5a. Diabetic ketoacidosis

5a.1 Background and Evidence Review

Pathophysiology

SGLT 2 inhibition induces glycosuria, which causes widespread changes in metabolism, including an increase in lipid mobilisation, free fatty acid oxidisation and increased plasma ketone levels, in particular β- hydroxybutyrate and acetoacetate (4, 5). States of relative insulin deficiency or reduced carbohydrate intake augment hepatic ketogenesis. Therefore factors such as infection, fasting or reduction in insulin levels can precipitate diabetic ketoacidosis (DKA) in people treated with SGLT 2 inhibitors. Risk is highest in those who require prescription of insulin, and incredibly low among people without DM.

Relative insulin deficiency and reduced carbohydrate intake, with concomitant carbohydrate deficit related to glycosuria contributes to normal or near normal glycaemia (6), making it possible for ketoacidosis to be present with normal or low capillary blood glucose levels (sometimes referred to as “euglycaemic” ketoacidosis) (5).

Large trial evidence

In the RSG and SMART-C meta-analysis of large randomised trials of SGLT 2 inhibition, the risk of DKA was increased approximately two-fold by SGLT 2 inhibition (3). Because the baseline risk of ketoacidosis was so low in people with type 2 diabetes, the excess absolute risk of DKA among people with CKD and diabetes was around 1 DKA event per 1000 person-years treated with an SGLT 2 inhibitor, an approximate order of magnitude less than the benefits in terms of kidney disease progression in this population (for whom 1000 person-years of treatment with an SGLT 2 inhibitor would on average prevent 11 kidney disease progression events)(see figure 5.1 above). Among the large SGLT 2 inhibitor trials, type 1 DM was generally excluded (with the exception of EMPA-KIDNEY, which recruited a small number of participants with type 1 DM (1)). More information regarding risk of DKA in people with type 1 DM is discussed in Section 7a.

People with chronic kidney disease

The CREDENCE trial included participants with CKD with estimated glomerular filtration rate (eGFR) of 30-90 mL/min/1.73m² and reported a statistically significant increased risk of DKA. Although the relative risk for DKA was 10.8 (95%CI 1.39-83.7), DKA was rare (canagliflozin 11/2200 vs placebo 1/2197), and the absolute excess risk in the canagliflozin group was ~2.0 per 1000 person years (7), meaning the absolute benefits clearly exceeded the DKA risk.

The DAPA-CKD trial included participants with eGFR between 25-75 mL/min/1.73m², 68% of whom had type 2 DM (8). There was no increased risk of DKA in the SGLT 2 inhibitor arm of this trial (dapagliflozin 0/2152 vs placebo 2/2152).

The EMPA-KIDNEY trial included 6609 participants with eGFR between 20-65 mL/min/1.73m² of whom 46% had diabetes (including 68 participants with type 1 diabetes) (1). DKA was numerically more frequent in the empagliflozin group (6/3304, vs 1/3305 in the placebo group), though there were too few events to assess statistical significance. One DKA event occurred in a participant with type 1 DM, while 4 events occurred in people with insulin-treated type 2 DM.

Physiologically, as eGFR reduces, the amount of glucose filtered by the glomeruli also reduces (9). The glucose-lowering action of SGLT 2 inhibition is therefore limited by eGFR and one can argue that this could protect people from the glycosuria-induced metabolic changes that increase the risk of DKA. However, it is also established that as eGFR reduces, insulin and sulphonylurea (SU) clearance is reduced and the risk of hypoglycaemia from these medications increases (10, 11). Reactive reductions in insulin doses in order to reduce the risk of hypoglycaemia could conceivably have contributed to the risk of DKA seen in CREDENCE and other trials.

People without diabetes melliltus

Starvation ketoacidosis can occur in people without diabetes, but only one such case has been reported during about 30,000 participant years of follow-up of people with heart failure or CKD treated with an SGLT 2 inhibitor in the reported trials within the meta-analysis (1, 3).

Factors increasing risk of DKA

The Association of British Clinical Diabetologists have identified characteristics of people with type 2 DM that may place them at greater risk of developing DKA when using SGLT 2 inhibitors (15). These characteristics are highlighted in Table 5a.1. People with these characteristics may benefit from ketone monitoring, therefore, we suggest discussing with the diabetes team prior to initiating SGLT 2 inhibitors in such people (See Recommendation 5a.2.2).

Table 5a.1. People with type 2/3 DM at higher risk of DKA

5a.1.2 Sick day guidance

Medicines and Healthcare products Regulatory Agency (MHRA) reports of DKA suggest that concomitant illnesses such as vomiting, dehydration, reduced food intake, infection or a surgical procedure preceded some DKA events (17). Similar to the MHRA reports, the U.S Food and Drug Administration (FDA) statement in 2015 stated that 50% of the cases presenting with DKA were associated with precipitating events (18).

Following appropriate sick day guidance may significantly reduce the risk of developing DKA on SGLT 2 inhibitors.

It is therefore important that individuals initiated on SGLT 2 inhibitors are given sick day guidance on what to do in these situations.

Sick day guidance is highlighted below:

• Hold SGLT 2 inhibitor if unwell, restricted food intake, or dehydration
• Individuals on insulin treatment should always be advised never to stop or significantly reduce their insulin as part of the sick day response
• SGLT 2 inhibitor treatment should be interrupted in people who are hospitalised for surgical procedures or serious medical illnesses
• Treatment should be restarted when the person’s condition has recovered

During periods of planned restricted food intake (for example, fasting for Ramadan), we suggest following the guidance by the ADA/EASD 2020 consensus update on the management of DM during Ramadan. If unwell during fasting, ketone testing should be considered, and for the elderly, those with CKD or those on diuretics, consider stopping or reducing dose of SGLT 2 inhibitor during the period of fasting (19).

5a.2 Recommendations for Implementations

1. We recommend that people with type 1 DM should only have SGLT 2 inhibitors initiated under the strict direction of the diabetes team (see section 7a) (Grade 1C).
2. We recommend that people with type 2 DM at greater risk of DKA (defined in Table 5a.1) should have SGLT 2 inhibitors initiated with caution after discussion with the diabetes team (Grade 1C).
3. We recommend SGLT 2 inhibitors are discontinued when a patient develops DKA (Grade 1A).
4. We suggest that after an episode of DKA and where a clear contributing factor has been identified, there should be discussion with the person and clinical team to establish whether the benefits of re-introducing an SGLT 2 inhibitor outweigh the risks (Grade 2D).
5. When initiating SGLT 2 inhibitors, we suggest that individuals should be advised on the signs and symptoms of DKA and be instructed to temporarily withhold SGLT 2 inhibitors and to seek immediate medical advice if symptoms develop (Grade 1C).
6. We recommend always offering advice on sick day guidance when initiating SGLT 2 inhibitors and reminding them of this at every medication review (Grade 1C).
7. We suggest that individuals taking SGLT 2 inhibitors should be advised against following a ketogenic diet (Grade 2C).
8. We suggest that for people who choose to intermittently fast (e.g. for Ramadan), and particularly for those who are elderly, on diuretics or have CKD, consider withholding SGLT 2 inhibitors for the duration of the fasting period and for those people with diabetes ketone testing should be undertaken if unwell (Grade 2D).

Rationale: The evidence from the studies reviewed indicates that DKA is a recognised complication in people treated with SGLT 2 inhibitors and that it is more commonly found in conjunction with dehydration or infection. DKA is also likely to occur more frequently in people who are insulin deficient which would include people with type 1 DM, people with type 2 DM with a relative insulin deficient phenotype, and situations where people on insulin have their insulin dose reduced substantially. These recommendations will allow clinicians to use SGLT 2 inhibitors in those who are likely to benefit from this treatment and yet also minimise the risk of the complication of DKA.

5a.3 Audit Measures

1. The proportion of people with CKD on SGLT 2 inhibitors with evidence of provision of sick day guidance.
2. The proportion of people with CKD in whom SGLT 2 inhibitors were withheld during acute illness, and the proportion appropriately re-initiated on recovery.