Main text

Metformin is the most widely used oral antihyperglycemic agent. Because it is eliminated unmodified in urine, patients with renal insufficiency can accumulate metformin and may develop lactic acidosis [1]. Recent guidelines only restrict the use of metformin in patients with severe chronic kidney disease (CKD) because the benefit is considered larger than the risk for lactic acidosis [2]. Lactate measurement has a central role in identifying and monitoring critical illness [3]. A better understanding of the impact of metformin on lactate levels could improve clinical assessment of the critically ill.

Data were collected by combining data from Danish nationwide medical databases with laboratory data [4]. This multicenter cohort included all adults (≥ 18 years) hospitalized and surviving 24 h of intensive care unit (ICU) treatment in northern Denmark between January 2010 and August 2017. We required ≥ 3 lactate measurements between 6 h before until 24 h after ICU admission, with ≥ 12 h between first and last measurement. Patients receiving dialysis before ICU admission were excluded.

Metformin use was defined as a filled prescription for metformin within 90 days before ICU admission [4]. CKD stage was assessed by the mean estimated glomerular filtration rate (eGFR) 365 days until 7 days before ICU admission [5]. Acute kidney injury (AKI) within 24 h after ICU admission was defined and staged according to the KDIGO creatinine criteria. Lactate trajectories over time for metformin users and nonusers were fitted by a mixed-effects model assuming unstructured covariance and including individual-level random intercept and slope. Time was modeled as a natural cubic spline with knot locations at − 1 h, + 4 h, and + 12 h relative to ICU admission. Time-by-group interaction was entered as a covariate, and analyses were subsequently stratified by eGFR level or AKI stage. Differences in maximum lactate level with 95% confidence intervals between metformin users and nonusers were model-based.

We studied 20,741 patients with a total of 209,394 lactate measurements, of whom 1905 (9%) patients used metformin (Table 1). Compared with nonusers, metformin users had a similar preadmission eGFR but had more often AKI stage 2 or 3. Metformin users had 0.61 (0.45–0.77) mmol/L higher maximum lactate levels than nonusers (Fig. 1a). This difference was highest for patients with eGFR ≤ 45 ml/min/1.73 m2 (1.06 [0.72–1.39] mmol/L; Fig. 1b). Differences in maximum lactate levels between metformin users and nonusers were more pronounced in patients with AKI stage 2 or 3 (Fig. 1c), with a difference of 0.30 (0.15–0.45) mmol/L for patients without AKI, and 0.12 (− 0.24 to 0.48), 1.00 (0.35–1.65), and 1.75 (1.03–2.47) mmol/L among patients with AKI stage 1, 2, or 3, respectively. The difference between metformin users and nonusers disappeared within 24 h of ICU admission. However, the time until this difference disappeared was longer for patients with moderate to severe CKD or AKI (Fig. 1).

Table 1 Characteristics of metformin users and nonusers
Full size table
Fig. 1
figure 1

Lactate levels for metformin users and nonusers according to estimated glomerular filtration rate and acute kidney injury stage. Mean lactate trajectories over time with 95% confidence interval for metformin users and nonusers were fitted by a mixed-effect model with individual-level random intercept and slope. Time after ICU admission was modeled as natural cubic spline with knot location at − 1 h, + 4 h, and + 12 h surrounding intensive care unit admission. a Total population. Subsequently, analyses were stratified according to b chronic kidney disease stage based on mean estimated glomerular filtration rate (eGFR) 1 year before ICU admission or c acute kidney injury (AKI) stage within 24 h of ICU admission

Full size image

In this large cohort of critically ill patients, metformin users had higher lactate levels than nonusers in the early phase of critical illness, which disappeared within 24 h of ICU admission. Importantly, the difference in lactate levels between metformin users and nonusers was higher in patients with more severe AKI, while the difference was almost similar across preadmission eGFR subgroups. This may be explained by reduced clearance of metformin or lactate, or both. A limitation is that blood metformin concentrations were unavailable to confirm this because such correlation was found in patients receiving renal replacement therapy for metformin-associated lactic acidosis [6].

The monitoring of lactate trajectories is recommended during critical illness [3]. Awareness of factors affecting this biomarker will improve its interpretation. We report that the association of metformin use with increased lactate levels is more pronounced in patients who develop AKI stage 2 or 3 than in patients without AKI or who develop AKI stage 1.