The DAPA‐DIET study: Metabolic response to Dapagliflozin combined with dietary carbohydrate restriction in patients with Type 2 Diabetes Mellitus and Obesity—A longitudinal cohort study

Abstract Objective The cardio‐renal benefits of sodium glucose‐like transporter 2 inhibitor (SGLT2i) therapies have been demonstrated in patients with and without type 2 diabetes. However, no studies have explored the long‐term metabolic effects of SGLT2i, combined with dietary carbohydrate restriction. Our primary objective was to describe long‐term changes in weight, energy expenditure, appetite and body composition after 12 months of Dapagliflozin therapy, with carbohydrate restriction, in people with type 2 diabetes and obesity. Our secondary objective was to assess changes in adiponectin and leptin. Method This was a 12‐month cohort study in a secondary care setting. Participants (n = 18) with type 2 diabetes (T2D) and class 3 obesity underwent baseline indirect calorimetry for determination of 24‐h energy expenditure, body composition, fasting serum leptin and adiponectin levels, and appetitive assessments. Following initiation of Dapagliflozin (and dietary carbohydrate restriction), measurements were repeated at monthly intervals up to 12 months. Results Mean starting weight of participants was 129.4 kg (SD 25.9), mean BMI 46.1 kg/m2 (SD 8.3) and mean HbA1c 53.9 mmol/mol (14.1). Seventeen participants completed the study; after 12 months of Dapagliflozin and dietary carbohydrate restriction, mean weight loss was 8.1 kg (SD 11.3 kg; p = .009). This was mediated by reduced fat mass (mean loss, 9.9 kg; SD 10.4 kg; p = .002) associated with reduced serum leptin at 12 months (mean reduction 11,254 pg/ml; SD 16,075; p = .011). There were no significant changes in self‐reported appetite, 24‐h energy expenditure or serum adiponectin during follow‐up. Conclusion In this study, combined Dapagliflozin therapy and carbohydrate restriction in patients with T2D and obesity resulted in a significant reduction of body weight and fat mass at 12 months without any discernible changes in energy expenditure or appetite. These results offer a scientific and clinical rationale to conduct an exploratory trial investigating the effects of a low carbohydrate diet combined with SGLT2 inhibitors in patients with T2D.


| INTRODUC TI ON
The Sodium-Glucose-like Transporter 2 Inhibitor (SGLT2i) class of drugs is an oral therapeutic option for patients with T2D. 1 The glycaemic effects of SGLT2i therapies occur through the promotion of urinary glucose excretion. 2 Recent data from large cardiovascular outcome trials highlight significantly improved cardiovascular and renal outcomes in patients with T2D treated with SGLT2i. 3 Furthermore, SGLT2i facilitate secondary weight loss. 4 Two proposed mechanisms for weight loss include (i) urinary excretion of calories from glycosuria 4 and (ii) fluid loss (supported by an observed 7.3% reduction in plasma volume from 12 weeks of SGLT2i therapy). 5 However, there is mixed evidence in the literature regarding the temporal pattern of weight loss with SGLT2i. Some studies suggest a plateauing effect at 24 weeks postinitiation. 6 Other studies show continual weight loss up to 102 weeks post-initiation. 7 Bailey et al. 8 showed a 'dual-phase' effect, with early rapid weight loss followed by a progressive reduction in body weight and decreased waist circumference in response to Dapagliflozin.
The mechanism(s) for the observed weight loss in response to SGLT2i therapies remains incompletely understood. Following therapeutic blockade of renal proximal tubule SGLT2 receptors, ~50% of renal glucose within the ultrafiltrate (approximately 80 g) is reabsorbed by SGLT1 receptors in the context of normoglycaemia.
Accordingly, an observed urinary glucose loss of 80-120 g/day with SGLT2i in patients with T2D would translate into a daily caloric deficit of 320-480 kcal, and a predicted weight loss of between 5.5 and 7 kg in 3-4 months, 4 exceeding the weight loss reported in the literature. 4,6,7 Although the discrepancy between observed and expected weight loss could manifest from up-regulation of SGLT1 receptors, it is feasible that adaptations to metabolic rate and appetite may attenuate SGLT2i-induced weight loss. 4 Given the paucity of long-term studies reporting on the metabolic effects of SGLT2i, our aim was to characterize the longer-term changes in metabolic rate, appetite and body composition with Dapagliflozin therapy combined with dietary carbohydrate restriction in participants with T2D and class 3 obesity.

| Research protocol
This was a longitudinal cohort study. 24-h profiles of metabolic rate at baseline and at 12-months (following initiation appetite. These results offer a scientific and clinical rationale to conduct an exploratory trial investigating the effects of a low carbohydrate diet combined with SGLT2 inhibitors in patients with T2D.

K E Y W O R D S
appetite, carbohydrate restriction, energy expenditure, fat mass, lean mass, metabolism, SGLT2 inhibitor, type 2 diabetes

Novelty statement
What Is Already Known?
• SGLT2i therapies lead to weight loss among patients with type 2 diabetes (T2D).

What this Study Has Found?
• Combination of the SGLT2i Dapagliflozin and carbohydrate restriction in patients with T2D and obesity resulted in a significant reduction of body weight, fat mass and serum leptin, with no discernible change in energy expenditure or appetite at 12 months.
What Are the Implications of the Study?
• Reduced carbohydrate diet with SGLT2i therapy among patients with T2D and obesity could lead to more clinically meaningful weight loss without detrimental effects on appetite and energy expenditure. These results offer a scientific and clinical rationale to conduct an exploratory trial investigating the effects of a low carbohydrate diet combined with SGLT2inhibitors in patients with T2D.
of Dapagliflozin) were performed on the Human Metabolism Research Unit (HMRU). Each participant had measurements of body weight, appetite and fasting blood tests at baseline and at monthly intervals until 10 months and then at 12 months.
Participants underwent assessment of body composition at baseline and at months 3, 6 and 12.

| Body composition assessment
'Body composition tracking system' (BOD POD® Life Measurement Inc, Concord, California, USA) was used for assessment of body composition. This uses whole-body air-displacement plethysmography, with whole-body densitometry to determine body composition. 9 BODPOD uses air instead of water to measure body volume (based on the principle of Boyle's Law) and is a reliable and valid technique for measuring body composition in individuals with obesity. 10

| Whole-body indirect calorimetry
For assessments of 24-h energy expenditure, we used whole-body indirect calorimetry, 11  The airtight respiratory chamber within HMRU has a large window, 2 airtight compartments where exchange of food, containers and other items can occur. Communication was enabled via intercom, but if intercom was not in use the chamber was sound proof. An air conditioning system maintained the temperature within the chamber at 24°C throughout all studies. Blood samples were obtained through a small opening in the door, whereby participants would place their arm through a plastic bag with a whole in to minimize air escaping, before the small port in the door was opened.
Activity was recorded with a Doppler radar.
We measured overall 24-h energy expenditure, including over-

| 24-h HMRU metabolic study
Participants entered the metabolic chamber at 9 AM, with break-

| SGLT2 inhibitor (Dapagliflozin) treatment
Following the baseline HMRU metabolic study, each participant commenced Dapagliflozin (10 mg, orally, once per day). All participants took Dapagliflozin therapy for the entire 12-month duration of the study (following their initial baseline HMRU study).

| Dietary intervention
Each participant received one-to-one dietary input provided by specialist weight management dieticians for the entire duration of the study. Specifically, participants were advised on a low carbohydrate diet (<100 g/day), with predominantly protein and fat (mainly from mono and polyunsaturated fats), and 400 g (5 portions) of vegetables/day. Records of food intake (including the estimated daily carbohydrate intake as <100 g, 100-300 g and >300 g) were encouraged and reviewed at each of the monthly assessments during a week.

| Biochemical assays
Fasting blood samples were collected into EDTA and serum separat-

| Statistical analysis
Based on existing randomized controlled studies and real-world data for Dapagliflozin, 15 Table S1.

| Dietary changes
Given that the majority of participants did not record accurate estimates of carbohydrate (CHO) intake, and the relatively small number of participants overall, to have reported these data would not have been clinically meaningful. Therefore, we took the decision not to include such data. The lack of estimate of dietary carbohydrate content was due to multiple reasons including impaired self-recall and inadequate compliance. There was no change in any measures of energy expenditure between baseline and 12-month assessments (including total energy expenditure, PPEE, REE and EIEE: data shown in Table 2). There was

| Body weight and body composition changes
Temporal changes in body weight: Mean body weights at baseline, 6 and 12 months were as follows: 129.4 kg (SD 25.9 kg), 124.2 kg (SD 26.2 kg) and 123 kg (SD 24.2 kg), respectively. Figure 2 shows cumulative changes to body weight and percentage weight loss based on all available data. Individual changes in body weight are included in Table S2.
A paired Student t-test was used to analyse the difference between baseline and 12-month follow-up data, for the 17 participants where data were available (data shown in Table 2

| Changes in appetite
Among all participants, there were 198 completed assessments of appetite. Compared with baseline data, there were no statistically significant differences in self-reported measures of appetite (including fasting hunger, satiety and fullness scores) at any of the followup assessments over 12 months (Figure 3).

| Leptin and adiponectin
Mean fasting serum adiponectin level at baseline (data available for 18 participants) and 12 months (data available for 17 participants) was 5711 ng/ml (SD 4275) and 6099 ng/ml (SD 3653), respectively ( Figure 4). Mean serum leptin level at baseline (data available for 18 participants) and 12 months (data available from 17 participants) was 51,713 pg/ml (SD 32,021) and 42,880 pg/ml (SD 26,005), respectively ( Figure 4). There was a significant reduction of 11,255 pg/ml at month 12 (p = .011 using paired Student t-test, Table 2).

| DISCUSS ION
We provide detailed characterization of metabolic and appetitive responses to 12 months of Dapagliflozin therapy in combination with dietary carbohydrate restriction in patients with T2D and class 3 obesity. We demonstrate significant and substantial reduction in body weight and fat mass, with significant reduction in serum levels of leptin, despite preservation of 24-h energy expenditure and selfreported appetite. To our knowledge, our study is the first to report a comprehensive 24-h metabolic assessment in response to SGLT2i therapy, beyond 12 weeks. Our observation of maintenance of constant metabolic rate with SGLT2i therapy (despite weight loss) reflects previous reports in the literature. 18 Explanations may include preserved muscle mass, and possible direct effects of Dapagliflozin on thecontrol of energy expenditure.
We demonstrated a mean body fat mass loss of 9.9 kg with Dapagliflozin over 12 months. However, the study size and design limits generalisability of our findings. One explanation is that we combined Dapagliflozin with a low carbohydrate diet (LCD).
Participants were encouraged to restrict their dietary intake of carbohydrates to 100 g per day, although none managed to adhere strictly to a LCD of this extent. EuDKA can affect patients with both Type 1 Diabetes (T1D) and T2D, and is characterized by the presence of ketones, acidosis and normal blood sugar level. It is a rare side effect of treatment with SGLT2i therapies, and was first described in patient with T2DM and Prader-Willi syndrome on a LCD by Hayami et al. 20 and in two patients with T2D following elective surgery. 21 As it is a potentially dangerous side effect of SGLT2i therapies, prompt testing for urinary or plasma ketones, with plasma pH is important in patients who present with malaise, abdominal pain, vomiting or respiratory distress. The risk of developing euDKA is much higher in patients with T1D, and therefore it is important to establish the correct subtype of Diabetes Mellitus for a patient who is started on a SGLT2i therapy. 21 A risk assessment should be performed by a diabetes clinician prior to the recommendation of a LCD concomitant with the administration of a SGLT2i therapy in T2D.
Existing evidence for SGLT2i therapies suggests either appetite enhancement 22 or no effect on appetite. 23 Among participants with T2D, Dapagliflozin (5 mg) was shown to increase appetite (at 3 months) for sugar-rich foods. 24 Absence of SGLT2i-induced appetite enhancement in our study is surprising as serum levels of appetite hormones often change following weight loss. 25  A recent meta-analysis (with 3-6 months follow-up) showed that therapy with SGLT2i therapies associate with a significant reduction in serum leptin by 290 pg/ml. 27 We demonstrated a much greater reduction in serum leptin in response to Dapagliflozin therapy (852 pg/ mL per month). One explanation for this discrepancy is the greater loss of body fat mass in our study compared with that in other studies. Our data on serum adiponectin changes are less easily explained.
Imbeault et al. 28 showed that serum adiponectin concentration increased following 8-9% weight loss. This was further supported by data reported by Lin et al. 29 with serum levels of adiponectin negatively correlating with changes in body fat mass. However, other studies have not shown any rise in serum (total) adiponectin levels following body weight loss, possibly due to a redistribution of adiponectin oligomers in response to weight loss, not apparent when total adiponectin level is measured. 30 Higher molecular weight adiponectin is a better predictor of glucose itolerance than total adiponectin. 31 In our study, it is possible that the reduction of body fat mass was insufficient in magnitude or duration for changes in serum (total) adiponectin to occur.
The main limitation of our study was a relatively small sample size. This was entirely pragmatic, reflecting the complexity of study with T2D and class 3 obesity resulted in a significant and substantial reduction of body weight, fat mass and serum leptin, but without any discernible compensatory changes in metabolic rate or appetite. Our study promotes a need for a randomized controlled trial investigating the metabolic effects of SGLT2i therapies combined with LCD in people withT2D and obesity.

ACK N OWLED G EM ENTS
We would like to acknowledge and thank AstraZeneca for their financial support of this investigator-initiated research study, as well as the staff and participants who took part in this study.

FU N D I N G I N FO R M ATI O N
Grants supporting this project: Investigator-initiated research study funded by AstraZeneca (Thomas M Barber was PI on this grant). Role of study funder: The study funder was not involved in the design of the study; the collection, analysis and interpretation of data; writing the report; and did not impose any restrictions regarding the publication of the report.

CO N FLI C T O F I NTE R E S T
None of the authors has any conflict of interest. This study was funded by AstraZeneca.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data are available on request from the corresponding author.