Classic diabetic ketoacidosis and the euglycemic variant: Something old, something new

Ketoacidosis in diabetes

When there is an absolute deficiency of insulin (as in type 1 diabetes) or severely augmented insulin resistance (by elevation of stress hormones and inflammatory cytokines in individuals predisposed to insulin resistance), the normal suppressive effect of insulin on glucagon is no longer operative. The unrestrained rise in glucagon disrupts the insulin-to-glucagon ratio in favor of glucagon, resulting in enhanced gluconeogenesis and release of free fatty acids. Fatty acid oxidation results in the formation of ketone bodies, which are secondary fuels used by organs and tissues such as the heart and muscles. In these tissues, beta hydroxybutyrate is converted back to acetoacetate, which is a substrate for the tricarboxylic acid (or Krebs) cycle that synthesizes and metabolizes acetyl-coenzyme A to yield adenosine triphosphate and carbon dioxide.

The rate-limiting steps in the ongoing metabolism of ketone bodies are nicotinamide adenine dinucleotide and its metabolites and succinyl coenzyme A. When these steps are overwhelmed, ketones accumulate and, because they are strong acids, induce acidosis. At the same time, the lack of insulin does not allow for the removal of glucose from the bloodstream, causing hyperglycemia and leading to osmotic diuresis, which promotes dehydration.

The EDKA variant

EDKA due to SGLT-2 inhibitors has a somewhat different etiologic basis: SGLT-2 inhibitors intrinsically stimulate the release of glucagon.² The mechanism is unclear; the sidebar “Glucagon release in EDKA: Plausible pathophysiologic mechanisms” outlines the proposed mechanisms.

The rise in glucagon changes the glucagon-to-insulin ratio and ensures that treatment with SGLT-2 inhibitors fosters a higher level of circulating ketones in the steady non-EDKA state.³ Any additional reasons for developing ketones, such as keto diets or “nothing by mouth after midnight” (who isn’t?) and prolonged fasting, can tip the balance into ketoacidosis. The effect of SGLT-2 inhibitors on glycosuria continues, limiting hyperglycemia while promoting dehydration.

Notably, the half-life of SGLT-2 inhibitors ranges between 13 and 17 hours, so their effect can persist for up to 3 days (4.5 half-lives), prolonging the duration of EDKA and the treatment regimen. In addition, SGLT-2 inhibitor–induced glycosuria is accompanied by natriuresis, which disrupts the electrochemical gradient in the tubular fluid and drives the reabsorption of negatively charged ketone bodies,⁴ further contributing to ketoacidosis.

Factors predisposing to EDKA

As outlined previously, SGLT-2 inhibitor therapy is always accompanied by an increase in ketonemia. Thus, any circumstances that would normally cause ketogenesis will further enhance ketonemia and may lead to ketoacidosis. These factors, which need to be reviewed with the patient, are as follows:

• Prolonged fasting (pre- and postoperatively)

• Ketogenic diets

• Anorexia and bulimia

• Alcohol intoxication

• Insulin pump malfunction

• Gastroenteritis and pancreatitis

• Gastroparesis.

In the event of an emergency that necessitates fasting and the patient has not been able to discontinue the SGLT-2 inhibitor, an intravenous drip with dextrose (glucose 4–5 g per hour) to offset starvation and either a basal insulin dose or an intravenous insulin infusion, depending on the clinical status of the patient, must be started to induce a favorable insulin-to-glucagon ratio and to stop ketogenesis. There are no studies on the appropriate dose; however, in our experience, basal insulin 0.15 to 0.25 units/kg, without exceeding 20 to 25 units, is adequate to minimize risk of both EDKA and hypoglycemia.

Dos

• Maintain hydration with water and a small amount of electrolyte solution

• Ensure 30% to 35% carbohydrate content in all meals and snacks

• Limit alcohol intake

• Know the identifying symptoms: fruity breath, thirst, polyuria and nocturia, nausea and vomiting, abdominal pain, confusion, and fever.

Don’ts

• If on basal insulin and self-managing, do not decrease the dose more than 20% without discussion with the supervising healthcare team

• No keto diets

• Do not take the SGLT-2 inhibitor starting 3 days before fasting longer than overnight.

Sick-day rules

In acute intercurrent illness with nausea, vomiting, or diarrhea, sip on calorie-dense electrolyte solutions (200 mL every 30 minutes), suck on hard candy, or consume a tablespoon of sugar every 15 to 20 minutes. If there is no improvement and symptoms persist, stop the SGLT-2 inhibitor and go to an emergency department or call the primary care clinician. Once in the emergency department, inform medical staff about the need to start a glucose-containing intravenous infusion.

Fluids

The primary goal of fluid administration is to restore tissue perfusion. The standard requirement is 15 to 20 mL/kg in the first hour, which is about 1 to 1.5 L.¹ Two priorities are addressed: dehydration and resolution of the component of hypernatremia caused by hemoconcentration before any significant electrolyte shifts happen between the intracellular and extracellular compartments. Normal saline remains the fluid of choice in the first hour, especially because the potassium level is initially not known and balanced electrolyte solutions contain potassium (albeit in low concentrations).

Fluid choice after the first bolus also depends on the corrected sodium level:

If the result reveals hyponatremia (serum sodium < 134 mmol/L), changing to a balanced electrolyte solution is helpful; however, in the face of normo- or hypernatremia, delaying the change to isotonic crystalloid solutions and using half normal saline may be more advisable.

The choice of which fluid is optimal after the bolus of normal saline and achieving an appropriate sodium level is open for discussion. The option of continuing normal saline vs changing to a balanced crystalloid solution such as Ringer’s lactate solution or a balanced electrolyte solution like Plasma-Lyte has been studied.^5,6 The higher chloride and osmolarity of normal saline are associated with the development of hyperchloremic metabolic acidosis.⁵ Balanced crystalloids can be used for volume expansion and are not associated with hyperchloremic acidosis. In a subgroup analysis of 172 adults with DKA from 2 large cluster-randomized clinical trials, the median time to DKA resolution was significantly shorter with balanced crystalloids (13.0 hours) compared with saline (16.9 hours),⁵ and balanced crystalloids resulted in less hyperchloremia and a faster recovery of bicarbonate levels.⁶

When the circulating glucose level is less than 250 mg/dL, giving a 5% dextrose solution containing balanced electrolytes is necessary to prevent hypoglycemia and to reduce the risk of cerebral edema caused by rapid correction of the extracellular compartment.⁷

In EDKA, hydration must be started with a dextrose-containing fluid to accomplish 4 crucial goals:

• Provide glucose to stop the ketogenic process

• Reinstitute secretion of endogenous insulin, if present, thereby starting to alter the insulin-to-glucagon ratio in favor of insulin, stopping the ketogenic process

• Rehydrate and replete solutes

• Counterbalance the effects of the infused insulin that was started to enhance recovery from ketoacidosis.

There are no specific studies related to fluid replacement in EDKA. However, in EDKA due to SGLT-2 inhibitors, hypernatremia is less of an issue because SGLT-2 inhibitors cause glycosuria, which also induces natriuresis. Theoretically, the persistence of acidosis caused by the continuing effect of SGLT-2 inhibitors in EDKA may make the use of a balanced electrolyte fluid even more beneficial.

Potassium

The serum potassium value is not an indicator of potassium status in either DKA or EDKA because in each form there is a total potassium deficit due to osmotic diuresis. Volume expansion, correction of acidosis, and insulin therapy all lower potassium levels. Therefore, once serum potassium levels are less than 5.3 mmol/L, potassium replacement is essential.¹ If hypokalemia is present at the time of diagnosis, potassium must be added to the initial fluid administered. Insulin should be started only after the serum potassium level is greater than 3.3 to 3.5 mmol/L to protect against arrhythmias, respiratory muscle weakness, and even death.

In SGLT-2 inhibitor–related EDKA, the SGLT-2 inhibitor generally lowers the risk of severe hyperkalemia in people with type 2 diabetes.⁸

Insulin therapy

Insulin therapy is optimally instituted after the first bolus of intravenous fluid. This delay allows for a more accurate estimation of hyperglycemia by correcting hemoconcentration and a more meaningful evaluation of the serum potassium level.

Intravenous insulin to treat DKA is preferable. Whereas hourly subcutaneous short-acting insulin analogs have been used successfully to treat mild to moderate DKA,^9,10 the consensus remains with intravenous therapy. Studies on the utility of an initial bolus followed by an infusion have shown no beneficial effect of a bolus dose on the rate of glucose decrease, the rate of anion-gap correction, or the hospital length of stay.¹¹ However, a bolus shows a trend toward more frequent development of hypoglycemia (6% vs 1%).

In EDKA, a simultaneous infusion of insulin and a dextrose-containing fluid is essential; insulin augments the effect of the infused glucose, as described above.

Bicarbonate

A systematic review of bicarbonate therapy in DKA showed adding bicarbonate produced a transient improvement in metabolic acidosis, but had no effect on glycemic improvement and also carried a risk of cerebral edema in children.¹² Even when used in severe acidotic DKA, bicarbonate therapy did not show any difference in time to resolution of acidosis, length of time intravenous insulin was required, potassium supplementation requirements, or hospital length of stay.¹³

There are no studies on bicarbonate administration in severe acidosis (pH < 6.9).¹² If pH is between 6.9 and 7.0 and bicarbonate therapy is being considered, 50 mmol of bicarbonate in 200 mL of sterile water with 10 mmol of potassium chloride can be given over 2 hours to achieve a pH greater than 7.0.¹ Adding potassium chloride simultaneously is dependent on ambient potassium levels because administering bicarbonate may increase risk for hypokalemia.

The discussion regarding bicarbonate therapy in EDKA is even more germane because bicarbonate levels may remain low until the SGLT-2 inhibitor is completely cleared (up to 4.5 half-lives, which is approximately 60–72 hours). Accordingly, it may be tempting to increase the bicarbonate level with exogenous bicarbonate therapy, but there is no proven benefit.^12,14

Phosphate

Hypophosphatemia during DKA is common and increases with severe acidosis. The increased loss of phosphate is a result of transcellular shift, osmotic diuresis, and reduced phosphate reabsorption in the renal proximal tubule due to acidosis and hyperglycemia. With administration of insulin and fluids, phosphate shifts into the intracellular compartment and a nadir of phosphate is reached at a median 16 hours into therapy.¹⁵ However, there appears to be no adverse effects from hypophosphatemia when left untreated, and no benefits from treatment have been observed.

There is a concern that very low phosphate levels (< 1 mg/dL) may contribute to cardiac and muscle weakness and respiratory depression. In such circumstances, potassium phosphate 20 to 30 mmol/L can be added to the replacement fluid to achieve a phosphate level just greater than 1 mg/dL.¹⁴ Overenthusiastic replacement not only shows no benefit, but also can result in significant hypocalcemia.

In EDKA from SGLT-2 inhibitors, low phosphate is uncommon because these agents usually raise phosphate levels by increasing renal tubular reabsorption of phosphate.¹⁶

Transitioning from intravenous insulin to subcutaneous insulin

The current recommendations are to start subcutaneous insulin when the serum glucose level is less than 200 to 250 mg/dL, the anion gap is less than 12, and the bicarbonate level is greater than 15 mmol/L.¹⁷ Under those circumstances, basal insulin needs to be administered at the preadmission dose (if known) or at 0.25 units/kg (if unknown) and should overlap with intravenous insulin for about 2 hours. Planning a transition without the use of basal insulin is harmful.

In addition, studies have shown benefits of early institution of basal insulin while using intravenous insulin to treat DKA, including less rebound hyperglycemia when intravenous insulin is discontinued, quicker resolution of DKA, reduced intravenous insulin requirements, and reduced hospital length of stay, with no increase in hypoglycemia or hypokalemia.¹⁷

In SGLT-2 inhibitor–related EDKA, the transition cutoffs need to be tempered because ketosis—and, therefore, a low bicarbonate level—may rebound after intravenous therapy is discontinued due to the persistent effect of the medication. Thus, it is important to ensure that the anion gap has normalized and the bicarbonate level is in the low-normal range before transitioning.