Dapagliflozin modulates plasma lipidomic profile and urinary metabolite excretion in type 2 diabetes

Abstract

Background

Dapagliflozin (DAPA) has shown major nephroprotective effects, improving kidney metabolism and oxigenation. Lipidomics and metabolomics are powerful tools for understanding such effects, providing a comprehensive look at how SGLT2 inhibitors might change the metabolic landscape beyond their primary glucose-lowering action. We investigated changes in plasma metabolomic/lipidomic profile and urinary excretion of metabolites that could occur independent of increased diuresis.

Methods

A two-armed, parallel-design, randomized clinical trial was conducted in subjects with type 2 diabetes and hypertension who received treatment with DAPA 10 mg/day or hydrochlorothiazide 12.5 mg/day for four weeks. Lipidomics and metabolomics were performed by high resolution mass spectrometry in fasting plasma and 24-hour urine samples collected before and after treatment.

Results

Compared to hydrochlorothiazide, DAPA significantly increased plasma isoleucine, methionine, citrate, β-hydroxybutyrate and decreased lactate. DAPA induced plasma lipid remodeling towards a significant raise in free fatty acids (FFAs) and some sphingomyelins and lysophosphatidylcholines containing these fatty acids. A significant change was observed in plasma medium- and short-chain acylcarnitines, positively correlated with changes in plasma FFAs and β-hydroxybutyrate. In addition, DAPA, but not hydrochlorothiazide, significantly increased 24-h urinary excretion of several amino-acids, lactate, TCA cycle metabolites, β-hydroxybutyrate and electrolytes, except for a decrease in malate excretion.

Conclusions

DAPA treatment has major effects on the plasma lipidomic and the urine metabolomic profiles, with significant increased renal excretion of several metabolites, especially amino-acids, independently of increased diuresis. These data offer insights into the complex metabolic pathways leading to kidney protection by SGLT2 inhibitors.

Clinical Trial Information

European Union Drug Regulating Authorities Clinical Trials No. 2015-004164-11.

Graphical abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s12933-025-03018-7.

Research Insights

What is currently known about this topic?

• Besides the increased diuresis, Dapagliflozin can modify the amino acid and lipid metabolism.

What is the key research question?

• What is the impact of 4-week dapagliflozin on plasma and urine metabolite and lipid profile?

What is new?

• Dapagliflozin raised amino acids and TCA substrates urinary excretion independently of diuresis.

• Dapagliflozin induced plasma lipid remodeling towards more unsaturated fatty acids.

• Dapagliflozin modulates acetylcarnitines associated with fatty acid transport and oxidation.

How might this study influence clinical practice?

Understanding the metabolic pathways modulated by dapagliflozin sheds light on its nephroprotective properties.

Background

Type 2 diabetes (T2D) is a serious, chronic disease characterized by elevated blood glucose concentrations and high cardiovascular risk affecting more than 500 million adults worldwide [1]. Sodium-glucose co‐transporter‐2 inhibitors (SGLT2i) are a class of anti-hyperglycemic agents that inhibit glucose reabsorption in the proximal tubules of the kidney, with established metabolic, cardiovascular and renal benefits [2]. The main action of SGLT2i is the decrease in glucose concentrations through its urinary excretion, but other effects have been described, like the increased diuresis as a consequence of glycosuria [3],a decrease in insulin levels and an increase in glucagon concentration. This leads to an increased lipolysis and free fatty acids (FFAs) mobilization [4], and the enhancement of β-oxidation and ketogenesis, with elevated plasma ketone bodies concentration as a consequence of increased FFAs to the liver [3, 5]. Moreover, SGLT2i increase endogenous glucose production [3] mainly through the stimulation of gluconeogenesis (GNG) and glycogenolysis, with a greater contribution of GNG [6], although the increase in endogenous glucose production (EGP) seems to be independent of the small raise of glucagon induced by gliflozins [7].

With the advancement of powerful technologies in biological research, as omics techniques like lipidomics and metabolomics, it is possible to investigate and identify metabolic changes associated with SGLT2i, to better understand not only their complex and multifaceted mechanism of action, but also to balance advantages and contraindications in a proper management of T2D and its complications [8].

However, the effect of SGLT2i on the metabolomic profile, especially on metabolites associated with the dysregulation of glucose and amino acid metabolism, or on lipids associated with increased cardiovascular risk [9], remains largely unexplored. Moreover, the alterations in the lipidomic profile are not necessarily associated to the degree of metabolic control [10].

Only a few studies have looked at the effect of SGLT2i on the metabolomic profile of plasma [11, 12] and urine [13–15] in patients with type 1 diabetes or T2D, vs. insulin treatment or placebo. The effect of SGLT2i on the lipidomic profile has been investigated only in rodents [16–18]. However, it remains to be elucidated whether these effects on the plasma metabolome are mediated by improved glycemia, decrease in body weight or by increased diuresis [3]. Thus, we evaluated the effect on glucose, amino acid and lipid metabolism of four weeks treatment with dapagliflozin (DAPA) in subjects with T2D with adequate glucose control and hypertension compared with the effects obtained with hydrochlorothiazide (HCT) as control.

Methods

Study protocol and subject characteristics

This is a post hoc analysis of a previously published study [19] in which subjects with T2D were randomized to treatment with an SGLT2i (Dapagliflozin, DAPA, 10 mg/die) or a thiazide diuretic (Hydrochlorothiazide, HCT, 12.5 mg/die) for 4 weeks. Each participant was instructed to keep a stable Na intake (92 mmol/day) and follow a standard isocaloric diet. The study protocol (EudraCT 2015-004164-11) has been previously reported in detail [19]. From the study cohort, which comprised n = 40 subjects (DAPA n = 20 and HCT n = 20), we included in this analysis all subjects for whom at least one paired (baseline and post-intervention) sample (either plasma, urine, or both) was available, i.e., 33 (DAPA n = 17 and HCT n = 16) having paired plasma samples and 30 (DAPA n = 17 and HCT n = 13) having paired 24 h urine samples collected before and after 4 weeks of treatment (Supplementary Figure S1).

Plasma lipids, metabolomic and lipidomic profile

Total cholesterol, high-density lipoprotein (HDL) cholesterol and low-density lipoprotein (LDL) cholesterol, was measured as previously described.

The metabolomics and lipidomics analyses, which include both circulating metabolites and lipids, were carried out by high resolution mass spectrometry as previously reported [10, 20]. Briefly, the lipidomic analyses were performed by liquid chromatography/quadrupole time-of-flight mass spectrometry with electrospray ionization (UHPLC-ESI-QTOF-MS, 1290 Infinity-6545 Agilent Technology, Santa Clara, CA, USA). The following lipid species were quantified: triacylglycerols (TAGs), ceramides (CERs), sphingomyelins (SMs), phosphocholines (PCs) and lysophosphatidylcholines (LPCs) using an Agilent ZORBAX Eclipse Plus C18 for. The spectra were analysed using the Agilent Mass Hunter Profinder B.08.00 software and concentrations of targeted lipids were calculated by relating the peak area of each lipid species to the peak area of the corresponding internal standard (TAG(C45:0), CER(18:1/17:0), SM(d18:1/17:0), PC(C34:0), LPC(C17:0), Avanti Polar Lipids, Alabaster, AL and Larodan, Solna, SE).

Acylcarnitines were analyzed by UHPLC-ESI-QTOF-MS using a BEH C18 column for chromatographic separation using labelled internal standard (CIL Cambridge, MA, USA) for quantification.

Polar metabolites concentrations and free fatty acid (FFA) composition were measured by gas chromatography-triple quadrupole (GC-QQQ Agilent Technologies, Santa Clara, CA, USA).

Urine metabolomic profile

For the quantification of urinary concentrations of amino acids and organic acids, 1 ml of urine was first centrifuged for 20 min at 13,000 rpm at 4 °C to remove sediments. Then, 100 µL of Urease type III (Sigma- Aldrich^® U1500) was added to each sample (50ul of urine) and incubated for 1 h at 37 °C.

Subsequently, the same internal standards in the same quantities used for plasma polar metabolites quantification, were added to urine with 700 µL of cold methanol and then centrifuged at 4 °C for 20 min at 13,000 g. The supernatant was evaporated to dryness under nitrogen and derivatized in two-steps, identically to plasma samples (methoxyamination and derivatization with TBDMS) and analyses were performed by GC-MS/MS as described above. Metabolite concentrations were then corrected for 24 h diuresis and expressed as µmol excreted during the 24 h.

Calculations and statistical analysis

Concentrations of very low-density lipoprotein (VLDL) cholesterol was calculated according to Sampsom et al. [21] using the formula: TAG/8.59 + TAG * Non-HDL/2250 – TAG/16500). eGFR was calculated through the CKD-EPI formula.

The DNL lipogenic index was calculated as the ratio between the plasma concentrations of palmitic acid/linoleic acid [10]. The indexes of fatty acid desaturation, a surrogate marker of hepatic stearoyl-CoA desaturase 1 activity/expression, were calculated as the ratio between the plasma concentrations of palmitoleic acid/palmitic acid (SCD1₁₆) and oleic acid/stearic acid (SCD1₁₈) [10].

The Glutamate-Serine-Glycine (GSG) index, which combines three amino acids involved in glutathione synthesis and hepatic lipotoxicity, was calculated as the ratio of glutamate/(serine + glycine) [20].

Descriptive statistics are presented as numbers with proportions for categorical variables and as the mean ± standard error for continuous variables. Change after the treatment were reported as Log₂-transformed fold changes [Log₂ (Post/Pre)]. Pairwise differences between groups were evaluated by Mann-Whitney’s test and pre vs. post comparisons were conducted by paired Wilcoxon signed rank test. Same statistical tests were used for heatmaps which have been created reporting data as mean within the groups of interest. For correlation analysis the non- parametric Spearman test has been used to evaluate the association among different variables. All statistical analysis was performed using R Statistical Software (version 4.0.5) and StatView, SAS Institute Inc.

Data and resource availability

Data are reported in the supplementary material and available from the guarantors of the work upon reasonable request.

Results

Clinical characteristics

Clinical characteristics of the study participants are shown in Table 1. They had a mean age of 60.7 ± 1.4 years, mean BMI 30.7 ± 1.1 kg/m², with a prevalence of the male sex. BMI was higher in the HTC group. Metabolic control was similar in the two groups. Liver enzymes (aspartate transaminase, AST, alanine transaminase, ALT, and gamma-glutamyl transferase, GGT) were within the normal range before and after the administration of the two drugs. Total cholesterol, HDL cholesterol, LDL cholesterol, VLDL and TAGs concentrations were within the normal range. At the end of the four weeks of treatment, BMI was slightly reduced by DAPA; as expected, diuresis increased in both groups, (Table 1), but not significantly for HCT because not potentiated by glycosuria as DAPA [22].

Table 1.

Clinical characteristics of the study participants at baseline and at the end of the 4 weeks of treatment

	DAPA (n = 17)		HCT (n = 16)		P-value	P-value
	Pre	Post	Pre	Post	aseline DAPA vs. HCT	Changes DAPA vs. HCT
Sex (M/F)	11/6		14/2
Age (years)	59.2 ± 2.1		62.2 ± 1.9		ns
HbA1c (%)	7.41 ± 0.34		6.93 ± 0.31		ns
HbA1c (mmol/mol)	58.1 ± 3.7		52.3 ± 3.3		ns
Glucose (mg/dl)	147 ± 8	139 ± 7	125 ± 7	131 ± 12	ns	ns
BMI (Kg/m²)	32.54 ± 1.61	32.22 ± 1.61*	28.79 ± 1.22	28.98 ± 1.12	ns	0.02
AST (IU/L)	18 ± 2	18 ± 2	20 ± 2	19 ± 1	ns	ns
ALT (IU/L)	24 ± 3	24 ± 3	26 ± 2	24 ± 2	ns	ns
GGT (IU/L)	33 ± 6	25 ± 4	32 ± 7	30 ± 5	ns	ns
Total Cholesterol (mg/dl)	168.8 ± 6.3	168.6 ± 5.3	164.2 ± 5.2	162.2 ± 7.1	ns	ns
LDL (mg/dl)	104.4 ± 7.0	104.5 ± 6.2	99.3 ± 4.5	98.1 ± 4.9	ns	ns
VLDL (mg/dl)	21.9 ± 2.2	23.6 ± 3.0	17.0 ± 2.1	18.4 ± 2.3	ns	ns
HDL (mg/dl)	49.5 ± 3.44	50.5 ± 4.4	55.0 ± 4.2	51.1 ± 3.5	ns	ns
Triglyceride (mg/dl)	127.5 ± 11.5	138.2 ± 16.3	101.9 ± 11.3	109.1 ± 12.2	ns	ns
Creatinine (mg/dl)	0.78 ± 0.04	0.75 ± 0.04	0.87 ± 0.04	0.86 ± 0.03	ns	ns
eGFR (ml/min/1.73 m 2 )	95.6 ± 2.77	97.54 ± 2.53	89.40 ± 2.49	88.82 ± 2.24	ns	ns
Diuresis (ml /24h)	1597.4 ± 159.3	2226.5 ± 120.5*	1806.7 ± 176.3	1853.3 ± 173.8	ns	< 0.001
GSG index	0.67 ± 0.08	0.59 ± 0.09*	0.51 ± 0.05	0.56 ± 0.05	ns	ns

Data are presented as mean ± standard error. * p value < 0.005 post vs. pre in each group

Changes in glucose, TCA cycle and ketogenesis metabolites in DAPA vs. HCT treatment

As expected, we observed a significant increase of glucosuria after the four weeks of treatment and compared with HCT (Fig. 1B); fasting plasma glucose showed a trend toward an improvement, although statistically not significant. Moreover, treatment with SGLT2i was associated with increased urinary excretion of several metabolites that was significantly different from the change observed after HCT (Fig. 1A); conversely, no changes in plasma or urinary electrolytes were observed, except for magnesium (Supplementary Tables S3 and S4). Treatment with DAPA determined a significant increase in plasma and urine β-hydroxybutyrate (BHB) concentration, compared with HCT (Fig. 1A and C).

Fig. 1.

A Heat map of changes in metabolites present both in plasma (umol/l) and in urine (umol excreted in the 24 h) samples. B Effects of DAPA vs. HCT on glucose concentrations in plasma (mg/dl) and 24-h urine (mg/24 h). C The effect of DAPA on metabolites involved in the main metabolic pathways comparing plasma (square) and 24 h urine (circle) concentrations. The fold changes from baseline in metabolite concentrations [Log₂(post/pre)] are represented by color codes [increase (red/pink) or decrease (blue/light blue)] in Panels 1 A and 1 C. (*) indicates which metabolites significantly changed from baseline. (o) indicates significant changes in response to treatment with DAPA vs. HCT. Abbreviations: TCA cycle, Tricarboxylic acid cycle; AAA, Aromatic Amino Acids; BCAA, Branched-Chain Amino Acids

Both plasma concentration and urinary excretion of citrate, a product of TCA cycle activity, were significantly increased compared with HCT and vs. baseline. Other metabolites like α-ketoglutaric acid, succinate, itaconic acid and α-hydroxybutyrate tended to increase, although not significantly, indicating a greater supply of glucose-derived metabolites for oxidation in the TCA cycle (Fig. 1A and C). We observed the same results also in urine, with an increase excretion of these metabolites but a decrease for malic acid (Fig. 1A and C). No correlation was observed between plasma and urinary metabolites.

Amino acid plasma profile and 24 h-urinary excretion for DAPA vs. HCT treatment

The analysis of the amino acid profile showed a significant increase only in plasma isoleucine and methionine concentrations after DAPA treatment, which were significantly different from HCT (Fig. 1A and C).

It is known that the raised EGP following SGLT2i administration is driven by increased gluconeogenesis [6]. Here, we observed a significant decrease in some of the gluconeogenic precursors such as plasma lactate and glutamic acid (Fig. 1A and C), although no change was observed in alanine or glutamine, other important glucogenic substrates. Another possible use of glutamate is in urea cycle [23] and for glutathione synthesis, with the combination of serine and glycine (Fig. 1C). However, this is less likely since DAPA decreased the GSG index (Table 1) which is a marker of glutathione synthesis in response to oxidative stress.

We observed a significantly higher urinary excretion of several amino-acids after DAPA treatment and compared to HCT treatment (Fig. 1A–C), likely indicating an energy loss. Although this might be related to the slight decrease in BMI, there was no correlation between pre vs. post differences in BMI and 24-hour urinary amino-acids, which instead correlated mostly with the pre vs. post difference in grams of glucose excreted (Supplementary Table S6 -S8).

We found a positive association between pre vs. post differences in urinary sodium and 24-hour urinary amino-acids after DAPA but only for alanine, glutamic acid and aspartic acid (Supplementary Table S6-S8). Furthermore, the excretion of these amino-acids was associated with sodium excretion at baseline (Supplementary Table S6). No association was observed between BMI and plasma/urinary amino-acids concentrations before or after DAPA treatment, ruling out any association between amino-acid excretion with obesity or weight loss, at least in these subjects.

Unlike sodium and other electrolytes, both plasma and urinary magnesium levels increased following DAPA, but not HCT treatment (Supplementary Table S3).

Changes in plasma free fatty acids concentrations and composition in DAPA vs. HCT treatment.

Previous studies have observed increased lipolysis and FFAs release and a compensatory increase in lipid oxidation and ketone production [3]. In this study there was an increase in plasma β-hydroxybutyrate but not in FFA concentration. Interestingly, it was observed an increase in unsaturated fatty acids (UFA) in the DAPA group (Fig. 2A) which changed the composition of the FFA towards a significantly lower proportion of saturated fatty acids (SFA) to unsaturated fatty acids ratio (SFA/UFA) after DAPA treatment. The lipogenesis (DNL) index, calculated as palmitic acid/linoleic acid [10], decreased significantly after DAPA while it was not altered by HCT (Supplementary Table S9). Index of fatty acid desaturation as stearoyl-CoA desaturase‐1 SCD1₁₆ (palmitoleic acid/palmitic acid) and SCD1₁₈ (oleic acid/stearic acid) increased significantly after DAPA treatment and/or compared with HCT treatment (Supplementary Table S9) in response to increased palmitoleic and oleic acid concentrations after DAPA.

Fig. 2.

Effect of DAPA and HCT on FFAs composition (Panel A) and on plasma acylcarnitine short- (SC) and medium-chain (MC) (Panel B). Panel C-E show the correlation (Pearson coefficients) between changes [Log₂(post/pre)] of total Free Fatty Acids (FFA), β-hydroxybutyrate (BHB) and acylcarnitines (AC); combined data of DAPA and HCT are reported. The fold changes from baseline in acylcarnitine and free fatty acids concentrations [Log₂(post/pre)] are represented by color codes [increase (red/pink) or decrease (blue/light blue)] in Panels 2 A and 2B. (*) indicates which metabolites significantly changed from baseline. (o) indicates significant changes in response to treatment with DAPA vs. HCT

The analysis of total FFAs and composition did not show any association with BMI at baseline in the DAPA group, in contrast with the post-treatment results where positive correlations were observed for all FFAs, except Myristic Acid. However, pre vs. post difference of total FFAs and single species concentrations did not show any association with pre vs. post difference of BMI, suggesting no effect on lipid metabolism associated with the higher BMI and its reduction before/after DAPA treatment respectively (Supplementary Table S10).

The change of total FFAs [Log₂ (post/pre) ] was, instead, positively correlated with the change in BHB concentration for data of DAPA and HCT combined (r = 0.654, p < 0.0001; Fig. 2C), as well as also considering the two treatments separated (DAPA: r = 0.545, p = 0.026; HCT: r = 0.705, p = 0.002) (Supplementary Fig. S2). FFAs are taken up by the liver and either oxidized in the mitochondria to produce BHB or esterified to TAGs within hepatocytes [24]. DAPA treatment increased hepatic fatty acid oxidation, as shown by the increase in β-hydroxybutyrate concentration, and the significant decrease in circulating TAGs (Fig. 1A, Supplementary Tables S5 and S11).

Acylcarnitines serve as carrier to transport FFAs into mitochondria for subsequent β-oxidation, with a role in regulating the balance of lipid metabolism [25] but their plasma accumulation can represent an impareid/incompleted fatty acid oxidation. In both groups we observed a decrease in acylcarnitines, both short-chain (SC) and medium-chain (MC), which was attenuated in the DAPA group (Fig. 2B). The change [Log₂ (post/pre)] in acylcarnitines were directly proportional to the change in FFA in DAPA and HCT data combined (r = 0.399, p < 0.03;Figure 2D) and positively correlated with the change in β-hydroxybutyrate (r = 0.353, p < 0.05; Fig. 2E). However, when the two treatments were analyzed separately, the correlation between acylcarnitines and FFAs remained significant only in the HCT group, but was lost in the DAPA group, while the correlation between β-hydroxybutyrate and acylcarnitines was lost in both treatments but maintained a positive trend (Supplementary Figure S2).

Before DAPA treatment, a positive correlation with BMI for TAG 48:0 and TAG 50:0, normally associated with de novo lipogenesis [10], was observed; however, such correlation was lost after treatment (Supplementary Table S10).

Plasma sphingomyelins (SMs) and their precursors ceramides (CERs) are released mainly by the liver into the bloodstream. In our study, we observed a general increment in plasma concentration for these species (Supplementary Table S11), with a significant increase after DAPA for a saturated dihydro-ceramide CER(d18:0/16:0) and the two sphingomyelins SM(d18:0/14:0) and SM(d18:0/16:0) and some unsaturated sphingomyelins (SM(d32:1); SM(d33:1); SM(d39:1); SM(d41:2); SM(d42:2)) and LPC C18:2. PCs increased but non-significantly, except for PC(O-38:5), which was significantly different compared with HCT (Fig. 3). These SMs and LPCs are associated with insulin sensitivity [26, 27]. Other, PCs-derivates; LPC C20:3, LPC C20:4 and LPC C22:6, increased in both groups. After DAPA treatment, a negative association between lipid concentrations and BMI was observed (Supplementary Table S10), particularly for CER(d18:1/26:0), CER (d18:1/24:1(15Z)), some SMs and PCs, but this association was lost when considering changes from baseline, suggesting BMI had no impact on the lipidomic profile (Supplementary Table S10).

Fig. 3.

Schematic diagram of plasma global lipids and lipidomic analysis in DAPA and HCT treatment. Color code shows the fold changes vs. baseline. The increase in lipids from baseline are indicated in red/pink. Only lipids significantly changed either in DAPA vs. HCT (°) or vs. baseline (*) have been reported

Discussion

The mechanisms underlying the clinical benefits of SGLT2i remain incompletely understood. In the current study we show that treatment with dapagliflozin influences the metabolomic profiles of both plasma and urine, independently of increased diuresis and of changes in BMI, which decreased slightly but significantly in the DAPA group. Additionally, it modifies the plasma concentrations of various lipids independently of changes in BMI. The absence of an association between changes in BMI and changes in plasma metabolites or lipid species suggests independence from this parameter, pointing out other potentially beneficial effects of SGLT2 inhibition beyond weight loss. Although previous studies have demonstrated important metabolic effect, particularly in overweight/obese individuals treated with SGLT2i [12, 15, 28], more recently researchers have confirmed such benefits also in subjects with normal or low BMI [29, 30].

Treatment with gliflozins appears to boost gluconeogenesis [6], potentially accounting for the observed reduction in plasma levels of important glucogenic substrates like lactate and glutamate, while dapagliflozin may also increase renal gluconeogenesis utilizing glutamate/glutamine [31, 32], alongside notable rises in other metabolites indicative of enhanced systemic availability. Other metabolites, including citrate, β-hydroxybutyrate, methionine and isoleucine, showed significant increase in plasma and 24 h urinary excretion, suggesting enhanced systemic availability and energy metabolism rather than a “compensatory” loss (Fig. 1).

Several other metabolites were increased in 24 h urine samples after DAPA treatment, especially amino acids, including branched chain amino acid (BCAA, leucine, isoleucine and valine), aromatic amino acids (AAA, phenylalanine, tyrosine and tryptophan), lysine, alanine, histidine, proline, aspartate, glutamate and lactate. This result confirms previous findings by Bletsa et al. [13] who evaluated the metabolomic changes in spot urine samples, collected in the morning, of 50 subjects treated with dapagliflozin 10 mg daily for 3 months compared with 30 subjects treated with insulin degludec, and excluded that dapagliflozin-induced changes in urine metabolome were dependent upon the improved glycemia. In our study we also did not find any association with glycemia (that did not change) or BMI, and we also excluded that changes in urine metabolome were associated with increased diuresis.

SGLT2 inhibition in the proximal tubule reduces the reabsorption of both glucose and sodium, thereby increasing their urinary excretion, and may also affect amino acid handling [13, 15]. Indeed, the kidney plays an important role for the amino acid reabsorption [33], and our findings suggest that SGLT2 inhibition is implicated in the modulation of this process. Although our data do not allow us to establish a definitive mechanistic link distinguishing direct from indirect effects, we observed a positive association between the changes in 24-hour urinary amino acid and glucose excretion following SGLT2 inhibition.

Experimental evidence supports both direct and indirect mechanisms. In mice, dapagliflozin reduced the renal expression of transmembrane protein 27 (TMEM27/collectrin) [8], a key regulator of selective amino acid transporters like the solute carriers (SLC) as Slc7a9, Slc6a19/20, and Slc7a9 for the transport of neutral and cationic amino acids. Dapagliflozin decreased also the activity of the anionic amino acid transporter regulated by Slc1a1 and reduced expression of SLC17a3 and SLC5a8, thereby indirectly impairing lactate transport [8, 34], (Fig. 4).

Fig. 4.

Proposed mechanism linking SGLT2 inhibition to altered amino acid handling in the kidney proximal tubule.

(modified from ref 8). SGLT2 inhibition may exert both direct and indirect effects on amino acid transport along the nephron. Dapagliflozin directly downregulates (ref 8) TMEM27/collectrin, a key regulator of several selective amino acid transporters, as Slc6a19/20, Slc1a1, Slc7a9, and SCL2a2 involved in the reabsorption of neutral, cationic, and anionic amino acids and glucose. Moreover, it decreases the expression of SLC17a3 and SLC5a8, which may indirectly influence lactate handling, and downregulate the activator of Na⁺/H⁺ exchanger NHE3. The resulting decrease in sodium reabsorption could alter the electrochemical gradient, further influencing amino acid transport processes that are dependent on sodium co-transport. Abbreviations: TMEM27, Transmembrane protein 27; NHE3, Sodium/hydrogen exchanger 3; SLC, Solute carrier family; GLUT2, Glucose transporter 2; AA, Amino Acids; Na+, Sodium; H+, Hydrogen; K+, Potassium.

Since amino acids in the proximal tubule are co-transported with sodium [33] and the SGLT2 interacts with sodium- and hydrogen-transporter–regulating proteins [8, 32, 34], SGLT2 inhibition reduces sodium reabsorption in the S1/S2 segments, thereby altering the luminal electrochemical gradient. This altered sodium gradient can modulate amino acid transporter activity, potentially leading to mild increases in urinary amino acid excretion (aminoaciduria). Consistent with this, we observed a positive baseline correlation between urinary sodium and amino acid concentrations (Supplementary Table S6). Thus, the high amino acid excretion might be due to the inhibition of the sodium-glucose cotransporter SGLT2 and/or to changes in the sodium cellular gradient on the proximal tubular renal cell membrane that might influence amino acids transport.

Unlike amino acids, plasma and urinary sodium levels remained unchanged following treatment, and no correlation was observed between changes in urinary excretion of sodium and most AAs. However, it should be noted that, unlike glucose, sodium can be reabsorbed throughout the entire tubule by several other transporters [35]. These results are in agreement with Boorsma et al. [36], where SGLT2 inhibition led to an increase in urine volume over the first 4 days without a corresponding rise in 24-hour urinary sodium excretion, and not only at 4 days, but also at 30 days, despite the immediate and sustained onset of glycosuria. We believe that this mechanism is not exclusively due to increased diuresis, since a similar effect was not observed in subjects treated with HCT.

Another possible mechanism for amino-acid reabsorption involves the Na(+)/H(+) exchanger (NHE3) [33], (Fig. 4), which can be indirectly affected by the SGLT2-downregulation of its activator PDZK1 [8]. Moreover, it could be linked to the electrochemical gradient and facilitates the movement of sodium ions across the cell membrane, which then indirectly drives the uptake of amino-acids [32, 33]. A reduction in sodium reabsorption can result in lower AAs reabsorption from lumen to blood, through the proximal tubule kidney cells membrane [37].

We cannot exclude that the increased amino acid excretion may reflect an increased catabolic activity, driven by a reduced insulin/glucagon ratio, which stimulates lipolysis and can also enhance protein breakdown in muscle [38]. This mechanism could contribute to changes in lean mass or skeletal muscle mass, especially in the older subjects, as suggested by a recent meta-analysis that analyzed the effect of SGLT2i compared to other anti-hyperglycemic agents [39]. However, evidence remains inconsistent, since not all studies found a reduction in muscle mass or strength after treatment with SGLT2 inhibitors [40]. Alternatively, increased renal amino acid metabolism and excretion might represent a compensatory mechanism to mitigate acidosis via production of ammonia, as previously reported [31].

In addition, DAPA significantly increased the 24-hour urinary excretion of other metabolites such as glycolic acid, citric acid, succinic acid, β-hydroxybutyrate and α-hydroxybutyrate, which are part of the Metabolomics Signature of Diabetic Kidney Disease (MSDKD) index, previously found to be decreased in subjects with chronic kidney disease (CKD) [41]. Again, the increase was not attributable to a diuretic effect, as the HCT-treated group did not show the same effect. This is in agreement with Mulder et al. [15] who found that dapagliflozin increased the MSDKD index in T2D individuals enrolled in a double-blind, randomised, placebo-controlled crossover study comparing dapagliflozin vs. placebo. Some of these metabolites are produced in the TCA cycle and are associated with improved mitochondrial function, suggesting this as a possible mechanism for renal protection. The decrease in urinary malic acid and increase in citrate excretion after DAPA observed here may also be in this direction and confirmed previous results in the renal cortex of mice treated with empagliflozin [42].

Another possible mechanism of renoprotection could be related to the reduction in oxidative stress previously demonstrated in vitro indeed [43]; we observed a significant reduction in the GSG index for DAPA (from 0.67 ± 0.08 to 0.59 ± 0.09; p-value = 0.02), which is a marker of increased glutathione metabolism due to oxidative stress [20].

Although we did not observe a significant change in fasting FFA concentrations, dapagliflozin altered the FFA composition, increasing the ratio of unsaturated to saturated fatty acids and the desaturation index, and decreasing the indices of de novo lipogenesis DNL, in agreement with previous studies [16, 44, 45]. We also measured the circulating acylcarnitines which are esters of L-carnitine and fatty acids and serve as carriers to transport long-chain fatty acids across the mitochondrial membrane for subsequent β-oxidation to provide energy for cell activities. Plasma short- and medium-chain acylcarnitines not only reflect dysregulated FFA oxidation, that contributes to metabolic disorders such as prediabetes, T2D and cardiovascular disease, but also predict residual cardiovascular risk [46]. Most of the measured acylcarnitines were decreased to a lesser extent in DAPA compared with HCT, as previously observed [23, 44]. This could be due to the increased β-oxidation and the improvement of mitochondrial status [44]; indeed, changes in acylcarnitines were positively related to changes in FFA and β-hydroxybutyrate concentrations.

We also evaluated changes in other lipids, observing a modest increase in sphingolipids and phosphocholines, mostly those containing unsaturated fatty acids, which seems to improve glucose tolerance and insulin secretion [27]. Conflicting results have been reported on the correlation between LPCs and metabolic disease. Our results showed a significant increase in unsaturated LPC in the DAPA group, as also previously shown in animal models [16, 42], particularly in unsaturated LPCs, which are associated with a better insulin sensitivity [26].

This study has several limitations: this is a post hoc analysis of a short-term, exploratory study, and requires confirmation in terms of long-term effects.We were unable to perform a metabolomic analysis 24 h after the first administration of the drug, which would have boosted the results. Moreover, the metabolomic method did not allow the quantification of certain metabolites associated with the metabolism of ketone bodies, like acetic acid; and data on muscle/fat body mass were not available.

Despite these limitations, this hypothesis-generating study offers novel insights through the simultaneous analysis of plasma and urine metabolomics, complemented by plasma lipidomics, revealing potential systemic changes in specific metabolic pathways. Furthermore, the characterization of the 24-h diuresis profile and direct comparison with the HCT group allowed us to consider the variation induced by changes in volume.

Conclusions

The effect of SGLT2 inhibition goes beyond the reduction of hyperglycemia and increase of fatty acid release and oxidation, as dapagliflozin: (a) increases plasma citric acid, (b) induces a remodeling of circulating lipids towards more unsaturated lipids, and (c) increases urinary excretion of several metabolites, in particular of amino-acids, but also of several organic acids such as lactate acid, citric acid and β-hydroxybutyric acid, reducing the risk of acidosis. Further longer, mechanistic studies are definitely needed to assess the long-term effects of gliflozins on lipids and metabolites, the role of such metabolites in kidney protection, and the possible effects on skeletal muscle, particularly in ageing, frail individuals.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

DAPA

Dapagliflozin

TCA

Tricarboxylic Acid cycle

BCAA leucine, isoleucine and valine

Branched chain amino acid

AAA phenylalanine, tyrosine and tryptophan

Aromatic amino acids

FFAs

Free fatty acids

TAGs

Triacylglycerols

CERs

Ceramides

SMs

Sphingomyelins

PCs

Phosphatidylcholines

LPCs

Lysophosphatidylcholines

GSG index

Glutamate-Serine-Glycine

AST

Aspartate transaminase

ALT

Alanine transaminase

GGT

Gamma-glutamyl transferase

Author contributions

SP: metabolomic and lipidomic profiling, data analysis, interpretation of results, wrote the first draft of the manuscript; FP: metabolomic profiling and data analysis; SS: statistical data analysis; FC: lipidomic profiling and interpretation of results ; AM: enrollment of patients and sample collection; AS: study ideation and study design, supervision of clinical studies, interpretation of results and critically review of the manuscript; AG: study ideation and study design, metabolomic/lipidomic profiling and interpretation of results, critically review of the manuscript.

Funding

Italian Ministry of University and Research (PRIN 2017: 20175L9H7H to AG; PRIN 2017: 20178YTNWC to AS); European Union’s Horizon Europe Research and Innovation Programme for the project “PAS GRAS: De-risking metabolic, environmental and behavioral determinants of obesity in children, adolescents and young adults” under grant agreement No. 101080329 to AG. Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union or European Health and Digital Executive Agency (HADEA), under the powers delegated by the European Commission. The initial part of this study was supported by an investigator-initiated, unrestricted research grant to A.S. from Astra Zeneca International.

Data availability

Data are reported in the supplementary material and available from the guarantors of the work upon reasonable request.

Declarations

Ethics approval and consent to participate

The study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki and was approved by the Ethics Committee of the University of Pisa and registered in the Registry of the Italian Drug Agency (Agenzia Italiana del Farmaco no. 772/2015. Written informed consent was obtained from all participants.

Consent for publication

Not applicable.

Competing interests

AG reports consultancy/advisory boards fee for Boehringer Ingelheim, Eli Lilly and Company, Merck Sharp & Dohme, Novo Nordisk, Pfizer, Regeneron Pharmaceuticals; received lecture fees from Boehringer Ingelheim, Merck Sharp & Dohme, Madrigal, Novo Nordisk, Eli Lilly and Company; educational grants from Eli Lilly and Company, Boehringer Ingelheim, Echosens, Mercodia, Madrigal. AS reports advisory boards fee/lecture fees for Astra Zeneca, Bayer, Boehringer Ingelheim, Lilly, Novo, Sankyo, Sanofi. The other authors report no conflicts of interests related to this manuscript.