Starvation ketoacidosis and refeeding syndrome

Abstract

Starvation ketoacidosis (SKA) is a rarer cause of ketoacidosis. Most patients will only have a mild acidosis, but if exacerbated by stress can result in a severe acidosis. We describe a 66-year-old man admitted with reduced consciousness and found to have a severe metabolic acidosis with raised anion gap. His body mass index (BMI) was noted to be within the healthy range at 23 kg/m²; however, it was last documented 1 year previously at 28 kg/m² with no clear timeframe of weight loss. While his acidosis improved with intravenous fluids, he subsequently developed severe electrolyte imbalance consistent with refeeding during his admission. Awareness of SKA as a cause for high anion gap metabolic acidosis is important and knowledge of management including intravenous fluids, thiamine, dietetic input and electrolyte replacement is vital.

Background

A high anion gap metabolic acidosis (HAGMA) can be caused by a variety of metabolic changes including uraemia, lactic acidosis and ingestion of substances such as ethylene glycol, methanol, salicylates, isoniazid and ketoacidosis. There are multiple metabolic acidoses which can result in accumulation of ketone bodies in the blood stream which can cause a diagnostic challenge to the clinician. Starvation ketoacidosis (SKA) is one of these metabolic acidoses, and there are case reports of low carbohydrate diet ketoacidosis.^{1 2}

Case presentation

A 66-year-old man was admitted as a stand-by to accident and emergency with worsening confusion and reduced responsiveness over the past 24 hours. He had last been seen 3 days ago and family reported no concerns. His medical history included previous severe major depression and psychosis, stroke, acute cholecystitis and psoriasis. His long-term treatment included lithium, lurasidone, omeprazole, simvastatin, aspirin, sertraline, lactulose and bisoprolol.

On presentation, he was initially febrile in the ambulance (38.3°C) which resolved with no therapy on arrival to hospital 36.1°C. Other observations included sinus tachycardia (heart rate 130–150 beats per minute), blood pressure 145/75 mm Hg, respiratory rate 17 breaths per minute and oxygen saturations 97% while breathing air. Apart from tachycardia, cardiac, respiratory and abdominal examination showed no significant abnormality. Glasgow Coma Score (GCS) was initially 10 (eyes 4, verbal 1, motor 5) and he was noted to be very tremulous with global increase in muscle tone.

Collateral history from his daughter was obtained and she believed he had been eating and drinking normally and he had no alcohol in his home. She noted there were no additional medications taken from his dosette box, and there were no cleaning fluids out of cupboards to suggest self-poisoning and he had no prior history of intentional overdose. She reported no significant family history.

Investigations

Biochemical investigations are shown in table 1. Initial venous blood gas showed HAGMA (anion gap >35.7 mmol/L) and serum osmolality was raised 323 mOsm/Kg. Other investigations of note showed random glucose was 10.2 mmol/L, HbA1c 42 mmol/mol (indicating pre-diabetes), lactate 2.3 mmol/L, urea 10.5 mmol/L, C-Reactive Protein (CRP) 18 mg/L and leucocytosis 28×10⁹ g/L (reference range: 4–10×10⁹ g/L) with 87% neutrophils. A new transaminitis was found with peak Alanine aminotransferase (ALT) 219 U/L, Aspartate aminotransferase (AST) 232 U/L with no evidence of biochemical obstruction. Adjusted calcium was initially 2.34 mmol/L and fell to 1.97 mmol/L 7 hours in to admission. Due to HAGMA and developing hypocalcaemia, ethylene glycol and methanol were analysed and subsequently not detected. He was found to have raised beta-hydroxybutyrate level of 8.85 mmol/L.

Table 1.

Results of biochemical investigations

	Reference intervals	Day 1 - admission	Day 1 +7 hours	Day 3
Sodium	133–146 mmol/L	139 mmol/L	136 mmol/L	136 mmol/L
Potassium	3.5–5.3 mmol/L	4.7 mmol/L	3.9 mmol/L	2.4 mmol/L
Chloride	95–108 mmol/L	103 mmol/L	108 mmol/L	106 mmol/L
Urea	2.5–7.8 mmol/L	10.5 mmol/L	11.1 mmol/L	4.7 mmol/L
Creatinine	40–130 umol/L	125 umol/L	118 umol/L	53 umol/L
Adjusted calcium	2.2–2.6 mmol/L	2.37 mmol/L	1.97 mmol/L	2.16 mmol/L
Phosphate	0.8–1.5 mmol/L	NA	0.87 mmol/L	0.28 mmol/L
Magnesium	0.7–1 mmol/L	1.01 mmol/L	0.86 mmol/L	0.8 mmol/L
Glucose	3.5–6 mmol/L	10.2 mmol/L	14.1 mmol/L	–
Beta-Hydroxybutyrate	0–0.42 mmol/L	8.85 mmol/L	–	5.85 mmol/L
H+	35–45 nmol/L	74 nmol/L	–	–
Bicarbonate	22–28 mmol/L	<5 mmol/L	7 mmol/L	20 mmol/L
Anion gap	–	>35.7 mmol/L	24.9 mmol/L	12.4 mmol/L
Osmolality	275-295 mOsm/kg	323 mOsm/kg	–	–
C-Reactive Protein	<10 mg/L	18 mg/L	36 mg/L	148 mg/L
Lactate	0.6–2.2 mmol/L	2.3 mmol/L	1 mmol/L	–
Paracetamol	–	<5 mg/L	–	–
Salicylate	–	<50 mg/L	–	–
Methanol	–	ND	–	–
Ethylene glycol	–	ND	–	–
Ethanol	–	<10 mg/dL	–	–
Lithium	0.4–1 mmol/L*	0.5 mmol/L	–	–
Alanine aminotransferase	<50 U/L	139 U/L	219 U/L	118 U/L
Aspartate aminotransferase	<40 U/L	158 U/L	232 U/L	37 U/L
Amylase	<100 U/L	49 U/L	54 U/L	–

All investigations were undertaken in the laboratory except for hydrogen ions which were measured on a GEM 5000 blood gas analyser with a venous blood sample.

*Reference range applies to bloods taken 12±1 hour after dose

ND, not detected; NA, not available

Chest X-ray showed no obvious abnormality and CT head showed no cerebral aetiology.

Differential diagnosis

Causes of HAGMA were considered and excluded as able. The patient’s blood toxicology was negative for ethanol and substances including salicylate, paracetamol, ethylene glycol and methanol. Urea was only mildly elevated excluding uraemia. Lactate was borderline raised but not the main driving cause of the acidosis. The patient had not been prescribed metformin, iron, isoniazid or sodium glucose co-transporter-2 inhibitors. Random glucose was initially 10 mmol/L and patient had previously likely type 2 diabetes in 2018 which appeared to have been reversed with lifestyle changes and HbA1c 5 months prior to admission was 38 mmol/mol. Later on admission, HbA1c was 42 mmol/mol. C peptide was performed which was 1599 pmol/L although this measurement was of limited relevance since current protocol would indicate C peptide measurement in patients with diabetes for >3 years.

Our working diagnoses became SKA or late adult presentation of inborn error of metabolism. While collateral history made starvation or low carbohydrate diet seem less likely, they could not be completely excluded as this man lives alone and therefore it was difficult to illicit the exact nature of his nutritional intake prior to admission. Urinary organic acids were sent to an external laboratory with a 14-day turnaround time. Later, results showed moderate-severe ketonuria with no other abnormal metabolites. 1,2-Propanediol was also present which is used as a solvent in many pharmaceutical formulations and was of no clinical significance.

Treatment

He was treated with intravenous fluids—initially 0.9% saline switched to predominantly 5% dextrose. His acidosis settled and GCS slowly returned to 15. Given the hypocalcaemia and unclear cause of HAGMA, fomepizole intravenously was given until negative toxicology for ethylene glycol was confirmed. Ethylene glycol is measured in an external laboratory with the assay available from Monday to Friday 9 am–5 pm and weekend mornings. It was arranged to be run the following morning and results were available within 24 hours of admission. Intravenous calcium was administered to correct his moderate hypocalcaemia.

Outcome and follow-up

Two days after admission, the patient developed a severe electrolyte imbalance with potassium dropping to a low of 2.4 mmol/L, phosphate 0.28 mmol/L and magnesium remained within normal limits. These electrolytes were replaced with intravenous potassium chloride and sodium glycerophosphate.

Dietetic review revealed his BMI was 23.3 kg/m² at admission; last measurement was in March 2020 where BMI was 28 kg/m² showing an 11.6% wt loss with no clear timeframe of weight loss established. Dietetic review suggested daily intake of 1560–1950 kcal/day and 65–98 g/day of protein would be his optimum intake. He was started on thiamine and a dietetic plan created with gradual increase in intake due to the potential refeeding risk. Using the Malnutrition Universal Screening Tool (MUST) was difficult initially as the patient was unclear about the timing of his weight loss but it highlighted that he was at least medium or high risk of malnutrition.

While his nutrition and electrolyte imbalance were being treated, his CRP began to rise to a peak of 334 mg/L and subsequent CT thorax, abdomen, pelvis showed evidence of acute calculous cholecystitis and possible element of pancreatitis. Lung findings were most likely in keeping with poor inspiration but could not exclude concurrent infection or aspiration. Amylase remained within reference range. Repeat chest X-ray showed no focal lung abnormality. Percutaneous ultrasound guided drainage of gallbladder found turbid foul smelling fluid with heavy Kliebsella oxytoca and Escherichia coli. He was treated with intravenous antibiotics and taken over by the surgical team.

Discussion

We describe a middle-aged man presenting with reduced consciousness due to severe acidosis who subsequently developed severe electrolyte disturbance as an inpatient. The patient initially had a HAGMA associated with ketosis and this improved with intravenous fluids. On day 2 of admission, severe hypokalaemia and hypophosphataemia were found requiring intravenous replacement. It was noted while his BMI was within the healthy range at 23 kg/m², 1 year previously he had a documented BMI of 28 kg/m².

The diagnosis of SKA was based on the improvement of his acidosis with intravenous fluids and his anion gap fell from >35.7 mmol/L to 24.9 mmol/L within 7 hours of admission. We also excluded other causes of metabolic acidoses including diabetic ketoacidosis and alcohol ketoacidosis given his raised beta-hydroxybutyrate levels. In addition, during admission the patient then developed severe electrolyte disturbance consistent with refeeding syndrome.

There are few case reports of SKA associated with severe metabolic acidosis. In healthy individuals, short-term starving results in mild ketosis. Low insulin levels promote lipolysis and delivery of fatty acids to the liver which undergo beta oxidation and form ketone bodies. In experimental studies acidoses have been mild and beta-hydroxybutyrate levels have been maximal (5–12 mmol/L) after several weeks of total fasting.^3–5 However a number of factors including stress may exacerbate the severity of metabolic acidosis.^{3 6} Some previous case reports describe severe SKA in the context of pregnancy, infant intensive care and bariatric surgery complication.^6–8 Starvation usually results in decreased levels of cortisol and catecholamines,³ however stress leads to increased ketogenesis due to raised hormones including catecholamines, cortisol and glucagon.^9–11 This man had a mildly raised CRP but marked leucocytosis at admission with peak CRP at day 4 of 336 mg/L and was diagnosed with acute cholecystitis and possible pancreatitis. This developing condition likely exacerbated his metabolic acidosis and also may have contributed by reducing his oral intake if he was experiencing abdominal pain and/or feeling unwell.

Sixty-four hours in to admission, biochemical investigations showed severe hypophosphataemia and hypokalaemia. The earliest biochemical change in refeeding is usually hypophosphataemia which typically occurs in the first 24–72 hours of refeeding.^{12 13} During refeeding there is an increased demand for phosphate due to increased carbohydrate metabolism, protein synthesis and insulin causes a shift of phosphate into the cells. Increased insulin secretion can lead to increased potassium uptake into cells and thus hypokalaemia.¹³ The patient was given thiamine, intravenous electrolyte replacement and fluids. Dieticians regularly reviewed his progress and formulated and adapted a nutritional rehabilitation diet as he was at a high refeeding risk.¹⁴ This coupled with his weight loss, therefore strengthened our diagnosis of SKA and subsequent refeeding syndrome.

Our patient described had a severe metabolic acidosis likely owing to SKA and exacerbated by stress. While a less common cause of HAGMA, SKA should always be considered in a patient with HAGMA and/or ketosis. MUST score or similar assessment should be carried out to help identify patients with malnutrition, at risk of malnutrition or obese and correct observations and/or dietetic input can be carried out.

Learning points.

• In patients with metabolic acidosis, an anion gap should be calculated to help aid diagnosis of underlying cause.

• Starvation ketoacidosis should be considered in high anion gap metabolic acidosis and/or ketosis.

• Patients should always be screened using Malnutrition Universal Screening Tool or appropriate screening tool to identify adults who are malnourished, at risk of malnutrition or obese.

Ethics statements

Patient consent for publication

Consent obtained directly from patient(s).