Editorial 

June 14, 2024

Sodium-Glucose Cotransporter 2 Therapy for Acute Organ Dysfunction in Critically Ill Patients

Hernando Gómez, Lennie P. G. Derde

JAMA. Published online June 14, 2024. doi:10.1001/jama.2024.10171

Multisystem organ dysfunction can follow several acute injury mechanisms, including trauma, sepsis, and shock and is the leading cause of death in critically ill patients.¹ Even after the initial cause of organ injury has been addressed, many organs remote from the initial injury remain “on strike” for days or even weeks, often despite little evidence of direct injury. There are no specific treatments to prevent or reverse this organ dysfunction, so supportive care is the cornerstone of treatment in the intensive care unit (ICU).

Long-term treatment with sodium-glucose cotransporter 2 (SGLT-2) inhibitors decrease cardiovascular death, heart failure–related hospitalizations, and progression of chronic kidney disease (CKD) regardless of the presence of diabetes or baseline renal function.^2,3 Data from large randomized trials also suggest that long-term SGLT-2 inhibition may reduce the incidence of acute organ dysfunction, raising the possibility of benefit in critically ill patients.^2,3 For example, in patients with CKD and significant albuminuria, dapagliflozin decreased the risk of acute decline in kidney function in a prespecified post hoc analysis of the Dapagliflozin and Prevention of Adverse Outcomes in Chronic Kidney Disease (DAPA-CKD) trial.⁴ Several meta-analyses report a decrease in acute kidney injury in patients with diabetes treated with SGLT-2 inhibitors compared with other oral medications for diabetes or placebo.^5,6

Typically, the benefits of SGLT-2 inhibitor therapy accrue after prolonged exposure. The potential mechanisms leading to organ protection are incompletely understood. In chronic SGLT-2 inhibitor treatment for patients with diabetes, the reduction in intraglomerular hypertension and hyperfiltration through the inhibition of the tubuloglomerular feedback is key.⁷ This mechanism may be protective in patients with sepsis with disrupted intrarenal hemodynamics, and critically ill patients may benefit from decreasing hyperfiltration because this may additionally decrease renal hypoxia through reduced tubular epithelial oxygen consumption. At a cellular level, SGLT-2–independent effects including a reduction in local inflammation and inhibition of mediators of maladaptive repair and fibrosis have been described.^8,9 Finally, SGLT-2 inhibition can initiate signaling through known protective pathways, including reduced oxidative stress, and the activation of mitochondrial autophagy and biogenesis.^10-12

Whether short-term therapy would yield benefit is unclear, but among patients with COVID-19 who were at risk of cardiometabolic disease, a short-term course of dapaglifozin (median days, 5.5) was well tolerated compared with placebo, although it did not improve outcomes.¹³ A pilot case-control study using empagliflozin in critically ill patients for a median of 8 days showed no difference in the development of relevant adverse events such as urinary tract infections, ketoacidosis, or hypoglycemia, suggesting SGLT-2 inhibitors are safe.¹⁴ However, no large randomized trials have tested whether the acute administration of SGLT-2 inhibitors ameliorates acute organ dysfunction in critically ill patients, and it remains unclear whether these medications are safe, which patients should be targeted, and whether short-term treatment achieves organ protection.

In this issue of JAMA, Tavares and colleagues¹⁵ report on the DEFENDER trial, a multicenter, randomized, open-label trial assessing whether the addition of dapagliflozin to standard care for up to 14 days compared with standard care alone improved outcomes in critically ill patients with acute organ dysfunction. The study was conducted in 22 centers in Brazil between November 2022 and August 2023. The primary outcome was a composite of hospital mortality, initiation of kidney replacement therapy, and ICU length of stay through 28 days. The authors used the win ratio, a statistical method that ranks each component of the composite outcome based on clinical relevance (ie, death is more important than ICU length of stay), and then uses this ranking to determine the occurrence (losses) or lack-thereof (wins) of the primary outcome. More than 500 participants were recruited, 248 randomized to dapagliflozin and 259 to standard care. Treatment with dapagliflozin did not result in a higher number of wins compared with standard care for the primary outcome (42.3% vs 41.9%; win ratio, 1.01; 95% CI, 0.90-1.13; P = .89). A bayesian secondary analysis showed that dapagliflozin may be associated with a lower adjusted risk of kidney replacement therapy initiation, with a posterior probability of 90%. There were no differences between treatment groups in the preplanned subgroup analyses, in the post hoc analyses, or in the investigator-reported serious adverse events.

The results of the DEFENDER trial are reassuring in terms of the safety of dapaglifozin in critically ill patients, but important questions remain. The authors enrolled a heterogeneous cohort of critically ill patients. Though multiple organ dysfunction is common in critical illness, the underlying pathogenesis may not be due to a common pathway. If SGLT-2 inhibition can improve organ dysfunction, it may do so only in a subset of patients with critical illness. For example, SGLT-2 inhibitors appear to have anti-inflammatory properties, and therefore may work most effectively in patients such as those with sepsis in whom excessive host inflammation contributes to organ dysfunction.^16,17 Absent a clearer understanding of which patients may benefit, one solution may be to enroll patients in a far larger trial, designed and powered to interrogate potential heterogeneity of treatment effect.¹⁸That dapaglifozin was well-tolerated and appeared safe provides valuable information in support of such a trial.

Arguably, the organ dysfunction most likely to have been improved by SGLT-2 inhibition therapy would have been acute kidney injury. The outcomes measured in this trial were the need for kidney replacement therapy and the related outcome of kidney replacement therapy–free days. Using the need for kidney replacement therapy as an outcome works well in study populations at high risk of moderate to severe acute kidney injury.¹⁹ However, this study enrolled patients at low risk of acute kidney injury (stage 1 as defined by the Kidney Disease: Improving Global Outcomes [KDIGO]).

Although the authors conducted a bayesian post hoc analysis that suggested a decreased risk of kidney replacement therapy, their prespecified frequentist analysis failed to demonstrate any difference (10.9% vs 15.2%, odds ratio [OR], 0.76; 95% CI, 0.51-1.18; P = .90).

Another challenge with kidney replacement therapy initiation as an outcome is that it may not relate to the severity and trajectory of organ dysfunction or the response to therapy, but rather, to the practice pattern of specific sites. The authors addressed this issue by conducting a stratified analysis of the win ratio by creatinine level at randomization and a post hoc analysis using major adverse kidney events during the first 5 days as the outcome. However, creatinine levels measured at a single time point do not discriminate between acute and chronic kidney disease and therefore provide little information about severity of the acute organ dysfunction. In addition, the outcome of major adverse kidney events is often assessed beyond day 30, in part because it is likely that at earlier time points (like day 5) some of the major adverse kidney events components (ie, death or need for kidney replacement therapy) may not yet have occurred.²⁰ Therefore, in the absence of other outcomes that characterize the type, severity, or trajectory of kidney dysfunction, initiation of kidney replacement therapy alone may have missed the potential beneficial effect of this therapy, and more subtle measures of kidney injury may have been preferable in this population.

The authors administered SGLT-2 inhibition therapy orally. Although critically ill patients often present with impaired gastrointestinal absorption, there is increasing recognition that many therapies can be effectively administered orally in the ICU setting. Hopefully, that is the case here, but it would have been helpful if the authors had been able to measure dapagliflozin concentrations or specific pharmacodynamic surrogates of drug action.

The authors deserve congratulations for a brave attempt to pave the way for novel therapies to protect organs from injury and to reverse organ dysfunction during critical illness. The DEFENDER trial is the first large, multicenter randomized trial demonstrating that the use of dapagliflozin is safe in a broad, heterogeneous population of critically ill patients. Despite the absence of an efficacy signal, these findings set the stage for future studies investigating the effectiveness of dapagliflozin in critically ill populations. However, the question remains as to whether the lack of effect reported in the trial is due to true lack of effect, the heterogeneity of the study population, inadequate surrogate measures of organ dysfunction, insufficient dapagliflozin uptake, or a combination of these factors. Given the lack of existing therapies to treat acute organ dysfunction, the indirect evidence supporting a role for SGLT-2 inhibitors in modifying the course of organ injury and dysfunction, particularly in the kidney, and the evidence from DEFENDER that the therapy appeared safe, there will hopefully be further investigations into the role of these therapies in critical illness.