Dapagliflozin does not impair microvascular hemorheology in type 2 diabetes mellitus: A multicenter randomized controlled trial (D‐PATH study)

ABSTRACT

Aims

Sodium–glucose cotransporter 2 inhibitors reduce cardiovascular and renal events beyond glucose lowering, but their hematocrit‐raising effects raise concerns about increased blood viscosity and microvascular impairment. We investigated whether dapagliflozin adversely affects microvascular hemorheology in patients with type 2 diabetes.

Materials and Methods

In this multicenter, open‐label, randomized controlled trial, 82 adults with type 2 diabetes received dapagliflozin (5 mg/day) or conventional therapy for 16 weeks. The primary endpoint was the change in whole blood passage time (WBPT), measured using a microchannel flow analyzer that simulates precapillary arterioles (7 × 7 μm), with a prespecified noninferiority margin of 6.0 s (15%). Secondary outcomes included apparent microvascular viscosity, adhesive leukocyte count, and oxidative stress markers. Serum erythropoietin (EPO) was evaluated in an ancillary analysis.

Results

WBPT changed minimally in the dapagliflozin group (+1.4 s; 95% CI, 0.0–2.9) and slightly decreased in the control group (−0.6 s; 95% CI, −1.9 to 0.6), yielding a between‐group difference of 2.0 s (95% CI, −0.2 to 3.9), with the upper bound below the prespecified noninferiority margin. Microvascular viscosity, hematocrit‐standardized viscosity, leukocyte adhesion, and oxidative stress markers remained stable in both groups. Dapagliflozin increased hematocrit and EPO, particularly in participants with lower baseline hematocrit, whereas those with hematocrit >45% showed minimal change.

Conclusions

Dapagliflozin did not impair microvascular hemorheology despite modest increases in hematocrit and EPO. These findings support the microvascular safety of dapagliflozin and reinforce its role in cardio–renal risk management in type 2 diabetes.

Graphical abstract illustrating the study design, physiological rationale, and key findings of the D‐PATH trial. Dapagliflozin, an SGLT2 inhibitor, lowers HbA1c and body weight while increasing hematocrit and erythropoietin levels. Although an elevated hematocrit could theoretically increase blood viscosity and impair microcirculatory flow, ex vivo assessment using a microchannel flow analyzer (MC‐FAN; 7,854 parallel channels with 7 × 7 μm square cross‐sections and 30 μm in length) demonstrated preserved microvascular hemorheology. Whole blood passage time (WBPT) remained within the predefined noninferiority margin (between‐group difference +2.0 s vs noninferiority margin of +6 s), with no significant changes in leukocyte adhesion or oxidative stress markers. In terminal arterioles, red blood cells (RBCs) align along the central flow axis, creating a plasma‐rich layer near the vessel wall. The velocity difference between RBCs and plasma reduces the apparent hematocrit and viscosity—a phenomenon known as the Fåhræus–Lindqvist effect. These findings provide the first direct evidence that physiological hematocrit elevation during SGLT2 inhibition preserves microvascular flow dynamics at the terminal arteriole level, a mechanism that may contribute to the observed cardio–renal protective effects of dapagliflozin in patients with type 2 diabetes mellitus.

INTRODUCTION

The global burden of type 2 diabetes mellitus (T2DM) continues to rise, with 1.31 billion individuals projected to be affected by 2050, posing a major public health challenge ¹ . Cardiovascular complications, including coronary artery disease, stroke, and heart failure, remain the leading causes of morbidity and mortality in patients with T2DM ² , ³ . Although intensive glycemic control effectively reduces microvascular complications, its impact on macrovascular outcomes is limited ⁴ , ⁵ , underscoring the need for novel therapeutic strategies.

Sodium–glucose cotransporter‐2 (SGLT2) inhibitors have emerged as key agents that improve cardiovascular and renal outcomes beyond glucose lowering by inhibiting glucose reabsorption in the proximal tubules ⁶ , ⁷ , ⁸ , ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ . Landmark trials, including EMPA‐REG OUTCOME ⁶ , DECLARE‐TIMI 58 ⁷ , DELIVER ⁹ , and EMPEROR‐Preserved ¹⁰ , demonstrated substantial reductions in heart failure hospitalization and cardiovascular mortality. Their renoprotective effects have also been confirmed in patients with ¹¹ , ¹² and without T2DM ¹³ .

A consistent physiological effect of SGLT2 inhibitors is a modest increase in hematocrit and hemoglobin levels ⁴ , ⁵ , ⁶ , ⁷ , ⁸ , ⁹ . Although initially attributed to hemoconcentration, accumulating evidence indicates stimulation of erythropoiesis through increased erythropoietin production and improved renal oxygenation ¹⁴ , ¹⁵ , ¹⁶ . Importantly, hematocrit elevation mediates a significant proportion of the cardiovascular benefits of SGLT2 inhibitors, particularly in individuals with chronic kidney disease–related anemia ¹⁴ . However, elevated hematocrit and blood viscosity have traditionally been associated with increased thrombotic risk and impaired microvascular flow ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ , ²¹ , raising concerns regarding potential adverse hemorheological effects.

Empagliflozin has been reported to increase whole‐blood viscosity at arterial shear rates, concomitant with a rise in hematocrit ²² . However, viscosity assessed under macrovascular conditions does not accurately reflect microvascular hemorheology. In small vessels, particularly those ≤8 μm in diameter, apparent viscosity decreases owing to the Fåhræus–Lindqvist effect, which reduces flow resistance and effective microvascular hematocrit ²³ , ²⁴ , ²⁵ . Consequently, macrovascular viscosity measurements are poor surrogates for microvascular perfusion, and no prospective randomized trial has directly evaluated whether SGLT2 inhibitor–induced erythrocytosis adversely affects microvascular blood fluidity or leukocyte adhesion.

Mechanistic data suggest that SGLT2 inhibitors preferentially reduce interstitial fluid volume while preserving circulating blood volume, potentially limiting viscosity‐related risks ²⁶ , ²⁷ . In addition, observational studies indicate no increase in arterial thrombotic events associated with SGLT2 inhibitor–related erythrocytosis ²⁸ .

To address this evidence gap, we conducted a multicenter randomized controlled trial to assess whether 16 weeks of dapagliflozin therapy is noninferior to conventional therapy with respect to microvascular hemorheology in adults with T2DM. Using a validated microchannel flow analyzer simulating precapillary arterioles (~8 μm), we evaluated whole blood passage time, microvascular viscosity, and leukocyte adhesion under physiologically relevant conditions ²⁹ . We hypothesized that dapagliflozin would not impair microvascular hemorheology, even in individuals with elevated hematocrit or poor glycemic control, and additionally examined oxidative stress as a mechanistic parameter.

MATERIALS AND METHODS

Study patients

The inclusion criteria were as follows: (a) age 20–75 years at the time of informed consent, (b) T2DM with HbA1c >57 mmol/mol (7.0%), (c) diabetic nephropathy stage ≤3b, and (d) receiving outpatient treatment for diabetes. The exclusion criteria were as follows: (a) type 1 diabetes mellitus; (b) history of hypersensitivity to dapagliflozin or any of its components; (c) severe ketosis, diabetic coma, or precoma; (d) severe infections, pre‐ or postoperative infections, or serious trauma; (e) severe liver dysfunction; (f) susceptibility to dehydration; (g) urinary tract or genital infection; (h) history of cerebrovascular disease within the past 3 months; (i) pregnancy or breastfeeding; (j) eGFR <30 ml/min/1.73 m²; (k) use of warfarin or novel oral anticoagulants (dabigatran, rivaroxaban, edoxaban, or apixaban); (l) current smoking; and (m) deemed unsuitable for the clinical study by investigators.

A total of 90 patients with T2DM were enrolled. This study was conducted in accordance with the Clinical Trials Act of Japan as a specified clinical trial, with approval from the relevant CRB and IRB. Although measurement of serum EPO concentrations was not included in the original clinical trial protocol submitted to the CRB, several reports published between 2019 and 2023 highlighted the importance of EPO‐mediated mechanisms underlying SGLT2 inhibitor–associated increases in hematocrit ¹⁴ , ¹⁵ , ¹⁶ . In light of these findings, we obtained approval from the IRB of Dokkyo Medical University Nikko Medical Center to retrospectively analyze serum EPO concentrations using stored residual serum samples from patients treated at the center (ethical license number: Nikko B25025).

Of the 90 enrolled patients, 45 were randomized to the dapagliflozin group, and 45 were randomized to the control group through central randomization (Figure 1). In the dapagliflozin group, one patient withdrew consent and two were excluded due to inappropriate measurements, leaving 42 patients in the full analysis set (FAS). In the control group, three patients withdrew consent, and two were excluded due to inappropriate measurements, leaving 40 patients in the FAS.

Figure 1.

Patient allocation flowchart. Flowchart illustrating patient enrollment and allocation. A total of 90 patients were randomized to the dapagliflozin group (n = 45) or the control group (n = 45). In the dapagliflozin group, one patient withdrew consent and two were excluded due to inappropriate measurement, resulting in 42 patients in the full analysis set (FAS). In the control group, three patients withdrew consent, and two were excluded for similar reasons, resulting in 40 patients in the FAS. After further exclusions for protocol deviation (one in the dapagliflozin group and two in the control group), 41 patients in the dapagliflozin group and 38 in the control group were included in the per‐protocol set (PPS).

Protocol

This study was an open‐label, randomized, comparative clinical trial designed to evaluate the effects of dapagliflozin on hemorheological parameters and oxidative stress. After written informed consent, eligibility and baseline hemorheological assessments were performed. Eligible participants were randomized using a web‐based minimization system with allocation factors of sex, age (≥70 or <70 years), hematocrit (≥40% or <40%), and whole blood passage time (WBPT ≥50 or <50 s, measured with the 7‐7‐7D microchannel chip).

Blood and urine samples were collected 16 weeks (±3 weeks) after treatment initiation. During the study, use of any SGLT2 inhibitor and oral anticoagulants (warfarin, dabigatran, rivaroxaban, edoxaban, or apixaban) was prohibited.

Patients and the public were not involved in the design, conduct, or reporting of this study.

Treatment regimens

Participants in the dapagliflozin group received dapagliflozin 5 mg/day in addition to standard therapy, with optional up‐titration to 10 mg/day if glycemic control was inadequate. The control group received standard therapy without SGLT2 inhibitors. Adjustment of concomitant medications was permitted at the discretion of the treating physicians.

Standard therapy

Standard therapy included dietary counseling, adequate hydration (>1.5 L/day), alcohol restriction, smoking cessation, regular walking exercise, and continuation of appropriate pharmacotherapy. Pioglitazone was avoided because of its potential effects on fluid retention. No mandatory treat‐to‐target or forced titration protocol was applied.

Assessment of whole blood rheology and leukocyte activity using an ex vivo microchannel model

Whole‐blood rheology in microvessels was assessed using an MC‐FAN device (BWAMCFAN; Kikuchi Microtechnology Co., Ltd., Ibaraki, Japan) equipped with a BK 7‐7‐7D chip containing 7,854 parallel microchannels (7 × 7 μm, 30 μm length). Under a constant negative pressure of −30 cm H₂O (approximate shear rate 2,940 s⁻¹), the system models precapillary arteriole flow and measures WBPT for a 0.1‐ml blood sample. The methodology and validation have been described previously ²⁹ , with intra‐ and interobserver variability of 7.3% and 9.1%, respectively ³⁰ . This system has been used in basic ³¹ , ³² , ³³ and clinical studies ³⁴ , ³⁵ , ³⁶ .

Blood samples collected in heparinized tubes were maintained at 37°C and analyzed within 30 min. Before each measurement, the saline passage time was determined for calibration ²⁹ . A predefined noninferiority margin of 6 s (15%) for WBPT was applied, as described in non‐inferiority margin derivation.

Microchannel flow images were recorded, and after 0.08–0.10 mL of blood passage, five randomly selected still images were analyzed offline. Adhesive leukocytes were defined as stationary cells with clearly demarcated borders and were counted accordingly ²⁹ , ³⁴ , ³⁵ , ³⁶ .

Apparent relative viscosity at the precapillary arteriole level was assessed using the passage time of 0.05 ml of blood. The ratio of whole‐blood to saline passage time approximates apparent relative viscosity, numerically corresponding to absolute viscosity (cP), as saline viscosity is defined as 1.0 cP ²⁹ .

Noninferiority margin derivation

The noninferiority margin and sample size were based on preliminary data from a previous study (Japan Diabetes Society Annual Meeting, 2019). In individuals with diabetes (n = 15), mean WBPT was 41.0 ± 5.9 s and increased to 42.6 ± 6.1 s after 2 months of dapagliflozin treatment (mean change +1.6 s). Assuming no change without dapagliflozin, a prespecified margin of +6 s (≈15%) was selected as the maximum clinically negligible increase. This margin exceeds the observed treatment effect, lies within measurement variability (SD ≈ 6.1 s), and is unlikely to reflect clinically meaningful deterioration in microcirculatory rheology, supporting its statistical and clinical appropriateness.

Sample size calculation

The sample size for each group was calculated using the following formula:

$$n_2=\frac{\left(Z_{1-\alpha }+Z_{1-n}\right){\delta }^2\left(1+\frac{1}{\kappa }\right)}{{\left(\in \mathit{ー\delta }\right)}^2};\epsilon ={\mu }_2-{\mu }_1;\kappa =\frac{n_1}{n_2};n_1=\kappa n_2.$$

where

• SD (σ) = 6.10

• Mean difference between groups (ε) = 1.60

• Noninferiority margin (δ) = 6.00

• Group size ratio (κ) = 1.00

• α = 0.025 (one‐sided), Z(0.975) = 1.96

• β = 0.10, Z(0.900) = 1.28

This yielded a sample size of n ₁ = 40.3 per group. Assuming a 10% dropout rate, the final target sample size was 90 participants (45 in each group).

Primary outcomes

The primary outcome was the change in WBPT (0.1 ml) from baseline to 16 weeks, assessed using an MC‐FAN (BWAMCFAN; Kikuchi Microtechnology Co., Ltd., Ibaraki, Japan) with a BK 7–7‐7D chip. Results are presented as mean changes with 95% confidence intervals (CI). A prespecified noninferiority margin of +6 s was applied; noninferiority was concluded if the upper limit of the two‐sided 95% CI for the between‐group difference was <6 s.

Secondary outcomes

Secondary outcomes included whole blood viscosity, hematocrit‐standardized viscosity, adhesive leukocyte count, and their changes from baseline to 16 weeks, all measured using the same MC‐FAN system. Additional laboratory assessments included complete blood count, lipid profile, high‐sensitivity C‐reactive protein, serum creatinine, serum reactive oxygen metabolites (d‐ROM), and biological antioxidant potential (BAP; FREE Carpe Diem, Diacron SRL, Grosseto, Italy). Subgroup analyses were performed according to baseline WBPT (above vs. below median) and baseline HbA1c (>68 vs. ≤68 mmol/mol; >8.0% vs. ≤8.0%).

Statistical analysis

All analyses were performed using IBM SPSS Statistics version 29 (IBM Corp., Armonk, NY, USA) and JMP 16.0 J software (SAS Institute, Cary, NC, USA). The primary analysis was conducted in the FAS, and a sensitivity analysis was performed in the per‐protocol set (PPS). Missing data were not imputed, and analyses were performed using available cases in the FAS (complete‐case analysis).

Primary outcomes

Changes in WBPT were analyzed using 95% CIs based on the t‐distribution, with a noninferiority margin of 6 s and a one‐sided significance level of 2.5%.

Secondary outcomes

Between‐group comparisons of continuous variables were performed using Student's t‐test, and categorical variables using Fisher's exact test. A two‐tailed P‐value of <0.05 was considered statistically significant. Subgroup analyses according to baseline hematocrit quartiles were exploratory, without adjustment for multiple comparisons.

Adverse events and safety monitoring

Adverse events were summarized and compared between groups using Fisher's exact test, with a two‐sided significance level of 5%. Adverse events were monitored throughout the study period at each study visit and were recorded based on spontaneous reporting by participants and clinical assessment by investigators. Prespecified safety endpoints included infections (e.g., urinary tract and genital infections), dehydration‐related events, and other clinically significant adverse events. All adverse events were evaluated for severity and potential relationship to the study treatment at the investigators' discretion.

Ancillary analysis

Changes in serum erythropoietin (EPO) from baseline to 16 weeks were analyzed within and between groups using Student's t‐test. Subgroup analyses by baseline hematocrit quartiles were exploratory, and P‐values were not adjusted for multiple comparisons.

RESULTS

A total of 90 patients were enrolled and randomized (45 to dapagliflozin and 45 to control). After accounting for consent withdrawal and exclusions due to inappropriate measurements, 82 patients comprised the FAS: 42 in the dapagliflozin group and 40 in the control group. The PPS was defined as participants in the FAS who completed the study without major protocol deviations and had evaluable primary endpoint data. Based on this definition, 79 patients comprised the PPS: 41 in the dapagliflozin group and 38 in the control group. Protocol deviations were identified in one patient in the dapagliflozin group and two patients in the control group. Medication adherence exceeded 95% in both groups. Baseline characteristics are shown in Table 1. Overall, demographic and clinical characteristics were well balanced between the two groups, with no clinically meaningful differences observed that would be expected to influence hemorheological outcomes.

Table 1.

Demographic, clinical, and laboratory characteristics of study patients with type 2 diabetes mellitus at baseline, and details of medications in the dapagliflozin and control groups (full analysis set, mean ± SD)

Characteristics	Dapagliflozin (n = 42)	Nondapagliflozin (n = 40)
Patients (male/female)	25/17	25/15
Age (years)	62.4 ± 8.9	64.7 ± 8.0
Diabetic duration (years)	11.0 ± 7.6	11.9 ± 8.3
Hypertension	33 (78.6%)	30 (75.0)
Dyslipidemia	28 (66.7%)	34 (85%)
Diabetic kidney disease
Stage 1	23 (54.8%)	25 (62.5%)
Stage 2	11 (26.2%)	12 (30.0%)
Stage 3	8 (19.0%)	3 (7.5%)
Previous history of CVD
Myocardial infarction	3 (7.1%)	7 (17.5%)
Angina	4 (9.5%)	10 (25.0%)
Cerebrovascular disease	1 (2.4%)	3 (7.5%)
Lower extremity arterial disease	0 (0.0%)	0 (0.0%)
Body weight (kg)	75.0 ± 14.9	68.7 ± 11.1
BMI (kg/m2)	28.1 ± 4.8	26.2 ± 3.9
Systolic BP (mmHg)	134.5 ± 13.1	133.7 ± 13.1
Diastolic BP (mmHg)	78.9 ± 9.8	75.7 ± 9.4
Heart rate (beats/min)	76.8 ± 10.7	75.6 ± 11.6
HbA1c (mmol/mol)	64.4 ± 11.5	62.6 ± 8.4
HbA1c (%)	8.1 ± 1.0	7.9 ± 0.8
Plasma glucose (mg/dL)	165.29 ± 56.5	165.45 ± 70.90
HDL‐c (mg/dL)	50.7 ± 12.2	49.1 ± 8.9
LDL‐c (mg/dL)	86.3 ± 30.3	85.3 ± 30.2
TG (mg/dL)	162.5 ± 106.9	134.6 ± 60.7
AST (IU/L)	35.6 ± 27.0	31.9 ± 21.5
ALT (IU/L)	34.0 ± 29.6	31.3 ± 24.3
Creatine kinase (U/L)	109.3 ± 88.8	126.2 ± 68.9
BUN (mg/dL)	16.3 ± 5.6	15.4 ± 4.8
Creatinine (mg/dL)	0.83 ± 0.23	0.81 ± 0.19
eGFR (ml/min/1.73 m2)	69.5 ± 18.5	70.6 ± 19.3
White blood cells (/μL)	6,914 ± 1832	6,332 ± 1798
Red blood cell (106/μL)	4.70 ± 0.6	4.72 ± 0.4
Hemoglobin (g/dL)	14.2 ± 1.6	14.3 ± 1.3
Hematocrit (%)	42.0 ± 4.8	42.3 ± 3.3
plt (104/μL)	22.8 ± 4.9	19.8 ± 5.3
BNP (pg/dL)	23.0 ± 52.6	23.3 ± 22.7
Hs‐CRP (mg/dL)	0.36 ± 0.84	0.31 ± 0.74
Urine specific gravity	1.017 ± 0.01	1.017 ± 0.01
d‐ROMs (UARR)	317.2 ± 52.4	319.5 ± 48.6
BPA (μmol/L)	2033.1 ± 234.4	2006.7 ± 145.3
WBPT (s)	41.8 ± 6.1	40.5 ± 3.0
Whole blood relative viscosity	3.04 ± 0.65	3.00 ± 0.25
Hematocrit‐corrected whole blood viscosity (at Hct 45%)	3.23 ± 0.56	3.22 ± 0.56
Adhesive leukocyte number/field	16.5 ± 15.7	13.9 ± 9.1
Medication
Insulin	7	8
Biguanide	23	29
GLP‐1 receptor antagonist	8	4
DPP‐4 inhibitor	32	22
Sulphonylurea/glinide	15	11
Thiazolidine	0	4
Alfa GI	9	2
Statin	24	21
ARB/ACE inhibitor	26	24
Calcium channel blocker	20	23
Diuretics	6	6
Beta blocker	10	5
Alpha blocker	5	4
Antiplatelet drugs	11	8

ACE inhibitor, angiotensin‐converting enzyme inhibitor; Alpha‐GI, α‐glucosidase inhibitor; ALT, alanine aminotransferase; ARB, angiotensin II receptor blocker; AST, aspartate aminotransferase; BMI, body mass index; BNP, brain natriuretic peptide; BUN, blood urea nitrogen; CVD, cardiovascular disease; DBP, diastolic blood pressure; d‐ROM, derivatives of reactive oxygen metabolites; DPP‐4i, dipeptidyl peptidase‐4 inhibitor; eGFR, estimated glomerular filtration rate; GLP‐1 RA, glucagon‐like peptide‐1 receptor agonist; HbA1c, glycated hemoglobin; HDL‐c, high‐density lipoprotein cholesterol; hs‐CRP, high‐sensitive C‐reactive protein; LEAD, lower extremity arterial disease; LDL‐c, low‐density lipoprotein cholesterol; PPG, postprandial glucose; SBP, systolic blood pressure; SU, sulfonylurea; TG, triglyceride; UARR, units of arbitrary reactive metabolites.

Primary outcome

Baseline WBPT was 41.8 ± 6.1 s in the dapagliflozin group and 40.5 ± 3.0 s in the control group. After 16 weeks, the corresponding values were 43.2 ± 6.7 s and 40.1 ± 3.6 s, respectively (Table 2). The mean change from baseline was +1.4 s (95% CI, 0.0 to 2.9) in the dapagliflozin group and −0.6 s (95% CI, −1.9 to 0.6) in the control group (Table 3). The between‐group difference was 2.0 s (95% CI, −0.2 to 3.9). As the upper bound of the confidence interval was below the prespecified noninferiority margin (+6 s), noninferiority was confirmed (Table 4).

Table 2.

Intergroup comparison of individual parameter values in the full analysis set (mean ± SD)

	Dapagliflozin group (n = 42)	Nondapagliflozin group (n = 40)	95% confidence interval	P
Body weight at 16 weeks	72.4 ± 15.7	68.9 ± 11.4
Change in body weight from baseline to 16 weeks	−2.5 ± 5.2	0.0 ± 2.1	−2.5 (−4.2, −0.7)	0.006
BMI at 16 weeks	27.5 ± 5.2	26.5 ± 3.8
Change in body weight from baseline to 16 weeks	−0.6 ± 2.8	−0.2 ± 0.0	−0.5 (−0.8, −0.2)	0.001
Systolic blood pressure at 16 weeks	130.0 ± 16.6	128.5 ± 15.8
Change in systolic blood pressure from baseline to 16 weeks	−5.0 ± 11.5	−5.2 ± 16.9	0.3 (−5.9, 6.5)	0.932
Diastolic blood pressure at 16 weeks	75.1 ± 12.0	72.6 ± 11.2
Change in diastolic blood pressure from baseline to 16 weeks	−4.2 ± 8.4	−3.1 ± 10.5	−1.1 (−5.1, 3.0)	0.623
Heart rate at 16 weeks	76.8 ± 10.7	75.6 ± 11.6
Change in heart rate from baseline to 16 weeks	−2.0 ± 10.3	2.6 ± 11.4	−3.5 (−8.5, 1.5)	0.076
HbA1c(mol/mol) at 16 weeks	57.7 ± 8.7	58.2 ± 9.3
Change in HbA1c from baseline to 16 weeks	−6.7 ± 6.0	−4.3 ± 8.4	−2.3 (−5.5, −0.9)	0.117
HbA1c at 16 weeks	7.4 ± 0.8	7.5 ± 0.8
Change in HbA1c (%) from baseline to 16 weeks	−0.6 ± 0.6	−0.4 ± 0.8	−0.2 (−0.5, 0.1)	0.117
Plasma glucose at 16 weeks	138.0 ± 36.4	158.4 ± 57.1
Change in plasma glucose from baseline to 16 weeks	−27.3 ± 50.8	−7.0 ± 54.9	−20.2 (−43.0, 2.6)	0.078
HDL‐C at 16 weeks	52.4 ± 14.6	49.9 ± 8.4
Change in HDL‐C from baseline to 16 weeks	2.1 ± 6.5	0.9 ± 5.3	−3.5 (−1.5, 3.6)	0.393
LDL‐C at 16 weeks	88.2 ± 27.9	82.7 ± 25.7
Change in LDL‐C from baseline to 16 weeks	2.5 ± 20.2	−3.3 ± 20.0	5.2 (−3.41, 13.9)	0.193
Serum level of TG at 16 weeks	127.3 ± 69.5	120.7 ± 54.2
Change in TG from baseline to 16 weeks	−37.2 ± 62.4	−11.8 ± 59.6	−29.1 (−56.0, −2.5)	0.069
Aspartate aminotransferase (AST) at 16 weeks	33.8 ± 28.0	27.7 ± 10.9
Change in AST from baseline to 16 weeks	−2.1 ± 10.9	−4.2 ± 18.8	3.3 (−4.9, 8.3)	0.537
Alanine aminotransferase (ALT) at 16 weeks	29.9 ± 29.8	24.2 ± 14.1
Change in ALT from baseline to 16 weeks	−4.7 ± 16.1	−7.1 ± 19.0	1.6 (−6.0, 9.2)	0.540
Creatine kinase at 16 weeks	160.4 ± 280.7	138.6 ± 70.8
Change in creatine kinase from baseline to 16 weeks	49.6 ± 276.6	11.7 ± 73.7	37.4 (−51.6, 126.3)	0.411
Blood urea nitrogen (BUN) at 16 weeks	18.8 ± 6.5	15.0 ± 3.9
Change in BUN from baseline to 16 weeks	2.6 ± 3.4	−0.4 ± 3.6	2.9 (1.4, 4.4)	0.00024
Creatinine at 16 weeks	0.86 ± 0.26	0.79 ± 0.19
Change in creatinine from baseline to 16 weeks	0.03 ± 0.09	−0.01 ± 0.07	0.04 (0.01, 0.07)	0.025
Estimated glomerular filtration rate (eGFR) at 16 weeks	67.7 ± 19.7	68.9 ± 18.3
Change in eGFR from baseline to 16 weeks	−1.9 ± 6.9	0.6 ± 8.0	−3.1 (−6.6, 0.4)	0.116
WBC at 16 weeks	6,730 ± 1,500	5,880 ± 1,470
Change in WBC from baseline to 16 weeks	−90 ± 1,140	−45 ± 110	−313.7 (−173.3, 798.6)	0.155
RBC at 16 weeks (*106/μL)	4.9 ± 0.6	4.7 ± 0.4
Change in RBC from baseline to 16 weeks (*106/μL)	0.2 ± 0.2	−0.1 ± 0.2	0.2 (0.1, 0.4)	0.0000020
Hemoglobin at 16 weeks	14.5 ± 1.5	13.9 ± 1.3
Change in Hemoglobin from baseline to 16 weeks	0.4 ± 0.7	−0.3 ± 0.5	0.6 (0.3, 0.8)	0.00002607
Hematocrit at 16 weeks	43.7 ± 4.7	41.5 ± 3.3
Change in Hematocrit from baseline to 16 weeks	1.8 ± 4.7	−0.6 ± 1.7	2.1 (1.2, 3.0)	0.00000011
Platelet at 16 weeks (*104/μL)	22.9 ± 5.1	20.3 ± 7.6
Change in platelet from baseline to 16 weeks (*104/μL)	0.0 ± 1.9	0.6 ± 4.8	−0.5 (−2.1, 1.1)	0.464
Log brain natriuretic peptide (BNP) at 16 weeks	1.05 ± 0.3	1.27 ± 0.4
Change in log BNP from baseline to 16 weeks	0.03 ± 0.3	0.3 ± 0.4	−1.0 (−1.3, −05)	0.008
hsCRP at 16 weeks	0.3 ± 0.8	0.5 ± 1.4
Change in hsCRP from baseline to 16 weeks	−1.0 ± 4.8	0.1 ± 1.5	0.3 (−0.8, 0.3)	0.220
Urine specific gravity at 16 weeks	1.024 ± 001	1.016 ± 0.001
Change in urine specific gravity from baseline to 16 weeks	0.008 ± 0.01	0.000 ± 0.01	0.01 (0.004, 0012)	0.00002326
d‐ROMs at 16 weeks	328.9 ± 39.0	326.33 ± 53.28
Change in d‐ROMs from baseline to 16 weeks	10.0 ± 43.9	6.8 ± 36.8	3.9 (−14.1, 22.0)	0.723
BAP at 16 weeks	2102.2 ± 219.0	2013.9 ± 193.8
Change in BAP from baseline to 16 weeks	68.4 ± 254.3	3.3 ± 187.2	66.8 (−32.8, 166.4)	0.194
Whole blood passage time at 16 weeks	43.2 ± 6.7	40.1 ± 3.6
Change in whole blood passage time from baseline to 16 weeks	1.4 ± 4.6	−0.6 ± 6.3	2.0 (−0.2, 3.9)	0.0360
Whole blood relative viscosity at 16 weeks	3.13 ± 0.39	2.96 ± 0.25
Change in whole blood relative viscosity from baseline to 16 weeks	0.101 ± 0.483	−0.036 ± 0.225	0.1 (−0.1, 0.3)	0.1170
Hematocrit‐corrected whole blood viscosity (at Hct 45%) at 16 weeks	3.23 ± 0.20	3.21 ± 0.61
Change in hematocrit‐corrected whole blood viscosity (at Hct 45%) from baseline to 16 weeks	0.0837 ± 0.756	−0.0075 ± 0.448	0.1 (−0.2, 0.4)	0.437
Erythropoietin (mlU/mL) at 16 weeks	15.3 ± 8.0	16.4 ± 8.0
Change in erythropoietin from baseline to 16 weeks	5.4 ± 8.7	1.3 ± 8.4	4.3 (0.4, 8.1)	0.02

ALT, alanine aminotransferase; AST, aspartate aminotransferase; BAP, biological antioxidant potential; BNP, brain natriuretic peptide; BUN, blood urea nitrogen; HbA1c, hemoglobin A1c; HDL‐C, high‐density lipoprotein cholesterol; Hct, hematocrit; hsCRP, high‐sensitivity C‐reactive protein; LDL‐C, low‐density lipoprotein cholesterol; RBC, red blood cell; TG, triglycerides; WBC, white blood cell; WBPT, whole blood passing time. Note: Bold values indicate statistically significant differences (P < 0.05).

Table 3.

Descriptive statistics of whole blood passage time as the primary endpoint in the full analysis set

Whole blood passage time (s)	Dapagliflozin group (n = 42)	Nondapagliflozin group (n = 40)
Baseline
Mean ± SD	41.8 ± 6.1	40.5 ± 3.0
16 weeks
Mean ± SD	43.4 ± 6.8	39.9 ± 3.7
Change from baseline to 16 weeks
Mean ± SD	1.6 ± 4.4	−0.5 ± 3.3

IQR, interquartile range; SD, standard deviation.

Table 4.

Noninferiority assessment of changes in whole blood passage time from baseline to 16 weeks in the full analysis set

	Dapagliflozin group (n = 42)	Nondapagliflozin group (n = 40)	Between‐group difference	Noninferiority assessment
Mean [95% CI]	1.6 [0.2, 3.0]	−0.5 [−1.6, 0.5]	2.2 [0.4, 3.9]	Noninferior

The noninferiority margin was pre‐specified at +6 s. Noninferiority was evaluated using a one‐sided significance level of 2.5%, corresponding to the upper bound of the two‐sided 95% CI for the between‐group difference. Noninferiority was concluded if the upper bound of the 95% CI for the treatment difference was <6 s. If the upper bound was ≥6 s, noninferiority was not established. CI, confidence interval.

Although a nominal two‐sided P‐value was calculated (P = 0.036), the study was not designed or powered to assess superiority; therefore, this result should be interpreted as descriptive. Results were consistent in the PPS (Tables 5 and 6).

Table 5.

Descriptive statistics of whole blood passage time as the primary endpoint in the per‐protocol set

Whole blood passage time (s)	Dapagliflozin group (n = 41)	Nondapagliflozin group (n = 36)
Baseline
Mean ± SD	41.6 ± 6.0	40.4 ± 3.0
16 weeks
Mean ± SD	43.2 ± 6.9	39.8 ± 3.6
Change from baseline to 16 weeks
Mean ± SD	1.7 ± 4.4	−0.5 ± 3.3

IQR, interquartile range; SD, standard deviation.

Table 6.

Noninferiority assessment of changes in whole blood passage time from baseline to 16 weeks in the per‐protocol set

	Dapagliflozin group (n = 41)		Nondapagliflozin group (n = 36)		Between‐group difference		Noninferiority assessment
Mean [95% CI]	1.7	[0.3, 3.1]	−0.5	[−1.7, 0.6]	2.2	[0.5, 4.0]	Noninferiority

CI, confidence interval.

The proportion of participants without clinically relevant WBPT prolongation (<6 s) was high in both groups—95.2% in the dapagliflozin group and 95.9% in the control group in the FAS (Table 7)—with similar findings in the per‐protocol analysis (95.1% vs 94.7%; Table 8). Whole blood viscosity and hematocrit‐standardized viscosity under precapillary arteriolar conditions remained stable in both groups, with no clinically meaningful changes observed (Table 2).

Table 7.

Comparison of the nonprolongation rate of whole blood passage time from baseline to 16 weeks in the full analysis set

	Dapagliflozin group (n = 42)		Nondapagliflozin group (n = 40)		P
Prolongation (≥6 s)	n = 2	4.8%	n = 2	5.0%	>0.999
Nonprolongation (<6 s)	n = 40	95.2%	n = 38	95.0%	>0.999

P‐value: Fisher's exact test.

Table 8.

Comparison of the nonprolongation rate of whole blood passage time from baseline to 16 weeks in the per‐protocol set

	Dapagliflozin group (n = 41)		Nondapagliflozin group (n = 38)		P
Prolongation (≥6 s)	n = 2	4.9%	n = 2	5.3%	>0.999
Nonprolongation (<6 s)	n = 40	95.1%	n = 38	94.7%	>0.999

P‐value: Fisher's exact test.

Secondary outcomes

The number of adhesive leukocytes per microscopic field did not significantly change from baseline to week 16 in either group (dapagliflozin: −2.5 ± 18.1; control: −4.0 ± 9.3; P = 0.21). Similarly, no significant within‐ or between‐group differences were observed in oxidative stress markers, including d‐ROMs and BAP (Table 2). These findings indicate that dapagliflozin had no measurable impact on leukocyte adhesion or oxidative balance within the study period.

Subgroup analyses

When stratified by baseline WBPT, noninferiority of dapagliflozin was consistently demonstrated in both subgroups, with between‐group differences of 0.8 s (95% CI, −2.0 to 3.7; P = 0.534) in the higher baseline WBPT subgroup and 3.0 s (95% CI, 0.5 to 5.3; P = 0.017) in the lower baseline subgroup (Figure 2). Similarly, stratification by baseline HbA1c showed consistent results, with between‐group differences of 1.4 s (95% CI, −1.7 to 4.7; P = 0.357) in participants with HbA1c >8.0% and 2.3 s (95% CI, −0.2 to 4.7; P = 0.034) in those with HbA1c ≤8.0% (Figure 2). In all subgroups, the upper bound of the confidence interval remained below the prespecified noninferiority margin, supporting the consistency of the primary finding across different baseline hemorheological and glycemic profiles.

Figure 2.

Noninferiority assessment of changes in whole blood passage time from baseline to 16 weeks by baseline whole blood passage time. Subgroup analysis stratified by baseline whole blood passage time (WBPT; above vs below the median) and baseline HbA1c (>8.0% vs ≤8.0%). Data are presented as mean differences (dapagliflozin—control) with 95% confidence intervals (CI). The prespecified noninferiority margin (+6 s) is shown as a dashed line. Noninferiority was concluded if the upper limit of the two‐sided 95% CI was <6 s. In participants with a baseline transit time above the median (n = 37), the between‐group difference (dapagliflozin—control) was 0.8 s (95% CI, −2.0 to 3.7, P = 0.534), with mean changes of −0.6 s (95% CI, −2.4 to 1.1) in the dapagliflozin group (n = 18) and −1.4 s (95% CI, −3.5 to 0.7) in the control group (n = 19). In participants with a baseline transit time below the median (n = 45), the between‐group difference was 3.0 s (95% CI, 0.5–5.3, P = 0.017), with mean changes of 3.0 s (95% CI, 0.9–5.0) in the dapagliflozin group (n = 24) and 0.0 s (95% CI, −1.3 to 1.3) in the control group (n = 21). In both subgroups, the upper limit of the two‐sided 95% CI was <6 s, indicating that dapagliflozin was noninferior in preserving microvascular hemorheology. In participants with HbA1c >68 mmol/mol (8.0%) (n = 27), the between‐group difference (dapagliflozin—control) was 1.4 s (95% CI, −1.7 to 4.7, P = 0.357), with mean changes of 1.2 s (95% CI, −0.9 to 3.3) in the dapagliflozin group (n = 16) and −0.3 s (95% CI, −3.0 to 2.5) in the control group (n = 11). In participants with HbA1c ≤68 mmol/mol (8.0%) (n = 55), the between‐group difference was 2.3 s (95% CI, ‐0.2 to 4.7), with mean changes of 1.6 s (95% CI, −0.5 to 3.6, P = 0.034) in the dapagliflozin group (n = 26) and −0.7 s (95% CI, −2.2 to 0.8) in the control group (n = 29). In both HbA1c strata, the upper limit of the two‐sided 95% CI for the between‐group difference was <6 s, indicating that dapagliflozin was noninferior in preserving microvascular hemorheology even in patients with poorly controlled diabetes. CI, confidence interval; HbA1c, glycated hemoglobin.

Hematocrit

Hematocrit increased significantly in the dapagliflozin group (42.0 ± 4.8% to 43.7 ± 4.7%; mean change +1.83 ± 4.67%), whereas a small, nonsignificant decrease was observed in the control group (42.3 ± 3.3% to 41.5 ± 3.3%; mean change −0.63 ± 1.65%). The between‐group difference in change was statistically significant (P < 0.0001) (Table 2).

Exploratory analyses by hematocrit quartile showed that increases were more pronounced in participants with lower baseline hematocrit (Q1–Q3) (Figure 3a–c), whereas no significant change was seen in the highest quartile (Q4) (Figure 3d).

Figure 3.

Changes in hematocrit at 16 weeks by baseline hematocrit quartile. Mean change in hematocrit from baseline to 16 weeks in the dapagliflozin and control groups, stratified by baseline hematocrit quartiles: Q1 (<39.5%), Q2 (39.5–41.9%), Q3 (42.0–45.4%), and Q4 (≥45.5%). Data are presented as mean ± SD. In the dapagliflozin group, hematocrit increased consistently in Q1–Q3 (Panels a–c) compared with the control group, whereas no significant changes were observed in Q4 in either group (panel d). All subgroup analyses were exploratory. P values were not adjusted for multiplicity. SD, standard deviation.

Oxidative stress markers

There were no significant changes in d‐ROM compounds or BAP from baseline to week 16 in either group. In addition, no significant between‐group differences were observed (Table 2), indicating no measurable effect of dapagliflozin on oxidative stress or antioxidant capacity during the study period.

Serum EPO

Serum EPO was analyzed in 76 patients using stored residual samples. In the dapagliflozin group, EPO rose from 10.1 ± 5.2 to 14.8 ± 7.7 mIU/mL (P < 0.05), whereas no significant change was observed in the control group (15.1 ± 7.6 to 16.4 ± 8.1 mIU/mL).

In stratified analysis, EPO increased significantly only in the lowest hematocrit quartile (Q1 < 39.5%) in the dapagliflozin group, with no appreciable changes in Q2–Q4 (Figure 4). These findings should be interpreted as exploratory.

Figure 4.

Changes in erythropoietin levels by hematocrit quartile. Change from baseline to 16 weeks in serum erythropoietin levels (mean ± SD) in both treatment groups, stratified by hematocrit quartiles: Q1 (<39.5%), Q2 (39.5–41.9%), Q3 (42.0–45.4%), and Q4 (≥45.5%). Comparisons between groups were performed within each quartile. In Q1 (Panel a), the increase in serum erythropoietin levels was significantly greater in the dapagliflozin group than in the control group (P = 0.035). In contrast, no statistically significant between‐group differences were observed in Q2–Q4 (Panels b–d). Data are shown as mean ± SD. All subgroup analyses were exploratory. P values were not adjusted for multiplicity. SD, standard deviation.

Safety

One case of trichophytosis occurred in the dapagliflozin group. No urinary tract infections were reported.

DISCUSSION

In this randomized controlled trial, 16 weeks of dapagliflozin therapy did not adversely affect hemorheological parameters at the precapillary arteriolar level in individuals with T2DM. Noninferiority was consistently demonstrated for the primary endpoint, WBPT, in the overall cohort and across all predefined subgroups, including participants with elevated baseline WBPT or suboptimal glycemic control. Although hematocrit and EPO levels increased modestly but significantly, there was no deterioration in microvascular blood fluidity, leukocyte adhesion, or oxidative stress markers. These results provide direct in vivo–ex vivo evidence that the dapagliflozin‐induced hematocrit elevation is not associated with adverse microvascular rheological effects.

The observed hematocrit increase aligns with previous studies suggesting that SGLT2 inhibitors stimulate erythropoiesis by enhancing renal cortical oxygenation and EPO production, rather than solely inducing hemoconcentration via osmotic diuresis ¹⁴ , ¹⁵ , ¹⁶ . In our study, EPO levels rose significantly following dapagliflozin treatment, particularly in individuals with lower baseline hematocrit, whereas participants with baseline hematocrit >45.5% exhibited minimal changes in EPO and hematocrit. This pattern is consistent with a physiologically regulated erythropoietic response subject to a homeostatic ceiling, rather than unregulated erythrocytosis. Such a ceiling effect is clinically reassuring, given concerns that marked hematocrit elevation could increase viscosity and thrombosis risk.

Hematocrit is a major determinant of whole blood viscosity and has been associated with cardiovascular risk in epidemiological cohorts ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ , ²¹ , ³⁷ . Recent studies with empagliflozin have demonstrated increases in blood viscosity and carotid wall shear stress in patients with T2DM ²² . In medium‐sized arteries (approximately 3–5 mm in diameter), increased viscosity may enhance endothelial wall shear stress, promoting eNOS expression and nitric oxide bioavailability, thereby supporting anti‐atherogenic endothelial function ¹¹ , ²² , ³⁸ . Thus, within physiological limits, SGLT2 inhibitor–induced hematocrit elevation may contribute to macrovascular protection by improving shear‐dependent endothelial signaling.

Importantly, the rheological effects of hematocrit elevation differ across the vascular tree. In small vessels, the Fåhræus and Fåhræus–Lindqvist effects reduce apparent viscosity as vessel diameter decreases, driven by axial migration of red blood cells and the formation of a cell‐free plasma layer near the vessel wall ²⁶ , ²⁷ , ²⁸ . These diameter‐dependent adaptations are particularly relevant in 8‐μm terminal arterioles, where microvascular resistance and tissue oxygen delivery are critically determined. In the present study, dapagliflozin did not prolong WBPT, nor did it increase whole blood viscosity or hematocrit‐standardized viscosity under precapillary arteriolar conditions, indicating that the modest hematocrit rise did not translate into increased microvascular flow resistance. Our data therefore support a model in which SGLT2 inhibitors may increase viscosity and shear stress in larger arteries—potentially enhancing shear‐mediated endothelial protection—while microvascular rheology remains preserved due to intrinsic hemorheological adaptations.

These observations are consistent with MRI‐based studies showing improved renal cortical oxygenation and perfusion with SGLT2 inhibition ³⁹ and may help explain the renal benefits observed in large outcome trials such as DAPA‐CKD ¹³ , in which dapagliflozin reduced the risk of kidney failure and major renal events across a wide range of baseline kidney function.

Beyond erythropoiesis and rheology, SGLT2 inhibitors exert pleiotropic effects that may contribute to vascular protection. Experimental and clinical evidence indicate that SGLT2 inhibition mitigates oxidative stress, improves mitochondrial function, and modulates inflammatory signaling pathways ¹⁶ , ²⁴ , ³⁸ , ⁴⁰ , ⁴¹ . In the present study, dapagliflozin did not significantly affect leukocyte adhesion or oxidative stress markers; however, these analyses were exploratory, and the study was not powered to detect differences in these endpoints.

Volume regulation is another relevant mechanism. In this trial, dapagliflozin induced modest weight loss without significant changes in blood pressure or pulse rate, suggesting selective reduction of interstitial rather than intravascular volume. This redistribution is consistent with prior reports and may contribute to the favorable hemodynamic profile of SGLT2 inhibitors ²⁵ , ⁴¹ , ⁴² , ⁴³ .

Clinically, these findings are important given concerns that hematocrit elevation may increase blood viscosity and thrombotic risk ¹⁴ , ²² , ²⁴ . Our data demonstrate that dapagliflozin increases hematocrit within a physiologically regulated range, without impairing microvascular blood fluidity or promoting leukocyte adhesion or oxidative stress. These results support dapagliflozin as a safe component of comprehensive cardio–renal risk management in T2DM, including patients with higher baseline hematocrit or suboptimal glycemic control.

Limitations

This study has several limitations. First, the 16‐week follow‐up may not capture long‐term hemorheological adaptations. Second, individuals with advanced chronic kidney disease (eGFR <30 ml/min/1.73 m²) were excluded, limiting generalizability. Third, the sample size was not sufficient to detect small changes in oxidative stress markers or rare adverse events. Fourth, the EPO analysis was not prespecified in the original CRB‐approved protocol and was conducted retrospectively using stored samples; therefore, these findings should be considered exploratory. Finally, the relatively low baseline WBPT in this cohort may limit extrapolation to patients with more advanced microvascular impairment.

CONCLUSION

Dapagliflozin modestly increases hematocrit and EPO without impairing microvascular blood fluidity, leukocyte rheology, or oxidative balance. These results support the hemorheological safety of SGLT2 inhibition and reinforce its role in cardio–renal risk management in T2DM, including in individuals with higher baseline hematocrit or suboptimal glycemic control.

FUNDING

This work was supported by AstraZeneca K.K. and Ono Pharmaceutical Co., Ltd. (grant number ESR‐15‐11214). The data and results of analyses will be reported to AstraZeneca K.K. and Ono Pharmaceutical Co., Ltd. after establishing a database. The funding sources were 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.

DISCLOSURE

The contract research funding was provided by AstraZeneca K.K. The rights and benefits of the patients have not been compromised by the conduct of this research. Dr. Nakatani reports receiving lecture fees from AstraZeneca K.K., Eli Lilly Japan K.K., Novo Nordisk pharma Ltd., Kowa Co. Ltd., and Bayer Yakuhin, Ltd. Dr. Otani reports receiving grant support from Sato Pharmaceutical Co., Ltd., Dr. Ueda reports receiving lecture fee from TAIHO Pharmaceutical Co., Ltd and Kowa Co. Ltd.; grant support from Kowa Co. Ltd., Bayer Yakuhin, Ltd., and Bristol Myers Squibb. Dr. Maejima reports receiving lecture fees from AstraZeneca K.K., Ono Pharmaceutical Co., Ltd., Nippon Boehringer Ingelheim Co., Ltd., Kowa Co. Ltd., Alynylam Japan, Daiichi Sankyo Co., Ltd., Mochida Pharmaceutical Co., Ltd. Bayer Yakuhin, Ltd., Sumitomo Pharma Co., Ltd., and Pfizer Japan Inc. Dr. Aso reports receiving grant/research funding from Ono Pharmaceutical Co., Ltd. Dr. Yasu reports receiving grant support and lecture fees from AstraZeneca K.K., Ono Pharmaceutical Co., Ltd., and Kowa Co., Ltd; grant support from MTG Co., Ltd.; and lecture fees from Actelion Pharmaceuticals Ltd., Sanofi S.A., Daiichi Sankyo Co., Ltd., and Takeda Pharmaceutical Co., Ltd. The other authors have no conflicts of interest to disclose.

Approval of the research protocol: The protocol for this research project has been approved by the Institutional Review Board (IRB) of Dokkyo Medical University Nikko Medical Center (ethical license number: Nikko 2015‐09) and the Certified Review Board (CRB) of the University of the Ryukyus (CRB7180002, approval code: jRCTs071180044). This study was conducted in accordance with the Declaration of Helsinki (as revised in 2013) and the relevant regulations of the Ministry of Education, Culture, Sports, Science, and Technology and the Ministry of Health, Labor, and Welfare in Japan (established on December 22, 2014, and partially revised on February 28, 2017). Patients were recruited from the Dokkyo Medical University Nikko Medical Center, University of the Ryukyus Hospital, and Yokokawa Clinic. This study was registered in the University Hospital Medical Information Network Clinical Trials Registry (UMIN‐CTR; identifier UMIN000025771, https://center6.umin.ac.jp/cgi‐open‐bin/ctr/ctr_view.cgi?recptno=R000029107). The enrollment period was from February 1, 2017 to November 30, 2018, and the observation period was from February 1, 2017 to May 31, 2019.

Informed consent: All participants provided written informed consent.

Approval date of registry and registration number of the study/trial: UMIN‐CTR; identifier UMIN000025771, approved on February 1, 2017.

Animal studies: N/A.

DATA AVAILABILITY STATEMENT

The single‐case data are publicly available from the University Hospital Medical Information Network Clinical Trials Registry (UMIN‐CTR; identifier UMIN000025771) at https://center6.umin.ac.jp/cgi‐open‐bin/ctr/ctr_view.cgi?recptno=R000029107.