Evaluating Safe Ketone Thresholds To Minimise Ketosis in People With Type 1 Diabetes Using Dapagliflozin

Sodium glucose cotransporter 2 (SGLT2) is responsible for approximately 90% of glucose re-absorption in the renal proximal tubules, so SGLT2 inhibition results in significant excretion of glucose in the urine. SGLT2 inhibition may play a key role in the prevention and management of the cardiovascular-kidney-metabolic (CKM) syndrome as this class of agent has been shown to benefit glucose levels (reducing HbA1c, and glycaemic variability without increasing severe or total hypoglycaemia), reduce heart failure, improve weight control, reduce cardiovascular mortality, and provide reno-protection in people with diabetes thereby addressing some of the fundamental issues contributing to the syndrome. While most data pertain to agents which inhibit SGLT2 in people with T2D, SGLT2 inhibition also has the potential to substantially benefit glucose control, weight management, and reduce complication development including that of the CKM syndrome in people with T1D. Dapagliflozin is a SGLT2 inhibitor. Data available in adults with T1D indicate that Dapagliflozin improves glucose control without increasing hypoglycaemia and reduces glycaemic variability. In addition, there are improvements in body weight and blood pressure. There is also substantial evidence in people with T2D that Dapagliflozin lowers mortality from cardiovascular causes and hospitalisations for heart failure. In terms of reno-protective effects, Dapagliflozin was demonstrated to slow down decline in renal function, reduce rates of renal failure and death from renal causes in T2D.These cardiorenal benefits may extend to people with T1D.

While SGLT2 inhibitors as adjunctive therapy to insulin may benefit people with T1D, in general their use has been associated with a significantly increased risk of diabetic ketoacidosis (DKA) which may occur in the absence of hyperglycaemia. DKA represents a life-threatening acute emergency of T1D, which occurs when insulin deficiency leads to lipolysis and accumulation of ketones, resulting in metabolic acidosis. It requires prompt identification and action, and delays in management are associated with an increase in mortality. For example, the pooled analysis of DEPICT-1 and DEPICT-2 participants showed that over 52 weeks of treatment, the incidence of adjudicated DKA was numerically higher with Dapagliflozin 5 mg/day or 10 mg/day than with placebo (4.0% and 3.5% vs. 1.1%). Therefore, SGLT2 inhibitors in general have failed to receive approval in the USA (FDA) and Australia (TGA) for use to control glucose levels in people with T1D due to an unacceptable increase in DKA. It is possible that SGLT2 inhibitor therapy may be viable in people with T1D if an intervention is able to shift the risk-benefit balance in favour of benefit e.g. if the risk of DKA could be recognized and addressed in a timely manner. The early recognition of impending DKA is of critical importance. However, symptoms of ketosis which include nausea, vomiting, fatigue, loss of appetite, malaise, weakness, and tachypnoea appear late and those associated with dehydration may be absent in the setting of normoglycaemia. Therefore, timely ketone measurement represents a cornerstone of management allowing therapeutic measures to be initiated.

A continuous ketone sensor (CKS) would address many of the shortcomings associated with current standard-of-care blood ketone testing using a handheld meter. Ketones can be quantified in interstitial fluid using a similar approach to that taken with widely commercially used continuous glucose monitor (CGM), using an enzymatic reaction. While feasibility of CKS devices has previously been demonstrated, the optimum thresholds for alerts, balancing the burden of alarms vs. timely intervention, have yet to be determined. Preliminary at home CKS data provided by Abbott indicate free-living people with T1D who are not using SGLTi spend <1% of the time with ketone levels >1.0mmol/L. Ketone levels in free-living with T1D using SGLT2 inhibitors remains unknown. To provide insights into appropriate ketone thresholds for alerts a literature search was undertaken in May 2024.

Studies included were published between January 2000 and April 2024 involving those aged ≥16 years of age which described capillary β-OHB levels in relation to suspected DKA and which reported sensitivity, specificity, negative predictive value, or positive predictive value. Eight of the 11 studies provided capillary β-OHB levels that excluded DKA; these ranged from 0.7 to ≥3.0 mmol/L 13-23 and the most recommended cut-off level was 1.0 mmol/L (n=3) followed by 1.5 mmol/L (n=2). The β-OHB cut-offs for the diagnosis of DKA ranged from 1.8 mmol/L to 3.5 mmol/L, with 3.0 mmol/L being most proposed by five of the studies. The investigators therefore concluded that capillary β-OHB cut off values of <1.0 mmol/L and <1.5 mmol/L exclude DKA. The investigators also conclude that a capillary β-OHB level ≥3.0 mmol/L exhibited high sensitivity and specificity in detecting DKA with levels between 1.5 mmol/L and 3.0 mmol/L representing a transition range with increasing risk.

The investigators propose that to pre-empt DKA in people with T1D using SGLTi that responses to CKS data could be initiated at levels of either 1.0mmol/L or 1.5mmol/L with alerts aligned accordingly. The lower alert may provide an additional safety margin though could come at the cost of a greater number of alarms and carbohydrates eaten that may not have necessarily been required. The investigators also suggest that there may be high DKA-risk subgroups (female sex, lower baseline body mass index (BMI), β-hydroxybutyrate levels pre-treatment, use of an insulin pump, and lower insulin requirements at baseline with dose reductions required post SGLT inhibitor commencement to avoid hypoglycaemia) who may benefit from the lower alarm thresholds facilitating earlier intervention. Finally, the risk for DKA may vary according to illness, fasting and high intensity exercise requiring alerts to be adjusted according to circumstances. However, data are required to inform protocols.