Worsening Metabolic Acidosis After Normal Saline Bolus in SGLT2 Inhibitor Use
The worsening metabolic acidosis (bicarbonate falling from 17 to 13 mmol/L) despite normal saline administration in a diabetic patient on canagliflozin (Invokana) is most likely due to ongoing euglycemic diabetic ketoacidosis (EDKA), not caused by the saline itself—the saline bolus revealed the severity of the underlying ketoacidosis by diluting the bicarbonate concentration while failing to suppress ongoing ketone production. 1, 2
Understanding the Clinical Picture
Your patient demonstrates:
- Persistent high anion gap acidosis (AG=15) indicating ongoing acid production, not dilutional acidosis 3
- Rising chloride (103→108 mmol/L) from normal saline administration, but this is expected and not the primary problem 3
- Falling bicarbonate (17→13 mmol/L) indicating worsening metabolic acidosis despite fluid resuscitation 3
Why This Is SGLT2 Inhibitor-Induced EDKA
SGLT2 inhibitors like canagliflozin cause euglycemic DKA through a unique mechanism:
- They promote glucosuria and shift metabolism toward ketone production even with normal glucose levels 1, 2
- The acidosis is often refractory to conventional DKA treatment including aggressive fluid resuscitation and bicarbonate boluses 1
- The persistent anion gap of 15 indicates ongoing ketoacid production that the saline bolus cannot correct 3, 2
Why the Saline Didn't Help (And What Actually Happened)
The normal saline bolus did NOT cause the worsening acidosis—it exposed it:
- Saline dilutes the existing bicarbonate concentration, making the acidosis appear worse numerically 3
- However, the unchanged anion gap of 15 proves this is not simple dilutional acidosis—there is ongoing acid production 3
- In true dilutional acidosis, the anion gap would decrease proportionally with bicarbonate 3
Immediate Management Algorithm
1. Confirm EDKA diagnosis:
- Measure beta-hydroxybutyrate (expect markedly elevated levels) 1, 2
- Check arterial blood gas to assess pH (likely <7.3 despite euglycemia) 3
- Verify glucose is not severely elevated (typically <250 mg/dL in EDKA) 1, 2
2. Initiate insulin therapy immediately:
- Start continuous insulin infusion even if glucose is normal or low 1
- This is the ONLY way to suppress ketone production 1
3. Implement high-calorie glucose infusion:
- Administer 10% dextrose at high rates (150-200 mL/hour) to prevent hypoglycemia while maintaining insulin infusion 1
- This represents a paradigm shift from traditional DKA management and is critical for SGLT2 inhibitor-induced EDKA 1
- Maintain tight glycemic control with insulin titration to keep glucose 150-200 mg/dL 1
4. Continue aggressive fluid resuscitation:
- Isotonic saline 15-20 mL/kg/h initially to restore perfusion 3
- Add potassium (20-30 mEq/L as 2/3 KCl and 1/3 KPO4) once urine output established 3
- Monitor potassium frequently as alkalinization drives potassium intracellularly 3
5. Bicarbonate therapy considerations:
- Generally NOT indicated unless pH <6.9-7.0 3
- In SGLT2 inhibitor-induced EDKA, bicarbonate boluses are often ineffective without addressing ongoing ketone production 1
- If used, goal is pH 7.2, not normalization 3
Critical Monitoring Parameters
- Venous blood gas every 2-4 hours to monitor pH and anion gap closure 3
- Serum potassium every 2-4 hours (risk of life-threatening hypokalemia during correction) 3
- Beta-hydroxybutyrate levels to confirm ketone clearance 1, 2
- Blood glucose hourly to maintain 150-200 mg/dL during high-dose glucose infusion 1
Common Pitfalls to Avoid
Do not stop insulin when glucose normalizes:
- The insulin is needed to suppress ketogenesis, not just to lower glucose 1
- Continue insulin with high-calorie glucose infusion until ketones clear 1
Do not attribute worsening bicarbonate solely to saline administration:
- The unchanged anion gap proves ongoing acid production 3
- Stopping fluids would worsen renal perfusion and acid clearance 3
Do not use traditional DKA protocols:
- SGLT2 inhibitor-induced EDKA requires simultaneous high-dose glucose and insulin 1
- Standard protocols that reduce dextrose as glucose normalizes will fail 1