Amelioration of pancreatic fat accumulation in Japanese type 2 diabetes patients treated with sodium‐glucose cotransporter 2 inhibitors: a retrospective study

Abstract Background Prior reports have suggested that pancreatic fat is related to type 2 diabetes. Sodium‐glucose co‐transporter‐2 (SGLT‐2) inhibitors are expected to reduce ectopic fat accumulation. Aim This study assessed the effect of SGLT‐2 inhibitors on pancreatic and liver fat accumulations in patients with type 2 diabetes. Materials and Methods Retrospective analyses of indices of pancreatic and liver fat accumulations were conducted in 22 type 2 diabetes outpatients who were receiving SGLT‐2 inhibitors for more than 12 weeks. The differences between the pancreatic (P) or liver (L) and splenic (S) computed tomography values were evaluated. Results Fatty pancreas was defined as P−S < −8 Hounsfield Unit (HU), and the number of patients with fatty pancreas was 11 (50%). Fatty pancreas significantly improved after SGLT‐2 inhibitor use (median, −20.8; IQR, −34.8 to −14.3 HU vs. median, −14.6; IQR, −29.5 to −7.8 HU; p = 0.041). Fatty liver was defined as L−S ≤ 3.9 HU, and the number of patients with fatty liver was 11 (50%). Fatty liver significantly improved after SGLT‐2 inhibitor use (median, −4.3; IQR, −23.0 to 3.0 HU vs. median, −0.7; IQR, −5.2 to 6.3 HU; p = 0.016). Conclusion Pancreatic fat and liver fat accumulations might be reduced after treatment with SGLT‐2 inhibitors in type 2 diabetes patients with intense cumulative fat depositions in these organs.


| CT protocols
The differences between the pancreatic (P) and splenic (S) CT values (PÀ S), the differences between the liver (L) and splenic CT values (LÀ S), visceral fat area (VFA) (cm 2 ), subcutaneous fat area (SFA) (cm 2 ), and the ratio of VFA to SFA (V/S ratio) were measured. Two patients whose VFA and SFA were considered to be inaccurate were excluded from the evaluation of these parameters because one patient was not imaged on the entire trunk, and VFA and SFA of the other patient were affected by artifacts of the wrist joint located on the umbilical region. To evaluate PÀ S and LÀ S, the regions of interest (ROIs) were specified using areas of 1.0 cm 2 in sites that did not contain major vessels and measured the Hounsfield units (HU; CT value) from one ROI each. A pancreatic CT value is defined as the mean CT value in three different pancreatic parts: head, body and tail. A liver CT value is defined as the mean CT value in three different liver parts: anterior and posterior right lobe and left lobe. A splenic CT value is defined as the mean CT value in three different splenic parts: upper, middle and lower.
From the aforementioned pancreatic CT value and splenic CT value, indices of pancreatic fat content were defined as PÀ S, which were calculated as previously shown. 10 Similarly, indices of liver fat content were defined as LÀ S, which were calculated as previously shown. 11 The CT values were analyzed using the software program Aquarius Net Viewer Version 4.4 (TeraRecon, Inc.). The VFA, SFA and V/S were measured using SYNAPSE VINCENT image analysis system (Fujifilm, Inc.).

| Statistical analysis
Normally distributed data are presented as the means � SD, and nonnormally distributed data are presented as the medians and interquartile ranges. Changes in continuous clinical parameters before and after the treatment period were tested using a paired t test, and changes in PÀ S and LÀ S before and after the treatment period were tested using the Wilcoxon signed-rank test for nonnormally distributed values. p values < 0.05 were regarded as statistically significant. All statistical analyses were performed with JMP Pro 14 software (SAS Institute Inc.). Table 1    Note: Values are mean � SD.
Abbreviations: ALT, alanine amino transferase; AST, aspartate amino transferase; BMI, body mass index; CPR, C-peptide; FIB4 index, age � AST (platelet count (109/L)) � √AST); GA, glycoalbumin; HDL-C, high density lipoprotein cholesterol; IPMN, Intraductal papillary mucinous neoplasm; IRI, immunoreactive insulin; LDL-C, low density lipoprotein cholesterol; LÀ S, the differences between the liver (L) and splenic (S) CT values; PÀ S, the differences between the pancreatic (P) and splenic (S) CT values; SFA, subcutaneous fat area; VFA, visceral fat area; V/S, the ratio of VFA to SFA; γGTP, γ-glutamyl transpeptidase.  Figure S1). In addi-  Figure S2). In addition, the analysis excluding three persons who had the past history of hepatocellular carcinoma also showed a significant improvement  Table 2). There was no significant correlation between the increase of LÀ S and the reduction in body weight (r ¼ À 0.02, p ¼ 0.95), BMI The total number of patients with fatty pancreas or fatty liver before treatment was 17. Concurrence of fatty pancreas and fatty liver was found in five patients. In other words, not all 11 patients with fatty pancreas overlapped with 11 patients with fatty liver.

| DISCUSSION
The present study showed a possible amelioration of pancreatic fat as well as liver fat accumulations with SGLT-2 inhibitor use in cases with intense cumulative fat depositions in these organs, with improvements in various clinical parameters. This is the first study that showed a possible effect of SGLT-2 inhibitor on pancreatic fat accumulation.
One of the possible mechanisms by which SGLT-2 inhibitors affect pancreatic and liver fat might be assumed to be associated with a reduction in body weight and visceral fat. The prior study in patients with a BMI above 30 kg/m 2 reported that after administration of canagliflozin, approximately 70% of the reduction in body weight was due to a decrease in fat mass, and the observed decrease in fat mass was due to a 13.7% reduction in VFA. 13 In this study, body weight, BMI and VFA significantly decreased. However, there were no significant correlations between the reduction in these parameters and the reduction in pancreatic or liver fat in patients with intense cumulative fat depositions in these organs. Honka et al. reported that although the decrease in pancreatic fat was detected after bariatric surgery, any association was not found with the degree of the change in body weight or visceral fat mass. 14 Patel et al. reported that a significant reduction in BMI of at least 5% did not lead to a significant decrease in pancreatic fat. 15 As to liver fat, there are several reports that analyze the association between the decrease in body weight and the decrease in liver fat in patients with SGLT-2 inhibitors use; Takase et al. reported that the change in BMI and VFA under the treatment with ipragliflozin tended to be associated with the change in fatty liver index. 16 Patel et al. also reported that a reduction in BMI of at least 5% is associated with significant decrease in liver fat and volume in patients with nonalcoholic steatohepatitis. 15 On the other hand, Kuchay et al. reported that liver fat reduction under the treatment with empagliflozin is irrespective of body weight reduction. 17 These are still controversial, however, considering these reports and the present study, the reduction in pancreatic and liver fat may be mediated by factors other than or additional to the reduction in body weight or VFA after the treatment with SGLT-2 inhibitors.
Another mechanism may be associated with a reduction in insulin levels, although this could not be affirmed in the study due to insufficient data. Insulin-signaling pathways play a significant role in adipogenesis. 18 Pancreatic islets transplanted into portal vein can induce focal hepatic steatosis in type 1 diabetes recipient, 19  rameters. This means that pancreatic fat accumulation is less associated with obesity-related parameters than liver fat. Some other reasons why patients with fatty pancreas were not identical to patients with fatty liver could be considered as follows; first, hepatic fat is mainly derived from intracellular fat deposition, while pancreatic fat is mainly derived from adipocytes infiltration, 23 second, fat loss in pancreas seems to be independent of that in liver in patients with bariatric surgery, suggesting tissue-specific mobilization of these ectopic fat stores, 24 finally, pancreatic fat has a significant correlation with age, 24,25 while liver fat does not, 24 which are consistent with the results of this study (data not shown).
There were several limitations in the present study. First, this is a retrospective observational study and the sample size was small. It can't be concluded that the results were solely derived from the effect of SGLT-2 inhibitors. Second, histological confirmation of fat accumulation in pancreas and liver could not been done in these patients. However, PÀ S and LÀ S is strongly associated with histological pancreatic and liver fat fraction. 10 Third, the effects of other diabetes therapy including insulin, sulfonylurea and glucagon-like The changes in indices of fat content and physical parameters between baseline and after the administration of SGLT-2 inhibitors in 11 patients with PÀ S < À 8 and 11 patients with LÀ S ≤ 3.9 Note: Values are mean � SD or median (25th-75th percentiles).
Abbreviations: BMI, body mass index; LÀ S, the differences between the liver (L) and splenic (S) CT values; PÀ S, the differences between the pancreatic (P) and splenic (S) CT values; SFA, subcutaneous fat area; VFA, visceral fat area; V/S, the ratio of VFA to SFA.