Metformin Improves Glycemic Control and Postprandial Metabolism and Enhances Postprandial Glucagon-Like Peptide 1 Secretion in Patients With Type 2 Diabetes and Heart Failure: A Randomized, Double-Blind, Placebo-Controlled Trial Free

This was a single-center trial consisting of two treatment periods. Participants were instructed to adhere to ordinary lifestyle habits and avoid changes in food intake, alcohol consumption, and exercise during the study duration.

The trial was conducted in accordance with the principles of the Declaration of Helsinki and the International Conference on Harmonization Good Clinical Practice guidelines and was approved by local authorities. An independent ethics committee approved the clinical protocol. The study was registered with ClinicalTrials.gov (NCT01690091), and all participants provided written informed consent before study entry.

The study was a randomized, double-blind, placebo-controlled, crossover trial of 7 months’ duration testing the effects of 3 months of metformin therapy versus placebo. Patients were randomly assigned in a 1:1 ratio to receive either metformin or placebo. After 3 months of treatment, there was a 1-month washout period, and participants were then switched to the other study arm for an additional 3 months.

Adults with HF and reduced left ventricular ejection fraction (LVEF) who were 40–70 years of age with a BMI of 20–35 kg/m², an A1C between 5.8 and 8.1% (40 and 54 mmol/mol), and a new diagnosis of type 2 diabetes (previously untreated) were enrolled. HF was defined as known diagnosis of HF of >6 months’ duration with a history of hospitalization for HF at any time in the past (including treatment with parenteral diuretics), stable pharmacotherapy for >30 days, chronic diuretic treatment (thiazides or furosemide), and LVEF <50%. Type 2 diabetes was defined as medical history and treatment of type 2 diabetes or A1C ≥5.7% (39 mmol/mol) plus fasting plasma glucose ≥100 mg/dL (5.6 mmol/L) in venous plasma or A1C ≥5.7% (39 mmol/mol) plus random plasma glucose ≥140 mg/dL (7.8 mmol/L) in venous plasma, or plasma glucose measured at 120 minutes during an oral glucose tolerance test (OGTT) ≥140 mg/dL (7.8 mmol/L). An OGTT was indicated if only one of the criteria was met. Adequate treatment was initiated in cases of impaired glycemic control (A1C ≥65 mmol/mol [6.5%] or hyperglycemia ≥270 mg/dL [15 mmol/L]).

Individuals with planned cardiac surgery during the study were excluded from the trial, as were those with uncontrolled thyroid disease, those treated with insulin or oral hypoglycemic agents 2 months before the enrollment, pregnant or breastfeeding women, women capable of conceiving without adequate contraception, individuals with anemia (hemoglobin <100 g/L), as well as those with an estimated glomerular filtration rate <42 mL/min/1.73 m², atrial fibrillation, drug or alcohol abuse, or the presence of any other health condition that could, in the investigators’ opinion, distort the yield and consistency of the data.

Participants were randomly assigned in a 1:1 ratio to receive either metformin or placebo for 3 months. Then, after a 1-month washout period (116 half-lives for metformin), the medication was changed to the other treatment for another 3 months. Randomization was based on a computer-generated randomization protocol.

During treatment, the initial dose of metformin (Siofor 1,000 mg [Berlin-Chemie AG Menarini Group, Berlin, Germany]) or placebo (1,000 mg [Favea, Koprivnice, Czech Republic]) was 500 mg/day. After 1 week, the dose was increased to 1,000 mg/day. After 14 days, the dose was increased to 2,000 mg/day administered in two doses in the morning and the evening. Placebo tablets were identical in appearance and could not be distinguished from the metformin tablets.

The primary outcome was insulin sensitivity (glucose metabolic clearance rate [MCR]), measured by a hyperinsulinemic-isoglycemic clamp. The secondary outcomes included markers of cardiac function assessed by echocardiography and spiroergometry, postprandial secretion of intestinal peptides measured during a standard meal test, glycemic control assessed by A1C, and concentrations of TMAO.

All participants underwent a standardized panel of metabolic and cardiovascular tests three times (at baseline and at the end of the intervention periods at 3 and 7 months) in a highly controlled inpatient setting, as well as an initial medical examination, blood sampling, meal test, echocardiography, spiroergometry, and hyperinsulinemic-isoglycemic clamp. All participants fasted for a minimum of 8 hours before each study visit. Symptoms were scored using the standardized Minnesota Living with Heart Failure (MLHF) questionnaire (12).

Height and weight were measured using a stadiometer calibrated scale accurate to 0.1 kg. Blood pressure was measured in a seated position using a digital monitor (Omron HEM907XL, Omron, Kyoto, Japan) three times at 5-minute intervals. A mean value was calculated from the last two measurements.

A teflon cannula (Venflon; Viggo, Helsingborg, Sweden) was inserted into an antecubital vein for the infusion of all test substances. A second cannula was inserted into a wrist vein for blood sampling. A stepwise primed-continuous insulin infusion (1 mU/kg body weight/min of Actrapid HM [Novo Nordisk, Copenhagen, Denmark]) was administrated to acutely raise and maintain the plasma concentration of insulin at ∼75 μU/mL. Blood glucose levels during the clamps were maintained at 5.5–8 mmol/L (99–144 mg/dL) by continuous infusion of 15% glucose. Arterialized blood glucose levels were determined every 5 minutes, and the infusion rate was adjusted accordingly. Insulin sensitivity was estimated as MCR (i.e., the amount of glucose [mg/kg body weight/minute] needed to maintain a stable blood glucose level during the last 20 minutes of the clamp).

Blood samples were drawn in the fasting state and then at 30, 60, 120, and 180 minutes after a standard breakfast (cheese gourmet sandwich, 180 g, containing 452.8 kcal, including 49.2 g carbohydrate, 18.5 g protein, and 18.8 lipids, including 6.8 g saturated fatty acids, 6.0 g monounsaturated fatty acids, and 5.0 g polyunsaturated fatty acids).

A detailed quantification of the systolic and diastolic function of the ventricles, analysis of myocardial deformation (speckle-racking), and quantification of left atrial volume were performed (System Vivid 7, GE, SW Echopac). Left ventricular end-systolic elastance (Ees) was determined by the single-beat technique of Chen based on measured stroke volume, ejection fraction, blood pressure, and systolic time intervals (13). Effective arterial elastance (Ea) was estimated from end-systolic pressure/stroke volume. End-systolic pressure was estimated as 0.9 times the systolic blood pressure; ventriculo-vascular coupling was estimated by Ees/Ea ratio as previously reported (14).

This procedure was performed twice on 2 consecutive days during each hospitalization. Quantification of cardiovascular reserve used peak oxygen consumption (V_maxEncore 29S, load 25W/25W for 3 minutes).

Plasma glucose was analyzed using the Beckman Analyzer glucose-oxidase method (Beckman Instruments, Fullerton, CA), and A1C was assessed by a Bio-Rad Haemoglobin A1c Column Test (Bio-Rad Laboratories, Munich, Germany). Plasma immunoreactive insulin and C-peptide concentrations were determined using insulin and C-peptide immunoradiometric assay kits (Immunotech, Prague, Czech Republic). Plasma lipid levels were measured using enzymatic methods (Roche, Basel, Switzerland). HDL cholesterol was measured after double precipitation with dextran and magnesium chloride. LDL cholesterol was estimated using the Friedewald equation if triglycerides were within normal limits. Inflammatory markers and gastrointestinal hormones were determined by multiplex immunoanalyses based on the xMAP technology using a Milliplex Map Human Metabolic Hormone Magnetic Bead Panel (HMHEMAG-34K; Millipore, Billerica, MA) and Luminex 100 IS instrument (Luminex Corp., Austin, TX). The concentration of methylglyoxal was determined after derivatization with 1,2-diaminobenzene using a high-performance liquid chromatography method with fluorescence detection. TMAO was quantified with the use of ultraperformance liquid chromatography coupled to electrospray ionization tandem mass spectrometry.

Sample size was calculated with a power analysis for the primary outcome based on previous clamp studies in similar populations (15,16). Assuming a 25% improvement in baseline glucose MCR (estimate 4.0 ± 1.5 mg/kg/min) with α = 0.05 and β = 0.80, the estimated sample size was n = 37.

Data reported here are expressed as mean ± SD unless otherwise indicated. The effects of metformin versus placebo on measured variables were tested by Friedman ANOVA. Changes were calculated for placebo and metformin treatment compared with pre-treatment levels, as well as the difference between metformin and placebo. Spearman correlation coefficients were used to express the relationship between variables. All tests were two-sided, and P <0.05 was considered statistically significant.

Of 539 patients screened, 45 qualified and were randomized, with 39 completing the study (Supplementary Figure S1). The baseline values of all tracked outcomes and their changes in response to metformin and placebo in all study completers are shown in Table 1. The mean age of the study participants was 58 ± 8.4 years, 11% (5 of 45) were women, the mean BMI was 32.3 ± 5.1 kg/m², and the mean LVEF was 33.6 ± 12.0%.

Insulin sensitivity, measured as glucose MCR during hyperinsulinemic-isoglycemic clamp, tended to increase on metformin by 0.73 ± 0.97 mg/kg/min, but this increase was not significantly different from placebo (P = 0.074).

Compared with placebo, metformin did not affect clinical symptoms quantified by the MLHF questionnaire (Table 1). No change was observed in left ventricular end-diastolic dimensions, indicating a neutral effect of metformin on the eccentric remodeling of the left ventricle (LV). LVEF did not change, but Ees, which reflects contractility, slightly decreased after metformin compared with placebo (P = 0.026). There were no differences in systemic arterial load, measured by Ea, or ventricular-vascular coupling between metformin and placebo. The ratio E/é (the ratio between early mitral inflow velocity and mitral annular early diastolic velocity), which reflects the left ventricular filling pressure, also remained unchanged after both treatments, and there was no effect of metformin on brain natriuretic peptide (BNP) or left atrial volume (Table 1).

Exercise tolerance, as assessed by peak oxygen consumption (VO₂), slightly decreased during both interventions, with no difference between the treatments (P = 0.118; Table 1). Peak diastolic blood pressure decreased during both interventions compared with baseline, with no difference between the treatments (P = 0.88).

Fasting plasma glucose and A1C decreased on metformin, and these changes were significant compared with placebo, with effect sizes −7 ± 20 mg/dL (−0.38 ± 1.1 mmol/L, P = 0.001) and −0.30 ± 0.49% (−3.39 ± 5.33 mmol/mol, P <0.001), respectively (Figure 1A).

No significant change in TMAO levels was observed during either treatment, and no association was detected between changes in TMAO and changes in metabolic or cardiac functions. Postprandial metabolic control was evaluated using a standard meal test, and data were expressed as total area under the curve (AUC). Metformin significantly decreased the plasma glucose AUC compared with placebo (effect size −2.7 ± 4.2 mmol/L/h, P <0.001) (Figure 1B), as well as insulin AUC (effect size −47 ± 95.8 mIU/L/h, P <0.001). Postprandial C-peptide did not change during metformin therapy compared with placebo. Postprandial glucagon levels increased during metformin therapy, but the change was not significantly different from placebo (P = 0.062). Postprandial GLP-1 and PYY concentrations significantly increased during metformin therapy compared with placebo (effect sizes for AUC 53 ± 62.5, P <0.001 [Figures 1C and 2A], and 107.5 ± 164, P <0.001 [Figures 1D and 2B], respectively).

In general, the patients were clinically stable and were not developing cardiac decompensation during metformin therapy or placebo. Five of the six patients that discontinued treatment did so during the placebo phase, and one discontinued treatment during the metformin phase.

This single-center, randomized, placebo-controlled crossover study examined the effect of 3-month treatment with metformin on glycemic control, insulin resistance, and cardiac function in patients with type 2 diabetes and chronic HF. Compared with placebo, metformin treatment improved glycemic control, reduced postprandial glycemic and insulin response to a standard test meal, and increased postprandial secretion of the incretins GLP-1 and PYY. No significant changes in cardiac function or TMAO were observed, and there was no significant association between serum TMAO concentration and any of the measured metabolic or cardiorespiratory outcomes.

Insulin sensitivity tended to increase on metformin, but this increase was not significantly different from placebo. This study confirmed the efficacy of metformin in improving glycemic control and glucose metabolism, as well as its safety in patients with HF. Administration of metformin was accompanied by a decrease in A1C, a reduction in the postprandial glucose load measured as glucose AUC during a standard meal test, an increase in postprandial GLP-1 and PYY secretion, and a trend toward higher insulin sensitivity, measured by a hyperinsulinemic clamp.

Metformin has been shown to exert its major effects through inhibition of liver glucose production. Enhanced glucose disposal, reflecting the whole body sensitivity, has also been noted in some studies (17). Decreased postprandial insulin levels with preserved C-peptide concentrations during metformin therapy suggest changes in insulin clearance (18).

Although the mechanism of action of metformin is not completely understood, it improves glycemic control, reduces weight, and tends to reduce insulin resistance. Metformin has been previously shown to increase insulin sensitivity in people with HF who do not have diabetes (19), although its insulin-sensitizing effects were not confirmed in another trial (20). Even in patients with HF without diabetes, insulin resistance represents a strong independent predictor of mortality and is associated with worse symptoms and more impaired exercise capacity (21).

Short-term 3-month therapy with metformin was safe and did not result in detectable changes in cardiovascular structure and exercise capacity, with the exception of a small reduction of LV contractility quantified by Ees. This change may reflect reduced myocardial oxygen consumption after metformin, recently documented by a positron emission tomography study in patients with HF (22). It is likely that cardioprotective effects of metformin (3,4) result from multiple small pleiotropic targets, rather than from a single detectable modification of cardiac structure and function. Favorable changes in cardiovascular structure or function could be observed only after a longer duration of metformin therapy. The lack of the effects of metformin on cardiac structure and exercise tolerance does not rule out the possibility of beneficial effects on clinical events in the long term. Similarly, therapy with β-blockers in patients with HF also may not improve peak VO₂ or BNP, but still leads to robust reduction in mortality over time. In a previous randomized study in people with HF without diabetes, metformin improved myocardial efficiency by reducing myocardial oxygen consumption (22), but there was no effect of metformin on Vo_2max and cardiac structure and function in another study in people with HF without diabetes (19). Similarly, metformin had no effect on LV size and function in people with type 2 diabetes with a history of hypertension (23).

To the best of our knowledge, this is the first study testing the effect of metformin on the gut microbiome–related metabolite TMAO and evaluating its relationship with glucose metabolism and insulin sensitivity measured by the clamp technique in humans. We found no effect of metformin on TMAO levels. Previously, increased TMAO levels in individuals with HF have been associated with a 3.4-fold increase in mortality risk independent of cardiorenal function (11). However, large intra- and inter-individual variations in TMAO levels have been documented (24). Furthermore, elevated TMAO levels are more likely to be observed in HF with preserved ejection fraction than in HF with reduced ejection fraction (24) and seem to be increased in decompensated HF and congestion (25). Our results indicate that beneficial effects of metformin in HF are unlikely to be mediated by changes in intestinal microbiome–related production of TMAO.

This study demonstrated that metformin acts at least in part as a GLP-1 mimetic, increasing postprandial GLP-1 concentrations. Metformin’s glucose-lowering effects may be partly explained by increased GLP-1 secretion (9), mainly resulting from its direct and AMP-activated protein kinase–dependent effect on GLP-1–secreting L cells in the small intestine (26). Metformin has been previously shown to increase the secretion of another anorectic gut hormone, PYY, which is also produced by intestinal L cells, and these effects are believed to promote the modest weight loss observed with metformin (27).

The present results suggest that the primary effect of metformin action resides in the gut rather than in the circulation (28). Metformin modulates multiple components of the incretin axis and enhances expression of the GLP-1 receptor and related insulinotropic islet receptors through a mechanism requiring peroxisome proliferator–activated receptor-α (29). Moreover, metformin enhances the expression of the genes encoding the receptors for both GLP-1 and glucose-dependent insulinotropic polypeptide (GIP) in experimental models and also increases the effects of GIP and GLP-1 on insulin secretion from pancreatic β-cells (30). Finally, metformin has been shown to improve glycemic control by increasing incretins independent of changes in gluconeogenesis in young people with type 2 diabetes, suggesting an enteroinsular mechanistic pathway (31). Confirming the beneficial effects of increased GLP-1 levels in this population, GLP-1 analogs seem to be a promising therapy option for people with HF and type 2 diabetes (32,33).

The main strengths of this study include its randomized, placebo-controlled design and multifaceted, in-depth cardiometabolic testing, including the 3-hour standard meal test, which provided insights into the postprandial dynamics of GLP-1 after metformin use that were in line with previous research (26).

The study was from a single center, and the sample size is modest, which may have reduced its power, but there was little signal of direct cardiac effect other than the modest reduction in Ees. The duration of treatment was relatively short, so we cannot address the effects of longer-term treatment with metformin. Although a larger sample size may have increased statistical significance of some variables such as MCR, the clinical significance of such an effect likely would have been negligible.

Metformin was found to be safe and to improve glycemic control in patients with HF and type 2 diabetes, but there were few direct effects of metformin on cardiovascular structure and function. Metformin did not affect the intestinal microbiome–derived toxic metabolite TMAO, but enhanced secretion of gut-derived incretins such as GLP-1, which may explain favorable metabolic effects with this agent.

The authors thank the study participants for their contributions. They also thank Dr. Věra Lánská (Institute for Clinical and Experimental Medicine, Prague, Czech Republic) for statistical advice and acknowledge the skillful technical assistance of Ms. Dana Lapešová and Ms. Dagmar Šišáková (Institute for Clinical and Experimental Medicine). Additionally, they thank Favea (Koprivnice, Czech Republic) for providing the placebo for the study at no charge.

This study was supported by the project National Institute for Research of Metabolic and Cardiovascular Diseases (Programme EXCELES, ID Project No. LX22NPO5104) and was funded by the European Union–Next Generation EU, by grant MZO 00023001 from the Institute for Clinical and Experimental Medicine, and by grant IGA NT/14250-3 from the Ministry of Health, Prague, Czech Republic.

No potential conflicts of interest relevant to this article were reported.

V.M., T.P., and H.K. designed and conducted the study and wrote the manuscript. E.H., K.V., J.V., J.B., O.K., T.C., M.S., L.T., J.K. Jr., and J.K. Sr. administered the intervention, collected the data, and reviewed the manuscript. B.A.B. helped with data interpretation and reviewed the manuscript. M. Haluzík provided supervision and funding and reviewed the manuscript. M. Hill performed the statistical analysis and reviewed the manuscript. H.K. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.