The course and prognosis of heart failure are closely tied to sodium and water retention, which is responsible for 90% of heart failure hospitalizations (13). Water and sodium retention result from reduced circulation, decreased renal blood flow, and elevated aldosterone when the heart's pumping capacity is compromised (14). Furthermore, the blood detained in vein vessels will seep into the interstitial fluid under the pressure of capillaries, and those penetrative fluids remain there when the volume of the fluids exceeds the surrounding cells' abilities of absorption (remarkably, the fluid in lower limbs) (15).

In terms of its mode of action, SGLT2i is distinct from traditional diuretics. With the application of SGLT2i, glucose reabsorption is hindered at the site of renal tubules, and therefore the unabsorbed portion will flow into the distal nephron. However, water can still be reabsorbed unimpededly as usual during this period. Thus, the reduction of the osmotic gradient leads to a decrease in water reabsorption, termed osmotic diuresis (16). One of the critical factors in a significantly reduced rate of heart failure deterioration is a decrease in circulation volume, which is caused by the natriuretic action of SGLT2i. But the eyes-catching role of diuresis in the process of heart failure has been undermined by another large clinical trial which concludes that patients with a high propensity for fluid retention show no remarkably extra benefits compared to the control group under the application of empagliflozin (17). Beyond this diuresis function, an additional independent element during the progression of heart failure may be the improvement of pulmonary artery diastolic pressure (18).

On the other hand, prior research has shown that diabetes patients have higher sodium levels in their skin and muscles (19). Moreover, tissue sodium is closely linked to ventricular hypertrophy regardless of blood pressure and water-sodium retention conditions. It is reported that the application of dapagliflozin for 6 weeks can decrease tissue sodium in patients with T2DM (20) compared to control groups, thereby reducing the risk of left ventricular hypertrophy and chronic heart failure. In contrast, in an emergency scenario, increased glycosuria rather than natriuresis, with SGLT2i utilization, contributes to improving the patient's state (21).

Excessive adipose tissue exerts vital functions in the onset and progression of heart failure (22). Several studies have revealed that SGLT2i plays a crucial part in weight shedding during its pharmacological action in both types of diabetes (23–25), and the baseline body mass index will not impair the effectiveness of SGLT2i, contrary to what is known as the “obesity paradox” (26).

Attributed to the calories loss caused by the excretion of glucose (27) and enhanced hypothalamic insulin responsiveness (28), the application of SGLT2i can significantly reduce the weight of patients, leading to the reduction of total fat mass, subdermal fat, visceral fat, and liver fat content (29–33). What's more, it has been revealed that SGLT2i can reduce the size of adipocytes in the perivascular adipose tissue (34) and be capable of inducing mitochondrial biogenesis through the AMPK/SIRT1 pathway and β3-adrenoceptor-cAMP-PKA signaling pathway, thereby increasing the energy consumption of adipocytes directly in vivo (35–37). Some researchers believe that it is possible to achieve pounds shedding by inducing the beigeing of fat (38), while beigeing refers to a specific metabolism remolding of fat tissue (39). Additionally, a recent study reveals that the activated innervation of intra-adipose sympathetic induced by SGLT2i in mice with a high-calorie diet also contributes to such an increased energy consumption (37, 40). Fibroblast growth factor 21 (FGF21) is a coordinator for the SGLT2i-dependent decline in adiposity and the trigger of lipolysis during weight loss (41). This mediator can perform a role in the activation of the nervous system, thereby achieving the biological effect of white/brown adipose tissue and the induction of thermogenesis (40, 42).

This weight loss impact does not last indefinitely; it progressively reaches a plateau between 24 and 52 weeks. These phenomena may be explained by the fact that adipose tissue has an anti-lipolytic function, and the decline in leptin levels has boosted compensatory eating (43–45). Such increased intake is a higher calorie intake (46) with a constant proportion of nutrients as previously, although those intake calories are ultimately countered by glucose loss (28, 47). Figure 2, which was produced using BioRender.com, shows intricate mechanisms. Therefore, although SGLT2i cannot permanently treat obesity, lowering body weight and fat percentage may still positively affect the cardiovascular system.

The blood volume in the body is tightly connected to blood pressure levels. Nevertheless, it has been observed that volume changes and urinary sodium excretion, under the application of SGLT2i, do not significantly lower blood pressure (48), suggesting that additional processes may have been at work throughout this blood pressure-lowering process.

In clinical trials, the 24 h ambulatory blood pressure shows a slight reduction among people suffering from nocturnal hypertension, diabetes, or salt sensitivity with the application of SGLT2i, thereby lowering the incidence of heart failure and the tolls of cardiovascular mortality (49). Such an efficacy may be achieved by regulating renal HIF-1α, impairing inflammation and oxidative stress (50), regulating the function of the paraventricular nucleus of the hypothalamus (PVN) (51), and sGC pathways (52). A meta-analysis of 43 RCT studies reveals that under SGLT2i treatment, patients' systolic blood pressure (SBP) decreases by a mean of 2.46 mmHg and diastolic blood pressure (DBP) by a mean of 1.46 mmHg (25, 53), while another meta-analysis confirms that SGLT2i significantly reduced 24 h dynamic SBP and dynamic DBP, and such an effect is concluded as SGLT2i-like effect (54). Therefore, SGLT2i generates cardiovascular benefits by reducing blood volume, lowering blood pressure, and improving ventricular load through its diuretic and natriuretic effects. In the meantime, it also exerts cardiovascular effects by regulating the central nervous system and affecting PVN.

An endogenous glycoprotein hormone called erythropoietin (EPO) can promote erythropoiesis (More details on this concept have been fully explicated in the reviews by Lappin, T.R. and Koury, M.J.) (55, 56). EPO is mostly produced by the kidney and in minor quantities by the liver. At the same time, a hypoxic environment will trigger its production. Moreover, the values of hematocrit and hemoglobin within normal ranges are negatively correlated to the incidence or prognosis of cardiovascular diseases (57–60). EPO and hematocrit significantly rise after using empagliflozin for 1–6 months, causing a continuous rise in blood cells that peaks in 2–3 months (29, 61, 62), and this tendency is independent of the initial levels of anemia (63).

These changes in EPO and hematocrit can be elaborated as follows: (1) Decreased plasma volume and blood concentration induced by osmotic diuresis (64), but some hold the contradictory view (2) Increased erythropoiesis and hematocrit via inhibiting hepcidin and regulating other iron regulatory proteins (65), (3) The application of SGLT2i can reduce the reabsorption of glucose and reduce the expensed ATP in Na⁺/K⁺ pumps, thereby decreasing the metabolic load of the EPO-related cells and improving the status of hypoxia. To a certain extent, such an improvement of hypoxia restores myofibroblasts back to erythropoietin-producing fibroblasts, leading to enhanced hematopoietic function and increased hematocrit (64, 66, 67), (4) Inhibition of SGLT2 will induce the activity of HIF and SIRT1. The HIF-2α activation in the kidney and liver can promote the production of erythropoietin (68) and strengthen the activation of SIRT1, resulting in modification of macrophage polarization which promotes anti-inflammatory phenotypes and impairs myocardial inflammation (69). (5) In vivo, it has also been observed that dapagliflozin can increase endogenous antioxidant enzymes and regulate fibrosis markers in renal tissues, thereby reducing the oxidative stress of rat kidneys.

Incessant contraction of the heart requires enough energy to support. Previous studies (70) have concluded that when the heart goes through the process of failing, abnormalities in cardiac energy metabolism, particularly in advanced stages, will occur, often manifested as increased glucose intake (mainly on anaerobic glycolysis) (71–73), reduced fatty acid oxidation, and decreased ketone oxidation, coincident with severely impaired mitochondrial function, leading to an overall depleting ATP content (More on the metabolic profile of the failing heart can be found in the reviews by Stanley et al. (70). Thus, it is essential to shift the substrate preference to satisfy such a pathological energy demand. However, such an alternation comes with a price—patients may be trapped in a vicious circle that worsens their more advanced heart failure.

The chief role SGLT2i plays in the metabolism of a failing heart is to improve the overall cardiac ATP production through the increased consumption of fatty acids, ketone bodies, and amino acids, with a concomitant decrease in glucose utilization (73, 74). Quantitatively, it can increase the overall ATP content by ~30% (75). Concerning glucose metabolism, SGLT2i-treated heart failure models show a reduced intake of cardiac glucose and a decreased level of metabolism-related enzymes, mainly shifting the substrate preference to ketone bodies, free fatty acids (73). Also, SGLT2i improves glucose tolerance and insulin resistance via several mechanisms such as inflammasome suppression (76), while the plasma insulin has been down-regulated and the glucagon shows a rise (37, 41, 77).

In mitochondrial dysfunction models, the improved potency of fatty acid oxidation, the reduced aggregation of fatty acid intermediates, as well as the improved synthetic efficiency of mitochondrial bioenergy are attributed to the function of SGLT2i, thereby enhancing the oxidation of the fatty acids in cardiac mitochondria and, as a result, avoiding mitochondrial failure (36, 78–80). A similar trend has also been observed in clinical trials (74). From another perspective, both in vitro and in vivo, SGLT2i has been confirmed to act as an activator of the AMPK signaling pathway to regulate metabolism, ameliorate inflammation, and maintain mitochondrial homeostasis (81–83).

SGLT2i can upregulate the plasma ketone levels, in part, mediated by the reduced plasma glucose levels and decreased insulin levels (41), coincident with a climb in free fatty acids (84, 85). As a result of the ongoing disorder of glucose, the failing heart becomes more dependent on ketone bodies to serve as extra fuel (86, 87). It will simulate a pseudo fasting state and give priority to the utilization of ketone bodies (29, 80, 88, 89). Such an effect of ketone bodies production is directly induced without other factors' mediation (90). SGLT2i also upregulates ketogenic enzymes and transporters in livers, kidneys, and intestines, which, accordingly, increases the level of beta-hydroxybutyric acid in the bodies and tissues (91). However, this promising hypothesis remains debatable and controversial due to the absence of further cogent and robust evidence (92). Some find that the efficiency of ketone bodies oxidation remains approximately unchanged under the application of SGLT2i and believe that the increased production of ATP benefits, in part, from the extra supplement of ketone bodies (75). Nevertheless, some researchers observe the contradictory results that bioavailability and oxidation have been increased by SGLT2i (93). This vague issue requires more research to tackle with. And such an increase in ketone bodies may, in turn, bring about worries about diabetic ketoacidosis, which is worthy of caution during its application.

As both a cause and a result of cardiovascular diseases, inflammatory cytokines, such as interleukin-1 (IL-1), interleukin-6 (IL-6), tumor necrosis factor (TNF), galectin-3, and others, have a very significant positive link with the pathophysiology and development of these conditions (94). The inflammatory profile of heart failure has been well-illustrated in the reviews by Murphy et al. (95) and Adamo et al. (96). Inflammasomes were initially reported as a kind of complex for caspase-activator (97). Moreover, now, they are referred to as multiprotein signaling complexes, including NOD-like receptor 1 (NLRP1), NOD-like receptor 3 (NLRP3), NOD-like receptor 6 (NLRP6), absent in melanoma 2 (AIM2), etc., which regulate inflammatory processes as well as anti-pathogen defenses (98, 99), inducing an increase in pro-inflammatory cytokines (100, 101). Those factors have a close connection with impaired blood vessels (102), disordered mitochondrial function (103), elevated oxidative stress, deteriorated hypertension (104), progressed atherosclerosis (99, 105), and advanced myocardial damage (106), etc.

Inflammatory cytokines and inflammasomes can both be suppressed by SGLT2i. To be more specific, it has been confirmed in rodent models that the crucial inflammasome NLRP3, a complex of caspase-1, IL-1β, and IL-18 cytokine-triggering factors (76, 101, 107), can be suppressed under the application of SGLT2i (108–110), which is, in part, calcium-mediated (105, 109). Several studies have shown that the AMPK pathway can restrain the increase of inflammasomes (111), and treatment with SGLT2i can bring about a climb of AMPK phosphorylation in LPS-treated cardio-fibroblasts (69), thereby avoiding the consequently augmented inflammation.

From another perspective, β-hydroxybutyrate also plays a significant role in regulating inflammasomes and inflammatory cytokines. Oscillated levels of β-hydroxybutyrate and insulin go hand in hand with inflammation (or inflammasome) levels. It has been established that SGLT2i can increase the β-hydroxybutyrate levels (discussed in the previous section), and this climb acts as a defender against endoplasmic reticulum stress-related inflammasomes via the activation of AMPK (111). However, some researchers believe this regulation is not AMPK-dependent. They reveal that the suppressive function of β-hydroxybutyrate on inflammasomes has nothing to do with fasting-related mechanisms such as AMPK, ROS, or other factors. Instead, it works by reducing the apoptosis-associated speck-like protein containing CARD (ASC) oligomerization and speck formation (112, 113), and this can be mediated through the G-protein-coupled receptor 109a (Gpr109a)–NLRP3 pathway associated with the increased influx of extracellular calcium (105). Furthermore, β-hydroxybutyrate is able to regulate mitochondrial protein acetylation, lowering the levels of NLRP3 and inflammatory cytokines, thus achieving the benefits in HFpEF models (103).

The most logical explanation for how SGLT2i reduces inflammation seems to be that inflammation is closely associated with hyperglycemia (114–117), and these negative effects will be lessened as soon as the glucose level falls. Recently, more attention has been paid to the regulation of insulin, which is regarded universally as an anti-inflammatory factor in most cases (118–120). However, recent studies report that in some specific circumstances, insulin can trigger the pro-inflammatory pattern of macrophages, thereby contributing to a climb in IL-β mediated by activated inflammasome and overproduction of ROS (121, 122). Such a function of insulin has been validated in a clinical setting by other research conducted by La Grotta et al. (123). Likewise, given the glucose-lowering function of SGLT2i, the insulin shows a decrease along with the dropped levels of blood glucose correspondingly, contributing to the avoidance of progressed inflammation and increased inflammasome. Moreover, alleviated insulin resistance, which has been mentioned above, also brings benefits to inflammation reduction, because the insulin resistance status possesses strong links with pro-inflammatory and inflammatory states (124–126). So, insulin-related mechanisms can at least partially mediate SGLT2i's ability to suppress inflammation.

Oxidative stress is a biological process characterized by high levels of free radicals that is accompanied by detrimental consequences (127, 128). It is well-known that a high level of blood glucose can induce oxidative stress intracellularly (129, 130), and such a stress-related function is negatively correlated with the level of HbA1c (131). SGLT2i can significantly contribute to the reduction of oxidative stress through a variety of mechanisms besides blood sugar lowering in its capacity as an anti-hyperglycemic drug.

Both in vivo and in vitro, it is reported that SGLT2i can attenuate cell apoptosis induced by endoplasmic reticulum stress (132–136). This process may be mediated by the curbed TGF-β/Smad pathway (137, 138) and the activated Nrf2/ARE signaling (139). Furthermore, SGLT2i use can eliminate intracellular ROS, block ROS-activated signaling pathways, and improve the hyperglycemic state in this process as well (140, 141). It can also normalize the size and quantity of mitochondria by regulating autophagy and reducing mitochondrial abnormalities attributed to myocardial infarction (142). What's more, SGLT2i is able to suppress inflammatory factors and oxidative stress by inhibiting the NO-sGC-cGMP pathway and curbing the polymerization of PKGIa (141, 143). Therefore, the methods above, as well as SGLT2i-mediated suppression of endoplasmic reticulum stress-induced apoptosis, can support cardioprotective function in heart failure.

In patients with T2DM, hyperglycemia and hyperinsulinemia often lead to sympathetic hyperexcitation, which consequently results in aggravated hypertension and an increased occurrence of other cardiovascular diseases (144–146). Contrary to our expectations, the heart rate, under the effects of those mechanisms above, including diuresis, decreased blood volume, and decreased blood pressure, has not shown a significant increase, suggesting that SGLT2i can reduce the heart rate of patients with a heart rate faster than 70 bpm (147). It may have the potential effect of inhibiting sympathetic nerve excitation. This role can be described as a “mediator” of SGLT2i to produce cardiovascular benefits.

In vivo, dapagliflozin significantly improves mice's blood pressure and endothelial function via reduced excitability of the sympathetic nervous system and down-regulated IL-6 (148). Interestingly, such a function goes hand in hand with the natriuretic effect of SGLT2i rather than its sugar-lowering function (149). At the same time, it is well-known that SGLT2 receptors are mainly enriched in the tissues of the central nervous system involved in autonomous regulation, and SGLT2i may affect the central nervous system in an uncharted way, exerting a sympathetic inhibition effect (51). In clinical trials of those who suffer from acute myocardial infarction with T2DM, the application of SGLT2i for 24 weeks improves sympathetic hyperexcitability and parasympathetic nerve function through changes in hemodynamics, myocardial energy metabolism, and function of the hepatic vagus nerve (150). On the other hand, from the perspective of cardiorenal syndrome, SGLT2i can also indirectly suppress sympathetic hyperexcitability by improving renal function.

However, there are limited clinical studies on how SGLT2i affects the sympathetic nervous system in the human body, and further clinical studies are required to be conducted.

It has been reported that SGLT2i can prevent and delay heart remodeling (135, 151–153), thereby protecting against the deterioration of heart failure.

In various animal models, SGLT2i shows the ability to prevent atrial remodeling, electrical remodeling (154, 155), and endothelial dysfunction (156–158). SGLT2i is confirmed to be capable of delaying the progression of left ventricular concentric hypertrophy in aorta cells by reducing sympathetic nerve tension and inflammation in the aorta (159). At the same time, the left ventricular systolic and diastolic functions have been enhanced, and the stiffness of cardiomyocytes has been improved (160, 161). Additionally, SGLT2i directly affects the phenotype and function of human cardiac myofibroblasts by weakening their activities and regulating cell-mediated collagen remodeling. Also, it can regulate the PERK-eIF2α-CHOP axis and activate the SIRT1 pathway (135, 162), thereby improving cardiac remodeling and heart failure.

On the other hand, SGLT2i can also exert a cardioprotective effect on patients suffering from myocardial infarction, preventing heart failure after that occurrence. It can reduce the infarct area (163, 164) independently of the blood sugar state, improving the heart function, remodeling, and metabolic state after myocardial infarction, etc. (93, 165, 166). However, cardiomyocytes chiefly express SGLT1 and seldom express SGLT2. It has been reported that specific knockdown of SGLT1 can reduce the infarct size in mice through function with EGFR (167), and SGLT1 knockdown can also ameliorate cardiac fibrosis and pyroptosis in vivo (168, 169), suggesting the potential ability of SGLT1 to provide cardiovascular benefits. So, is there a possibility that SGLT1/2i will have more beneficial impacts compared with SGLT2i? The SOLOIST-WHF trial shows that SGLT1/2 brings great benefits to the patients (which has also been detailed in Table 1), but whether SGLT1/2i is superior to SGLT2i or not remains unclear. Moreover, what is worth mentioning is that dual inhibitors of SGLT1/2 have been reported to exacerbate cardiac dysfunction after myocardial infarction in rats (170), which contradicts the assumption and clinical trials. The underlying reason for such a phenomenon requires additional preclinical research.

It has been revealed that empagliflozin can curb the activity of NHE1 (Na⁺/H⁺ Exchange Protein 1) in the hearts of mice, rats, and rabbits, thereby reducing the concentration of sodium and calcium in cardiomyocytes (171–173). Also, it can function as a direct cardiac effect to inhibit the NHE of the isolated intact heart and delay the occurrence of ischemic contracture of the heart in the absence of insulin (174). Similarly, empagliflozin can reduce the calcium sensitivity of human myocardial myofilament (175), improving the diastolic dysfunction of the human myocardium and inhibiting NHE in human bodies. By and large, SGLT2i successfully reduces the amount of Na⁺ entering cells by inhibiting NHE1 and improves the contractile dysfunction in heart failure by normalizing the intracellular pH, thereby producing cardiovascular benefits. However, there are some opposite voices that the activity of cardiac NHE1 will not be influenced by empagliflozin or other SGLT2i, as under the application of therapeutic doses, empagliflozin shows no regulatory effect on the concentration of Na⁺ (176). They believe that the role of SGLT2i in heart failure should not be interpreted as being mediated by myocardial NHE1 or intracellular Na⁺. In general, the function of SGLT2i for NHE is controversial since the clinical research data is very scarce, and more evidence is still required.

Adipose tissues secrete a large amount of biologically active substances collectively known as adipokines, which have been fully described in the reviews by Lelis et al. (177) and Kim et al. (178). Adipokines are in a harmonious balance—adiponectin and SFPR5 can down-regulate the expression of many pro-inflammatory mediators and prevent a variety of obesity-related endocrine or cardiovascular diseases (179, 180). Simultaneously, leptin can stimulate inflammatory responses and up-regulate pro-inflammatory elements such as TNF- and IL-6, hastening the onset and progression of cardiovascular disease (181, 182). Leptin and adiponectin exert contrary effects on subclinical inflammation and insulin resistance. Under the pathological condition of obesity, hypertrophic lipocytes and immunocytes in adipose tissue will activate and accelerate the chronic pro-inflammatory response and regulate the secretion of adipokines and other regulatory factors, with a deterioration of cardiometabolic diseases (183). Meta-analysis shows that SGLT2i can effectively reduce circulating leptin levels, increase circulating adiponectin levels (184), and inhibit the excretion of IL-6 and TNF-α in animal aortas (137, 159, 185), thereby producing cardiac vascular protection and delaying the progression of heart failure. Furthermore, SGLT2i can also reduce the size of epicardial adipose tissue (186) in patients by improving systemic micro-inflammation (187) and alleviating adipose-related vascular diseases (188), benefiting heart failure patients.

Myeloid angiogenic cells are abbreviated to MAC, which means vascular endothelial progenitor cells, and play a vital role in the pathological process of atherosclerosis. Its quantity is negatively correlated with cardiovascular events, serving as a biomarker (189). The regulation of SGLT2i on such progenitor cells expression is controversial, and some studies confirm that treatment with SGLT2i can up-regulate the expression of vascular progenitor cells (190). However, some studies believe that SGLT2i does not increase circulating stem cells and endothelial progenitor cells, while such an increase of endothelial progenitor cells may be indirectly achieved through the improvement of blood glucose level, and SGLT2i has no direct effect on endothelial progenitor cells (191, 192). In conclusion, the relationship between SGLT2i and endothelial progenitor cells is an emerging direction, and the results of existing studies are inconsistent and controversial.