Changes in pancreatic exocrine function, fat and fibrosis in diabetes mellitus: analysis using MR imaging

Abstract

Objective

To evaluate the relationships between hemoglobin A1c (HbA1c) levels with exocrine pancreatic function using cine-dynamic magnetic resonance cholangiopancreatography (MRCP) and the pancreatic parenchyma using fat-suppressed T1 mapping and the proton density fat fraction (PDFF).

Methods

Patients who underwent 3T-MRI and HbA1c measurement were retrospectively recruited. MRI included cine-dynamic MRCP with a spatially selective inversion-recovery (SS-IR) pulse, fat-suppressed Look-Locker T1 mapping and multiecho 3D Dixon-based PDFF mapping. The pancreatic exocrine secretion grade on cine-dynamic MRCP, T1 values, and PDFF were analyzed in non-diabetic (n = 32), pre-diabetic (n = 44) and diabetic (n = 23) groups defined using HbA1c.

Results

PDFF was weakly correlation with HbA1c (ρ = 0.30, p = 0.002). No correlations were detected between HbA1c and secretion grade (ρ = - 0.16, p = 0.118) or pancreatic parenchymal T1 (ρ = 0.13, p = 0.19). The secretion grade was comparable between the three groups. The T1 value was higher in diabetic (T1 = 1006.2+/- 224.8 ms) than in non-diabetic (T1 = 896.2+/- 86.3 ms, p = 0.010) and pre-diabetic (T1 = 870.1+/- 91.7 ms, p < 0.010) patients. The PDFF was higher in diabetic (FF = 11.8+/- 8.7 %) than in non-diabetic (FF = 6.8+/- 4.2 %, p = 0.014) patients.

Conclusion

Pancreatic exocrine function, T1, and FF showed no correlation with HbA1c. Pancreatic T1 and fat fraction is increased in patients with Type 2 diabetes mellitus.

Advances in knowledge

This study demonstrates unaffected exocrine function in pre-diabetes and diabetes and confirms that pancreatic parenchymal T1 and FF are increased in patients with diabetes.

Introduction

Exocrine and endocrine pancreatic function are closely related anatomically and physiologically. It has been reported that pancreatic exocrine dysfunction is present in 30 to 50% of patients with Type 2 diabetes mellitus (T2DM) on direct and indirect pancreatic exocrine function tests. ^1,2 Impaired glucose tolerance (IGT) is known to cause pancreatic exocrine dysfunction, pancreatic fibrosis, and pancreatic fat deposition. ^3,4 Patients with T2DM also have abdominal complaints such as bloating, steatorrhea, dyspepsia, malnutrition, and weight loss that impair quality of life. ^5,6 These abdominal complaints are thought to arise in part from pancreatic exocrine dysfunction. There are various reports of the mechanisms and factors in T2DM that induce pancreatic exocrine dysfunction, pancreatic fibrosis, and pancreatic fat deposition, but there is a lack of comprehensive multiparametric analyses of these parameters in patients with disease.

Gold-standard pancreatic exocrine function tests include direct hormone-stimulated tests (secretin and/or secretin-cholecystokinin stimulation tests), but they are invasive, expensive, and time-consuming, and they are not widely available. ⁷ Further, while indirect methods (fecal chymotrypsin [FC] test, fecal elastase-1 [FE-1] test, and N-benzoyl-L-tyrosyl-p-aminobenzoic acid [BT-PABA] test) are non-invasive, they are also time-consuming and complicated. ^8–10 In addition, secretin and FE1 products are not commercially available in Japan. Thus, in the clinical setting, pancreatic exocrine dysfunction cannot be easily diagnosed. Recent studies have shown that cine-dynamic magnetic resonance cholangiopancreatography (MRCP) with a spatially selective inversion recovery (SS-IR) pulse enables direct and non-invasive visualization of the flow of pancreatic secretions. ^11–14 With a significant correlation between pancreatic juice secretion grade estimated by cine-dynamic MRCP with an SS-IR pulse and the BT-PABA test, it has been proposed that this method might have the potential to aid in the assessment of pancreatic exocrine function. ^12,14 Pancreatic parenchymal T1 and proton density fat fraction (PDFF) have been proposed as non-invasive measures of pancreatic fibrosis and fat deposition as pancreatic biopsy is not routinely performed in standard clinical practice, because of the potential for significant complications. Recently, T1 mapping has been used to assess the degree of fibrosis in several organs, particularly myocardium, ¹⁵ liver, and pancreas. ^16–19 Noda et al reported that pancreatic T1 values indirectly reflect the degree of pancreatic fibrosis and correlate with hemoglobin A1c (HbA1c) levels. ²⁰ One method to quantify tissue fat content by imaging is a chemical shift-based technique with measurement of the PDFF on MRI. ^21,22 Recent studies have shown that PDFF-based techniques for quantifying tissue fat are highly accurate in various organs. ^23–25

Thus, the purpose of this study was to evaluate the relationships between HbA1c levels and pancreatic exocrine function, pancreatic fat deposition, and pancreatic fibrosis using 3T-MRI.

Methods and materials

Patient selection

Our institutional review board approved this retrospective study and waived the requirement for patient informed consent. Review of the medical records at Kawasaki Medical School identified 349 consecutive adult patients (age over 20 years) between November 2016 and December 2017 with known or suspected hepatobiliary or pancreatic disease based on either their clinical history or previously performed abdominal ultrasonography or previously performed CT, who underwent 3T-MRI with a pancreatobiliary protocol. Of these patients, those who had an HbA1c test within 2 months before or after MRI were identified. Of the 349 patients, 230 were excluded for the following reasons: no HbA1c data (n = 57); presence of intraductal papillary mucinous neoplasm (n = 142), chronic pancreatitis (n = 6), or pancreatic ductal adenocarcinoma (n = 7); surgery for pancreatic ductal adenocarcinoma (n = 3), mucinous cystic neoplasm (n = 1), or solid pseudopapillary neoplasm (n = 1); suspected solid pseudopapillary neoplasm (n = 1), serous neoplasm (n = 1), or pancreatic neuroendocrine tumor (n = 1); and unstable breath-holding during MRI (n = 10). In addition, 20 patients (under 50 years of age) were excluded to adjust the age distribution among the three groups. A total of 99 consecutive patients who satisfied the inclusion criteria (40 men, 59 women; mean age, 67.5 years; age range, 51–85 years) were included in the study. All patients underwent 3T-MRI with a pancreatobiliary protocol that included cine-dynamic MRCP with an SS-IR pulse, fat-suppressed T1 mapping, and multiecho 3D DIXON.

Using the American Diabetes Association criteria, ²⁶ the patients were classified into the following groups according to their HbA1c levels: non-diabetic group (n = 32), HbA1c < 5.7%; pre-diabetic group (n = 44), 5.7% ≤ HbA1c < 6.5%; and diabetic group (n = 23), HbA1c ≥ 6.5% (Figure 1).

Figure 1.

Flow chart of patient selection. HbA1c, hemoglobin A1c

MRI technique

All MRI examinations were performed using a 3T-MR system (Ingenia 3 T CX Quasar Dual; Philips Medical Systems, Best, The Netherlands) with 32-channel phased-array coils (anterior and posterior coils). All patients fasted for at least 3 h before the examination. Immediately before scanning, each patient ingested 250 ml of Bothdel Oral Solution 10 (36 mg of manganese chloride tetrahydrate; Kyowa Hakko Kirin, Tokyo, Japan) as a negative contrast material to reduce signals from the bowel on T ₂ weighted imaging.

Cine-dynamic MRCP with an SS-IR pulse was performed as follows. As a reference image, a thick-slab 2D-MRCP image was obtained under breath-holding, using a fast advance spin echo sequence to depict the main pancreatic duct (MPD) in the coronal plane, using the imaging parameters shown in Table 1. An SS-IR pulse (inversion time, 2200 ms; width, 20 mm) was then applied over the head of the pancreas, perpendicular to the MPD. The pulse nulls signals from water, such as that present in pancreatic juice, so that static pancreatic juice in the area of the SS-IR pulse appears dark. However, inflow of secreted pancreatic juice into the area under observation is observed as a high-intensity signal within the area of the SS-IR pulse. MRCP with an SS-IR pulse was performed every 15 s for 5 min (imaging time including the inversion time, 8 s). A total of 20 single-shot images were obtained in a cine-dynamic manner.

Table 1.

MRI parameters of the four sequences evaluated in the current study

Parameter	2D-MRCP with a SS-IR pulse	Fat-suppressed T1 mapping (Look-Locker acquisition with PROSET 1: 2: 1)	T 1 weighted axial gradient-recalled-echo	Multiecho 3D DIXON (3D mDIXON quant)
Field of view (mm2)	350 ⋅ 350	350 ⋅ 294	350 ⋅ 300	350 ⋅ 287
Slice thickness (mm)	50	8	3	3
Matrix	288 ⋅ 288	128 ⋅ 128	224 ⋅ 177	144 ⋅ 100
TR (ms)	8000	6.0	3.2	5.9
TE/ δ TE (ms)	600/-	2.9/-	1.14/2.0	1.02/0.7
Inversion time (ms)/ width (mm)	2200/20	/	/	/
Flip angle (°)	90	7	10	3
Number of slices	1	1	69	56
Bandwidth (Hz/pixel)	384.1	498.2	1992.8	2515.7
NEX	1	1	1	1
Minimum TI delay (ms)	-	55.8	-	-
Acquisition time (s)	8 ⋅ 20 (times)	20	19.0	15.4
Breath-holding	○	○	○	○
SENSE (phase) ⋅ (slice)	2.0	2.5	2.3 ⋅ 1.0	2.0 ⋅ 1.2

MRCP, magnetic resonance cholangiopancreatography; SS-IR, spatially selective inversion recovery; TE, echo time; TR, repetition time.

The parameters for fat-suppressed T1 mapping using the Look-Locker acquisition with PROSET 1:2:1 sequence was as shown in Table 1. It is known that the T1 value of the organs could be affected by fat deposition. ²⁷ Increased fat deposition in the pancreatic parenchyma is observed in many conditions, including aging and obesity. As the T1 relaxation time of adipose tissue is significantly shorter than that of pancreatic parenchyma, increased fat deposition in the pancreatic parenchyma could alter the true pancreatic parenchymal MR signal. A previous study has shown that the fat-suppression technique in T1 mapping is essential for evaluating the T1 value of pancreatic parenchyma. ²⁸ The parameters for the T ₁ weighted axial gradient-recalled-echo sequence (T ₁WI) were as shown in Table 1. The parameters for the multiecho 3D DIXON (3D mDIXON quant) sequence were as shown in Table 1. Parametric PDFF maps were automatically generated on the imager software from the 6-echo mDIXON examination.

Image analysis

Two fellowship-trained radiologists with 15 years and 11 years of experience in abdominal MRI (Akihiko Kanki and Kazuya Yasokawa, respectively), who were blinded to the patients’ sex, age, and other clinical information, performed the image analysis. They retrospectively and independently reviewed twice each the cine-dynamic MRCP and PDFF map on a standard picture archiving and communication system (PACS) (Synapse EX, Fujifilm Corporation, Tokyo, Japan), and T1 map and T ₁WI on a workstation with dedicated post-processing software (off-line Philips Research Integrated Development Environment (PRIDE) software; Philips Healthcare, Best, Netherlands). Each patient’s images were presented to the readers in a random order under the control of the study coordinator (Hiroki Nakamura).

The cine-dynamic MRCP images were evaluated for the secretion grade of pancreatic juice, which was scored according to the distance of movement of pancreatic juice in the pancreatic duct within the 20 mm area of the SS-IR pulse, using the following 5-point scale: Grade 0, no flow; Grade 1, ≤ 5 mm; Grade 2, 6–10 mm; Grade 3, 11–15 mm; and Grade 4, >15 mm (Figures 2 and 3). The secretion grade in each set of images was defined as follows: (total of the grading scores in 20 images)/20) ^12–14 (Figures 2 and 3). Next, the readers measured the T1 value of pancreatic parenchyma on the single-slice T1 map image using a region of interest (ROI) placed on the head of the pancreas using the T ₁W images as a reference. The largest possible circular or oval ROI was placed on a homogeneous region of the pancreatic parenchyma, avoiding the pancreatic duct, vessels, retroperitoneal fat, and artifacts (Figures 2 and 3). Finally, the PDFF (%) was measured using an ROI placed on the head, body, and tail of the pancreas on parametric PDFF map images. The largest possible circular or oval ROI was placed, avoiding the pancreatic duct, vessels, retroperitoneal fat, and artifacts. The PDFF (%) in each patient used the mean value of the head, body, and tail (Figures 2 and 3). The secretion grade, the T1 value and the PDFF (%) were measured twice each; two readers independently measured each parameter, and the mean value was subsequently calculated.

Figure 2.

A 66-year-old male with an HbA1c value of 5.3%. (a) Cine-dynamic MRCP with an SS-IR pulse. All grades are observed (Grades 0, 1, 2, 3, and 4). The secretion grade is 2.00. Secretory inflow of pancreatic juice is visualized as a high-intensity signal within the area of the SS-IR pulse. (b) Fat-suppressed, T ₁ weighted image. The arrow indicates the head of the pancreas for the ROI placement. (c) T1 map showing a mean pancreatic T1 value of 822.5 ms for an ROI placed on the head of the pancreas, using the fat-suppressed T ₁ weighted image (b, arrow) as a reference. (d) PDFF maps showing a mean pancreatic PDFF of 4.36% for an ROI placed on the head, body, and tail of the pancreas. HbA1c, hemoglobin A1c; MRCP, magnetic resonance cholangiopancreatography; PDFF, proton density fat fraction; ROI, region of interest; SS-IR, spatially selective inversion recovery.

Figure 3.

A 73-year-old female with an HbA1c value of 8.1%. (a) Cine-dynamic MRCP with an SS-IR pulse. Grades 0 and 1 are observed, but Grades 2, 3, and 4 are not observed. The secretion grade is 0.08. (b) Fat-suppressed T ₁ weighted image. The arrow indicates the head of the pancreas for the ROI placement. (c) T1 map showing a mean pancreatic T1 value of 981.5 ms for an ROI placed on the head of the pancreas, using the fat-suppressed T ₁ weighted image (b, arrow) as a reference. (d) PDFF maps showing a mean pancreatic PDFF of 11.94% for an ROI placed on the head, body, and tail of the pancreas. HbA1c, hemoglobin A1c; MRCP, magnetic resonance cholangiopancreatography; PDFF, proton density fat fraction; ROI, region of interest; SS-IR, spatially selective inversion recovery.

Interobserver and intraobserver reproducibility

The intraclass correlation coefficient (ICC) was calculated to evaluate the intraobserver reproducibility (ICC-1) and interobserver reproducibility (ICC-2) of the secretion grade of pancreatic juice, the pancreatic parenchymal T1 value and the PDFF of pancreatic parenchyma. The entire cohort were for the secretion grade of pancreatic juice and ROI segmentation by two radiologists who both had more than 10 years of experience of interpreting abdominal images. The performances of two radiologists (** and **) were used to assess the interobserver ICC. Then, the same procedure was repeated 1 week later for both radiologists. An ICC greater than 0.7 was considered to represent good agreement.

Statistical analysis

Based on previous studies, a secretion grade <0.70 on cine-dynamic MRCP was used as the cut-off value for pancreatic exocrine dysfunction. ^12,14 Statistical analyses were performed using SPSS for Windows v. 22.0 software (SPSS, Chicago, IL). The HbA1c, the T1 value of pancreatic parenchyma, the secretion grade of pancreatic juice on cine-dynamic MRCP, and the PDFF of pancreatic parenchyma obtained in this study were all non-normally distributed in the Shapiro-Wilk test. Correlations among the HbA1c level, T1 value, secretion grade, and PDFF were assessed using Spearman’s rank correlation coefficient (ρ) analysis. The Kruskal-Wallis test was used to assess the significance of differences in the T1 value of pancreatic parenchyma, the secretion grade of pancreatic juice on cine-dynamic MRCP, and the PDFF of pancreatic parenchyma among the three groups. If the p-value of the Kruskal-Wallis test showed a significant difference (p < 0.05), pairwise comparisons between the two groups were performed using the Mann–Whitney U test with Bonferroni correction.

Results

Based on the classification of the patients into groups using the American Diabetes Association criteria ²⁶ according to HbA1c levels, 32 patients were included in non-diabetic group, 44 in pre-diabetic group, and 23 in diabetic group. Age was 65.2 ± 8.0 years for non-diabetic group, 68.8 ± 8.9 years for pre-diabetic group, and 68.2 ± 7.7 years for diabetic group. There were no significant differences in age in any of the group comparisons.

The ICCs of the secretion grade of pancreatic juice, the pancreatic parenchymal T1 value and the PDFF of pancreatic parenchyma are summarized in Table 2. Similar good reproducibility was demonstrated for each parameter.

Table 2.

Reproducibility of the secretion grade of pancreatic juice, the T1 value and the PDFF of pancreatic parenchyma

Parameter	ICC-1a	ICC-1b	ICC-2
The secretion grade of pancreatic juice	0.949	0.994	0.976
95% confidence interval	0.925–0.966	0.992–0.996	0.965–0.984
p	<0.001	<0.001	<0.001
The T1 value of pancreatic parenchyma	0.944	0.945	0.831
95% confidence interval	0.917–0.962	0.920–0.963	0.759–0.884
p	<0.001	<0.001	<0.001
The PDFF of pancreatic parenchyma	0.981	0.996	0.986
95% confidence interval	0.972–0.987	0.994–0.997	0.980–0.991
p	<0.001	<0.001	<0.001

ICC, intraclass correlation coefficient; ICC-1, intraobserver reproducibility; ICC-2, interobserver reproducibility; ICC-1a, Reader 1 (11 years of experience); ICC-1b, Reader 2 (15 years of experience); PDFF, proton density fat fraction.

Age was not significantly correlated with HbA1c (ρ = 0.19, p = 0.060), secretion grade (ρ = -0.13, p = 0.19), pancreatic parenchymal T1 value (ρ = 0.087, p = 0.39), or PDFF (ρ = 0.16, p = 0.11).

HbA1c values were not significantly correlated with secretion grade on cine-dynamic MRCP with an SS-IR pulse (ρ = -0.16, p = 0.118). The secretion grade of each group is shown in Figure 4a. The secretion grade was 0.98 ± 0.82 for non-diabetic group, 1.04 ± 0.73 for pre-diabetic group, and 0.69 ± 0.75 for diabetic group. There were no significant differences in the secretion grade in three-group comparisons (non-diabetic group vs pre-diabetic group; p = 1.000, non-diabetic group vs diabetic group; p = 0.514, and pre-diabetic group vs diabetic group; p = 0.251). A secretion grade <0.70 was used as the cut-off value for pancreatic exocrine dysfunction ^12,14 ; pancreatic exocrine dysfunction was found in 53.1% (17/32) of the non-diabetic group, 38.6% (17/44) of the pre-diabetic group, and 65.2% (15/23) of the diabetic group. There were no significant differences in the prevalence of exocrine dysfunction in a three-group comparison (p = 0.104). HbA1c values were not significantly correlated with the pancreatic T1 value from T1 mapping (ρ = 0.13, p = 0.19). The pancreatic T1 value for each group is shown in Figure 4b. The T1 value was significantly higher in diabetic group (1006.2 ± 224.8 ms) than in pre-diabetic group (870.1 ± 91.7 ms) and non-diabetic group (896.2 ± 86.3 ms) (p < 0.001, and p = 0.010, respectively). There was no significant difference in the pancreatic T1 values between non-diabetic group and pre-diabetic group (p = 1.00). Only PDFF was significantly, but weakly, correlated with HbA1c (ρ = 0.30, p = 0.002). The PDFF for each group is shown in Figure 4c. The PDFF was significantly higher in diabetic group (11.82 ± 8.70%) than in non-diabetic group (6.84 ± 4.15%) (p = 0.014). There were no significant differences in the PDFF between non-diabetic group and pre-diabetic group (9.42 ± 6.13%) (p = 0.245) and between pre-diabetic group and diabetic group (p = 0.423).

Figure 4.

Secretion grade, pancreatic T1 value, and PDFF by group according to the American Diabetes Association criteria. (a) The secretion grade shows no significant differences in any of the groups. A secretion grade <0.70 was used as the cut-off value for pancreatic exocrine dysfunction. Pancreatic exocrine dysfunction is found in 53.1% (17/32) of non-diabetic, 38.6% (17/44) of pre-diabetic group, and 65.2% (15/23) of diabetic group. (b) The pancreatic T1 value is significantly higher in diabetic group than in pre-diabetic group and non-diabetic (p < 0.001, and p = 0.010, respectively). (c) The PDFF is significantly higher in diabetic group than in non-diabetic (p = 0.014). Note: The horizontal lines indicate significant differences between groups. PDFF, proton density fat fraction.

Discussion

T2DM is known to cause pancreatic exocrine dysfunction, pancreatic fibrosis, and pancreatic fat deposition. ^1–4,29 However, there have been various reports of the relationships between HbA1c and exocrine pancreatic function and pancreatic parenchymal changes. In the present study, MRI was used to evaluate the relationship between the HbA1c level and pancreatic exocrine dysfunction estimated by cine-dynamic MRCP with an SS-IR pulse, pancreatic fibrosis estimated by T1 mapping, and pancreatic fat deposition by PDFF.

Secretion grade, calculated by cine-dynamic MRCP with an SS-IR pulse, has been introduced in clinical practice as a simple, physiological, and non-invasive method that has the potential to aid in the evaluation of pancreatic exocrine function. ^12–14 In the present study, no significant correlation was found between HbA1c and secretion grade. Terzin et al ³⁰ reported that although exocrine pancreatic insufficiency demonstrated using fecal pancreatic elastase-1 (PE-1) levels is more frequent in T2DM patients with poor glycemic control, the PE-1 level was not correlated with HbA1c. There were no significant differences in the secretion grade in any of the group comparisons. Pancreatic exocrine dysfunction (a secretion grade <0.70) was observed in 65.2% of diabetic group, but also in 38–53% of non-diabetic group (53.1%) and pre-diabetic group (38.6%). Torigoe et al ³¹ reported that the age-related decrease in pancreatic exocrine secretion evaluated by cine-dynamic MRCP with an SS-IR pulse may appear in middle-elderly people without pancreatic diseases. In the present study, the subjects were over 50 years of age, and pancreatic exocrine dysfunction may have been strongly affected by aging. The prevalence of pancreatic exocrine dysfunction in the T2DM group in the present study was higher than in previous reports. ¹ The difference in the prevalence of pancreatic exocrine dysfunction may be due to age, since the mean age of the T2DM group in the present study was higher than in previous reports (mean age; 68.2 vs 53.8 years). Furthermore, the difference in the prevalence of pancreatic exocrine dysfunction may be due to the difference in the method of evaluating pancreatic exocrine dysfunction, although there have been no previous reports evaluating the relationship between pancreatic exocrine secretion evaluated by cine-dynamic MRCP with an SS-IR pulse and FE-1 concentration.

No significant correlations were found between HbA1c and the pancreatic T1 value. Noda et al reported that HbA1c and the pancreatic T1 value using T1 mapping showed a significant positive correlation. ^20,32 The pancreatic parenchyma is known to lose acinar cells and undergo replacement by fibrosis with age. In previous reports, age ranged from 31 to 88 and 42 to 83 years, suggesting that T1 values may also include the effects of aging. On the other hand, in the present study, the T1 value of the pancreatic parenchyma was considered to be less affected by aging, because the subjects were over 50 years of age. Moreover, the T1 value of the pancreatic parenchyma using fat-suppressed T1 mapping in our study was considered to reflect fibrosis of the pancreatic parenchyma without the effect of pancreatic fat deposition. However, the pancreatic T1 value was significantly higher in diabetic group than in non-diabetic group and pre-diabetic group, although no differences in the pancreatic T1 value were seen in any other comparisons. Reduced T1 signal intensity of pancreatic parenchyma has been linked to the loss of acinar cells (which contain protein-rich cytoplasm), and their replacement by fibrosis. ^33,34 A previous study reported that, in T2DM patients, marked acinar atrophy and pancreatic fibrosis significantly reduced islet mass. ³⁵ Noda et al ³² reported that the pancreatic T1 value was significantly higher in T2DM than in non-diabetes subjects and pre-diabetes, which is consistent with the present results. The pancreatic T1 value is significantly increased in diabetic group compared to non-diabetic group and pre-diabetic group, but pancreatic exocrine dysfunction is already present in 38–53% of non-diabetic and pre-diabetic group. These findings suggest that the pancreatic T1 in T2DM may reflect pancreatic fibrosis rather than acinar cell depletion. In addition, it was suggested that it is necessary to include the duration of T2DM process when evaluating the pancreatic T1 value.

In the present study, only the PDFF was significantly, but weakly, correlated with HbA1c. T2DM is known to be associated with ectopic fat deposition in the pancreas. On the other hand, Saisho et al ³⁶ reported that, although pancreatic fat increases with age and obesity, it was not increased in T2DM in a postmortem analysis of adults. In addition, a recent study with measurement of the PDFF using MRI has shown that fatty pancreas with adjustment for age, sex, and BMI was not related to pre-diabetes or T2DM. ³⁷ In the present study, in PDFF comparisons, significant differences were seen between non-diabetic group and diabetic group. However, because BMI was not measured, the present results may have been influenced by an inability to exclude obesity.

The present study has several limitations. First, there might have been selection bias because of the retrospective study design. Second, histopathological evidence of pancreatic parenchymal changes such as pancreatic fibrosis and pancreatic fat deposition was not obtained. On the other hand, a recent study reported that the sensitivity and specificity of PDFF obtained from multiecho 3D DIXON in detecting histologic steatosis were 95.0 and 100%, respectively. ²⁵ However, the sensitivity and specificity of PDFF measurements in the pancreas remains unknown because PDFF sequences designed for towards the measurement of hepatic fat. Furthermore, whether the measured changes in pancreatic fat are due to increased intracellular fat, parenchymal atrophy or both, or indeed the relationship and significance of these very different pathophysiological processes in the evolution of metabolic disease and obesity are largely unknown. ³⁸ Therefore, additional study including a pathological correlation is needed to evaluate the relationship with the pancreatic T1 value and PDFF. Finally, stimulation tests, which are the gold-standard for evaluating pancreatic exocrine function, were not used. However, previous reports have shown a significant correlation between pancreatic juice secretion grade evaluated by cine-dynamic MRCP with an SS-IR pulse and the BT-PABA test, suggesting that it may be a potential replacement for indirect pancreatic exocrine function tests. ^12–14

In conclusion, pancreatic exocrine dysfunction, pancreatic parenchymal fibrosis, and pancreatic fat deposition showed no correlation with HbA1c level. Our results also suggest that pancreatic parenchymal fibrosis and pancreatic fat deposition may be higher in T2DM. Future work with methodological control of patient age, disease duration and pattern of T2DM disease when evaluating T1 values and PDFF would be important. The evaluation of the pancreatic exocrine function, pancreatic fat deposition, and pancreatic fibrosis using 3T-MRI may contribute to mechanistic implications for T2DM, which in turn could be useful in better understanding disease pathophysiology.