Ketogenesis: Pathway, Rate-Limiting Enzyme & DKA vs AKA

Ketogenesis is how the liver turns surplus acetyl-CoA from fatty acid β-oxidation into ketone bodies (acetoacetate, β-hydroxybutyrate and acetone) inside hepatocyte mitochondria when insulin is low and glucagon is high. The rate-limiting enzyme is mitochondrial HMG-CoA synthase, not HMG-CoA reductase. The liver exports ketones but can't use them because it lacks succinyl-CoA:3-ketoacid CoA transferase (SCOT, "thiophorase"), and red blood cells can't use them because they have no mitochondria. Runaway ketogenesis causes high anion gap metabolic acidosis: diabetic ketoacidosis (DKA), usually with high glucose (lower, sometimes normal, on SGLT2 inhibitors), and alcoholic ketoacidosis (AKA), with low or normal glucose and a high NADH/NAD⁺ ratio that makes β-hydroxybutyrate dominate, so urine dipsticks underestimate the ketosis.

The pathway in five steps

1. Lipolysis. Low insulin plus high glucagon, epinephrine and cortisol switch on hormone-sensitive lipase in adipose tissue. Free fatty acids travel to the liver; glycerol feeds gluconeogenesis.
2. Entry and β-oxidation. Fatty acyl groups enter mitochondria on the carnitine shuttle, with CPT-1 as the gate, and are cut into acetyl-CoA, generating NADH and FADH₂ (see beta-oxidation and the carnitine shuttle).
3. The TCA cycle backs up. Oxaloacetate is pulled into gluconeogenesis, and high NADH shifts what is left toward malate, so acetyl-CoA outruns the oxaloacetate it needs to enter the cycle.
4. Ketone synthesis. Thiolase joins two acetyl-CoA into acetoacetyl-CoA; HMG-CoA synthase (mitochondrial isoform, HMGCS2; rate-limiting, and the first irreversible step of ketogenesis) adds a third to make HMG-CoA; HMG-CoA lyase splits HMG-CoA into acetoacetate and acetyl-CoA.
5. Interconversion. β-hydroxybutyrate dehydrogenase reduces acetoacetate to β-hydroxybutyrate using NADH, so more NADH means more β-hydroxybutyrate. Some acetoacetate loses CO₂ spontaneously to form acetone, which is exhaled.

Regulation: what flips on in fasting and DKA

Insulin is the main brake on ketogenesis. When it falls and counter-regulatory hormones rise, every control point opens:

The same ACC and malonyl-CoA switch runs fatty acid synthesis in the fed state, so synthesis and β-oxidation never run at full speed together. In DKA, insulin deficiency plus counter-regulatory hormones drive unrestrained lipolysis and hepatic fatty acid oxidation (2024 consensus).

Who makes ketones and who burns them

• The liver makes them but can't use them. Hepatocytes lack SCOT (thiophorase), the enzyme that transfers CoA from succinyl-CoA to acetoacetate. Without it, acetoacetate can't be reactivated to acetoacetyl-CoA, so the liver exports ketones.
• Users: brain, heart, skeletal muscle and kidneys. In their mitochondria, β-hydroxybutyrate is oxidized back to acetoacetate, SCOT makes acetoacetyl-CoA, and thiolase splits it into two acetyl-CoA for the TCA cycle.
• The brain in starvation. Ketones cross the blood-brain barrier readily and become the brain's main alternative fuel in prolonged fasting, which is what lets humans survive long starvation: less glucose is needed, so less muscle protein is broken down for gluconeogenesis.
• Red blood cells never use them. No mitochondria means no SCOT and no TCA cycle, so RBCs depend on glucose through glycolysis.
• Flip side: inherited SCOT deficiency (OXCT1, autosomal recessive) causes ketoacidotic attacks and persistent ketosis, because the body makes ketones it can't burn.

Starvation ketosis and the low-ketone trap

Starvation ketosis is the normal response to not eating: insulin is low but present, so ketones rise without the runaway acidosis of DKA. A history of eating under 500 kcal/day suggests it (2024 consensus).

The trap is the opposite pattern. A fasting child who becomes hypoglycemic without ketones (hypoketotic hypoglycemia) can't make enough acetyl-CoA, which points to a fatty acid oxidation disorder. MCAD deficiency is the most common β-oxidation defect: symptoms start after about 2-3 months of age, fasting as short as 12 hours can trigger them, and medium-chain acylcarnitines appear in plasma (Merck).

DKA vs alcoholic ketoacidosis

Why NADH matters in AKA. Metabolizing ethanol generates NADH. A high NADH/NAD⁺ ratio drives acetoacetate to β-hydroxybutyrate and pushes pyruvate to lactate and oxaloacetate to malate, starving gluconeogenesis and explaining the low glucose. Give thiamine before glucose to prevent Wernicke encephalopathy; dextrose works because it raises insulin, which shuts off lipolysis and ketogenesis. Merck reserves insulin for possible atypical DKA or glucose above 300 mg/dL.

DKA management in brief (2024 consensus): isotonic saline or a balanced crystalloid, fixed-rate IV insulin (0.1 U/kg/h in severe DKA), potassium replacement once K⁺ falls below 5.0 mmol/L (20-30 mmol per liter of IV fluid, aiming for 4-5 mmol/L), and, if K⁺ is below 3.5 mmol/L, no insulin until potassium replacement brings it above 3.5 mmol/L. Merck uses a slightly lower cut-off: withhold insulin until K⁺ is at least 3.3 mEq/L. Serum K⁺ often starts normal or high despite a significant total-body deficit, because insulin deficiency and acidosis shift potassium out of cells. Bicarbonate is considered only if pH is below 7.0 (consensus and Merck). DKA has resolved when blood ketones are under 0.6 mmol/L and venous pH is at least 7.3 or bicarbonate at least 18 mmol/L. Compare DKA vs HHS.

The nitroprusside trap

• Nitroprusside, used in urine dipsticks and some serum assays, reacts with acetoacetate (and acetone), not β-hydroxybutyrate.
• Early in DKA and in AKA, it underestimates the ketosis.
• During treatment, β-hydroxybutyrate is converted back to acetoacetate, so the dipstick can look worse while the patient improves.
• Sulfhydryl drugs such as captopril can give false positives.
• Fix: measure blood β-hydroxybutyrate to diagnose and monitor (2024 consensus).

Euglycemic DKA and SGLT2 inhibitors

• About 10% of DKA presents with glucose below 200 mg/dL. SGLT2 inhibitors now account for most of these cases; pregnancy, reduced food intake, insulin given before arrival, alcohol and liver failure are other causes (2024 consensus).
• SGLT2 inhibitors (empagliflozin, dapagliflozin, canagliflozin, ertugliflozin) lower glucose by spilling it into urine, so glucose looks reassuring while too little insulin lets lipolysis and ketogenesis run.
• The current empagliflozin label warns that glucose may be below 250 mg/dL and lists risk factors: too little insulin, acute febrile illness, reduced calorie intake or a ketogenic diet, surgery, volume depletion, pancreatic disorders and alcohol misuse. Stop the drug if ketoacidosis is suspected, and withhold it for at least 3 days before surgery or procedures with prolonged fasting (the ertugliflozin label says at least 4 days).
• Treat as DKA, adding dextrose with the insulin from the start (glucose is already under 250 mg/dL) so insulin can continue until ketones clear.

How USMLE tests ketogenesis

• "Rate-limiting enzyme of ketogenesis?" HMG-CoA synthase (mitochondria). HMG-CoA reductase is the statin target in cytosolic cholesterol synthesis.
• Fruity breath with deep, rapid breathing = acetone plus Kussmaul respirations compensating for the acidosis.
• High anion gap, glucose around 70, alcohol history, "small" urine ketones = AKA with β-hydroxybutyrate predominance. Practice the gap with the anion gap breakdown.

Practice question

A 44-year-old man with alcohol use disorder comes to the emergency department with 2 days of vomiting and abdominal pain. He drank heavily for a week, then stopped eating and drinking when the vomiting began. He has no history of diabetes. He is tachycardic, with dry mucous membranes. Glucose is 76 mg/dL, sodium 136 mEq/L, chloride 96 mEq/L, bicarbonate 12 mEq/L (anion gap 28) and lactate 3.1 mmol/L; serum ethanol is undetectable and the osmolal gap is normal. Urine dipstick ketones are "small", but serum β-hydroxybutyrate is 7.0 mmol/L. Which best explains the discrepancy between the two ketone results?

• A. Acetone, which the dipstick cannot detect, is the predominant ketone body
• B. A high NADH/NAD⁺ ratio has shifted acetoacetate to β-hydroxybutyrate, which nitroprusside does not detect
• C. Hepatocytes lack succinyl-CoA:3-ketoacid CoA transferase
• D. Increased malonyl-CoA is inhibiting CPT-1
• E. Hypoglycemia has stimulated insulin release, suppressing ketogenesis

Answer: B. Ethanol metabolism raises NADH, which drives β-hydroxybutyrate dehydrogenase toward β-hydroxybutyrate. Nitroprusside reacts with acetoacetate, not β-hydroxybutyrate, so the dipstick reads "small" despite severe ketoacidosis. This is alcoholic ketoacidosis: give thiamine, then dextrose in saline.

• A: β-hydroxybutyrate, not acetone, dominates, and nitroprusside does react with acetone.
• C: True, and it is why the liver exports ketones, but it has nothing to do with what the dipstick measures.
• D: Backwards. Low insulin and high glucagon switch ACC off, malonyl-CoA falls and CPT-1 is disinhibited.
• E: Insulin is low in AKA, and an anion gap of 28 shows ketogenesis is running hard. Giving dextrose does raise insulin, which is exactly why it is the treatment.

Sneaky variant: a 58-year-old woman with type 2 diabetes on empagliflozin who has eaten little since surgery has glucose 180 mg/dL, pH 7.18 and β-hydroxybutyrate 5.5 mmol/L. That is euglycemic DKA: stop the SGLT2 inhibitor and give fluids, IV insulin plus dextrose, and potassium.

Key takeaways

• Ketogenesis happens in liver mitochondria: acetyl-CoA → acetoacetyl-CoA → HMG-CoA (HMG-CoA synthase, rate-limiting) → acetoacetate (HMG-CoA lyase) → β-hydroxybutyrate or acetone.
• Low insulin, high glucagon: lipolysis on, ACC off, malonyl-CoA down, CPT-1 open, β-oxidation up.
• The liver lacks SCOT and RBCs lack mitochondria; brain, heart, muscle and kidneys burn ketones.
• DKA (2024): glucose 200 mg/dL or more or known diabetes, β-hydroxybutyrate 3.0 mmol/L or more, pH below 7.30 or bicarbonate below 18; delay insulin while K⁺ is below 3.5 mmol/L (Merck: 3.3 mEq/L).
• AKA: low or normal glucose, high NADH/NAD⁺, extra β-hydroxybutyrate; thiamine, then dextrose-saline.
• Dipsticks detect acetoacetate, not β-hydroxybutyrate, so measure blood β-hydroxybutyrate.
• SGLT2 inhibitors cause most euglycemic DKA; withhold them before surgery (at least 3 days for empagliflozin, 4 for ertugliflozin).

Sources

• Biochemistry, Ketogenesis — StatPearls, NCBI Bookshelf (2025)
• Hyperglycaemic crises in adults with diabetes: a consensus report — Diabetologia (2024)
• Acute Complications of Diabetes Mellitus — Merck Manual Professional Edition (2025)
• Alcoholic Ketoacidosis — Merck Manual Professional Edition
• Alcoholic Ketoacidosis — Medscape (2022)
• JARDIANCE (empagliflozin) prescribing information — DailyMed, NLM (2026)
• STEGLATRO (ertugliflozin) prescribing information — DailyMed, NLM (2026)
• HMGCS2, mitochondrial HMG-CoA synthase (P54868) entry — UniProt
• Succinyl-CoA:3-ketoacid CoA transferase deficiency — MedlinePlus Genetics
• Fatty Acid and Glycerol Metabolism Disorders — Merck Manual Professional Edition
• Fuel Metabolism in Starvation — Annual Review of Nutrition (2006)