Diffusion-weighted imaging in normal pancreas: Apparent diffusion coefficient test-retest repeatability

Background: The assessment of apparent diffusion coefficient (ADC) repeatability is important to body diffusion weighted imaging (DWI). However, data onto evaluating the repeatability of ADC values for normal pancreas are limited. The aim of this study was to evaluate the ADC test-retest repeatability of the normal pancreas. Methods: Twenty-six healthy volunteers (mean 47.6 years; 13 men) were included and scanned twice with reposition using a DWI sequence at 3.0-T. ADCs of the pancreatic head, body and tail for two DWIs were measured by two readers. The mean ADCs of the pancreatic head, body and tail were calculated as the global pancreatic ADC. Test-retest repeatability and agreement of ADC measurement were evaluated by the Bland-Altman analysis, intra-class correlation coefficient (ICC) and coefficient of variation (CV). Results: The global pancreatic ADC showed the best test-retest repeatability (mean difference ± limits of agreement were 0.05 ± 0.25×10 -3 mm 2 difference ± limits of agreement between two tests were 0.03 ± 0.47, 0.05 ± 0.42 and 0.06 ± 0.31 (×10 -3 mm 2 /s), respectively (ICCs, 0.81, 0.52 and 0.68; CVs, 9%, 8% and 8%). Both intra-observer repeatability and inter-observer reproducibility were acceptable for whole pancreas between ADC measurements of the two DWIs. Conclusions: ADC values of the pancreas are technically challenging and repeatability is variable. Cautions should be taken when interpreting longitudinal clinical changes in ADC values of the normal pancreas because the measurements do have an inherent variability by locations.

difference ± limits of agreement between two tests were 0.03 ± 0.47, 0.05 ± 0.42 and 0.06 ± 0.31 (×10 -3 mm 2 /s), respectively (ICCs, 0.81, 0.52 and 0.68; CVs, 9%, 8% and 8%). Both intra-observer repeatability and inter-observer reproducibility were acceptable for whole pancreas between ADC measurements of the two DWIs. MRI is widely used to detect and to differentiate pancreatic diseases [1]. Specifically, as a quantitative MRI technology, diffusion weighted imaging (DWI) with derived apparent diffusion coefficient (ADC) was introduced to quantify water diffusion in vivo [2]. With a quasi-exponential growth of research applications as well as clinical practice, DWI provides an additional information on the pancreas as a supplement to conventional techniques such as T1/T2-weighted imaging [3].
As a biomarker, ADC is helpful in the characterization of pancreatic diseases including chronic pancreatitis, cystic and solid pancreatic tumors [2]. Some previous studies have reported that pancreatic cancer had lower mean ADC value than the normal pancreas [4][5][6][7][8][9][10][11]. Ideally, changes in ADC value should principally reflect the composition and/or cellularity of tissue, whilst at the same time not being susceptible to variations associated with measurement repeatability. Therefore, in order to detect meaningful difference in ADC, it is desirable that the uncertainty of the ADC measurement should be lower than the difference between the normal and abnormal pancreatic tissues.
It is actually an important point to assess ADC repeatability in body DWI [2]. Data on evaluating the repeatability of ADC for normal pancreas are limited: one study only reported the mean coefficient of variations (CV) of 10.6% for ADC in the normal pancreatic body at 3.0-T [12]. As most of the studies have used normal pancreas as control group when comparing ADC in different pancreatic entities, the ADC test-retest repeatability in the normal pancreas was few reported, especially for different anatomical regions of the pancreas. To better to understand the magnitude in change of ADC and obtain useful metrics for ADC, the aim of this study was to investigate the ADC test-retest repeatability in the normal pancreas.

Clinical Study
This prospective study was approved by our ethics review board. Before MRI examination, informed consents were signed by the volunteers. Twenty-six volunteers (mean age 47.6±10.5 years; range 24-67 years; 13 men, 13 women) were enrolled in the study at January 6 and 7, 2018. Exclusion criteria were set as subjects with pancreatic disease, diabetes, hepatic cirrhosis history and any contraindication of MRI examination.

Data analysis
Two radiologists issued the diagnostic reports of MRI examination for all volunteer and did not find any abnormalities of the pancreas. The head, body and tail regions of the pancreas [13] were clearly displayed on the DW images in all subjects.
To evaluate the variability, two observers (both with over 8 years experience in radiology) measured the ADCs of pancratic head, body and tail for the two DWIs. One of the observers (as observer 1) measured ADCs again with an interval of four weeks for the two DWIs. Mean ADC values were obtained from an oval or round region of interest (ROI, Figure 1) on the ADC maps. The area of ROI was between 31-119 mm 2 (mean 71.6 mm 2 ). Vessels, ducts and common bile duct were avoided with reference to T2WI/T1WI in the measurements of ADC. The mean ADC values of the pancreatic head, body and tail was calculated as the global pancreatic ADC.

Statistical analysis
Statistical analyses were performed using MedCalc, version Intra-/inter-observer variability and test-retest repeatability of ADC measurements for each anatomical region of the pancreas and whole pancreas were analyzed by Bland-Altman analysis [14], CV

Results
The typical images of test-retest DWIs and ADC measurements were demonstrated in Figure 1. The mean ADC values of different regions and the whole pancreas for the two DWIs were summarized and shown in Table 1. were also found in the reposition DWI.

Test-retest repeatability
For the test-retest DWIs, the bias and LOAs between two ADC There were no significant differences in ADCs between two DWIs at head, body or whole pancreas. Figure 4 showed the Bland-Altman analyses results.

Discussion
Our study demonstrated that ADCs of the normal pancreas and test-retest repeatability were dependent on the different anatomical regions of pancreas. Only the repeatability of mean ADC of whole pancreas were acceptable, because the test-retest bias of ADC measurements less than ± 0.10×10 -3 mm 2 /s and the LOAs were less than ± 0.30×10 -3 mm 2 /s [16]. As a biomarker, the evaluation of repeatability in ADC measurements is important to diagnose diseases or to detect meaningful change in treatment. ADC repeatability has already been evaluated over time for breast [17], lung [18,19] and liver [20,21], but only few studies have evaluated the test-retest ADC repeatability in the pancreas. Rosenkrantz et al found that test-retest ADC repeatability was moderate in normal pancreatic body both at 1.5-T and 3.0-T [12]. However, the test-retest ADC repeatability was not acceptable to three anatomical regions of pancreas in our study. The inconsistent findings between two studies may be caused by the parameters differences in DWI, specifically in b values. In our study, two b values were used for DWI, and the choosing of the b values is according to the recommendation of ISMRM-Sponsored workshop [22]. Other two studies [23,24] reported that there was no significant effect of MRI systems on ADCs reproducibility of normal pancreas (including head, body and tail) over a short-term of normal pancreas at 1.5-T and 3.0-T for a given individual, respectively. Barral, et al [25] investigated the reproducibility and variations in ADCs for normal pancreas using repeated measurements method both at 1.5-and 3.0-Tesla, and found that intra-and inter-observer ADC measurements were acceptable and ADCs in different segments of pancreas were homogeneous at 3.0-Tesla. In the current study, we also evaluated the intra-/inter-observer variability of ADC measurements of there regions of normal pancreas at 3.0-T, as well as the whole pancreas on ADC measurements for test-retest DWIs.
However, the intra-observer repeatability of mean ADCs in pancreatic body was not acceptable. We also found that the ADC values were the lowest in the pancreatic tail, which was similar as being showed in some studies [26,27]. In addition to the features of tissues, DWI acquisition parameters, including field strength, b-values selection, respiratory compensation acquisition, and post-processing approach may influence the ADCs of pancreas [28].
The inconsistent findings in our study might be the differences in b-values and field strength. Additionally, the free-breathing technique used in our study might have some effects on the findings. The finding of a significantly lower ADC in the pancreatic tail in the second ADC measurement compared to the first one may be due to the effects of these factors. To limit the possible influence of field heterogeneity, b values and acquisition methods on the ADC measurements, a normalized ADC method is a promising tool [29][30][31]. An average value of 3 delineations will be more robust than the three small delineations. -Availability of data and materials Please contact author for data requests.

-Competing interests
The authors declare that they have no competing interests.   Y-axis: bias of ADC measurements; x-axis: the mean ADCs.