactions to insulin formulation components (protamine in NPH, zinc) in addition to the local mitogenic effects described earlier.5
Insulin is the only antihyperglycemic agent proven safe in pregnancy and remains the standard of care for both T1DM and gestational diabetes mellitus (GDM) requiring pharmacological treatment. Insulin does not cross the placenta in physiologically significant amounts; fetal glucose comes from maternal circulation by facilitated diffusion, and fetal insulin is secreted by fetal pancreatic beta cells in response to fetal glucose levels. Uncontrolled maternal hyperglycemia causes fetal hyperinsulinemia (the Pedersen hypothesis), driving macrosomia, neonatal hypoglycemia, respiratory distress, and stillbirth risk. Recommended insulin preparations in pregnancy include all human insulins and the analogs lispro and aspart (Pregnancy Category B, extensive safety data); glargine and detemir have less data but are used in clinical practice with apparent safety; degludec is not recommended in pregnancy due to insufficient data. Insulin requirements typically increase substantially in the second and third trimesters as placental hormones (human placental lactogen, cortisol, progesterone, prolactin) produce progressive insulin resistance, often doubling the pre-pregnancy dose requirement by the third trimester, and then fall abruptly with delivery of the placenta.15
Saltiel AR, Kahn CR. Insulin signalling and the regulation of glucose and lipid metabolism. Nature. 2001;414(6865):799–806.
doi:10.1038/414799aRorsman P, Braun M. Regulation of insulin secretion in human pancreatic islets. Annu Rev Physiol. 2013;75:155–179.
doi:10.1146/annurev-physiol-030212-183754Nauck MA, Meier JJ. Incretin hormones: their role in health and disease. Diabetes Obes Metab. 2018;20(Suppl 1):5–21.
doi:10.1111/dom.13129Korenblat KM, Fabbrini E, Mohammed BS, Klein S. Liver, muscle, and adipose tissue insulin action is directly related to intrahepatic triglyceride content in obese subjects. Gastroenterology. 2008;134(5):1369–1375.
doi:10.1053/j.gastro.2008.01.075Hirsch IB. Insulin analogues. N Engl J Med. 2005;352(2):174–183.
doi:10.1056/NEJMra040832Havelund S, Plum A, Ribel U, et al. The mechanism of protraction of insulin detemir, a long-acting, acylated analog of human insulin. Pharm Res. 2004;21(8):1498–1504.
doi:10.1023/b:pham.0000036926.54824.37Becker RHA, Dahmen R, Bergmann K, Lehmann A, Jax T, Heise T. New insulin glargine 300 Units/mL provides a more even activity profile and prolonged glycemic control at steady state compared with insulin glargine 100 Units/mL. Diabetes Care. 2015;38(4):637–643.
doi:10.2337/dc14-0006Frid AH, Kreugel G, Grassi G, et al. New insulin delivery recommendations. Mayo Clin Proc. 2016;91(9):1231–1255.
doi:10.1016/j.mayocp.2016.06.010Duckworth WC, Bennett RG, Hamel FG. Insulin degradation: progress and potential. Endocr Rev. 1998;19(5):608–624.
doi:10.1210/edrv.19.5.0349Riddle MC, Rosenstock J, Gerich J; Insulin Glargine 4002 Study Investigators. The treat-to-target trial: randomized addition of glargine or human NPH insulin to oral therapy of type 2 diabetic patients. Diabetes Care. 2003;26(11):3080–3086.
doi:10.2337/diacare.26.11.3080American Diabetes Association Professional Practice Committee. Standards of Care in Diabetes—2024. Diabetes Care. 2024;47(Suppl 1):S1–S321.
doi:10.2337/dc24-SINTNICE-SUGAR Study Investigators; Finfer S, Chittock DR, Su SY, et al. Intensive versus conventional glucose control in critically ill patients. N Engl J Med. 2009;360(13):1283–1297.
doi:10.1056/NEJMoa0810625Cryer PE. Mechanisms of hypoglycemia-associated autonomic failure in diabetes. N Engl J Med. 2013;369(4):362–372.
doi:10.1056/NEJMra1215228Heise T, Nosek L, Bottcher SG, Hastrup H, Haahr H. Ultra-long-acting insulin degludec has a flat and stable glucose-lowering effect in type 2 diabetes. Diabetes Obes Metab. 2012;14(10):944–950.
doi:10.1111/j.1463-1326.2012.01638.xBlumer I, Hadar E, Hadden DR, et al. Diabetes and pregnancy: an Endocrine Society clinical practice guideline. J Clin Endocrinol Metab. 2013;98(11):4227–4249.
doi:10.1210/jc.2013-2465