Metformin for sepsis-associated AKI: a protocol for the Randomized Clinical Trial of the Safety and FeasibiLity of Metformin as a Treatment for sepsis-associated AKI (LiMiT AKI)

Abstract

Introduction

Acute kidney injury (AKI) is a common complication of sepsis associated with increased risk of death. Preclinical data and observational human studies suggest that activation of AMP-activated protein kinase, an ubiquitous master regulator of energy that can limit mitochondrial injury, with metformin may protect against sepsis-associated AKI (SA-AKI) and mortality. The Randomized Clinical Trial of the Safety and FeasibiLity of Metformin as a Treatment for sepsis-associated AKI (LiMiT AKI) aims to evaluate the safety and feasibility of enteral metformin in patients with sepsis at risk of developing SA-AKI.

Methods and analysis

Blind, randomised, placebo-controlled clinical trial in a single-centre, quaternary teaching hospital in the USA. We will enrol adult patients (18 years of age or older) within 48 hours of meeting Sepsis-3 criteria, admitted to intensive care unit, with oral or enteral access. Patients will be randomised 1:1:1 to low-dose metformin (500 mg two times per day), high-dose metformin (1000 mg two times per day) or placebo for 5 days. Primary safety outcome will be the proportion of metformin-associated serious adverse events. Feasibility assessment will be based on acceptability by patients and clinicians, and by enrolment rate.

Ethics and dissemination

This study has been approved by the Institutional Review Board. All patients or surrogates will provide written consent prior to enrolment and any study intervention. Metformin is a widely available, inexpensive medication with a long track record for safety, which if effective would be accessible and easy to deploy. We describe the study methods using the Standard Protocol Items for Randomized Trials framework and discuss key design features and methodological decisions. LiMiT AKI will investigate the feasibility and safety of metformin in critically ill patients with sepsis at risk of SA-AKI, in preparation for a future large-scale efficacy study. Main results will be published as soon as available after final analysis.

Trial registration number

NCT05900284.

Correspondence to Dr Hernando Gómez; [email protected]

STRENGTHS AND LIMITATIONS OF THIS STUDY

• Sepsis-associated acute kidney injury is an important clinical problem that could be addressed by repurposing an inexpensive, widely used medication with a track record of safety.

• The strong scientific foundation and preliminary data demonstrate biological plausibility and support the study hypothesis.

• The study design includes randomisation, concealment of treatment allocation and placebo control.

• The small planned sample size of the study will limit the conclusions related to efficacy.

• As a single-centre study, the results will need to be confirmed in a larger-scale multicentre study.

Introduction

Sepsis, a dysregulated immune response to an infection that results in life-threatening organ dysfunction, represents a large proportion of global deaths^{1 2} and is the leading cause of acute kidney injury (AKI). Sepsis-associated AKI (SA-AKI) accounts for roughly half of all cases of AKI,^{3 4} complicates the course of more than half of patients with sepsis,⁵ and is independently associated with increased mortality and morbidity.^{3 4 6–10} However, there are no specific treatments for patients with SA-AKI, and current recommendations are limited to supportive measures.³

We have shown that activation of the ubiquitous cellular energy master regulator AMP-activated protein kinase (AMPK) by metformin protects against AKI and death in an experimental model of sepsis,^{11 12} while pharmacological inhibition or genetic deletion of AMPK increases the severity of SA-AKI and mortality.^{11 12} At the cellular level, metformin modulates mitochondrial metabolism by transiently blocking complex I of the electron transport chain (ETC). Blockade of the ETC leads to an increase in the levels of AMP relative to the intracellular content of ATP. AMP competes with ATP and binds to the CBS3 site within the regulatory module of the heterotrimeric protein, causing a conformational change that favours the phosphorylation of the threonine residue by the upstream modulator, liver kinase B1, leading to the allosteric activation of AMPK.^13–15 It is unclear how AMPK or metformin may exert protection during sepsis. Potential mechanisms of AMPK-induced protection include cellular energy conservation, the activation of mitochondrial quality control processes such as mitophagy and biogenesis, limiting mitochondrial dysfunction,¹⁶ a key mechanism of organ dysfunction in sepsis.^{17 18} In addition, sepsis is known to impair renal fatty acid oxidation (FAO),¹⁹ which is one of the preferred metabolic pathways to derive energy in the renal tubules,^{20 21} thereby switching metabolism toward aerobic glycolysis and hindering mitochondrial respiration.^{12 19 22} Inhibition of aerobic glycolysis, and restoration of FAO and oxidative phosphorylation have been shown to be protective during sepsis.²³ AMPK is a potent promotor of mitochondrial β oxidation, the most important form of FAO, which may explain the protective effect during sepsis. Remote ischaemic preconditioning has been shown to protect from subsequent kidney injury by activating AMPK and inducing transient cell cycle arrest in tubular epithelial cells.²⁴ AMPK’s protective effect may be explained by its capacity to induce G1/S cell cycle arrest through phosphorylation of p53.²⁵ Finally, it is also possible that AMPK may protect from SA-AKI through non-renal mechanisms such as limiting endothelial and microvascular dysfunction,^{26 27} improved myocardial performance leading to improved systemic perfusion²⁸ or modulating the inflammatory response.²⁹

These findings appear to translate to the bedside because observational studies have shown a similar association.^{30 31} In the largest study to date, we showed in patients with sepsis that in-hospital exposure to metformin was associated with a ~30% decrease in the risk of moderate to severe AKI, ~50% decrease in the risk of death by day 90 and with a fourfold increase in the odds of recovery from AKI if already present.³² However, evidence remains observational, and a causal link in humans has not been established.

Based on this scientific premise, our overarching hypothesis is that treatment with metformin can decrease tubular stress and ultimately decrease the risk of developing or progressing to severe SA-AKI (figure 1) in patients with subclinical (ie, only biomarker positive without clinical criteria) or clinical SA-AKI (ie, positive Kidney Disease: Improving Global Outcomes (KDIGO) criteria).³ Accordingly, the objectives of the Randomized Trial of the Safety and FeasibiLity of Metformin as a Treatment for sepsis-associated AKI (LiMiT AKI) are to (1) investigate the safety and feasibility of administration of metformin to patients with sepsis, (2) investigate the effect of metformin on the risk of developing SA-AKI or progressing to severe SA-AKI, and (3) inform a future efficacy trial.

Figure 1. Graphical representation of the study hypotheses for safety and efficacy. AKI, acute kidney injury.

Methods and analysis

Design

LiMiT AKI will be a double blind, randomised, placebo-controlled clinical trial (ClinicalTrials.gov Registry NCT05900284). Execution and reporting of this trial will conform to the Standard Protocol Items for Randomized Trials (SPIRIT) statement.³³ The SPIRIT checklist and the full study protocol are presented in the online supplemental files 1 and 4.

Eligibility criteria

We will enrol adult patients with sepsis and with available oral or enteral access, within 48 hours of meeting Sepsis-3 criteria.¹ Detailed inclusion and exclusion criteria are presented in table 1. This study is designed as an interventional trial aiming to find a treatment for acute tubular stress and kidney injury. Preventive strategies for SA-AKI are problematic, largely because close to 50% of patients arriving in the emergency room with sepsis already have AKI.¹⁰ In addition, while the KDIGO classification has the unified definition of AKI, it still relies on serum creatinine, which is a delayed marker of kidney injury. The advent of novel biomarkers like tissue inhibitor of metalloproteinase 2 (TIMP2)/insulin-like growth factor-binding protein 7 (IGFBP7) have shown that significant tubular injury may be occurring hours before a rise in creatinine³⁴ and thus, it is very likely that patients with sepsis who are not diagnosed with AKI by KDIGO criteria may already be developing tubular injury at a subclinical level. In recognition of this scenario, the Acute Disease Quality Initiative (ADQI) consensus conference on biomarkers has proposed a new category, subclinical AKI or stage 1S, where tubular injury can occur as evidenced by positive (TIMP2·IGFBP7) in the absence of traditional KDIGO criteria.³⁵ The majority of patients with sepsis without AKI who will develop SA-AKI will do so within the first 24 hours,⁵ and thus, many of these patients may already have stage 1S upon admission. Based on these data, we decided against only including patients with SA-AKI, as we hypothesised that treatment with metformin early in the course of sepsis can limit tubular stress and injury, and therefore decrease the risk of progression to severe forms of SA-AKI (ie, from KDIGO stage 1S to 1 or from 2 to 3) in patients with subclinical (ie, only TIMP2·IGFBP7 positive without clinical criteria) or clinical AKI (ie, positive KDIGO criteria).

Table 1

Inclusion and exclusion criteria

Inclusion criteria	Age 18 years or older
	Diagnosis of sepsis by Sepsis-3 criteria1
	Available enteral access (oral or feeding tube)
Exclusion criteria	Estimated glomerular filtration rate <30 mL/min/1.73 m2 calculated by CKD-EPI 2021 equation, within 30 days prior to entering study57
	Not expected to survive more than 24 hours
	Advanced directive to withhold life-sustaining treatment
	Metformin use in the last 30 days from admission (assessed by medical or refill prescription history, and by medication reconciliation)
	The treating clinician believes that participation in the trial would not be in the best interests of the patient
	Known or suspected pregnancy
	On mechanical circulatory support of any kind
	History of allergy to metformin listed in medical records or from patient or family report
	Severe metabolic acidosis with venous or arterial pH <7.20, with PaCO2 <45 or PvCO2 <50 mm Hg at the time of enrolment

CKD-EPI, Chronic Kidney Disease Epidemiology Collaboration; PaCO2, Arterial partial pressure of carbon dioxide; PvCO2, Venous partial pressure of carbon dioxide.

Location

The study will be conducted in Presbyterian hospital at the University of Pittsburgh Medical Center (UPMC), Pittsburgh, Pennsylvania.

Blinding and concealment of allocation

We will guarantee concealment using an automated, centralised system that will randomly assign participants to treatment groups and a unique participation ID. Patients, treating physicians, outcome assessors, data collectors and analysts will be blinded to treatment group allocation. We will assign a third party to handle any unblinded information, including delivering specific samples to co-investigators for processing and who otherwise will not be involved in any aspect of the study’s execution.

Investigational drug services at UPMC will provide a placebo that looks identical to the low-dose and the high-dose metformin preparations and will deliver study treatments following the allocation sequence. For patients with feeding tubes, investigational drug services will also deliver either metformin or placebo tablets, crushed in a ‘Silent Knight’ pouch (McKesson, Irving, Texas, USA) to the bedside nurse along with instructions to reconstitute using 10 mL of normal saline 0.9%.

Study arms

Patients will be allocated to one of three arms to receive enteral low-dose (500 mg) or high-dose (1000 mg) metformin, or placebo two times per day for 5 days starting on the day of enrolment. Doses have been selected based on prior studies involving critically ill patients undergoing cardiac surgery or with traumatic brain injury.^{36 37} Figures 2 and 3 show the study flow and methodological details, including sampling, randomisation, follow-up and analysis.

Figure 2. Schematic representation of the anticipated patient flow through the study including biosampling schedule and study drug administration. AKI, acute kidney injury.

Figure 3. Study flow representing methodological details including screening, consent and enrolment, randomisation, initiation of study procedures and follow-up. EHR, electronic health record; ICUs, intensive care units; LAR, legally authorised representative.

Randomisation

Eligible patients will be randomised to any of the three study arms in a 1:1:1 ratio using a computer-generated permuted block scheme with block size of balancing interval, varying randomly from blocks of 3 to 6 subjects.

Background care

All patients will receive background care as per the treating team. The study personnel will only provide the study drug to the bedside nurse, collect data and biological specimens, and in no way will interfere with care of the patient.

Statistical analyses

Sample size calculation

LiMiT AKI will test the primary hypothesis that the proportion of metformin-associated serious adverse effects (mSAEs) are different between the metformin groups and control. We have considered four key mSAEs and ranked them in order of clinical priority as follows: hyperlactatemia, metabolic acidosis, hypoglycaemia and gastrointestinal (GI) intolerance (see below for specific definitions). The analysis of the primary outcome will be based on the mSAEs that are blindly adjudicated to be associated with the study intervention by the study’s steering committee. We have chosen to use mSAEs as a composite outcome because in critical illness, serious adverse effects (SAEs) are very common which increases the signal-to-noise ratio in the data, making it less likely to find meaningful differences.³⁸ In addition, traditional reporting of composite outcomes may be problematic because these methodologies are very sensitive to the first (and likely the most frequent) event, and risk dismissing the presentation of other less frequently occurring events. For this study, this would result in mSAE rates driven by the presence of GI intolerance (the most frequent mSAE) while ignoring the potential effect of the intervention on other more clinically relevant mSAEs that can lead to death like hyperlactatemia and metabolic acidosis. To overcome these limitations, we will use the win ratio to estimate the difference in mSAEs between study groups.³⁹ By ranking mSAEs based on their clinical importance, the most frequent but less life-threatening mSAE (GI discomfort) will only be compared between pairs of patients after the most clinically relevant mSAEs (ie, hyperlactatemia and metabolic acidosis) have been compared.³⁹

We calculated a sample size of 60–80 to have a power of 80% to detect a win ratio of 2.5:3.0 with various proportions of ties for all mSAEs based on a 1:1:1 metformin low dose:metformin high dose:placebo control recruitment ratio (table 2). The rates of GI disturbances in patients with diabetes treated with metformin ranged from 1.5 to 2.5 times that of patients using other medications, and higher when compared with placebo.⁴⁰ We considered that the addition of the other potential adverse events (ie, hyperlactatemia, acidosis, hypoglycaemia) can increase the rate of events in the metformin arm beyond 2.8. While the risk ratio is calculated based on the risk of events in the intervention group over the risk of events in the control group, the win ratio is calculated as the ratio of the events in the intervention group (wins for metformin), over the events in the control group (losses for metformin). Thus, while the risk ratio and the win ratio are not the same, the win ratio will resemble the risk ratio of the composite of adverse events occurring in patients receiving metformin when compared with placebo. Because the risk ratio for GI disturbances can be as high as 2.8 when compared with placebo, we considered that the addition of the other potential metformin-associated adverse events (ie, hyperlactatemia, acidosis, hypoglycaemia) can increase the rate of events in the metformin arm beyond this risk ratio, and therefore, that a win ratio of 2.5:3 is in line with the literature and will account for all high-risk adverse events potentially associated with the study intervention.

Table 2

Sample size calculation using the win ratio (WR) with multiple possibilities of ties

Proportion of ties	Detectable WR
	WR=2.5	WR=2.7	WR=3
0.1	60	52	41
0.2	74	64	52
0.3		78	64
0.4			80

Interim analysis

Data will be submitted to an independent Data and Safety Monitoring Board (DSMB) at 50% recruitment (40 patients) or based on stopping rules as described below, to perform the only planned interim analysis.

Intention to treat

All outcomes will be analysed under the intention-to-treat (ITT) principle. Safety outcomes will be also assessed per-protocol analysis. If the rate of major protocol deviations (eg, non-compliance, crossover or loss-to-follow-up) is greater than 5%, we will also perform a complier average causal effect analysis to address the limitations of ITT.⁴¹

Primary outcome

Our primary safety outcome is the number of patients who complete the treatment with study drug (at either dose) without any mSAE. We will report the number and proportion of patients with mSAE per group adjudicated by the steering committee to be associated with the study intervention and the win ratio. For the analysis of the primary outcome, we will compare mSAEs between patients exposed to metformin (including patients from the low-dose and high-dose arms) and patients receiving placebo. Blood samples for monitoring mSAEs will be obtained every 12 hours for the first 7 days in the study (online supplemental table 1).

We will use the following definitions for each mSAE: (a) hyperlactatemia, plasma lactate ≥8 mmol/L; (b) metabolic acidosis, arterial or venous pH ≤7.20 with an arterial partial pressure of carbon dioxide (PaCO₂)<45 mm Hg or PvCO₂<50 mm Hg; (c) hypoglycaemia, serum or capillary glucose ≤60 mg/dL; (d) GI intolerance, defined as the need to hold enteral nutrition due to nausea/vomiting, the need for anti-nausea medication like ondansetron, prochlorperazine, two times in a 6-hour period, or more than five bowel movements/day, after the administration of at least three doses of metformin. We chose a lactate of ≥8 mmol/L first, because it is within the lactate concentration range of 2.2–12.7 mmol/L, reported in patients with sepsis exposed to metformin,³¹ and because ~94% of the these patients will have a lactate level <5 mmol/L and ~99% a lactate <10 mmol/L.⁴² Second, because the incremental rate of risk of death is similar between lactate levels of 5 and 8 mmol/L, but steeper after 10 mmol/L.⁴² GI intolerance will be monitored by reviewing electronic medical records and discussing with the treating team. The presence of nausea, emesis or diarrhoea will be assessed both in the chart and by direct communication with the bedside nurse two times per day by the study team. If GI intolerance persists for >24 hours and is deemed to be associated with the study drug by clinical adjudication, we will proceed to dose de-escalation as follows: patients allocated to a high dose will shift to the low-dose regimen, and those in the low-dose regimen will shift to 250 mg two times per day. After 24 hours of monitoring, treatment will be withheld if GI intolerance persists, and the symptoms are deemed to be associated with the study drug by the steering committee. If symptoms abate, the intervention will be continued at this reduced dose to complete 5 days. For all other non-GI mSAEs, interruption of the study intervention will only occur if blinded clinical adjudication by the steering committee determines a potential association with the study drug.

Secondary outcomes

Safety

Any SAE, defined as life-threatening, resulting in death, or prolonged hospitalisation, disability, or permanent damage, will be adjudicated and differences will be evaluated between groups.^{43 44} If an SAE is considered to be potentially associated with use of metformin, emergency unblinding will be considered. Emergency unblinding may be performed by the investigator in case this knowledge is essential for management or welfare of the patient. If the blind is broken for any reason, the investigator will proceed with notification of DSMB and recording date and reason in the appropriate case report form.

Feasibility

We will assess feasibility stratified by five categories as shown in online supplemental table 2. Our benchmark for feasibility will consist of (1) recruitment of ≥2.5 patients per month, (2) adherence to study protocol ≥90%, (3) complete data acquisition in >95%, (4) follow-up to discharge >90% of patients and (5) missing data per variable for the primary outcome <5%.

Recruitment, retention and adherence

We will quantify process evaluation outcomes such as the proportion of eligible patients, the proportion of eligible patients who are randomised, attrition rate, adherence rates in relation to treatment allocation and study procedures and the number of patients who complete treatment.

Acceptability

We will identify perceived barriers by physicians and patients’ legally authorised representative (LAR) in the implementation of metformin as a treatment in patients with sepsis. A study team member will obtain feedback from physicians and LARs who decline to participate in the study in the form of one open question: What is the primary reason for declining to participate or opting out? Data will be tabulated and presented by themes after analysis by the research team. Our benchmark for acceptability will be ≥75% of treating physicians willing to allow patient enrolment, of ≥50% of patients or LAR willing to discuss the study with the research team.

Data accrual and follow-up

We will quantify the number of patients with complete data collection, the proportion of data loss per collected variable, the proportion of patients lost to follow-up and the reason for such loss.

Efficacy

We will determine the effect size of treatment with metformin on kidney tubular stress and platelet mitochondrial function, as follows:

Kidney tubular stress: measured using a commercial assay (bioMerieux, Marcy-l’Etoile, Auvergne-Rhone-Alpes, France) that combines two Food and Drug Administration (FDA)-approved urinary biomarkers (TIMP2·IGFBP7). We will measure (TIMP2·IGFBP7) from baseline to day 3 in all patients and calculate the area under the curve defined by the sum of the concentration of three measurements multiplied by time in hours from baseline to day 3.

Platelet mitochondrial function: will be assessed in a subset of patients by evaluating the platelet bioenergetic profile, platelet ETC complex expression and activity and AMPK activation.

The platelet bioenergetic profile: will be assessed as described by Cardenes et al.⁴⁵ After isolating platelets by differential centrifugation in the presence of prostaglandin I2, the platelet bioenergetic profile will be assessed using the XF24 Seahorse metabolic analyser (Agilent, Santa Clara, California, USA) to quantify basal oxygen consumption rate, proton leak respiration and maximal respiratory capacity as described previously.⁴⁵

The platelet mitochondrial ETC complex (I–V) expression: will be measured by densitometric quantification of western blots using a LICOR system normalised to integrin αIIb. The activity of complexes I–V and citrate synthase will be measured by spectrophotometric kinetic analysis as in Cardenes et al.⁴⁵

Platelet AMPK testing: platelet AMPK will be tested as a surrogate for systemic AMPK modulation by metformin. Platelet total AMPK and phosphorylated AMPK (ie, activated) will be measured by western blot by densitometric quantification using a LICOR system normalised to β-actin as described before.¹²

Mortality: in-hospital mortality will be assessed until discharge or 30 days, whichever occurs first.

AKI incidence, staging, recovery and duration: we hypothesise that the use of metformin early in the course of sepsis would serve the purpose of limiting tubular stress and injury, and therefore decrease the risk of progression to severe forms of SA-AKI (ie, from KDIGO stage 1S–subclinical–to stage 1, or from 2 to 3). We will compare the development of AKI daily for 7 days and at discharge or 30 days from enrolment, whichever comes first, between groups. AKI will be defined as per KDIGO criteria, using the worst daily creatinine and urine output values.⁴⁶ SA-AKI will be defined as per the 28th ADQI consensus report as meeting KDIGO criteria for AKI within 7 days of fulfilling Sepsis-3 criteria.³ We will compare the rates of transient AKI, persistent AKI and acute kidney disease, and AKI recovery as previously defined,^47–49 and the need for initiation of any modality of renal replacement therapy between groups.

Pharmacokinetic study

We will obtain complete pharmacokinetic profiles in a subset of 20 random patients exposed to metformin (10 patients from the low and high-dose groups each). We will exclude patients with jejunal tube feeding because the bulk of metformin absorption occurs in the small intestine and is negligible in the stomach and colon.⁵⁰ Quantification of plasma metformin concentrations will be processed at the University of Pittsburgh Small Molecule Biomarker Core. Serial drug concentrations will be used to assess single-dose absorption (day 1), followed by trough accumulation (days 2–4) and elimination (days 5–7) after multiple dosing. Metformin pharmacokinetic parameters, including area under the plasma concentration-time curve over the dosing interval, maximum plasma concentration, apparent oral clearance, volume of distribution, elimination rate constant and the corresponding terminal elimination half-life, will be calculated using non-compartmental methods.

Follow-up

Patients will be followed up to hospital discharge or 30 days, whichever occurs first.

Ethics and dissemination

The study protocol is approved by the University of Pittsburgh Institutional Review Board (IRB no. STUDY22120032), without a required Investigational New Drug exemption from the FDA. Eligible patients or LAR will be counselled on the trial procedures and opt to participate or decline enrolment (please see informed consent in the online supplemental file 2). Two independent groups were established to provide oversight to the study, the steering committee and the DSMB.

Patient and public involvement

Patients and/or the public were not involved in the design, conduct, reporting or dissemination plans in this study.

Discussion

Designing LiMiT AKI presented several challenges. First, while metformin carries a small, although real risk of hyperlactatemia and metabolic acidosis, the spectre of metformin’s frequent association with severe lactic acidosis remains ingrained in the medical community despite a safety track record of more than 60 years. This concern is at least in part anchored on the unfavourable reputation that phenformin, a close relative of metformin, earned due to the frequent association with lethal lactic acidosis which led to its removal from the market in the 1970s. Based on clinical studies demonstrating safety in at-risk populations,^{51 52} the FDA recently expanded its recommendation to use metformin in patients with mild to moderate kidney dysfunction. While there is no precedent of the use of metformin in sepsis, a robust body of observational data suggests that exposure to metformin is associated with lower mortality, regardless of the magnitude of hyperlactatemia when compared with patients not exposed to metformin.^{30 31 42} Based on this, but cognisant of the small but potential hazard in patients with shock and AKI, we designed the study to minimise the patient’s risk by (1) delivering our intervention as a short 5-day course, (2) excluding patients with severe acute and chronic kidney disease, and (3) adopting a rigorous, frequent monitoring schedule (online supplemental table 1).

Second, we wrestled with the definition of our primary safety outcome. Critical illness is fraught with SAE, and thus the noise generated by the high event rate when using all SAEs may not allow for differences to be teased apart between study groups. To address this issue, we used a ‘signal-to-noise’ adverse event improvement strategy³⁸ used in prior trials^{53 54} that reduces the ‘noise’ by only focusing on adverse events known to be associated with the intervention in hand (ie, mSAEs for metformin). Furthermore, recognising that mSAEs can still occur for reasons other than metformin (ie, lactic acidosis during a seizure), we designed the primary analysis to only include mSAEs that are adjudicated as being associated with the study treatment by a blinded adjudication panel of experts. An additional consideration is that, in terms of overall risk, developing severe metabolic acidosis may be more relevant than GI discomfort and thus, we decided to use the win ratio, which is a method that enables a stepwise analysis that combines outcomes incorporating risk hierarchy.⁵⁵ Of the existing methods capable of this type of analysis, we chose the win ratio to overcome important limitations of traditional methodologies assessing composite outcomes. In particular, traditional methodologies will count the first occurring event within the composite outcome (ie, GI discomfort), and dismiss whether or not other events occurred (ie, hyperlactatemia). The win ratio prevents this by allowing the investigator to rank each component of the composite outcome by clinical importance, and then sequentially and hierarchically counts the occurrence of each component starting with the most clinically relevant.⁵⁵

Third, there is no parenteral formulation for metformin. Critically ill patients with sepsis and/or shock may have GI dysmotility and changes in absorption and/or volume of distribution resulting in unpredictable systemic exposure of metformin.⁵⁶ We addressed these unknowns using four strategies: (1) we only included patients with available enteral access; (2) we designed a detailed pharmacokinetic study that investigates the three critical determinants of drug exposure and response—metformin absorption, elimination and accumulation; (3) we paired the pharmacokinetic study with two pharmacodynamic tests, one at the systemic level by assessing metformin-induced platelet mitochondrial ETC blockade, and a second one at the organ level, by assessing tubular kidney stress biomarkers and clinical SA-AKI; and (4) we included monitoring for GI dysfunction as part of the primary safety outcome.

Fourth, our choice of estimated glomerular filtration rate (eGFR) as an exclusion criterion was pragmatic, but also because it is a better method of gauging risk of drug toxicity than KDIGO stage, which is better at gauging risk of death and extent of kidney injury. Consider for example two patients. Patient A is a young female with a baseline serum creatinine of 0.5 mg/dL and an eGFR >120 mL/min/1.73 m², who presents with sepsis and serum creatinine of 1.5 mg/dL (KDIGO stage 3) but her eGFR is still 50. Conversely, patient B is an elderly woman with underlying chronic kidney disease whose baseline serum creatinine is 1.5 mg/dL and an eGFR 35, who now presents with a serum creatinine of 2.3 (stage 1) but her eGFR is now 22. In point of fact, since these are both patients with AKI and not in steady state, the true GFR might be much lower for both patients, but in either case, eGFR is a better measure of their risk of toxicity than KDIGO stage. Indeed, patient B would be much higher risk despite being in stage 1 compared with patient A in stage 3.

Fifth, we anticipate that metformin will not be effective in every patient with sepsis, but rather, in patients in whom mitochondrial dysfunction and/or metabolic reprogramming away from oxidative phosphorylation are a culprit of tissue injury and organ dysfunction. We designed the study to include a biorepository of plasma, urine and platelets that will allow us to address this question by (a) characterising the metabolic endotypes (ie, specific mechanisms leading to organ dysfunction, in this case, mitochondrial dysfunction or metabolic reprogramming) that may be most responsive to metformin treatment; and (b) by evaluating potential biomarkers that can identify the subset of patients in whom metformin is most effective.

In summary, the available preclinical and observational evidence, the strong biological plausibility and the long-standing safety track record suggest metformin is a promising therapeutic for SA-AKI. It is on this backdrop that LiMiT AKI becomes the next step necessary to inform the design of a future, targeted efficacy randomised clinical trial that will determine if clinicians will have, in metformin, a specific treatment for SA-AKI for the first time. The main results from LiMiT AKI will be published as soon as available.

The authors wish to thank the LiMiT AKI team including Ali Smith, Howard Stein, Scot King, Stephanie Montgomery, Kelly Urbanek, Michael Muir, Darla Mcgivern, Karen Nieri and Angelina Kendi for their hard work. Also, the BDMC team in the Department of Critical Care Medicine, Thomas Nolin for his contributions to the design of the PK studies and IDS at UPMC, Cynthia Nutter and Jen Kozar.

Ethics statements

Patient consent for publication

Not applicable.

Footnote

Contributors Manuscript preparation—IES, NH, DTH and HG. Protocol planning and writing—DTH, SLK-G, RMR, SS, TDN, XC, JM, C-CHC, XL and HG. Study design and planning—DTH, SLK-G, RMR, SS, TDN, XC, JM, C-CHC, XL, JK and HG.

Funding This work was funded by the National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases (1R01DK133142-01A1) and by bioMérieux (N/A). IES has received NIH funding through T32HL007820.

Competing interests None declared.

Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

Provenance and peer review Not commissioned; externally peer reviewed.

Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.

References

¹ Singer M, Deutschman CS, Seymour CW, et al. The third International consensus definitions for sepsis and septic shock (Sepsis-3). JAMA 2016; 315: 801–10. doi:10.1001/jama.2016.0287

² Rudd KE, Johnson SC, Agesa KM, et al. Global, regional, and national sepsis incidence and mortality, 1990-2017: analysis for the global burden of disease study. Lancet 2020; 395: 200–11. doi:10.1016/S0140-6736(19)32989-7

³ Zarbock A, Nadim MK, Pickkers P, et al. Sepsis-associated acute kidney injury: consensus report of the 28th acute disease quality initiative workgroup. Nat Rev Nephrol 2023; 19: 401–17. doi:10.1038/s41581-023-00683-3

⁴ Hoste EAJ, Bagshaw SM, Bellomo R, et al. Epidemiology of acute kidney injury in critically ill patients: the multinational AKI-EPI study. Intensive Care Med 2015; 41: 1411–23. doi:10.1007/s00134-015-3934-7

⁵ White KC, Serpa-Neto A, Hurford R, et al. Sepsis-associated acute kidney injury in the intensive care unit: incidence, patient characteristics, timing, trajectory, treatment, and associated outcomes. A multicenter, observational study. Intensive Care Med 2023; 49: 1079–89. doi:10.1007/s00134-023-07138-0

⁶ Liu J, Xie H, Ye Z, et al. Rates, predictors, and mortality of sepsis-associated acute kidney injury: a systematic review and meta-analysis. BMC Nephrol 2020; 21: 318. doi:10.1186/s12882-020-01974-8

⁷ Bagshaw SM, Uchino S, Bellomo R, et al. Septic acute kidney injury in critically ill patients: clinical characteristics and outcomes. Clin J Am Soc Nephrol 2007; 2: 431–9. doi:10.2215/CJN.03681106

⁸ Cruz MG, Dantas J de O, Levi TM, et al. Septic versus non-septic acute kidney injury in critically ill patients: characteristics and clinical outcomes. Rev Bras Ter Intensiva 2014; 26: 384–91. doi:10.5935/0103-507X.20140059

⁹ Al-Jaghbeer M, Dealmeida D, Bilderback A, et al. Clinical decision support for in-hospital AKI. J Am Soc Nephrol 2018; 29: 654–60. doi:10.1681/ASN.2017070765

¹⁰ Kellum JA, Chawla LS, Keener C, et al. The effects of alternative resuscitation strategies on acute kidney injury in patients with septic shock. Am J Respir Crit Care Med 2016; 193: 281–7. doi:10.1164/rccm.201505-0995OC

¹¹ Escobar DA, Botero-Quintero AM, Kautza BC, et al. Adenosine monophosphate-activated protein kinase activation protects against sepsis-induced organ injury and inflammation. J Surg Res 2015; 194: 262–72. doi:10.1016/j.jss.2014.10.009

¹² Jin K, Ma Y, Manrique-Caballero CL, et al. Activation of AMP-activated protein kinase during sepsis/inflammation improves survival by preserving cellular metabolic fitness. FASEB J 2020; 34: 7036–57. doi:10.1096/fj.201901900R

¹³ Zhou G, Myers R, Li Y, et al. Role of AMP-activated protein kinase in mechanism of metformin action. J Clin Invest 2001; 108: 1167–74. doi:10.1172/JCI13505

¹⁴ Foretz M, Guigas B, Bertrand L, et al. Metformin: from mechanisms of action to therapies. Cell Metab 2014; 20: 953–66. doi:10.1016/j.cmet.2014.09.018

¹⁵ Rena G, Hardie DG, Pearson ER. The mechanisms of action of metformin. Diabetologia 2017; 60: 1577–85. doi:10.1007/s00125-017-4342-z

¹⁶ Herzig S, Shaw RJ. AMPK: Guardian of metabolism and mitochondrial homeostasis. Nat Rev Mol Cell Biol 2018; 19: 121–35. doi:10.1038/nrm.2017.95

¹⁷ Brealey D, Karyampudi S, Jacques TS, et al. Mitochondrial dysfunction in a long-term rodent model of sepsis and organ failure. Am J Physiol Regul Integr Comp Physiol 2004; 286: R491–7. doi:10.1152/ajpregu.00432.2003

¹⁸ Singer M. The role of mitochondrial dysfunction in sepsis-induced multi-organ failure. Virulence 2014; 5: 66–72. doi:10.4161/viru.26907

¹⁹ Li Y, Nourbakhsh N, Pham H, et al. Evolution of altered tubular metabolism and mitochondrial function in sepsis-associated acute kidney injury. Am J Physiol Renal Physiol 2020; 319: F229–44. doi:10.1152/ajprenal.00390.2019

²⁰ Forbes JM, Thorburn DR. Mitochondrial dysfunction in diabetic kidney disease. Nat Rev Nephrol 2018; 14: 291–312. doi:10.1038/nrneph.2018.9

²¹ Bhargava P, Schnellmann RG. Mitochondrial energetics in the kidney. Nat Rev Nephrol 2017; 13: 629–46. doi:10.1038/nrneph.2017.107

²² Feingold KR, Wang Y, Moser A, et al. LPS decreases fatty acid oxidation and nuclear hormone receptors in the kidney. J Lipid Res 2008; 49: 2179–87. doi:10.1194/jlr.M800233-JLR200

²³ Yang L, Xie M, Yang M, et al. PKM2 regulates the Warburg effect and promotes HMGB1 release in sepsis. Nat Commun 2014; 5: 4436. doi:10.1038/ncomms5436

²⁴ Rossaint J, Meersch M, Thomas K, et al. Remote ischemic preconditioning causes transient cell cycle arrest and renal protection by a NF-kappaB-dependent Sema5B pathway. JCI Insight 2022; 7: e158523. doi:10.1172/jci.insight.158523

²⁵ Jones RG, Plas DR, Kubek S, et al. AMP-activated protein kinase induces a P53-dependent metabolic checkpoint. Mol Cell 2005; 18: 283–93. doi:10.1016/j.molcel.2005.03.027

²⁶ Escobar DA, Botero-Quintero AM, Kautza BC. Adenosine monophosphate-activated protein kinase activation protects against sepsis-induced organ injury and inflammation. J Surg Res 2015; 194: 262–72. doi:10.1016/j.jss.2014.10.009

²⁷ Castanares-Zapatero D, Bouleti C, Sommereyns C, et al. Connection between cardiac vascular permeability, myocardial edema, and inflammation during sepsis. Critical Care Medicine 2013; 41: e411–22. doi:10.1097/CCM.0b013e31829866dc

²⁸ Inata Y, Piraino G, Hake PW, et al. Age-dependent cardiac function during experimental sepsis: effect of pharmacological activation of AMP-activated protein kinase by AICAR. Am J Physiol Heart Circ Physiol 2018; 315: H826–37. doi:10.1152/ajpheart.00052.2018

²⁹ Huang J, Liu K, Zhu S, et al. AMPK regulates immunometabolism in sepsis. Brain Behav Immun 2018; 72: 89–100. doi:10.1016/j.bbi.2017.11.003

³⁰ Liang H, Ding X, Li L, et al. Association of preadmission metformin use and mortality in patients with sepsis and diabetes mellitus: a systematic review and meta-analysis of cohort studies. Crit Care 2019; 23: 50. doi:10.1186/s13054-019-2346-4

³¹ Tan K, Simpson A, Huang S, et al. The association of premorbid metformin exposure with mortality and organ dysfunction in sepsis: a systematic review and meta-analysis. Crit Care Explor 2019; 1: e0009. doi:10.1097/CCE.0000000000000009

³² Gómez H, Del Rio-Pertuz G, Priyanka P, et al. Association of metformin use during hospitalization and mortality in critically ill adults with type 2 diabetes mellitus and sepsis. Crit Care Med 2022; 50: 935–44. doi:10.1097/CCM.0000000000005468

³³ Chan A-W, Tetzlaff JM, Altman DG, et al. SPIRIT 2013 statement: defining standard protocol items for clinical trials. Ann Intern Med 2013; 158: 200–7. doi:10.7326/0003-4819-158-3-201302050-00583

³⁴ Kashani K, Al-Khafaji A, Ardiles T, et al. Discovery and validation of cell cycle arrest biomarkers in human acute kidney injury. Crit Care 2013; 17: R25. doi:10.1186/cc12503

³⁵ Ostermann M, Zarbock A, Goldstein S, et al. Recommendations on acute kidney injury biomarkers from the acute disease quality initiative consensus conference: a consensus statement. JAMA Netw Open 2020; 3: e2019209. doi:10.1001/jamanetworkopen.2020.19209

³⁶ Taheri A, Emami M, Asadipour E, et al. A randomized controlled trial on the efficacy, safety, and pharmacokinetics of metformin in severe traumatic brain injury. J Neurol 2019; 266: 1988–97. doi:10.1007/s00415-019-09366-1

³⁷ El Messaoudi S, Nederlof R, Zuurbier CJ, et al. Effect of metformin pretreatment on myocardial injury during coronary artery bypass surgery in patients without diabetes (Metcab): a double-blind, randomised controlled trial. Lancet Diabetes Endocrinol 2015; 3: 615–23. doi:10.1016/S2213-8587(15)00121-7

³⁸ Cook D, Lauzier F, Rocha MG, et al. Serious adverse events in academic critical care research. CMAJ 2008; 178: 1181–4. doi:10.1503/cmaj.071366

³⁹ Pocock SJ, Ariti CA, Collier TJ, et al. The win ratio: a new approach to the analysis of composite endpoints in clinical trials based on clinical priorities. Eur Heart J 2012; 33: 176–82. doi:10.1093/eurheartj/ehr352

⁴⁰ Nabrdalik K, Skonieczna-Żydecka K, Irlik K, et al. Gastrointestinal adverse events of metformin treatment in patients with type 2 diabetes mellitus: a systematic review, meta-analysis and meta-regression of randomized controlled trials. Front Endocrinol (Lausanne) 2022; 13: 975912. doi:10.3389/fendo.2022.975912

⁴¹ Peugh JL, Strotman D, McGrady M, et al. Beyond intent to treat (ITT): a complier average causal effect (CACE) estimation primer. J Sch Psychol 2017; 60: 7–24. doi:10.1016/j.jsp.2015.12.006

⁴² Posma RA, Frøslev T, Jespersen B, et al. Prognostic impact of elevated lactate levels on mortality in critically ill patients with and without preadmission metformin treatment: a Danish registry-based cohort study. Ann Intensive Care 2020; 10: 36. doi:10.1186/s13613-020-00652-0

⁴³ FDA. What is a serious adverse event? [ U.S. Food and Drug Administration ]. 2023. Available: https://cacmap.fda.gov/safety/reporting-serious-problems-fda/what-serious-adverse-event

⁴⁴ James EC, Dunn D, Cook AD, et al. Overlap between adverse events (AEs) and serious adverse events (SAEs): a case study of a phase III cancer clinical trial. Trials 2020; 21: 802. doi:10.1186/s13063-020-04718-z

⁴⁵ Cardenes N, Corey C, Geary L, et al. Platelet bioenergetic screen in sickle cell patients reveals mitochondrial complex V inhibition, which contributes to platelet activation. Blood 2014; 123: 2864–72. doi:10.1182/blood-2013-09-529420

⁴⁶ Khwaja A. KDIGO clinical practice guidelines for acute kidney injury. Nephron Clin Pract 2012; 120: c179–84. doi:10.1159/000339789

⁴⁷ Hoste E, Bihorac A, Al-Khafaji A, et al. Identification and validation of biomarkers of persistent acute kidney injury: the RUBY study. Intensive Care Med 2020; 46: 943–53. doi:10.1007/s00134-019-05919-0

⁴⁸ Chawla LS, Bellomo R, Bihorac A, et al. Acute kidney disease and renal recovery: consensus report of the acute disease quality initiative (ADQI) 16 workgroup. Nat Rev Nephrol 2017; 13: 241–57. doi:10.1038/nrneph.2017.2

⁴⁹ Kellum JA, Sileanu FE, Bihorac A, et al. Recovery after acute kidney injury. Am J Respir Crit Care Med 2017; 195: 784–91. doi:10.1164/rccm.201604-0799OC

⁵⁰ Graham GG, Punt J, Arora M, et al. Clinical pharmacokinetics of metformin. Clin Pharmacokinet 2011; 50: 81–98. doi:10.2165/11534750-000000000-00000

⁵¹ Crowley MJ, Diamantidis CJ, McDuffie JR, et al. Clinical outcomes of metformin use in populations with chronic kidney disease, congestive heart failure, or chronic liver disease: a systematic review. Ann Intern Med 2017; 166: 191. doi:10.7326/M16-1901

⁵² Roumie CL, Chipman J, Min JY, et al. Association of treatment with metformin vs sulfonylurea with major adverse cardiovascular events among patients with diabetes and reduced kidney function. JAMA 2019; 322: 1167–77. doi:10.1001/jama.2019.13206

⁵³ Ranieri VM, Thompson BT, Barie PS, et al. Drotrecogin alfa (activated) in adults with septic shock. N Engl J Med 2012; 366: 2055–64. doi:10.1056/NEJMoa1202290

⁵⁴ National Heart L, Moss M, Huang DT, et al. Early neuromuscular blockade in the acute respiratory distress syndrome. N Engl J Med 2019; 380: 1997–2008. doi:10.1056/NEJMoa1901686

⁵⁵ Yu RX, Ganju J. Sample size formula for a win ratio endpoint. Stat Med 2022; 41: 950–63. doi:10.1002/sim.9297

⁵⁶ Morales Castro D, Dresser L, Granton J, et al. Pharmacokinetic alterations associated with critical illness. Clin Pharmacokinet 2023; 62: 209–20. doi:10.1007/s40262-023-01213-x

⁵⁷ Inker LA, Eneanya ND, Coresh J, et al. New creatinine- and Cystatin C-based equations to estimate GFR without race. N Engl J Med 2021; 385: 1737–49. doi:10.1056/NEJMoa2102953