See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/327419010

Evaluation of the effects of dapagliflozin, an SGLT2 inhibitor, on hepatic

steatosis and fibrosis by transient elastography in patients with type 2
diabetes and non‐alcoholic fatty...

Article in Diabetes Obesity and Metabolism · September 2018

DOI: 10.1111/dom.13520

CITATIONS READS

131 755

11 authors, including:

Suzuki Kunihiro Kanako Kato

Dokkyo Medical University Dokkyo Medical University
56 PUBLICATIONS 1,811 CITATIONS 12 PUBLICATIONS 262 CITATIONS

SEE PROFILE SEE PROFILE

Teruo Jojima Toshie Iijima

Dokkyo Medical University Dokkyo Medical University
54 PUBLICATIONS 989 CITATIONS 34 PUBLICATIONS 514 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

SGLT2i pevents development of live caarcinogenesis from non-alcoholic steatohepatitis in a mouse model of NASH under diabetic state View project

All content following this page was uploaded by Yoshimasa Aso on 24 April 2020.

The user has requested enhancement of the downloaded file.

Received: 2 July 2018 Revised: 27 August 2018 Accepted: 30 August 2018
DOI: 10.1111/dom.13520

ORIGINAL ARTICLE

Evaluation of the effects of dapagliflozin, a sodium-glucose co-

transporter-2 inhibitor, on hepatic steatosis and fibrosis using
transient elastography in patients with type 2 diabetes and
non-alcoholic fatty liver disease
Masanori Shimizu MD1* | Kunihiro Suzuki MD2* | Kanako Kato MD1 | Teruo Jojima MD1 |
Toshie Iijima MD1 | Toshimitsu Murohisa MD3 | Makoto Iijima MD3 |
Hidehiro Takekawa MD4 | Isao Usui MD1 | Hideyuki Hiraishi MD3 | Yoshimasa Aso MD

1
Department of Endocrinology and
Metabolism, Dokkyo Medical University, Aims: To investigate the effects of dapagliflozin on liver steatosis and fibrosis evaluated in
Tochigi, Japan patients with type 2 diabetes and non-alcoholic fatty liver disease (NAFLD).
2
Oyama East Clinic, Tochigi, Japan Materials and methods: In a randomized, active-controlled, open-label trial, 57 patients with
3
Department of Gastroenterology, Dokkyo type 2 diabetes and NAFLD were randomized to a dapagliflozin group (5 mg/d; n = 33) or a con-
Medical University, Tochigi, Japan
trol group (n = 24) and were treated for 24 weeks. Hepatic steatosis and fibrosis were assessed
4
Centre of Medical Ultrasonics, Dokkyo
using transient elastography to measure controlled attenuation parameter (CAP) and liver stiff-
Medical University, Tochigi, Japan
ness, respectively.
Correspondence
Yoshimasa Aso MD, Department of Results: Baseline liver stiffness measurement (LSM) was positively correlated with several
Endocrinology and Metabolism, Dokkyo markers and scoring systems for liver fibrosis. In week 24, there was a significant decrease in
Medical University, Mibu, Tochigi 321-0293, CAP from 314  61 to 290  73 dB/m (P = 0.0424) in the dapagliflozin group, while there was
Japan.
Email: yaso@dokkyomed.ac.jp
no significant change in the control group. In addition, LSM tended to decrease from

Funding information
9.49  6.05 to 8.01  5.78 kPa in the dapagliflozin group. In 14 patients from this group with
The work was supported by Ono. LSM values ≥8.0 kPa, indicating significant liver fibrosis, LSM decreased significantly from
14.7  5.7 to 11.0  7.3 kPa (P = 0.0158). Furthermore, serum alanine aminotransferase and
γ-glutamyltranspeptidase levels decreased in the dapagliflozin group, but not in the control
group, and visceral fat mass was significantly reduced in the dapagliflozin group.
Conclusions: Based on these findings, the sodium-glucose co-transporter-2 inhibitor dapagliflo-
zin improves liver steatosis in patients with type 2 diabetes and NAFLD, and attenuates liver
fibrosis only in patients with significant liver fibrosis, although the possibility cannot be excluded
that a reduction in body weight or visceral adipose tissue by dapagliflozin may be associated
with a decrease of liver steatosis or fibrosis.

KEYWORDS

dapagliflozin, liver fibrosis, non-alcoholic fatty liver disease, transient elastography, type
2 diabetes

1 | I N T RO D UC T I O N alcoholic steatohepatitis (NASH), which can cause liver fibrosis and

progresses to cirrhosis with a risk of hepatocellular carcinoma in 10%
Non-alcoholic fatty liver disease (NAFLD) is the most common chronic to 20% of patients.2,3 Type 2 diabetes is a major risk factor for NAFLD
1
liver disease, affecting 17% to 46% of the population worldwide. and/or NASH, as the prevalence of NAFLD is as high as 40% to 50%
Approximately 20% to 30% of patients with NAFLD also have non- among these patients.4 Early detection of severe steatosis and clini-
cally significant fibrosis would be useful to identify patients who may
*M.S. and K.S. contributed equally to this work. have aggressive NAFLD and therefore need further evaluation. As a

Diabetes Obes Metab. 2019;285–292. wileyonlinelibrary.com/journal/dom © 2018 John Wiley & Sons Ltd 285
286 SHIMIZU ET AL.

high proportion of patients with NAFLD are asymptomatic for a long the investigators were encouraged to manage their patients according
period and have normal or only slightly abnormal liver function tests, to local guidelines in order to achieve optimum glycaemic control. The
non-invasive methods for early identification of severe steatosis and control group received standard treatment for type 2 diabetes, and if
advanced fibrosis/cirrhosis are needed5 to allow early treatment of the HbA1c target (<7.0%) was not achieved after ~3 months, treat-
NAFLD in high-risk patients with type 2 diabetes. ment with antidiabetic drugs was uptitrated, excluding SGLT2 inhibi-
Sodium-glucose co-transporter-2 (SGLT2) inhibitors are a new tors. In the control group, additional antidiabetic drugs were newly
class of oral antidiabetic drug that reduce hyperglycaemia indepen- initiated during the study period: three patients with dipeptidyl
dently of insulin secretion by promoting the urinary excretion of glu- peptidase-4 inhibitors, two patients with α-glucosidase inhibitors, one
6
cose. In patients with diabetes who received treatment with the patient with glinides, and one patient with basal insulin.
SGLT2 inhibitor dapagliflozin, the majority of weight loss was The diagnosis of NAFLD was made on the basis of liver dysfunc-
accounted for by fat loss, with significantly greater reduction in the tion (persistent elevation of alanine transaminase [ALT] ≥ the upper
volume of both abdominal visceral adipose tissue (VAT) and subcuta- limit for our laboratory), the presence of fatty liver on ultrasonogra-
7
neous adipose tissue (SAT) with dapagliflozin than with placebo. phy, low daily alcohol intake (<30 g for men and < 20 g for women),
There have been some recent reports that SGLT2 inhibitors can sup- and exclusion of other liver diseases, such as chronic hepatitis B
press the development of NAFLD and/or NASH in rodent models,8–11 and C, autoimmune hepatitis, primary biliary cirrhosis, haemochroma-
and SGLT2 inhibitors have also been shown to improve histological tosis and Wilson's disease. All of the participants gave informed con-
hepatic steatosis or steatohepatitis in obese mice or rats with type sent to this study and it was approved by the institutional review
2 diabetes8–11; however, only two prospective clinical studies have board of Dokkyo Medical University. The study was registered with
investigated the influence of SGLT2 inhibitors on hepatic steatosis in University Hospital Medical Information Network (UMIN) Clinical Tri-
12,13
patients with type 2 diabetes and NAFLD, and the antifibrotic als Registry (UMIN000022155).
effect of SGLT2 inhibitors has not been examined in these patients.
Transient elastography is an ultrasonography-based method of 2.1 | Methods
elastography,14 which allows simultaneous evaluation of hepatic stea-
Transient elastography was performed using a FibroScan with the
tosis by measuring the controlled attenuation parameter (CAP) and
standard 3.5 MHz M probe (Echosen, Paris, France) to measure CAP
liver fibrosis by measuring liver stiffness (LS), which is strongly corre-
and LS simultaneously in the same cylinder of liver parenchyma
lated with the stage of liver fibrosis assessed by concurrent liver
(1 × 4 cm). CAP is a measure of ultrasonic attenuation at 3.5 MHz on
biopsy.15 Thus, measurement of CAP and LS by transient elastography
the FibroScan signal that is used to assess the severity of liver steato-
might be an appropriate screening tool for liver fibrosis and steatosis
16,17 sis and is expressed in dB/m.19 FibroScan simultaneously assesses LS
in patients with diabetes, and may be more accurate than bio-
by measuring the propagation of an elastic shear wave through the
markers or scoring systems for detection of significant fibrosis and
liver parenchyma,20 and it is expressed in kPa, with higher values indi-
cirrhosis.18
cating greater stiffness.13,19 An LS measurement (LSM) ≥8.0 kPa was
Accordingly, we employed transient elastography to determine
shown to indicate significant liver fibrosis in a study in the general
the effects of dapagliflozin, an SGLT2 inhibitor, on hepatic fibrosis and
French population,15 and this value was used to indicate the existence
steatosis in patients with type 2 diabetes and NAFLD.
of liver fibrosis in the present study. The median value of 10 measure-
ments was used to define LSM. The intra- and inter-assay coefficients
2 | RESEARCH DESIGN AND METHODS of variation of LSM were 3.2% and 3.3%, respectively. A scan failure
was defined as the inability to obtain 10 valid measurements in a sin-
We studied 63 patients with type 2 diabetes and NAFLD who were gle patient. No patient in the present study was considered to have
referred to the diabetes outpatient clinic of Dokkyo Medical Univer- experienced scan failure in the present study.
sity Hospital. Patients were eligible for enrolment if they had type Dual bioelectrical impedance analysis was used to measure VAT
2 diabetes combined with NAFLD, were aged ≥20 years, and had a (Dual Scan; Omron Healthcare Company, Limited, Kyoto, Japan). This
glycated haemoglobin (HbA1c) level of 6.0% to 12.0% on stable ther- instrument calculates the cross-sectional area of intra-abdominal fat
apy with one to three oral antidiabetic agents with or without insulin (VAT and SAT) at the umbilicus, based on measurement of electrical
for at least 3 months. potentials after application of small electrical currents to two different
This study was performed according to a prospective, random- body spaces.21 VAT measured by dual bioelectrical impedance is equal
ized, open-label, blinded endpoint design. Patients were randomly to that measured by abdominal CT, which is the “gold standard” for
allocated in a 1:1 ratio to receive either dapagliflozin or the standard determination of VAT.22 The intra- and inter-assay coefficients of var-
treatment without SGLT2 inhibitors. Randomization was stratified iation of VAT were 6.3% and 6.8%, respectively.
according to sex, age and body mass index (BMI) using the minimiza- The serum level of high-molecular weight (HMW) adiponectin
tion method described by Pocock and Simon. Each patient was fol- was measured using a sandwich ELISA, with a monoclonal antibody
lowed for 24 weeks and was reviewed every month. The dose of for human HMW adiponectin, as described previously.23
dapagliflozin was fixed at 5 mg/day, which is the standard dose for Serum leptin and type 4 collagen 7S levels were determined using
treatment of type 2 diabetes in Japan. Throughout the study, all radioimmunoassay kits (Human leptin RIA kit; Millipore Corporation,
patients received standard-of-care treatment for type 2 diabetes and St. Louis, Missouri; type 4 collagen 7S kit, SCETI MEDICAL LABO,
SHIMIZU ET AL. 287

Tokyo, Japan). The serum ferritin concentration was measured with the Mann–Whitney U test. Correlations were determined by linear
an electrochemiluminescence assay (ECLlusys ferritin; Roche-Diag- regression analysis or Spearman rank correlation test. Statistical ana-
nostics, Tokyo, Japan), while serum hyaluronic acid was determined lyses were carried out by using GraphPad Prism 7 software (GraphPad
by the latex agglutination method (LPIA-ACE HA, LSI Medience, Software, Inc., La Jolla, California), and P values <0.05 were taken to
Tokyo, Japan). indicate statistical significance.
The NAFLD fibrosis risk score predicts the severity of hepatic We calculated that a sample of 61 patients would be required for
fibrosis based on six variables according to the following formula: 90% power at a significance level of 0.05 to detect a difference in the
score = −1.675 + 0.037 × age (years) + 0.094 × BMI (kg/m2) + 1.13 mean of CAP of 25 on the assumptive SD of 30.24
× impaired fasting glucose/diabetes (yes = 1, no = 0) + 0.99 × aspar-
tate aminotransferase (AST)/ALT ratio- 0.013 × platelet count (109/
3 | RE SU LT S
L)-0.66 × albumin (g/dL). The NAFIC score was also calculated as the
sum of the following three clinical variables: serum ferritin ≥200 ng/
A total of 63 patients were screened and underwent randomization to
mL (female) or ≥ 300 ng/mL (male) = 1 point; serum fasting insulin
receive dapagliflozin (n = 35) or the standard treatment (n = 28;
≥10 μU/mL = 1 point; and serum type IV collagen 7 s ≥ 5.0 ng/mL = 2
Figure 1). In the dapagliflozin group, 33/35 patients completed the
points. Furthermore, the FIB-4 index was calculated with the follow-
trial, while 24/28 patients completed it in the standard treatment
ing formula: (age × AST [units/L]) /platelet count (×109/L) × (ALT
group.
[units/L])1/2.
At baseline, the two groups were well balanced with respect to
demographic characteristics and laboratory data (Table 1).
2.2 | Outcomes In the 57 patients who completed the study, baseline CAP was

The primary endpoint was change in CAP from baseline to 24 weeks positively correlated with baseline body weight, BMI, VAT, subcutane-

of treatment. The key secondary endpoint was change in LSM from ous adipose tissue (SCT), homeostatic model assessment of insulin

baseline to 24 weeks of treatment, while other secondary endpoints resistance (HOMA-IR), AST, ALT and NAFIC score, while CAP showed

were change in HbA1c, VAT, liver enzymes (AST, ALT and a negative correlation with the serum level of HMW adiponectin

γ-glutamyltranspeptidase [GGT]), and various markers and scores for (Table S1). By contrast, baseline LS measurement (LSM) was positively
correlated with baseline VAT, AST, ALT, GGT, type 4 collagen 7S, hya-
hepatic fibrosis.
luronic acid, and Mac-2 binding protein. Moreover, LSM was closely
associated with the three liver fibrosis scores (the FIB-4 index, NAFLD
2.3 | Statistical analysis
score, and NAFIC score). There was a significant positive correlation
Data are expressed as mean  SD or median with interquartile range. between CAP and LSM in all 57 patients (r = 0.4199, P = 0.0016).
Differences between groups were analysed using Student's paired In the dapagliflozin group, VAT, SCT and body weight all showed
t test or the unpaired t test, while between-group differences in non- a significant decrease at the end of the treatment period, while no
parametric data were analysed by Wilcoxon's matched-pairs test or changes in these variables were found in the standard treatment

63 patients with type 2 diabetes and NAFLD enrolled and randomized

Aggressive non-SGLT2 inhibitor treatment (control) n = 28 Additional treatment with SGLT2 inhibitors (dapagliflozin) n = 35

Newly added treatment as follows; Added 5 mg dapagliflozin to preceding therapy

All patients: Intensifying diet and exercise
3 patients received DPP-4 inhibitors
2 patients received α-glucosidase inhibitors
1 patient received basal insulin
1 patient received glinides

4 patients not completed

1 patient due to congestive heart failure
1 patient due to surgery for Charcot joint 2 patients not completed
1 patient no longer met study criteria 1 patient no longer met study criteria
1 patient lost to follow-up 1 patient lost to follow-up

24 patients completed the trial 33 patients completed the trial

FIGURE 1 Study design. A total of 57 patients completed the study (33 in the dapagliflozin group and 24 in the control group). DPP-4, dipeptidyl
peptidase-4; NAFLD, non-alcoholic fatty liver disease; SGLT2, sodium-glucose co-transporter-2
288 SHIMIZU ET AL.

TABLE 1 Baseline demographic, clinical, and laboratory data for patients with type 2 diabetes and non-alcoholic fatty liver disease treated with
dapagliflozin (5 mg/d) or standard therapy (control)
Dapagliflozin Control P
Patients (men/women) 33 (19/14) 24 (15/9) 0.7883
Age, ys 56.2  11.5 57.1  13.8 0.7823
Body weight, kg 73.9  16.1 76.4  13.9 0.5521
BMI, kg/m2 27.6  4.7 28.3  3.5 0.3166
VAT, cm2 108.7  42.9 125.7  38.2 0.1334
SCT, cm 2
249.5  82.5 226.7  89.9 0.3501
Body water, kg 37.3  9.1 36.8  6.7 0.8374
Skeletal muscle mass, kg 27.8  7.5 27.1  5.4 0.7001
FPG, mmol/l 7.66  3.003 7.59  2.29 0.9257
HbA1c, mmol/mol 63.5  15.3 60.6  6 0.0748
HOMA-IR 3.58 (2.41, 6.10) 3.92 (1.96, 5.47) 0.7106
LDL cholesterol, mmol/l 2.71  0.89 2.65  0.61 0.4354
Triglycerides, mmol/l 1.36 (1.02, 1.90) 1.22 (1.02, 2.00) 0.9052
HDL cholesterol, mmol/l 1.29  0.33 1.19  0.32 0.2934
AST, U/L 28.0 (20.5, 49.8) 26.0 (20.3, 32.0) 0.4652
ALT, U/L 38.0 (21.5, 61.0) 33.0 (24.5, 46.5) 0.5187
GGT, U/L 47.0 (28.0, 88.3) 37.5 (20.0, 62.3) 0.1799
Uric acid, mg/dL 4.83  1.24 5.16  1.33 0.3072
Albumin, g/dL 4.32  0.42 4.32  0.34 0.9921
Haematocrit, % 44.3  3.9 44.2  3.7 0.9415
Platelets (×109/mL) 24.1  6.4 27.8  8.1 0.0609
eGFR, mL/min/1.73m 2
79.4  15.8 76.9  19.0 0.5881
Leptin, ng/mL 9.75 (6.20, 17.4) 13.0 (8.33, 25.5) 0.2240
HMW adiponectin, μg/mL 1.08 (0.56, 4.11) 1.31 (0.44, 2.53) 0.4959
Ferritin, ng/mL 82.1 (33.1, 177.5) 69.8 (30.0, 157.8) 0.6705
Type 4 collagen 7S, ng/mL 4.40  1.30 3.71  0.91 0.0316
Hyaluronic acid, ng/mL 31.4 (20.9, 60.4) 28.2 (18.6, 51.) 0.3938
Mac-2 binding protein 0.71 (0.41, 1.10) 0.59 (0.45, 0.81) 0.2643
Fib-4 index 1.32 (0.74, 2.10) 0.98 (0.62, 1.33) 0.0860
NAFLD fibrosis score −0.66 (−1.91, −0.05) −1.43 (−2.56, −0.45) 0.1719
NAFIC score 1.00 (0.00, 2.00) 1.00 (0.00, 1.00) 0.2338
CAP, dB/m 314.1  61.0 306.0  34.3 0.5659
LSM, kPa 7.20 (5.22, 13.6) 6.10 (4.83, 9.43) 0.2102
Metformin/DPP-4 inhibitor/SU/insulin, n 31/16/12/10 21/14/11/4 0.6413

Abbreviations: ALT, alanine transaminase; AST, aspartate aminotransferase; BMI, body mass index; CAP, controlled attenuation parameter; DPP-4, dipepti-
dyl peptidase-4; eGFR, estimated glomerular filtration rate; FPG, fasting plasma glucose; GGT, γ-glutamyltranspeptidase; Hb, haemoglobin; HMW,
high-molecular weight; HOMA-IR, homeostatic model assessment of insulin resistance; LSM, liver stiffness measurement; NAFLD, non-alcoholic fatty liver
disease; SCT, subcutaneous adipose tissue; SU, sulfonylurea; VAT, visceral adipose tissue.
Data are presented as mean  SD or the median (interquartile range) unless otherwise indicated.

group (Table 2). HbA1c decreased significantly, from 8.37  1.48% at and increased the haematocrit at 24 weeks. HMW adiponectin
baseline to 7.36  1.22% after 24 weeks of dapagliflozin treatment. showed a significant increase at 24 weeks in the dapagliflozin group,
HbA1c also decreased in the standard treatment group, but the but not in the standard treatment group. Serum ferritin was decreased
change was not significant. HOMA-IR showed a significant decrease at 24 weeks in the dapagliflozin group, but not in the placebo group.
after 24 weeks in the dapagliflozin group, but not in the standard CAP was significantly decreased after 24 weeks in the dapagliflozin
treatment group. Changes in BMI, VAT and HbA1c in the control ver- group (Figure 2A), while there was no change in CAP in the standard
sus the dapaglifloizn group were 0.0 (−0.55, 0.50) versus −0.8 (−1.25, treatment group, and the percent reduction in CAP from baseline to
−0.07), −2.0 (−13.0, 6.0) versus −10.0 (−17.0, 0.5) cm , and − 0.3
2
24 weeks was significantly larger in the dapagliflozin group than in
(−0.5, 0.5) versus −0.8 (−1.3, −0.5)%, respectively. the standard treatment group (92.4  18.7% vs. 102.2  13.2%;
After 24 weeks, there was a significant decrease in AST, ALT and P = 0.0429 [Figure 2B]). LSM was also decreased after 24 weeks in
GGT in the dapagliflozin group, while there were no changes in liver the dapagliflozin group (9.49  6.05 to 8.01  5.78 kPa), but the
enzymes in the standard treatment group (Table 2). Consistent with change was not significant (P = 0.0539). We divided the 33 patients
previous studies, dagagliflozin trreatment reduced the uric acid level who completed dapagliflozin treatment into subgroups with or
TABLE 2 Changes in clinical parameters, hepatic steatosis and fibrosis, and mean differences between groups in type 2 diabetic and NAFLD patients treated with dapagliflozin (the dapagliflozin groups) or
those with standard treatment (the control group)
Dapagliflozin Control Difference b/w groups
Baseline Wk 24 Baseline Wk 24
SHIMIZU ET AL.

P P P
N 33 33 - 24 24 - -
Body weight, kg 73.6 (61.9, 80.8) 70.7 (60.0, 79.2) 0.0004 76.4  13.8 75.8  12.8 0.4911 0.0375
BMI, kg/m2 27.6  4.7 26.9  5.0 0.0006 28.7  3.5 28.6  3.6 0.4930 0.0513
VAT, cm2 108.7  42.9 101.4  39.2 0.0068 125.7  32.2 120.0  40.1 0.1795 0.3692
SCT, cm2 226.7  90.0 215.5  80.6 0.0354 249.5  82.5 250.8  91.3 0.6991 0.6205
Total body water, kg 37.3  9.1 36.2  9.5 0.0057 36.8  6.7 36.5  6.4 0.9762 0.0286
Skeletal muscle mass, kg 27.8  7.5 26.9  7.8 0.0580 27.1  5.4 26.8  5.0 0.3828 0.0060
FPG, mmol/l 7.66  3.00 6.81  1.95 0.1057 7.59  2.29 8.01  2.53 0.2218 0.2964
HbA1c, mmol/mol 63.5  15.3 56.9  14.0 <0.0001 60.6  6 53.11  4.6 0.1414 0.0949
HOMA-IR 3.58 (2.41, 6.10) 2.66 (1.49, 4.85) 0.0076 3.92 (1.96, 5.47) 3.56 (2.17, 5.78) 0.9240 0.0407
LDL cholesterol, mmol/l 2.71  0.89 2.74  0.95 0.6398 2.65  0.61 2.83  0.61 0.2842 0.8232
Triglycerides, mmol/l 1.50  0.62 1.29  0.58 0.0842 1.64  1.01 1.60  0.95 0.3512 0.3821
HDL cholesterol, mmol/l 1.29  0.33 1.42  0.38 0.0011 1.19  0.32 1.24  0.28 0.1126 0.0773
AST, U/L 28.0 (20.5, 49.8) 27.5 (17.3, 31.8) 0.0018 29.8  12.8 27.4  9.6 0.3353 0.0837
ALT, U/L 38.0 (21.5, 61.0) 26.5 (16.3, 42.5) <0.0001 33.0 (24.5, 46.5) 32.0 (25.0, 49.3) 0.4493 0.0212
GGT, U/L 47.0 (28.0, 88.3) 27.0 (20.5, 61.5) 0.0003 37.5 (20.0, 62.3) 32.0 (22.3, 50.0) 0.4584 0.0041
Albumin, g/L 4.32  0.42 4.48  0.35 0.0127 5.16  1.33 5.12  1.23 0.7527 0.2829
Uric acid, mg/dL 4.80  1.24 4.46  1.20 0.0236 5.16  1.33 5.12  1.23 0.7527 0.1549
Hematocrit, % 44.3  3.9 45.6  4.3 0.0002 44.2  3.7 44.8  3.2 0.3864 0.0945
Platelets, ×109/mL 24.1  6.4 24.1  8.0 0.2278 27.8  8.1 28.2  10.0 0.8350 0.4705
eGFR, mL/min/1.73m2 79.4  15.8 82.3  17.1 0.1435 76.9  19.0 80.8  24.0 0.1586 0.8900
HMW adiponectin, μg/mL 1.08 (0.56, 4.11) 1.62 (0.91, 4.67) 0.0002 1.31 (0.44, 2.53) 1.79 (0.39, 2.55) 0.3894 0.0045
Leptin, ng/mL 9.75 (6.20, 17.4) 10.8 (7.73, 15.7) 0.7243 13.0 (8.3, 25.5) 15.2 (9.7, 24.7) 0.2453 0.9772
Ferritin, ng/mL 79.5 (32.4, 150) 46.5 (22.6, 106) <0.0001 63.0 (30.0, 157.8) 76.7 (36.3, 116.0) 0.7270 0.0037
Type 4 collagen 7S, ng/mL 4.00 (3.53, 4.90) 4.00 (3.32, 4.78) 0.4707 3.71  0.91 4.15  1.06 0.0295 0.0222
Hyaluronic acid, ng/mL 31.4 (20.9, 60.4) 30.6 (21.3, 78.2) 0.8941 28.2 (18.6, 51.8) 30.7 (20.1, 64.3) 0.5392 0.6274
Mac-2 binding protein 0.71 (0.41, 1.10) 0.68 (0.47, 1.00) 0.3543 0.65  0.27 0.67  0.29 0.9335 0.9861
Fib-4 index 1.32 (0.74, 2.10) 1.27 (0.77, 1.91) 0.7207 1.11  0.64 1.17  0.70 0.9286 0.9155
NAFLD fibrosis score −0.66 (−1.91, −0.05) −0.78 (−1.66, −0.22) 0.2988 −1.41(−2.50, −0.46) -1.12 (−2.24, −0.40) 0.5226 0.2923
NAFIC score 1.0 (0.0, 2.0) 0.0 (0.0, 1.0) 0.2026 1.0 (1.0, 4.0) 1.0 (1.0, 4.0) 0.7930 0.5301
CAP, dB/m 314.6  61.0 290.3  72.7 0.0424 306.0  34.3 311.3  37.3 0.6253 0.0479
LSM, kPa 9.49  6.05 8.01  5.78 0.0539 7.40  3.76 7.85  4.18 0.8655 0.2217

Abbreviations: ALT, alanine transaminase; AST, aspartate aminotransferase; BMI, body mass index; CAP, controlled attenuation parameter; eGFR, estimated glomerular filtration; FPG, fasting plasma glucose; GGT, γ--
glutamyltranspeptidase; Hb, haemoglobin; HMW, high-molecular weight; HOMA-IR, homeostatic model assessment of insulin resistance; LSM, liver stiffness measurement; NAFLD, non-alcoholic fatty liver disease;
SCT, subcutaneous adipose tissue; VAT, visceral adipose tissue.
Data are presented as mean  SD or the median (interquartile range) unless otherwise indicated.
289
290 SHIMIZU ET AL.

without significant liver fibrosis, which were stratified according to a (β = 0.530, P = 0.004) and LDL cholesterol (β = 0.411, P = 0.033) were
baseline LSM ≥8.0 or < 8.0 kPa. As shown in Table S2, AST, ALT and independent determinants of change in CAP, while only change in
GGT levels were significantly higher in the patients with significant HDL cholesterol was an independent determinant of changes in LSM
fibrosis than in those without fibrosis. Serum markers of liver fibrosis (β = 0.417, P = 0.030).
and fibrosis scores were also significantly higher in the patients with
significant fibrosis. In the 14 patients with a baseline LSM ≥8.0 kPa,
dapagliflozin treatment significantly decreased LSM from
4 | DI SCU SSION
14.7  5.7 kPa at baseline to 11.0  7.3 kPa after 24 weeks
This was the first investigation into the effects of the SGLT2 inhibitor
(P = 0.0158; Figure 2C). Furthermore, the magnitude of percent
dapagliflozin on hepatic steatosis and fibrosis, as evaluated by tran-
reduction in LS from baseline to 24 weeks was significantly greater in
sient elastography with the FibroScan, in patients with type 2 diabetes
the dapagliflozin group (14 patients with a baseline LSM ≥ 8.0 kPa)
and NAFLD. We demonstrated significant improvement in CAP
than in the control group (77.9 [55.8, 112] vs. 95.3 [74.5, 135]%;
(an indicator of hepatic steatosis) in the dapagliflozin group after
P = 0.0479 [Figure 2D]).
24 weeks of treatment, while there was no improvement in the con-
Next, we investigated the associations between changes in CAP
trol group, and the percent reduction in CAP from baseline to
or LSM and changes in clinical variables in the dapagliflozin group. We
24 weeks was significantly larger in the dapagliflozin group than in
found no significant correlations between change in CAP and changes
the control group. There has only been two previous prospective
in body weight, BMI or VAT in the dapagliflozin group (Table S3).
investigations into the effect of SGLT2 inhibitors on hepatic steatosis
There was a significant positive correlation between change in CAP
that employed imaging methods such as CT and MRI.12,13 In the pre-
and change in HbA1c (r = 0.4288, P = 0.0203). Like CAP, no signifi-
sent study, we investigated the effect of dapagliflozin by measuring
cant correlations were observed between change in LSM and changes
CAP, a new quantitative index of hepatic steatosis.19 Recent studies
in body weight, BMI, VAT or HbA1c in the dapagliflozin group
have shown that CAP is significantly correlated with both percentage
(Table S3). To identify independent determinants of changes in CAP
of steatosis and grade of steatosis evaluated by liver biopsy.19,25 Fat
or LSM after treatment with dapagliflozin in patients with type 2 dia-
attenuates the propagation of ultrasonography, and CAP is deter-
betes and NAFLD, we performed stepwise regression analysis with
mined by quantitation of ultrasonic attenuation at the central fre-
forward selection that included changes in body weight, VAT, HbA1c,
quency of the FibroScan M probe (3.5 MHz).26 A large-scale
LDL cholesterol, HDL cholesterol and ALT. Changes in HbA1c prospective study demonstrated the accuracy of CAP for diagnosis of
NAFLD.19 In the present study, we found a significant correlation of
(A) (B) baseline CAP with baseline VAT and with liver function tests (ALT,
P=0.0424
P=0.0429
Changes in CAP at 24 weeks

400 150 AST and GGT), confirming that CAP is a marker of hepatic steatosis.
350
Ito et al.12 previously reported that treatment with ipragliflozin,
(% of baseline)

300
CAP (dB/m)

100
250
200
another SGLT2 inhibitor, reduced hepatic steatosis, which they evalu-
150
100
50 ated from the liver-to-spleen attenuation ratio on CT. Very recently,
50 Kuchay et al13 also demonstrated that empagliflozin reduced liver fat
0 0
Before After Dapagliflozin Control content, evaluated by MRI-derived proton density fat fraction in
Dapagliflozin 5 mg/d
patients with type 2 diabetes and NAFLD. Magnetic resonance elasto-
(C) (D) graphy or MRI-derived proton density fat fraction has a better diag-
P=0.0158 P=0.0479
20 200
Changes in LS at 24 weeks

nostic accuracy for assessing significant fibrosis than transient

Liver stiffness (kPa)

(% of baseline)

15 150
elastography, although the former techniques are expensive and time-
10 100 consuming to implement in clinical practice for screening liver fibro-
5 50 sis.27 The results of the study by Kuchay et al. are consistent with our
0 0 finding that SGLT2 inhibitor therapy can reduce hepatic steatosis in
Before After Dapagliflozin Control
patients with type 2 diabetes and NAFLD, but the mechanisms under-
Dapagliflizin 5 mg/d
lying improvement of hepatic steatosis (hepatic triglyceride content)
FIGURE 2 A, Change in controlled attenuation parameter (CAP) after by dapagliflozin remain to be elucidated. According to a meta-analysis,
treatment with dapagliflozin 5 mg/d in patients with type 2 diabetes ≥5% weight loss improved hepatic steatosis and ≥ 7% weight loss also
and non-alcoholic fatty liver disease (NAFLD). B, Magnitude of
improved histological findings of NAFLD, although fibrosis was
percent change in CAP from baseline to 24 weeks of treatment with
dapagliflozin (5 mg/d) or control (standard treatment) in patients with unchanged.28 The present study showed no significant correlation
type 2 diabetes and NAFLD. Data are expressed as the between change in CAP and changes in body weight or VAT from
mean  SD. C, Change in liver stiffness measurement (LSM) after baseline to 24 weeks of dapagliflozin treatment, suggesting that dapa-
treatment with dapagliflozin (5 mg/d) in patients with type 2 diabetes gliflozin improved liver dysfunction and steatosis by a mechanism
and NAFLD who had significant baseline liver fibrosis (LSM
unrelated to reduction of VAT or body weight. One possible explana-
≥8.0 kPa). D, The magnitude of percent change in LSM from baseline
tion is that dapagliflozin inhibited de novo lipogenesis in the liver,
to 24 weeks of treatment with dapagliflozin (5 mg/d) in patients with
type 2 diabetes and NAFLD who had significant baseline liver fibrosis since, in our previous study using a mouse model of NASH with diabe-
(LSM ≥8.0 kPa) or the control (standard treatment) tes, we demonstrated significantly lower expression of fatty acid
SHIMIZU ET AL. 291

synthase and acetyl-CoA carboxylase 1, two genes involved in de ferritin (r = 0.5632, P = 0.0010) in the dapagliflozin group. It has been
29
novo lipogenesis, in mice treated with empagliflozin than in mice reported that serum ferritin is an independent predictor of histological
receiving vehicle or linagliptin.8 Fatty acid synthase is a key enzyme in severity and advanced fibrosis in patients with NAFLD,34 suggesting
the hepatic biosynthesis of fatty acids and is believed to determine that serum ferritin level may reflect liver inflammation and fibrosis in
the maximal liver capacity for producing fatty acids by de novo lipo- NAFLD/NASH. Taken together, these results suggest that dapagliflo-
genesis, because it catalyses the last step in the fatty acid biosynthetic zin may attenuate the development of significant liver fibrosis in high-
pathway.29 Thus, dapagliflozin may contribute to improvement of risk patients with type 2 diabetes and NAFLD.
hepatic steatosis in patients with type 2 diabetes and NAFLD by inhi- The present study has several limitations. Firstly, our control
biting fatty acid production through promotion of urinary glucose group was not a placebo group, and may therefore have been subject
excretion. However, a very recent study has investigated the impact to variations in treatment, including varying levels of medication
of bariatric surgery-induced weight loss on NAFLD by FibroScan in intensification. Secondly, we did not perform liver biopsy to confirm
morbidly obese subjects, showing a significant improvement in CAP
liver fibrosis. Although liver biopsy is the gold standard for assessment
and LSM.30 The possibility cannot be excluded, therefore, that CAP or
of liver fibrosis, there is accumulating evidence that transient elasto-
LSM reduction by dapagliflozin may be associated with a reduction in
graphy by FibroScan is strongly correlated with the stage of liver
body weight or VAT. Nevertheless, as we found a significant positive
fibrosis assessed by concurrent liver biopsy.15 Thirdly, significant liver
correlation between change in CAP and change in HbA1c in the dapa-
fibrosis (an LSM ≥8.0 kPa) was defined based on a previous study in
gliflozin group, the reduction in CAP may be attributable to an
the French general population,15 which is probably not representative
improvement in glycaemic control. A previous study also showed that
of Japanese people with type 2 diabetes. Fourthly, abdominal obesity
changes in fasting plasma glucose and HbA1c after treatment with
had a potential impact on the LSM readings; a high rate of a scan fail-
ipragliflozin were correlated positively with change in fatty liver index
ure with FibroScan occurs in obese people, because the thick subcuta-
in Japanese patients with type 2 diabetes.31
neous fat between the probe and liver attenuates the propagation of
The present study provides the first evidence that an SGLT2
shear wave. Fortunately, scan failure arose in none of participants in
inhibitor can prevent progression of hepatic fibrosis in patients with
type 2 diabetes and NAFLD who have pre-existing significant liver the present study. Lastly, reduction in liver fibrosis in the present

fibrosis, defined as LSM >8.0 kPa.15 Measurement of LSM by tran- study was only observed over a short period (24 weeks).

sient elastography was reported to be an easy non-invasive method In conclusion, this was the first study to investigate the effects of

for reliably estimating the severity of liver fibrosis in patients with dapagliflozin on hepatic steatosis and fibrosis in patients with type
type 2 diabetes and NAFLD.17 In addition, LSM was found to be an 2 diabetes and NAFLD. The results showed that dapagliflozin
accurate and reproducible parameter for detecting advanced liver improves both liver steatosis and fibrosis in these patients. It is neces-
fibrosis in patients with NAFLD that was comparable with liver sary to keep in mind that a reduction in body weight or VAT by dapa-
32 gliflozin may be associated with a decrease in liver steatosis or
biopsy. In the present study, baseline LSM showed a strong positive
correlation with several laboratory markers and scoring systems for fibrosis. Transient elastography allows identification of patients with
liver fibrosis, including type 4 collagen, FIB-4 index, and NAFLD fibro- type 2 diabetes who are at risk of developing aggressive NAFLD by
sis score. LSM ≥8.0 kPa is used as the threshold for excessive stiff- early detection of severe steatosis and significant fibrosis. It may be
ness, which indicates clinically important liver disease,15 based on a possible to prevent the development of liver cirrhosis and/or hepato-
large-scale study of the general population in France that showed this cellular carcinoma in these high-risk patients by administration of
value was an accurate predictor of liver fibrosis on biopsy.17 Another SGLT2 inhibitors, such as dapagliflozin.
large population-based study (the Rotterdam study) showed that 5.6%
of the participants had an LSM ≥8.0 kPa and that elevation of LSM
was strongly associated with diabetes and hepatic steatosis.33 In the ACKNOWLEDGMENTS
present study, LSM decreased in all 33 patients treated with dapagli- The authors thank Drs T. Minaguchi and T. Arisaka, Department of
flozin, although the change was not significant; however, there was a Gastroenterology, Dokkyo Medical University, for their technical
significant reduction in LSM in the 14 patients with a baseline LSM assistance.
value ≥8.0 kPa. This suggests that dapagliflozin may attenuate the
development of significant liver fibrosis in patients with type 2 diabe-
tes who have significant fibrosis. CONFLIC T OF INT E RE ST
The mechanisms leading to improvement of hepatic fibrosis by Y.A. has received speaker fees from AstraZeneca and Ono.
dapagliflozin remain unclear. We previously obtained histological evi-
dence that empagliflozin treatment improved hepatic steatosis and
Author contributions
inflammation, as well as fibrosis, in a mouse model of NASH and dia-
betes.8 These findings suggest that SGLT2 inhibitor therapy may pre- M.S. and K.S. contributed to the study design, data collection, and
vent the progression of hepatic fibrosis by reducing inflammation in drafting of the manuscript. K.K. and K.M. helped with data collection.
the liver.8 In fact, serum ferritin was decreased significantly by treat- T.J., T.I., M.I. and H.T. contributed to the discussion and reviewed the
ment with dapagliflozin in the present study, and we found a signifi- manuscript. I.U. reviewed/edited the manuscript. Y.A. researched data
cant positive correlation between the changes in ALT and serum and wrote and reviewed/edited the manuscript.
292 SHIMIZU ET AL.

ORCID 20. Foucher J, Chanteloup E, Vergniol J, et al. Diagnosis of cirrhosis by

transient elastography (FibroScan): a prospective study. Gut. 2006;55:
Yoshimasa Aso https://orcid.org/0000-0003-0610-1999 403-408.
21. Shiga T, Hamaguchi T, Oshima Y, et al. A new simple measurement
system of visceral fat accumulation by bioelectrical impedance analy-
RE FE R ENC E S sis. In: Döossel O et al., eds. World Congress on Medical Physics and Bio-
1. European Association for the Study of the Liver (EASL), European medical Engineering. IFMBE Proceedings. Vol. 25/7. Munich, Germany:
Association for the Study of Diabetes (EASD) and European Associa- Springer-Verlag; 2009:338-341.
tion for the Study of Obesity (EASO). EASL–EASD–EASO Clinical 22. Ida M, Hirata M, Odori S, et al. Early changes of abdominal adiposity
Practice Guidelines for the management of non-alcoholic fatty liver detected with weekly dual bioelectrical impedance analysis during cal-
disease. J Hepatol. 2016;64:1388-1402. orie restriction. Obesity (Silver Spring). 2013;21:E350-E353.
2. Wong VW, Wong GL, Choi PC, et al. Disease progression of 23. Aso Y, Yamamoto R, Wakabayashi S, et al. Comparison of serum
non-alcoholic fatty liver disease: a prospective study with paired liver high-molecular weight (HMW) adiponectin with total adiponectin con-
biopsies at 3 years. Gut. 2010;59:969-974. centrations in type 2 diabetic patients with coronary artery disease
3. Ascha MS, Hanouneh IA, Lopez R, Tamimi TA, Feldstein AF, Zein NN. using a novel enzyme-linked immunosorbent assay to detect HMW
The incidence and risk factors of hepatocellular carcinoma in patients adiponectin. Diabetes. 2006;55:1954-1960.
with nonalcoholic steatohepatitis. Hepatology. 2010;51:1972-1978. 24. Lee YH, Kim JH, Kim SR, et al. Lobeglitazone, a novel thiazolidine-
4. Williamson RM, Price JF, Glancy S, et al. Edinburgh Type 2 Diabetes dione, improves non-alcoholic fatty liver disease in type 2 diabetes: its
Study Investigators. Prevalence of and risk factors for hepatic steatosis efficacy and predictive factors related to responsiveness. J Korean
and nonalcoholic Fatty liver disease in people with type 2 diabetes: the Med Sci. 2017;32:60-69.
Edinburgh Type 2 Diabetes Study. Diabetes Care. 2011;34:1139-1144. 25. Myers RP, Pollett A, Kirsch R, et al. Controlled attenuation parameter
5. Vilar-Gomez E, Chalasani N. Non-invasive assessment of (CAP): a noninvasive method for the detection of hepatic steatosis
non-alcoholic fatty liver disease: clinical prediction rules and based on transient elastography. Liver Int. 2012;32:902-910.
blood-based biomarkers. J Hepatol. 2018;68:305-315. 26. Sasso M, Beaugrand M, de Ledinghen V, et al. Controlled attenuation
6. Tahrani AA, Barnett AH, Bailey CJ. SGLT inhibitors in management of parameter (CAP): a novel VCTE™ guided ultrasonic attenuation mea-
diabetes. Lancet Diabetes Endocrinol. 2013;1:140-151. surement for the evaluation of hepatic steatosis: preliminary study
7. Bolinder J, Ljunggren Ö, Kullberg J, et al. Effects of dapagliflozin on and validation in a cohort of patients with chronic liver disease from
body weight, total fat mass, and regional adipose tissue distribution in various causes. Ultrasound Med Biol. 2010;36:1825-1835.
patients with type 2 diabetes mellitus with inadequate glycemic con- 27. Huwart L, Sempoux C, Vicaut E, et al. Magnetic resonance elastogra-
trol on metformin. J Clin Endocrinol Metab. 2012;97:1020-1031. phy for the noninvasive staging of liver fibrosis. Gastroenterology.
8. Jojima T, Tomotsune T, Iijima T, Akimoto K, Suzuki K, Aso Y. Empagliflo- 2008;135:32-40.
zin (an SGLT2 inhibitor), alone or in combination with linagliptin (a DPP-4 28. Musso G, Cassader M, Rosina F, Gambino R. Impact of current treat-
inhibitor), prevents steatohepatitis in a novel mouse model of non- ments on liver disease, glucose metabolism and cardiovascular risk in
alcoholic steatohepatitis and diabetes. Diabetol Metab Syndr. 2016;8:45. non-alcoholic fatty liver disease (NAFLD): a systematic review and
9. Tahara A, Kurosaki E, Yokono M, et al. Effects of SGLT2 selective meta-analysis of randomised trials. Diabetologia. 2012;55:885-904.
inhibitor ipragliflozin on hyperglycemia, hyperlipidemia, hepatic stea- 29. Girard J, Perdereau D, Foufelle F, Prip-Buus C, Ferré P. Regulation of
tosis, oxidative stress, inflammation, and obesity in type 2 diabetic lipogenic enzyme gene expression by nutrients and hormones. FASEB
mice. Eur J Pharmacol. 2013;715:246-255. J. 1994;8:36-42.
10. Komiya C, Tsuchiya K, Shiba K, et al. Ipragliflozin improves hepatic stea- 30. Garg H, Aggarwal S, Shalimar, et al. Utility of transient elastography
tosis in obese mice and liver dysfunction in type 2 diabetic patients irre- (fibroscan) and impact of bariatric surgery on nonalcoholic fatty liver
spective of body weight reduction. PLoS One. 2016;11(3):e0151511. disease (NAFLD) in morbidLy obese patients. Surg Obes Relat Dis.
11. Qiang S, Nakatsu Y, Seno Y, et al. Treatment with the SGLT2 inhibitor 2018;14:81-91.
luseogliflozin improves nonalcoholic steatohepatitis in a rodent model 31. Takase T, Nakamura A, Miyoshi H, Yamamoto C, Atsumi T. Amelioration
with diabetes mellitus. Diabetol Metab Syndr. 2015;7:104. of fatty liver index in patients with type 2 diabetes on ipragliflozin: an
12. Ito D, Shimizu S, Inoue K, et al. Comparison of ipragliflozin and piogli- association with glucose-lowering effects. Endocr J. 2017;64:363-367.
tazone effects on nonalcoholic fatty liver disease in patients with Type 32. Wong VW, Vergniol J, Wong GL, et al. Diagnosis of fibrosis and cir-
2 diabetes: a randomized, 24-week, open-label, active-controlled trial. rhosis using liver stiffness measurement in nonalcoholic fatty liver dis-
Diabetes Care. 2017;40:1364-1372. ease. Hepatology. 2010;51:454-462.
13. Kuchay MS, Krishan S, Mishra SK, et al. Effect of empagliflozin on liver 33. Koehler EM, Plompen EP, Schouten JN, et al. Presence of diabetes
fat in patients with type 2 diabetes and nonalcoholic fatty liver dis- mellitus and steatosis is associated with liver stiffness in a general
ease: a randomized controlled trial (E-LIFT trial). Diabetes Care. 2018; population: the Rotterdam study. Hepatology. 2016;63:138-147.
41:1801-1808. 34. KowdLey KV, Belt P, Wilson LA, et al. NASH Clinical Research Net-
14. Sandrin L, Fourquet B, Hasquenoph JM, et al. Transient elastography: work. Serum ferritin is an independent predictor of histologic severity
a new noninvasive method for assessment of hepatic fibrosis. Ultra- and advanced fibrosis in patients with nonalcoholic fatty liver disease.
sound Med Biol. 2003;29:1705-1713. Hepatology. 2012;55:77-85.
15. Roulot D, Costes JL, Buyck JF, et al. Transient elastography as a
screening tool for liver fibrosis and cirrhosis in a community-based SUPPOR TI NG I NFORMATION
population aged over 45 years. Gut. 2011;60:977-984.
16. Kwok R, Choi KC, Wong GL, et al. Screening diabetic patients for
Additional supporting information may be found online in the Sup-
non-alcoholic fatty liver disease with controlled attenuation parameter porting Information section at the end of the article.
and liver stiffness measurements: a prospective cohort study. Gut.
2016;65:1359-1368.
17. Roulot D, Roudot-Thoraval F, NKontchou G, et al. Concomitant How to cite this article: Shimizu M, Suzuki K, Kato K, et al.
screening for liver fibrosis and steatosis in French type 2 diabetic
Evaluation of the effects of dapagliflozin, a sodium-glucose co-
patients using Fibroscan. Liver Int. 2017;37:1897-1906.
18. Castera L. Diagnosis of non-alcoholic fatty liver disease/non-alcoholic transporter-2 inhibitor, on hepatic steatosis and fibrosis using
steatohepatitis: non-invasive tests are enough. Liver Int. 2018;38(suppl transient elastography in patients with type 2 diabetes and
1):67-70. non-alcoholic fatty liver disease. Diabetes Obes Metab. 2019;
19. de Lédinghen V, Vergniol J, Capdepont M, et al. Controlled attenua-
tion parameter (CAP) for the diagnosis of steatosis: a prospective
285–292. https://doi.org/10.1111/dom.13520
study of 5323 examinations. J Hepatol. 2014;60:1026-1031.

View publication stats