Abstract
Background
Sodium-glucose cotransporter 2 inhibitors (SGLT2is) have cardioprotective effects without acting directly on the myocardium.
Objectives
The purpose of the study was to evaluate the impact of SGLT2i on myocardial glucose utilization.
Methods
This retrospective propensity-matched cohort study examined subjects who underwent whole-body 18F-fluorodeoxyglucose (18F-FDG) positron emission tomography for oncologic purposes between 2016 and 2024. A 1:1 propensity match analyzed positron emission tomography-derived myocardial FDG uptake, maximum myocardial standardized uptake value (SUV), and total myocardial glycolysis in SGLT2i-treated vs untreated patients.
Results
Among 6,747 subjects, 187 were actively prescribed SGLT2i and matched with 187 who were not. Mean age was 68.4 ± 9.7 years, 57.5% were males, 81.8% had diabetes, and 41.4% had heart failure. Myocardial FDG uptake (to quantify glucose utilization) was absent in 64.7% of patients on SGLT2i and 37.4% not on SGLT2i (P < 0.001), despite higher prescan glucose in those receiving SGLT2i (124 vs 110 mg/dL). SUVmax (5.4 vs 5.8, P = 0.004), volume of myocardial FDG uptake (0.8 cm3 vs 4.6 cm3, P = 0.001), and total myocardial glycolysis (1.6 g vs 20.6 g, P < 0.001) were significantly lower in patients treated with SGLT2i vs those untreated. SGLT2i use blunted insulin-induced increases in total myocardial glycolysis (P-interaction 0.036). There was no difference in background SUVmax and SUVmean in the blood pool and liver.
Conclusions
SGLT2i use was associated with a greater suppression of myocardial glucose without affecting background tissue glycolysis despite higher prescan glucose levels. These findings add to the understanding of the effect of SGLT2i on the myocardium and may have implications with respect to cardiac imaging protocols that require glucose manipulation.
Key words: cardiomyopathy, diabetes, fatty acids, heart failure, ketone body, metabolic syndrome
Central Illustration
Sodium-glucose cotransporter 2 inhibitors (SGLT2is) work by blocking glucose reabsorption in the proximal renal tubule, increasing glucose excretion through urine.1 This process results in a decrease in both plasma glucose and insulin levels, leading to increased lipid oxidation and production of free fatty acids and ketone bodies. This ketotic state is similar to fasting and also triggers autophagy and lysosomal activity. Ketone bodies are an efficient fuel source that typically requires less oxygen to generate adenosine triphosphate than glucose. While the myocardium preferentially uses long-chain fatty acids for energy metabolism, it has metabolic flexibility to increase glucose utilization in the fed state and ketone bodies during periods of prolonged fasting or low carbohydrate dietary intake.2,3
Given the ketotic properties of SGLT2is, the primary objective of this study was to evaluate the impact of SGLT2i on myocardial glucose utilization. We hypothesized that treatment with these medications is associated with lower myocardial glucose utilization as imaged by 18F-fluorodeoxyglucose (18F-FDG) positron emission tomography (PET) compared with patients not on treatment. We studied patients who underwent FDG PET for oncologic purposes, as these patients are not required to have a specific dietary preparation or prolonged fasting before imaging.
Methods
This retrospective propensity-matched cohort study included 6,747 subjects who underwent an FDG PET scan for oncological diagnostic purposes at our institution between January 2016 and March 2024. Patients with myocardial infarction within 90 days of FDG PET imaging were not included in this analysis. Baseline characteristics, including a history of heart failure (HF), diabetes, insulin use, and SGLT2i use, were captured.
Imaging protocol
The whole-body FDG PET imaging protocol was in keeping with standards for oncologic imaging.4 Patient preparation included a 4-hour fast. Diabetic patients were instructed to withhold their antidiabetic medications, including insulin, for at least 4 hours before the scan. Whole-body images were acquired using Siemens PET/computed tomography Biograph Vision, mCT 64, or mCT 16 systems after receiving a target of 12 millicurie FDG. The time interval between FDG administration and the start of emission scanning was 55 to 75 minutes. A low-dose transmission computed tomography scan was also acquired for attenuation correction. Visual uptake of FDG in the myocardium was graded as no uptake, mild diffuse or focal uptake, and heavy diffuse uptake. Maximum standardized uptake value (SUVmax) and mean were calculated in the blood pool (left atrium) and liver, and SUVmax was calculated in the left ventricular myocardium. Myocardial total glycolysis (also called cardiac metabolic activity) was also calculated, which is a volumetric parameter that measures the metabolic activity of the left ventricular myocardium above 1.5 times the blood pool SUVmax as the background. The target-to-background ratio was also calculated as myocardial SUVmax divided by blood pool SUVmean. All analyses were performed using Syngo.via software (Siemens Healthcare).
Statistical analysis
The goal of the statistical analysis was to identify the effect of SGLT2i on cardiac and noncardiac glucose utilization. To this end, a propensity-matched analysis was performed using propensity scores with nearest neighbor 1-to-1 matching without replacement. Variables used for the match were selected a priori based on their potential to influence the choice of receiving an SGLT2i: age, sex, body mass index, race, diabetes, insulin use, coronary artery disease, systolic HF, diastolic HF, hypertension, hyperlipidemia, atrial fibrillation, and year before or after 2020. The year 2020 was chosen because this was when the first SGLT2i was approved for use in HF. Standardized mean differences were calculated to evaluate the balance in measured variables between the groups, and variables with values below 0.10 were considered well matched.
Using propensity score matching to mitigate baseline disparities, FDG PET variables were compared between patients prescribed and not prescribed SGLT2i. Continuous outcome data are presented as mean ± SD if normally distributed and analyzed using a t test or median (IQR) if non-normally distributed and analyzed using the Wilcoxon rank-sum test. Categorical data are expressed as numbers and percentages and analyzed using a chi-square test. Stacked bar graphs and violin plots are presented to visualize data. A multivariable linear regression model was generated to test the effects of SGLT2i use on total myocardial glycolysis adjusted by covariates. The linearity assumption was tested by examining the residual plot, and covariate interactions with SGLT2i use were also tested. Analyses were performed using Stata v14.2. The study was approved by the Institutional Review Board and Ethics Committee of Saint Luke’s Health System, and informed consent was waived.
Results
There were 6,747 patients who underwent FDG PET studies for oncologic purposes during the study period, 187 of whom had been prescribed SGLT2i at the time of the PET scan (126 empagliflozin, 40 dapagliflozin, 20 canagliflozin, and 1 ertugliflozin). These patients were matched with 187 who were not prescribed SGLT2i (Central Illustration), and the outputs of the matching process are found in Supplemental Figures 1 and 2 with standardized percent bias below 10% for all variables after matching. The mean age in the entire cohort was 68.4 ± 9.7 years, and 57.5% were males (Table 1). Approximately 8 in 10 patients had diabetes (1 in 4 of whom were on insulin), 4 in 10 carried a diagnosis of HF, and 8 in 10 had FDG PET scans performed in 2020 or later. A cancer diagnosis was present in 99 (53%) patients on SGLT2i and 102 (57%) patients not on SGLT2i, 18 patients had previously known metastases (13 on SGLT2i), and 31 patients were actively receiving chemotherapy (15 on SGLT2i).
Central Illustration.
Impact of Sodium-Glucose Cotransporter 2 Inhibitor on Myocardial Glucose Utilization
Summary of important findings in this study, which demonstrated reduced myocardial glucose utilization in patients treated with SGLT2i vs those untreated. SUV = standardized uptake value; other abbreviations as in Figure 1.
Table 1.
Baseline Characteristics
| SGLT2i (n = 187) | No SGLT2i (n = 187) | Standardized Mean Difference | |
|---|---|---|---|
| Age, y | 68.4 ± 9.2 | 68.4 ± 10.3 | −0.002 |
| Male | 109 (58.3%) | 106 (56.7%) | −0.032 |
| Body mass index, kg/m2 | 31.4 ± 7.5 | 31.5 ± 7.6 | 0.020 |
| White race | 151 (80.7%) | 152 (81.3%) | 0.014 |
| Diabetes | 151 (80.7%) | 155 (82.9%) | 0.055 |
| Insulin use | 43 (23.0%) | 46 (24.6%) | 0.038 |
| Coronary artery disease | 93 (49.7%) | 94 (50.3%) | 0.011 |
| Systolic heart failure | 23 (12.3%) | 27 (14.4%) | 0.063 |
| Diastolic heart failure | 51 (27.3%) | 54 (28.9%) | 0.036 |
| Hypertension | 158 (84.5%) | 154 (82.4%) | −0.057 |
| Hyperlipidemia | 154 (82.4%) | 152 (81.3%) | −0.028 |
| Atrial fibrillation | 81 (43.3%) | 73 (39.0%) | −0.087 |
| Year of FDG PET | |||
| 2020-2024 | 153 (81.8%) | 147 (78.6%) | −0.097 |
| 2016-2019 | 34 (18.2%) | 40 (21.4%) |
FDG = fluorodeoxyglucose; PET = positron emission tomography; SGLT2i = sodium-glucose cotransporter 2 inhibitor.
Prescan glucose was higher in those receiving SGLT2i vs those who were not (124 vs 110 mg/dL) (Table 2). There were no differences in background uptake, as blood pool and liver SUVmax and SUVmean were similar in those treated with SGLT2i vs those untreated. However, myocardial visual uptake was significantly lower in the patients treated with SGLT2i and was absent in 64.7% of treated patients vs 37.4% of those untreated (Table 2, Figure 1A). Examples of patients with no myocardial uptake, mild/focal uptake, and heavy/diffuse uptake are shown in Figure 2.
Table 2.
FDG PET Findings
| SGLT2i (n = 187) | No SGLT2i (n = 187) | P Value | |
|---|---|---|---|
| FDG dose, mCi | 12.5 ± 1.6 | 12.7 ± 1.6 | 0.10 |
| Blood pool SUVmax | 2.9 (2.4-3.5) | 2.8 (2.2-3.4) | 0.15 |
| Prescan glucose, mg/dL | 124 (99-147) | 110 (92-134) | 0.015 |
| Blood pool SUVmean | 2.1 (1.9-2.4) | 2.1 (1.8-2.4) | 0.10 |
| Liver SUVmax | 3.7 (3.0-4.5) | 3.6 (2.8-4.5) | 0.15 |
| Liver SUVmean | 2.5 (2.1-2.9) | 2.5 (2.1-2.8) | 0.33 |
| Myocardial visual uptake | <0.001 | ||
| No uptake | 121 (64.7%) | 70 (37.4%) | |
| Mild/focal uptake | 42 (22.5%) | 56 (29.9%) | |
| Heavy/diffuse uptake | 24 (12.8%) | 61 (32.6%) | |
| Myocardial SUVmax | 5.4 (3.8-7.3) | 5.8 (4.3-9.4) | 0.004 |
| Myocardial TCG, g | 1.6 (0.1-40.0) | 20.6 (0.1-413.7) | <0.001 |
| Myocardial FDG volume, cm3 | 0.8 (0.0-12.4) | 4.6 (0.0-103.0) | <0.001 |
SUV = standardized uptake value; TCG = total cardiac glycolysis; other abbreviations as in Table 1.
Figure 1.
Cardiac Glucose Utilization With Sodium-Glucose Cotransporter 2 Inhibitor
SGLT2i use was associated with a greater frequency of no or mild cardiac glucose utilization on FDG PET scans (panel A). The violin plots (panel B) demonstrate the total cardiac glycolysis (extent and severity of FDG uptake in the myocardium) was lower in patients on SGLT2i vs those who were not, with a distribution clustered around lower values and fewer outliers with extremely high myocardial glucose utilization. Displayed in these violin plots are the medians (black circle), IQR (gray box), whiskers (gray line), and kernel density (red and blue plots). FDG = fluorodeoxyglucose; PET = positron emission tomography; SGLT2i = sodium-glucose cotransporter 2 inhibitor.
Figure 2.
Case Examples
Examples of patients with no uptake (A), mild/focal uptake (B), and heavy/diffuse uptake (C).
Quantitative measurements of myocardial glucose utilization, left ventricular myocardial SUVmax, volume of FDG uptake, and total glycolysis were also significantly lower in patients treated with SGLT2i vs untreated (Table 2, Figure 1B). A multivariable linear regression model including all variables in the propensity match showed that insulin use, systolic HF, and diastolic HF were significantly associated with increased total myocardial glycolysis, while SGLT2i was associated with decreased total glycolysis (Supplemental Table). There was a significant interaction between SGLT2i and insulin use, suggesting that the effects of insulin on increased myocardial glycolysis are blunted by SGLT2is (P = 0.036). The time of day that the scan was performed was not associated with any markers of FDG uptake (total glycolysis 0.06 g [−0.34, 0.22], P = 0.673). Additionally, in patients with visual evidence of myocardial FDG uptake, SGLT2i use was associated with lower myocardial SUVmax (−2.2 [−4.3, −0.1], P = 0.038), total glycolysis (−376.5 g [−749.4, −3.5], P = 0.048), and target-to-background ratio (−1.4 [−2.8, 0.1], P = 0.041).
Discussion
This study examined a propensity-matched cohort of patients who underwent an FDG PET scan for oncologic purposes and found that patients who were treated with SGLT2i were more likely to have suppressed myocardial glucose utilization compared with those who were untreated, despite higher prescan glucose levels. SGLT2i also blunted the effects of insulin on increasing myocardial glucose utilization. Background glucose uptake in the blood pool and liver was similar among groups. These findings add to the mechanistic understanding of SGLT2i treatment effect on the myocardium and have implications for FDG PET studies requiring suppression of myocardial glucose utilization.
Though SGLT2i do not act directly on the myocardium, they deliver powerful positive effects on the heart. Large randomized clinical trials have shown benefits with these medications for HF events regardless of ejection fraction or the presence of diabetes.5 A rapid benefit in HF symptoms with SGLT2i is seen in as early as 12 weeks.6 These early benefits are likely related to natriuresis and diuresis, improved vascular resistance, reduced blood pressure, and changes in tissue sodium handling. Longer-term benefits exist and may stem from a shift toward ketone body metabolism, reduced serum uric acid levels, reduced inflammation, decreased oxidative stress, and increased autophagy.1,7
The reduction in glucose utilization in the heart noted in this study is likely from the longer-term effects of SGLT2i, such as increased lipolysis and ketone body formation. Interestingly, background glucose utilization in the blood pool and liver was similar in those prescribed vs not prescribed SGLT2i. The myocardium generates 60% to 70% of its adenosine triphosphate through free fatty acid oxidation in the fasting state, and glucose is prioritized in the fed state through GLUT1 and GLUT4 transporters.3 SGLT2i treatment appears to transition myocardial energy utilization toward nonglucose sources further. A small randomized trial of 13 subjects who underwent C11-acetate and FDG PET studies found that fatty acid utilization was similar, but glucose utilization was reduced in patients after 4 weeks of treatment with empagliflozin.8 Future studies should continue to assess whether reduced myocardial glucose utilization may be at play in the cardioprotective effects of SGLT2i.
These physiologic processes are exploited during cardiac imaging with FDG PET to assess viability or inflammation. In cases where myocardial viability is in question, glucose and insulin are provided to the patient to drive glucose into the myocardial cells, promoting its utilization as a substrate.3 Viable myocardial regions will take up glucose under these conditions, while fibrotic/nonviable regions will not. Since activated macrophages and other inflammatory cells take up glucose, imaging for inflammation has the converse goal as it relates to glucose manipulation. In these scenarios, a high-fat, no-carbohydrate diet and prolonged fasting are used to suppress physiologic myocardial glucose utilization. Patients who achieve a more ketotic state as measured by beta-hydroxybutyrate levels are more likely to have myocardial glucose suppression on FDG PET scans.2 Intravenous heparin is sometimes given to help convert myocardial energy consumption from glucose to free fatty acids and ketone bodies. Unfortunately, some patients continue to have “false positive” studies due to the inability to completely suppress physiologic myocardial glucose utilization. A retrospective study including 449 patients undergoing FDG PET scans for sarcoidosis (65 on SGLT2i) found that diagnostic scans were more frequently achieved in the SGLT2i group (98.8%) compared with the non-SGLT2i group (91.9%).9 However, this study did not use propensity matching, and all subjects underwent rigorous dietary preparation prior to FDG PET scanning in order to suppress myocardial glucose. Nevertheless, SGLT2i did appear to be associated with further reduction in physiologic myocardial FDG uptake. These data,8,9 combined with our study, suggest that SGLT2i could offer a novel approach to enhance the accuracy of FDG PET scans when assessing for myocardial inflammation. Further study is needed to determine how long a patient must be treated with SGLT2i before the effects of suppressed myocardial glucose utilization are seen. Additionally, whether or not SGLT2i should be stopped prior to viability testing should also be studied.
Study limitations
Despite propensity matching, this is a retrospective, nonrandomized study. Exact fasting times and dietary logs at the time of FDG PET were not consistently available. Given the relatively lenient dietary preparation recommended for oncologic studies, myocardial FDG uptake can be variable. Test-retest reproducibility was not done in this study due to the retrospective nature but may be an area for future prospective investigation. Additionally, as most patients in this and prior studies had a history of diabetes, it is unclear whether or not SGLT2i use in nondiabetics will have a similar effect on myocardial glucose utilization. The effects of other medications on myocardial glucose metabolism were not tested. Furthermore, patients with obstructive coronary artery disease may have altered glucose metabolism, and this was not able to be assessed during this study with ischemic testing. Though patients with recent myocardial infarction were not included in this study, a history of coronary artery disease was used as a variable for propensity matching.
Conclusions
This propensity-matched study of subjects who underwent an FDG PET scan for oncologic purposes demonstrated that SGLT2i use was associated with lower myocardial but not background glucose utilization. SGLT2i administration also mitigated insulin-induced increases in myocardial glycolysis. These findings add to the understanding of the effect of SGLT2i on the myocardium and may have implications with respect to cardiac imaging protocols that require glucose manipulation.
Perspectives.
COMPETENCY IN MEDICAL KNOWLEDGE: The use of SGLT2is is associated with lower myocardial glucose utilization compared with those not prescribed SGLT2i. These findings support the notion that the cardioprotective effects of SGLT2i are associated with metabolic remodeling rather than direct myocardial targeting, aligning with their benefits observed in large outcome trials in HF and diabetic populations.
TRANSLATIONAL OUTLOOK: Future studies should explore how SGLT2i-induced alterations in myocardial substrate utilization influence cardiac energetics and imaging interpretation, particularly in protocols that depend on predictable myocardial glucose uptake.
Funding support and author disclosures
Dr Sperry has received consulting fees from Spectrum Dynamics. Dr Sauer reports research funding support from the Heart Failure Society of America (HFSA), the American Heart Association, AstraZeneca, Boehringer Ingelheim, Bayer, CSL Vifor, Pfizer, Rivus Pharmaceuticals, 35Pharma, Novo Nordisk, Edwards Lifesciences, and Story Health; consulting fees for Abbott, Boston Scientific, Edwards Lifesciences, Impulse Dynamics, Acorai, General Prognostics, Story Health, Amgen, and Bayer; and stock ownership with ISHI and Pulsli (both privately held digital health companies). Dr Bateman has received research grant support from Bracco, GE Healthcare, Jubilant Drax Image, and Spectrum Dynamics; has served as a consultant to GE Healthcare and Synektik; and has an equity interest in Cardiovascular Imaging Technologies. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
Acknowledgment
The authors would like to acknowledge Philip G. Jones, MS, for his assistance with the statistical planning.
Footnotes
The authors attest they are in compliance with human studies committees and animal welfare regulations of the authors’ institutions and Food and Drug Administration guidelines, including patient consent where appropriate. For more information, visit the Author Center.
Appendix
For supplemental figures and a table, please see the online version of this paper.
Supplementary data
References
- 1.Cowie M.R., Fisher M. SGLT2 inhibitors: mechanisms of cardiovascular benefit beyond glycaemic control. Nat Rev Cardiol. 2020;17(12):761–772. doi: 10.1038/s41569-020-0406-8. [DOI] [PubMed] [Google Scholar]
- 2.Vidula M.K., Selvaraj S., Rojulpote C., et al. Relationship of ketosis with myocardial glucose uptake among patients undergoing FDG PET/CT for evaluation of cardiac sarcoidosis. Circ Cardiovasc Imaging. 2024;17(8) doi: 10.1161/CIRCIMAGING.124.016774. [DOI] [PubMed] [Google Scholar]
- 3.Sperry B.W. Opportunities to improve image quality in PET myocardial viability imaging in diabetics. J Nucl Cardiol. 2021 doi: 10.1007/s12350-021-02792-7. [DOI] [PubMed] [Google Scholar]
- 4.Graham M.M., Wahl R.L., Hoffman J.M., et al. Summary of the UPICT protocol for18 F-FDG PET/CT imaging in oncology clinical trials. J Nucl Med. 2015;56(6):955–961. doi: 10.2967/jnumed.115.158402. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.McGuire D.K., Shih W.J., Cosentino F., et al. Association of SGLT2 inhibitors with cardiovascular and kidney outcomes in patients with type 2 diabetes: a meta-analysis. JAMA Cardiol. 2021;6(2):148. doi: 10.1001/jamacardio.2020.4511. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Spertus J.A., Birmingham M.C., Nassif M., et al. The SGLT2 inhibitor canagliflozin in heart failure: the CHIEF-HF remote, patient-centered randomized trial. Nat Med. 2022;28(4):809–813. doi: 10.1038/s41591-022-01703-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Selvaraj S., Patel S., Sauer A.J., et al. Metabolic effects of the SGLT2 inhibitor dapagliflozin in heart failure across the Spectrum of ejection fraction. Circ Heart Failure. 2024;18 doi: 10.1161/CIRCHEARTFAILURE.124.011980. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Lauritsen K.M., Nielsen B.R.R., Tolbod L.P., et al. SGLT2 inhibition does not affect myocardial fatty acid oxidation or uptake, but reduces myocardial glucose uptake and blood flow in individuals with type 2 diabetes: a randomized double-blind, placebo-controlled crossover trial. Diabetes. 2021;70(3):800–808. doi: 10.2337/db20-0921. [DOI] [PubMed] [Google Scholar]
- 9.Selvaraj S., Vidula M.K., Claggett B., et al. Diagnostic utility of SGLT2 inhibitors to facilitate myocardial glucose suppression during evaluation of myocardial inflammation. JACC Cardiovasc Imaging. 2024;17(7):827–829. doi: 10.1016/j.jcmg.2024.02.011. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.