Philippe Garteiser^{1}, Vinciane Rebours^{1,2}, Sabrina Doblas^{1}, Gwenael Pagé^{1}, André Bado^{1}, Valérie Paradis^{1,3}, Maude Le Gall^{1}, Anne Couvelard^{1,4}, and Bernard E Van Beers^{1,5}

In obesity, pancreas is affected by fatty infiltration and fibrosis. Bariatric surgery is one of the only therapies which demonstrably improves pancreas status in obesity. In the present work we investigated the mechanical properties at several frequencies, the PDFF and the T2* of pancreatic explants in an obese rat model of bariatric surgery. Bariatric surgery reversed the MRI parameters to values not significantly different than controls. MRI parameters closely matched the reference histology observations. MRE and multiparametric MRI may be used to monitor pancreatic status and treatment response in an obese rat model of bariatric surgery.

*
Animals: *Three groups of male Wistar rats were
composed: obese rats fed with high fat diet (HFD) for 6 months (n = 9); rats
fed with normal diet (ND) for 6 months (n = 6) and obese rats which underwent bariatric surgery at 3 months, followed by ND for 14
days (n=5).
After MR imaging, the pancreas were analyzed with hematoxylin and eosin staining for the
quantification of fibrotic complexes and fat deposits, and Perls staining for
the evaluation of iron deposition.

*MRE:* MRE was performed using a three-directional spin-echo sequence at
400Hz, 600Hz and 800Hz to accommodate the softness of pancreatic tissue(4), with 300µm isotropic resolution. Viscoelastic parameters (storage
modulus, G' and loss modulus, G'' at all three frequencies, and their frequency
dispersion coefficient, γ, expressed as the exponent of a fit to a power law)
were obtained by algebraic inversion of the wave propagation equation.

*R2*
and proton density fat fraction (PDFF):*
acquisition consisted of a gradient echo sequence (TE=1.65ms+n·0.92ms,
n=0···15) with 250µm in-plane resolution, 1mm slice thickness, 300kHz
bandwidth, 950ms TR, 15° flip angle and 4 signal averages. T2* (expressed as
R2*) and fat fraction were extracted by fitting the magnitude image to a model
accounting for T2* decay and signal interferences from water and the 7 most
abundant lipid peaks.

The storage modulus at 600Hz was
significantly different between normal, obese and surgery groups (1.23[0.92-1.33]kPa,
1.71[1.27-2.06] kPa and 1.46[1.38-1.78] kPa, respectively, Kruskall-Wallis p=0.0037,
figure 1). The storage modulus was significantly higher in obese rats versus
controls (p < 0.01). After surgery the storage modulus decreased to a value
that was not significantly different from controls. No significant differences
were found between groups regarding the loss modulus at any of the three tested
frequencies. The frequency dispersion coefficient of the storage modulus was
significantly different between control, obese and surgical groups (1.53[1.34-1.97],
1.21[1.04-1.42] and 1.55[1.15-1.65], respectively, Kruskall-Wallis p=0.02,
figure 2). After surgery the wave dispersion coefficient reverted to a value
that was not significantly different from the control group.
When pooling the measurements obtained from
the three experimental groups, both the storage modulus G' at 600Hz and the
storage modulus dispersion coefficient γ_{G'} were significantly
correlated to the number of fibrotic complexes (r=0.61, p=0.0059 and r=-0.52,
p=0.0216, respectively, figure 3).

The PDFF were 0.00[0.00-0.01]%, 0.39[0.00-1.59]%, and 0.20[0.00-0.49]% in the control, obese and surgical groups, respectively (Kruskal-Wallis p=0.014, figure 4), and were significantly correlated to the number of adipocytes per unit area (r = 0.71, p=0.0005). The R2* were 83.92[63.88-93.38] Hz, 104.39[94.14-164.42] Hz and 83.80[64.57-90.20] Hz for the control, obese and surgery groups, respectively (Kruskal-Wallis p=0.0008), and were significantly correlated to the number of Perls positive spots (figure 5).

Multiple regressions indicated that the number of endofibrotic complexes was the sole histologic factor associated to storage modulus at 600Hz while the frequency dispersion of the storage modulus was not associated to any single histologic factor. The number of adipocytes was the only histologic factor associated with PDFF, while the amount of Perls staining was the sole histologic factor associated with R2*.

France Life Imaging network

Imageries Du Vivant network

1. Rebours V, Gaujoux S, d'Assignies G, et al. Obesity and Fatty Pancreatic Infiltration Are Risk Factors for Pancreatic Precancerous Lesions (PanIN). Clin Cancer Res. 2015;21(15):3522-8.

2. Garteiser P, Doblas S, Cavin J-B, et al. Abstract 1906: Pancreatic disease in obesity: observations on fat content, relaxometric properties and mechanical properties in the rat ex vivo. Powerpitch. Proceedings of the 2016 ISMRM conference. 2016.

3. Mingrone G, Panunzi S, De Gaetano A, et al. Bariatric-metabolic surgery versus conventional medical treatment in obese patients with type 2 diabetes: 5 year follow-up of an open-label, single-centre, randomised controlled trial. Lancet. 2015;386(9997):964-73.

4. Shi Y, Glaser KJ, Venkatesh SK, Ben-Abraham EI, Ehman RL. Feasibility of using 3D MR elastography to determine pancreatic stiffness in healthy volunteers. Journal of magnetic resonance imaging: JMRI. 2015;41(2):369-75.

Figure 1: Storage modulus (G') at 600Hz
measured in control, obese and bariatric surgery groups. Significant difference
isobserved in obese group, in which storage modulus ishigher. Surgery markedly
improves the storage modulus to values that are not significantly different
from those of controls.

Figure 2: Frequency
dispersion coefficient (γ) for storage modulus measured in control,
obese and bariatric surgery groups. Decreased value is observed in the obese
group, while surgery partially reverses the value to levels close to the
controls.

Figure 3: Correlations between storage
modulus (G') at 600Hz (left panel) and frequency dispersion coefficient of storage modulus (right panel) and the number
of fibrotic complexes. Significant correlations are seen in the total cohort
formed by the three experimental groups (storage modulus: Spearman r=0.61,
p=0.0059; frequency dispersion coefficient: Spearman r=-0.52, p=0.0216).

Figure 4: Proton density fat fraction measured
in the control, obese and surgery groups.

Figure 5: R2* relaxation
rate measurements correlate to number of Perls positive spots in the cohort
formed by the three experimental groups (Spearman r=0.77, p=0.0001).