Magnetic Resonance Imaging 34 (2015) 144–151

Contents lists available at ScienceDirect

Magnetic Resonance Imaging journal homepage: www.mrijournal.com

Gastric carcinoma: Evaluation with diffusion-tensor MR imaging and tractography ex vivo☆ Ichiro Yamada a,⁎, Keigo Hikishima b, c, Naoyuki Miyasaka d, Keiji Kato e, Kazuyuki Kojima e, Tatsuyuki Kawano f, Eisaku Ito g, Daisuke Kobayashi g, Yoshinobu Eishi g, Hideyuki Okano b a

Department of Diagnostic Radiology and Nuclear Medicine, Graduate School, Tokyo Medical and Dental University, Tokyo, Japan Department of Physiology, Keio University School of Medicine, Tokyo, Japan Central Institute for Experimental Animals, Kanagawa, Japan d Department of Pediatrics, Perinatal and Maternal Medicine, Tokyo Medical and Dental University, Tokyo, Japan e Department of Gastric Surgery, Tokyo Medical and Dental University, Tokyo, Japan f Department of Esophageal Surgery, Tokyo Medical and Dental University, Tokyo, Japan g Department of Pathology, Graduate School, Tokyo Medical and Dental University, Tokyo, Japan b c

a r t i c l e

i n f o

Article history: Received 29 January 2015 Revised 20 August 2015 Accepted 17 October 2015 Available online xxxx Keywords: Stomach Gastric carcinoma Diffusion-tensor imaging Diffusion-tensor tractography Diffusion-weighted imaging MR imaging

a b s t r a c t Purpose: To determine the feasibility of diffusion-tensor imaging (DTI) and tractography as means of evaluating the depth of mural invasion by gastric carcinomas. Materials and methods: This study was approved by our institutional review board, and written informed consent was obtained from each patient. Twenty gastric specimens containing a carcinoma were studied with a 7.0-T MR imaging system equipped with a four-channel phased-array surface coil. DTI was performed by using a field of view of 50–60 mm × 25–30 mm, matrix of 256 × 128, section thickness of 1 mm, b value of 1000 s/mm 2, and motion-probing gradients in seven noncollinear directions. The MR images were compared with the histopathologic findings as the reference standard. Results: In all 20 carcinomas (100%) the diffusion-weighted images, apparent diffusion coefficient (ADC) maps, fractional anisotropy (FA) maps, λ1 maps, and direction-encoded color FA maps made it possible to identify the same depth of tumor invasion of the gastric wall as observed during histopathologic examination. The λ1 maps provided the best contrast between the carcinomas and the layers of the gastric wall. The carcinomas also had lower ADC values and lower FA values than the normal gastric wall; thus, the carcinomas were clearly demarcated from the normal gastric wall. Tractography images were also useful for determining the depth of tumor invasion of the gastric wall. Conclusions: DTI and tractography are feasible means of evaluating gastric specimens and provide excellent diagnostic accuracy for evaluating mural invasion by gastric carcinomas. © 2015 Elsevier Inc. All rights reserved.

1. Introduction The prognosis of gastric carcinoma patients strictly depends on the histopathologic stage of the carcinoma, and accurate preoperative staging has a definitive impact on selection of the optimal therapy for gastric carcinoma [1,2]. Local staging of gastric carcinoma is currently performed on the basis of the computed tomography (CT) and endoscopic ultrasound (EUS) findings. While the advent of multidetector CT has contributed to the recently reported improved ☆ Conflicts of Interest and Sources of Funding: For all the authors listed, no conflicts are declared. Dr. Yamada has received the Grant-in-Aid for Scientific Research (C) of MEXT, Japan (23591753). ⁎ Corresponding author at: Department of Diagnostic Radiology and Nuclear Medicine, Graduate School, Tokyo Medical and Dental University, 1-5-45 Yushima, Bunkyo-ku, Tokyo 113-8519, Japan. Tel.: +81 3 5803 5310; fax: +81 3 5803 0147. E-mail address: [email protected] (I. Yamada). http://dx.doi.org/10.1016/j.mri.2015.10.007 0730-725X/© 2015 Elsevier Inc. All rights reserved.

accuracy of CT as a means of local staging of gastric carcinoma [3–5], it is still difficult to accurately determine the depth of invasion of the gastric wall by means of CT, because the poor soft-tissue contrast makes it impossible to resolve the layers of the gastric wall even with this new technology. EUS also entails many inherent problems, including technical failures in stenotic tumors and artifactual interface echoes in the gastric wall [4–6]. EUS is also highly operator-dependent, and there is a recognized learning curve [7]. The high-frequency probe used to achieve high spatial resolution has a rather limited sonographic range, and it is difficult to visualize the relationship between the stomach and the surrounding tissues by EUS [4–7]. Magnetic resonance (MR) imaging has been reported to be capable of visualizing mural invasion by gastric carcinomas and is recognized as an alternative to CT and EUS [8,9], but conventional MR imaging is still incapable of resolving the individual layers of the

I. Yamada et al. / Magnetic Resonance Imaging 34 (2015) 144–151

gastric wall [10–18]. Yamada et al. [19,20], however, have recently demonstrated that diffusion-tensor imaging (DTI) and tractography are useful for resolving the individual layers of the esophageal wall and evaluating the depth of invasion of esophageal carcinomas. To the best of our knowledge, however, there have been no reports of using DTI and tractography to evaluate the depth of mural invasion by gastric carcinomas. The purpose of this study was to prospectively examine surgical specimens of stomachs that contained a gastric carcinoma by DTI and tractography, and to assess the diagnostic accuracy of DTI and tractography as means of evaluating the depth of mural invasion by gastric carcinomas. 2. Materials and methods 2.1. Study population This study was approved by our institutional review board, and written informed consent in regard to the purpose of this study and the use of clinical and histopathologic data was obtained from each patient. We studied 20 surgical specimens of the stomach, each of which contained a tumor that was histopathologically diagnosed as adenocarcinoma. The specimens were obtained from 20 consecutive gastric carcinoma patients who were surgically treated at our institution between August 2013 and February 2014. Sixteen of the 20 patients were male, and four were female. Patient ages at the time of surgery ranged from 50 years to 86 years (mean age: 68 years ± 11 [standard deviation]). The specimens of these patients were included in our previous work on high-spatial-resolution MR imaging [21]. All 20 specimens were imaged after fixation in 10% formalin. We did not examine stomachs in vivo in this series. 2.2. Imaging technique DTI was performed using a 7.0-T MR imaging unit (BioSpec 70/ 16; Bruker BioSpin, Ettlingen, Germany) equipped with actively shielded gradients that had a maximum strength of 700 mT/m. A four-channel phased-array surface coil was used to make all measurements. The orientation of DTI was set longitudinally along the long axis of the resected segment of the stomach. DTI data sets were acquired by using a diffusion-weighted spinecho pulse sequence based on a Stejskal–Tanner diffusion preparation. The imaging parameters were: repetition time, 3000 ms; echo time, 25 ms; field of view, 50–60 × 25–30 mm; matrix, 256 × 128; section thickness, 1 mm without intersection gaps; voxel size, 0.195–0.234 × 0.195–0.234 × 1 mm (0.038–0.055 mm 3); number of excitations, two; b value, 0 s/mm 2 (for a reference b0 image without diffusion weighting) or 1000 s/mm 2 (gradient lobe separation = 14 ms); and motion-probing gradients (MPG), in seven noncollinear directions. The acquisition time was 102 min. We used only seven noncollinear gradient directions because we performed the DTI by using the spin-echo sequence that is very time-consuming. We thought that echo-planar imaging (EPI) might cause considerable distortion artifacts, although it allows much more gradient directions in a limited time. The seven gradient directions are more than the six gradient directions that are at least required for DTI. Thus we performed the seven-direction DTI in this study. We used the spin-echo sequence for DTI in this study because it is more resistant to susceptibility artifacts than the EPI. However, EPI acquisition may be necessary for in-vivo DTI of gastric carcinoma because it enables high-speed DTI acquisition. The b value of 1000 s/mm 2 is currently a de facto standard for clinical DTI in all of the MR imaging machine vendors. Furthermore, we chose the same b value of 1000 s/mm 2 as in our previous studies

145

[19,20] because it had been found to be useful for evaluating specimens of esophageal carcinoma. Thus we decided to use the b value of 1000 s/mm 2 in this study. We performed DTI alone in this study because our purpose was to compare the DTI findings with the histopathologic findings, and thus no other MR images (such as T1-weighted images and T2-weighted images) were acquired. 2.3. Image processing The DTI data were processed with Diffusion Toolkit software (Massachusetts General Hospital, MA, USA) [22] using monoexponential fitting. The three eigenvalues (λ1 ≥ λ2 ≥ λ3) and corresponding eigenvectors were calculated for each voxel. The following formulas were used to calculate the apparent diffusion coefficient (ADC) and fractional anisotropy (FA) for each voxel from the eigenvalues: ADC ¼ ðλ1 þ λ2 þ λ3 Þ=3 ¼ hDi 1=2

FA ¼ ð3=2Þ

n

2

2

2

ðλ1 −hDiÞ þ ðλ2 −hDiÞ þ ðλ3 −hDiÞ

o1=2   2 2 2 1=2 = λ1 þ λ2 þ λ3

where bD N is mean diffusivity. Isotropic diffusion-weighted images (DWI), ADC maps, FA maps, λ1 maps, λ2 maps, and λ3 maps were generated on a voxel-by-voxel basis. We also generated direction-encoded color FA maps that showed anisotropy in different colors according to the direction of the major axis. Color FA maps make it possible to differentiate fiber tracts based on their orientation. Principal colors (green, blue, and red) were assigned to the three orthogonal orientations with reference to the gastric wall: green, indicating orientation along the long axis of the stomach; blue, indicating orientation along the short axis parallel to the gastric wall; and red, indicating orientation along the short axis perpendicular to the gastric wall. In the present study, we used a spin-echo sequence, not an EPI sequence, so we thought that there was no eddy current caused by high-speed switching. Concerning the eddy current caused by the MPG, we confirmed that the shift from the b0 image was 0.128 mm at most in the in-plane resolution of 0.195 mm, and thus it was less than two thirds of a voxel. Based on these reasons, we performed the DTI data processing without image coregistration in this study. Tractography images were computed with TrackVis software (Massachusetts General Hospital, MA, USA) [22]. The eigenvector (e1) associated with the largest eigenvalue (λ1) was assumed to represent the local coherent fiber orientation. We performed a deterministic tractography based on the Fiber Assignment by Continuous Tracking (FACT) algorithm [23] by tracking the principal eigenvector (e1) in each voxel. The angle of deflection threshold between contiguous voxels was set at 35°. However, to minimize false-negative fiber tracking, we did not set an FA threshold to track fibers [24]. The software performed fiber tracking by reading in all of the track data of the gastric specimen, and it displayed all fiber tracts contained in the specimen. Then we used the two regions of interest (two-ROI) method to selectively display specific fiber tracts [25]. We set two regions of interest on opposite sides of the specific fiber tracts in the specimen. For the fiber tracts oriented along the long axis and short axis of the stomach, the size of the regions of interest was approximately equal to the thickness of each layer multiplied by the width and length of the displayed tractography image, respectively (Fig. 1F). 2.4. Image analysis An independent, blinded evaluation of DTI in each surgical specimen was performed by two observers (I.Y., N.M., 23 and 19

146

I. Yamada et al. / Magnetic Resonance Imaging 34 (2015) 144–151

Fig. 1. Images of the normal gastric wall. (A) DWI (3000/25) depicts the normal gastric wall as consisting of the following seven layers: epithelium and lamina propria mucosae (Epi-LPM; high SI), muscularis mucosae (MM; high SI), submucosa (SM; low SI), inner oblique muscle or middle circular muscle (IOM-MCM; high SI), intermuscular connective tissue (IMCT; low SI), outer longitudinal muscle (OLM; high SI), and subserosa and serosa (SS; low SI). (B) ADC map (3000/25) depicts the following seven layers: epithelium and lamina propria mucosae (low ADC), muscularis mucosae (low ADC), submucosa (high ADC), inner oblique muscle or middle circular muscle (low ADC), intermuscular connective tissue (high ADC), outer longitudinal muscle (low ADC), and subserosa and serosa (high ADC). (C) FA map (3000/25) depicts the following seven layers: epithelium and lamina propria mucosae (low FA), muscularis mucosae (high FA), submucosa (low FA), inner oblique muscle or middle circular muscle (high FA), intermuscular connective tissue (low FA), outer longitudinal muscle (high FA), and subserosa and serosa (low FA). (D) λ1 map (3000/25) depicts the following seven layers: epithelium and lamina propria mucosae (intermediate λ1), muscularis mucosae (intermediate λ1), submucosa (high λ1), inner oblique muscle or middle circular muscle (intermediate λ1), intermuscular connective tissue (high λ1), outer longitudinal muscle (intermediate λ1), and subserosa and serosa (high λ1). (E) Direction-encoded color FA map (3000/25) depicts the following seven layers: epithelium and lamina propria mucosae (variable color), muscularis mucosae (green), submucosa (black), inner oblique muscle or middle circular muscle (blue), intermuscular connective tissue (black), outer longitudinal muscle (green), and subserosa and serosa (black). Green indicates the long-axis direction of the stomach; blue, the short-axis direction parallel to the gastric wall; and red, the short-axis direction perpendicular to the gastric wall. (F) Direction-encoded color FA map (3000/25) showing the ROIs used for tractography of the muscularis mucosae (yellow rectangles) and the outer longitudinal muscle (orange rectangles). In this longitudinal image, however, we can not show the ROIs used for tractography of the inner oblique muscle or middle circular muscle. (G) Histologic section of the normal gastric wall shows the epithelium and lamina propria mucosae, muscularis mucosae, submucosa, inner oblique muscle or middle circular muscle, intermuscular connective tissue, outer longitudinal muscle, and subserosa and serosa. (Hematoxylin–eosin stain; original magnification, ×20.)

I. Yamada et al. / Magnetic Resonance Imaging 34 (2015) 144–151

years experience in reading MR images, respectively) who had no knowledge of the results of the histopathologic examination. When the observers could not fully agree on the findings, a consensus was achieved by means of discussion. Thus we could not provide interrater reliability test results in this study. However, this disagreement occurred in one (5%) of the 20 specimens in the present study. DWI, ADC maps, FA maps, λ1 maps, and direction-encoded color FA maps were reviewed for the presence, signal intensity (SI), uniformity, and thickness of each layer of the gastric wall. In addition, the contour and SI of the carcinoma were also analyzed. The degree of carcinoma penetration into the gastric wall was categorized according to the deepest layer that was invaded: mucosa, submucosa, muscularis propria, or subserosa and serosa. The MR imaging criteria used by the observers to determine the depth of carcinoma invasion were: (a) discrete mass(es) in the layers, (b) abnormal SI within the thickened layers, (c) focal abnormal SI within the layers, and (d) mucosal ulceration with surrounding or underlying mass(es) [8]. For the ADC maps, FA maps, and λ1 maps, regions of interest were placed on the carcinoma and on the following layers of the gastric wall: epithelium and lamina propria mucosae, muscularis mucosae, submucosa, inner oblique muscle or middle circular muscle, intermuscular connective tissue, outer longitudinal muscle, and subserosa and serosa. The size of the regions of interest was approximately equal to the section area of the carcinoma and to the thickness of each layer, and one of the observers (I.Y.) shaped and placed the regions of interest. The number of the regions of interest was three or four for the carcinoma and for each layer per specimen, and the mean value of the three or four regions of interest in the carcinoma and in each layer of the gastric wall was calculated. All quantitative measurements of DTI reported were made by analyzing the regions of interest using ImageJ 1.47 software (available at http://imagej.nih.gov/ij). The TrackVis software was used to superimpose tractography images on DWI, ADC maps, FA maps, and λ1 maps in order to identify the layers of the gastric wall. Fiber tracts on the tractography images were evaluated for the presence, orientation, color, continuity, uniformity, and thickness of each layer of the gastric wall. The DTI findings in the 20 gastric specimens were compared with the histopathologic findings, which served as the reference standard. The MR images were compared with specific histopathologic sections on a slice-by-slice level, and correlations were made by visual inspection. Concerning matching the MR images and the specimens, spatial correlation was achieved by identifying anatomic landmarks (eg, gastric contour, blood vessels) that were depicted. 2.5. Histologic preparations and examination After MR imaging, each surgical specimen was sectioned longitudinally so that the orientation of the sections corresponded to the orientation of the MR images. The sectioned specimens were embedded in paraffin and cut into 6-μm-thick slices with a microtome. These slices were then stained with hematoxylin–eosin (H-E) stain, elastica–van Gieson (EVG) stain, and periodic acid–Schiff (PAS) stain. An experienced pathologist (E.I., 15 years experience in histopathology) who did not have any knowledge of the MR imaging findings identified carcinoma invasion into each layer of the gastric wall. 2.6. Statistical analysis Means ± standard deviations of the ADC values, FA values, and λ1 values in the carcinoma and in each layer of the gastric wall were calculated. All statistical analyses were performed with a commercial software package (IBM SPSS Statistics, version 20; IBM SPSS Japan, Tokyo, Japan). The differences in DTI parameters between the

147

Table 1 Findings on DWI, ADC maps, FA maps, λ1 maps, and color FA maps of the gastric carcinomas and the layers of the gastric wall at DTI. Tissue

DWI

ADC Maps

FA Maps

λ1 Maps

Color FA Maps

Carcinoma Epithelium-LPM MM Submucosa IOM-MCM IMCT OLM Subserosa-Serosa

High High High Low High Low High Low

Low Low Low High Low High Low High

Low Low High Low High Low High Low

Low Intermediate Intermediate High Intermediate High Intermediate High

Variable Variable Green Black Blue Black Green Black

Note. LPM = Lamina propria mucosae, MM = Muscularis mucosae, IOM = Inner oblique muscle, MCM = Middle circular muscle, IMCT = Intermuscular connective tissue, OLM = Outer longitudinal muscle. Green indicates the long-axis direction of the stomach; blue, the short-axis direction parallel to the gastric wall; red, the shortaxis direction perpendicular to the gastric wall; and black, low anisotropy values.

carcinoma and the layers of the gastric wall were statistically analyzed by the Dunnett test. A P value less than 0.05 was considered evidence of a statistically significant difference. 3. Results 3.1. DWI, ADC maps, FA maps, λ1 maps, and direction-encoded color FA maps of the gastric carcinomas and the layers of the gastric wall The DWI, ADC maps, FA maps, λ1 maps, and direction-encoded color FA maps of all 20 specimens (100%) clearly depicted the gastric wall as consisting of the following seven layers: epithelium and lamina propria mucosae, muscularis mucosae, submucosa, inner oblique muscle or middle circular muscle, intermuscular connective tissue, outer longitudinal muscle, and subserosa and serosa (Table 1, Fig. 1A–E). These seven layers clearly corresponded to the layers of the gastric wall observed in histologic sections (Fig. 1G). As shown in Table 1, the gastric carcinomas were seen as high SI on DWI, as low values on ADC maps, FA maps, and λ1 maps, and as variable colors on direction-encoded color FA maps. The carcinomas on the DWI appeared as areas of higher SI than the submucosa, intermuscular connective tissue, and subserosa and serosa. On the ADC maps the carcinomas had lower ADC values than the submucosa, intermuscular connective tissue, and subserosa and serosa. On the FA maps the carcinomas also had lower FA values than the muscularis mucosae, inner oblique muscle or middle circular muscle, and outer longitudinal muscle. On the λ1 maps, however, the carcinomas had lower λ1 values than all of the layers of the gastric wall, and thus the λ1 maps provided the best contrast between the carcinoma and the layers of the gastric wall. The reason they provided the best contrast was that the submucosa, intermuscular connective tissue, and subserosa and serosa had high λ1 values, and the epithelium and lamina propria mucosae, muscularis mucosae, inner oblique muscle or middle circular muscle, and outer longitudinal muscle had intermediate λ1 values (Table 1). 3.2. ADC values, FA values, and λ1 values of the gastric carcinomas and the layers of the gastric wall As shown in Table 2, the ADC values of the gastric carcinomas were significantly lower than the ADC values of the submucosa, intermuscular connective tissue, and subserosa and serosa (P b 0.001), and were not significantly different from the ADC values of the epithelium and lamina propria mucosae, muscularis mucosae, inner oblique muscle or middle circular muscle, and outer longitudinal muscle. The FA values of the carcinomas, on the other hand,

148

I. Yamada et al. / Magnetic Resonance Imaging 34 (2015) 144–151

were significantly lower than the FA values of the epithelium and lamina propria mucosae, muscularis mucosae, inner oblique muscle or middle circular muscle, and outer longitudinal muscle (P b 0.001, except the epithelium and lamina propria mucosae [P = 0.015]), and not significantly different from the FA values of the submucosa, intermuscular connective tissue, and subserosa and serosa. The λ1 values of the carcinomas, however, were significantly lower than the λ1 values of all of the layers of the gastric wall (P b 0.001, except the epithelium and lamina propria mucosae [P = 0.016] and the muscularis mucosae [P = 0.029]). These findings explain why the λ1 maps provided the best contrast between the carcinomas and the layers of the gastric wall. 3.3. Evaluation of gastric carcinoma invasion on DTI Histopathologic examination of the 20 gastric carcinomas showed that three carcinomas were confined to the mucosa (stage T1a), ten carcinomas had invaded the submucosa (stage T1b), three carcinomas had invaded the muscularis propria (stage T2), and four carcinomas had invaded the subserosa and serosa (stage T3 or T4). The depth of invasion of the gastric wall by all 20 carcinomas (100%) was clearly demonstrated by the DWI, ADC maps, FA maps, λ1 maps, and direction-encoded color FA maps. Since the λ1 maps provided the best contrast between the carcinomas and the layers of the gastric wall, the λ1 maps were the most useful means of assessing tumor invasion of the gastric wall. There were no false-positive or false-negative cases with respect to the assessments of invasion of the different gastric layers in this study. However, since there was disagreement in one case (5%) that was resolved by means of consensus, it may be unlikely that the technique would actually have perfect diagnostic performance. Carcinomas confined to the mucosa appeared as discrete thickenings of the mucosa that had lower λ1 values than other parts of the gastric wall. Carcinomas that had invaded the submucosa appeared as irregular masses with low λ1 values that had made the muscularis mucosae discontinuous (Fig. 2A and B). Carcinomas that had involved the muscularis propria appeared as low-λ1 masses that had partially replaced the muscularis propria but had not penetrated through the muscularis propria (Fig. 3A and B). Carcinomas that had extended into the subserosa and serosa appeared as low-λ1 masses that had completely made the muscularis propria discontinuous and invaded the subserosa and serosa (Fig. 4A and B). 3.4. Evaluation of gastric carcinoma invasion on tractography images The three-dimensional tractography images of all 20 specimens (100%) clearly depicted the following four layers: epithelium and

Table 2 ADC values, FA values, and λ1 values of the gastric carcinomas and the layers of the gastric wall at DTI. Tissue

ADC Values (×10−3 mm2/s)

FA Values

Carcinoma Epithelium-LPM MM Submucosa IOM-MCM IMCT OLM Subserosa-Serosa

0.51 0.61 0.59 1.46 0.64 1.16 0.63 1.66

0.20 0.25 0.54 0.15 0.75 0.19 0.70 0.24

± 0.08 ± 0.03 ± 0.03 ± 0.24 ⁎ ± 0.03 ± 0.18 ⁎ ± 0.04 ± 0.26 ⁎

± 0.03 ± 0.02 ± 0.05 ± 0.03 ± 0.04 ± 0.03 ± 0.08 ± 0.04

λ1 Values (×10−3 mm2/s) ⁎ ⁎ ⁎ ⁎

0.58 0.78 0.76 1.88 1.53 1.60 1.55 2.41

± 0.04 ± 0.03 ± 0.04 ± 0.22 ± 0.13 ± 0.29 ± 0.10 ± 0.23

⁎ ⁎ ⁎ ⁎ ⁎ ⁎ ⁎

Note. LPM = Lamina propria mucosae, MM = Muscularis mucosae, IOM = Inner oblique muscle, MCM = Middle circular muscle, IMCT = Intermuscular connective tissue, OLM = Outer longitudinal muscle. ⁎ Significantly different from the corresponding value of the gastric carcinoma.

lamina propria mucosae (as variable-color fiber tracts), muscularis mucosae (as green fiber tracts), inner oblique muscle or middle circular muscle (as blue fiber tracts), and outer longitudinal muscle (as green fiber tracts) (Figs. 2C, 3C, and 4C). The tractography images were also found to be useful for determining the depth of tumor invasion of the gastric wall. When the carcinoma was confined to the mucosa, all four of the above layers appeared to be intact. When the carcinoma had invaded the submucosa, the muscularis mucosae (green fiber tracts) was observed to have been made discontinuous by the tumor (Fig. 2C). When the carcinoma had involved the muscularis propria, the muscularis propria (blue fiber tracts and/or green fiber tracts) was observed to have been partially replaced by the tumor (Fig. 3C). When the carcinoma had extended into the subserosa and serosa, the muscularis propria (both blue fiber tracts and green fiber tracts) was observed to have been completely made discontinuous by the tumor (Fig. 4C). 4. Discussion The DWI, ADC maps, FA maps, λ1 maps, and direction-encoded color FA maps generated from the DTI data set obtained in all 20 specimens (100%) in the present study clearly depicted the normal gastric wall as consisting of seven layers, and the seven layers matched the tissue layers of the gastric wall observed histologically, thereby clearly demonstrating that DTI is capable of depicting the tissue layers of the gastric wall more accurately than other currently available imaging modalities [3–7]. Yamada et al. [19] have proposed a new schematic illustration of the normal esophageal wall to explain the DTI and tractography findings of the normal esophageal wall. Based on their schematic illustration, alternating levels of cellularity (cell density), anisotropy, and directionality between the adjacent layers of the normal gastric wall may also account for their clear depiction on the ADC maps, FA maps, and direction-encoded color FA maps, respectively. The gastric carcinomas on the λ1 maps appeared darker than all of the layers of the gastric wall, and thus the λ1 maps provided the best contrast between the carcinomas and the layers of the gastric wall. The λ1 values of the gastric carcinomas were found to be statistically significantly lower than the λ1 values of all of the layers of the gastric wall, and the fact that the gastric carcinomas had the lowest λ1 values therefore explains why the λ1 maps provided the best contrast between the carcinomas and the layers of the gastric wall. Yamada et al. [19,20] have reported finding that the λ1 values of the solid high-cellularity layers were considerably higher than the ADC values of the solid high-cellularity layers because of local coherent fiber orientation, while the λ1 values of the carcinomas and the loose low-cellularity layers were only slightly higher than their respective ADC values because of the absence of local coherent fiber orientation. This means that the diffusion ellipsoids in the voxels of the solid high-cellularity layers are markedly long and narrow in shape, and thus their λ1 values (axial diffusivity) are considerably larger than their ADC values (mean diffusivity). These differences are likely to have been responsible for the gastric carcinomas having the lowest λ1 values, thereby accounting for the fact that the λ1 maps provided the best contrast between the carcinomas and the layers of the gastric wall. In general, the cellularity (cell density) does not correlate with axial diffusivity (λ1 values), and the ADC values and FA values reflect the size and shape of the diffusion ellipsoids, respectively. In the gastric wall, the solid high-cellularity layers include smooth muscle fibers that are arranged in coherent orientation, and thus they exhibit low ADC values, high FA values, and intermediate λ1 values. The loose low-cellularity layers include loose connective tissue that does not contain coherent fiber structure, and thus they exhibit high ADC values, low FA values, and high λ1 values. In this study, the

I. Yamada et al. / Magnetic Resonance Imaging 34 (2015) 144–151

149

Fig. 2. Gastric carcinoma that has invaded the submucosa. (A) λ1 map (3000/25) clearly shows a mass lesion (arrows) that has lower λ1 values than other parts of the gastric wall. (B) FA map (3000/25) shows that the mass lesion has made the muscularis mucosae discontinuous (arrows). (C) Tractography image (3000/25) shows that the muscularis mucosae (green fiber tracts) has been made discontinuous by the tumor (arrows). Green indicates the long-axis direction of the stomach; blue, the short-axis direction parallel to the gastric wall; and red, the short-axis direction perpendicular to the gastric wall. (D) Corresponding histopathologic section shows carcinoma that has invaded the submucosa, which has made the muscularis mucosae discontinuous (arrows). (Hematoxylin–eosin stain; original magnification, ×10.)

Fig. 3. Gastric carcinoma that has involved the muscularis propria. (A) λ1 map (3000/25) clearly shows a mass lesion (arrows) that has lower λ1 values than other parts of the gastric wall. (B) FA map (3000/25) shows that the mass lesion has partially replaced the muscularis propria (arrows). (C) Tractography image (3000/25) shows that the muscularis propria (blue fiber tracts) has been partially replaced by the tumor (arrows). Green indicates the long-axis direction of the stomach; blue, the short-axis direction parallel to the gastric wall; and red, the short-axis direction perpendicular to the gastric wall. (D) Corresponding histopathologic section shows carcinoma that has involved the muscularis propria (arrows). (Hematoxylin–eosin stain; original magnification, ×10.)

150

I. Yamada et al. / Magnetic Resonance Imaging 34 (2015) 144–151

Fig. 4. Gastric carcinoma that has extended into the subserosa and serosa. (A) λ1 map (3000/25) clearly shows an irregular-shaped mass lesion (arrows) with low λ1 values. (B) FA map (3000/25) shows that the mass lesion has completely made the muscularis propria discontinuous and invaded the subserosa and serosa (arrows). (C) Tractography image (3000/25) shows that the muscularis propria (both blue fiber tracts and green fiber tracts) has been completely made discontinuous by the tumor (arrows). Green indicates the long-axis direction of the stomach; blue, the short-axis direction parallel to the gastric wall; and red, the short-axis direction perpendicular to the gastric wall. (D) Corresponding histopathologic section shows carcinoma that has extended into the subserosa and serosa (arrows). (Hematoxylin-eosin stain; original magnification, ×1.)

gastric carcinomas were found to have low ADC values, low FA values, and low λ1 values, the combination of which findings was different from all of the layers of the gastric wall. Our findings demonstrated that DTI makes it possible to determine the depth of invasion by gastric carcinomas by identifying areas in the gastric wall that have both low ADC values and low FA values and/or that have low λ1 values. Since the λ1 maps provided the best contrast between the carcinomas and the layers of the gastric wall, the λ1 maps were found to be the most useful for assessing tumor invasion of the gastric wall. However, the tractography images were also found to be useful for determining the depth of tumor invasion of the gastric wall. Gaige et al. [26] reported using DTI and tractography in vivo to visualize the structural anatomy of the tongue, and their findings suggested that in-vivo DTI and tractography might be feasible in the stomach as well. Since the involuntary gastric motion may impede the clinical application of DTI and tractography to gastric carcinoma, it must be sufficiently controlled before their practice. Conventional MR imaging has recently been used to preoperatively stage gastric carcinoma. Previous studies have reported using multichannel phased-array coils, parallel imaging techniques, and more powerful gradient systems in gastric carcinoma patients [10–16], and Heye et al. [17,18] have developed an endoluminal MR imaging technique that uses endoluminal radiofrequency (RF) coils as a means of staging gastric carcinoma. We therefore think that using these high-speed techniques or endoluminal RF coil techniques would make it possible to perform DTI and tractography in gastric carcinoma patients in vivo. There were several limitations to our study. First, the study was performed ex vivo, and the specimens were imaged after fixation in formalin. However, since previous studies of other organs have shown

that the diffusion anisotropy indices measured in fixed and in-vivo tissues provided essentially the same information regarding the underlying microstructure [27–31], using DTI and tractography in fixed tissues would appear to be valid means of evaluating the diffusion anisotropy of tissues in vivo. The relative anisotropy (RA) may be less affected by tissue fixation, but we used the FA in this study because the FA is a much more important parameter in deterministic tractography than the RA. The FA threshold may affect tractography results substantially, but we did not set the FA threshold in this study to minimize false-negative fiber tracking [24]. The same studies also showed that although the ADC values of fixed tissues were lower than in vivo, the relative ADC difference between different tissue types in vivo was preserved in fixed tissues, and that the difference in relative ADC values between fixed tissues and in-vivo tissues was not statistically significant [29,30,32]. We therefore think that the data obtained in the present study are valid for DTI and tractography of tissues in vivo as well as in formalin-fixed tissues. The second limitation was that the imaging time in this study was considerably long (102 min), and shortening scan time would be necessary to extend our method to in-vivo DTI and tractography. At present, clinical application of whole body 7-T MR imaging to gastric carcinoma may be technically difficult because of motion effects (patient motion, respiratory movements, and peristalsis) and the deep location of the stomach in the abdomen. Modifications of the pulse sequences, development of faster MR imaging techniques, or application of higher field strengths in clinical settings may be required to make its application more technically feasible in the future. The third limitation was that the number of specimens examined in this study was rather limited, leaving reproducibility an open question. To overcome this limitation, we should continue to collect more gastric DTI and tractography data. Our ultimate goal is routine

I. Yamada et al. / Magnetic Resonance Imaging 34 (2015) 144–151

application of DTI and tractography as quantitative tools for accurate preoperative staging of gastric carcinoma, which is decisive in selecting the optimal therapy for gastric carcinoma. In conclusion, the results of the present study have demonstrated that DTI and tractography are able to clearly depict the tissue layers of the gastric wall ex vivo. DTI and tractography ex vivo provide excellent diagnostic accuracy for evaluating mural invasion by gastric carcinomas. DTI and tractography may therefore make it possible to noninvasively diagnose the depth of mural invasion by gastric carcinomas in patients.

[15]

[16] [17]

[18]

[19]

References [1] Ahn HS, Lee HJ, Hahn S, Kim WH, Lee KU, Sano T, et al. Evaluation of the seventh American Joint Committee on Cancer/International Union Against Cancer Classification of gastric adenocarcinoma in comparison with the sixth classification. Cancer 2010;116:5592–8. [2] Edge SB, Byrd DR, Compton CC, Fritz AG, Greene FL, Trotti A, editors. AJCC cancer staging manual. 7th ed. New York: Springer; 2010. p. 117–26. [3] Kumano S, Okada M, Shimono T, Kuwabara M, Yagyu Y, Imaoka I, et al. T-staging of gastric cancer of air-filling multidetector-row CT: comparison with hydromultidetector-row CT. Eur J Radiol 2012;81:2953–60. [4] Feng XY, Wang W, Luo GY, Wu J, Zhou ZW, Li W, et al. Comparison of endoscopic ultrasonography and multislice spiral computed tomography for the preoperative staging of gastric cancer — results of a single institution study of 610 Chinese patients. PLoS One 2013;8:e78846. [5] Wakelin SJ, Deans C, Crofts TJ, Allan PL, Plevris JN, Paterson-Brown S. A comparison of computerised tomography, laparoscopic ultrasound and endoscopic ultrasound in the preoperative staging of oesophago-gastric carcinoma. Eur J Radiol 2002;41:161–7. [6] Yoshinaga S, Oda I, Nonaka S, Kushima R, Saito Y. Endoscopic ultrasound using ultrasound probes for the diagnosis of early esophageal and gastric cancers. World J Gastrointest Endosc 2012;4:218–26. [7] Schlick T, Heintz A, Junginger T. The examiner's learning effect and its influence on the quality of endoscopic ultrasonography in carcinoma of the esophagus and gastric cardia. Surg Endosc 1999;13:894–8. [8] Yamada I, Saito N, Takeshita K, Yoshino N, Tetsumura A, Kumagai J, et al. Early gastric carcinoma: evaluation with high-spatial-resolution MR imaging in vitro. Radiology 2001;220:115–21. [9] Yamada I, Takeshita K, Saito N, Yoshino N, Tetsumura A, Kumagai J, et al. Evaluation of gastric cancer by high-resolution three-dimensional CISS MR imaging in vitro. Clin Imaging 2009;33:354–60. [10] Choi JI, Joo I, Lee JM. State-of-the-art preoperative staging of gastric cancer by MDCT and magnetic resonance imaging. World J Gastroenterol 2014;20:4546–57. [11] Liu S, He J, Guan W, Li Q, Zhang X, Mao H, et al. Preoperative T staging of gastric cancer: comparison of diffusion- and T2-weighted magnetic resonance imaging. J Comput Assist Tomogr 2014;38:544–50. [12] Lei C, Huang L, Wang Y, Huang Y, Huang Y. Comparison of MRI and endoscope ultrasound detection in preoperative T/N staging of gastric cancer. Mol Clin Oncol 2013;1:699–702. [13] Hallinan JT, Venkatesh SK. Gastric carcinoma: imaging diagnosis, staging and assessment of treatment response. Cancer Imaging 2013;13:212–27. [14] Maccioni F, Marcelli G, Al Ansari N, Zippi M, De Marco V, Kagarmanova A, et al. Preoperative T and N staging of gastric cancer: magnetic resonance imaging

[20]

[21]

[22]

[23] [24]

[25]

[26]

[27]

[28]

[29]

[30] [31]

[32]

151

(MRI) versus multi detector computed tomography (MDCT). Clin Ter 2010;161: e57–62. Anzidei M, Napoli A, Zaccagna F, Di Paolo P, Zini C, Cavallo Marincola B, et al. Diagnostic performance of 64-MDCT and 1.5-T MRI with high-resolution sequences in the T staging of gastric cancer: a comparative analysis with histopathology. Radiol Med 2009;114:1065–79. Kwee RM, Kwee TC. Imaging in local staging of gastric cancer: a systematic review. J Clin Oncol 2007;25:2107–16. Heye T, Kuntz C, Dux M, Encke J, Palmowski M, Autschbach F, et al. CT and endoscopic ultrasound in comparison to endoluminal MRI: preliminary results in staging gastric carcinoma. Eur J Radiol 2009;70:336–41. Heye T, Kuntz C, Dux M, Encke J, Palmowski M, Autschbach F, et al. New coil concept for endoluminal MR imaging: initial results in staging of gastric carcinoma in correlation with histopathology. Eur Radiol 2006;16:2401–9. Yamada I, Hikishima K, Miyasaka N, Tokairin Y, Kawano T, Ito E, et al. Diffusiontensor MRI and tractography of the esophageal wall ex vivo. J Magn Reson Imaging 2014;40:567–76. Yamada I, Hikishima K, Miyasaka N, Kawano T, Tokairin Y, Ito E, et al. Esophageal carcinoma: ex vivo evaluation with diffusion-tensor MR imaging and tractography at 7 T. Radiology 2014;272:164–73. Yamada I, Miyasaka N, Hikishima K, Kato K, Kojima K, Kawano T, et al. Gastric carcinoma: ex vivo MR imaging at 7.0 T-correlation with histopathologic findings. Radiology 2015;275:841–8. Wang R, Benner T, Sorensen AG, Wedeen VJ. Diffusion Toolkit: a software package for diffusion imaging data processing and tractography [abstr]. Proceedings of the Fifteenth Meeting of the International Society for Magnetic Resonance in Medicine. Berkeley: Calif: International Society for Magnetic Resonance in Medicine; 2007. p. 3720. Mori S, van Zijl PC. Fiber tracking: principles and strategies — a technical review. NMR Biomed 2002;15:468–80. van der Jagt PK, Dik P, Froeling M, Kwee TC, Nievelstein RA, ten Haken B, et al. Architectural configuration and microstructural properties of the sacral plexus: a diffusion tensor MRI and fiber tractography study. Neuroimage 2012;62:1792–9. Konomi T, Fujiyoshi K, Hikishima K, Komaki Y, Tsuji O, Okano HJ, et al. Conditions for quantitative evaluation of injured spinal cord by in vivo diffusion tensor imaging and tractography: preclinical longitudinal study in common marmosets. Neuroimage 2012;63:1841–53. Gaige TA, Benner T, Wang R, Wedeen VJ, Gilbert RJ. Three dimensional myoarchitecture of the human tongue determined in vivo by diffusion tensor imaging with tractography. J Magn Reson Imaging 2007;26:654–61. Sun SW, Neil JJ, Song SK. Relative indices of water diffusion anisotropy are equivalent in live and formalin-fixed mouse brains. Magn Reson Med 2003;50: 743–8. Sun SW, Neil JJ, Liang HF, He YY, Schmidt RE, Hsu CY, et al. Formalin fixation alters water diffusion coefficient magnitude but not anisotropy in infarcted brain. Magn Reson Med 2005;53:1447–51. Xu J, Humphrey PA, Kibel AS, Snyder AZ, Narra VR, Ackerman JJ, et al. Magnetic resonance diffusion characteristics of histologically defined prostate cancer in humans. Magn Reson Med 2009;61:842–50. Bourne R, Kurniawan N, Cowin G, Sved P, Watson G. 16 T diffusion microimaging of fixed prostate tissue: preliminary findings. Magn Reson Med 2011;66:244–7. Bourne RM, Kurniawan N, Cowin G, Sved P, Watson G. Microscopic diffusion anisotropy in formalin fixed prostate tissue: preliminary findings. Magn Reson Med 2012;68:1943–8. Williams MC, Does MD, Price RR. ADC mapping of whole excised human prostate [abstr]. Proceedings of the Twelfth Meeting of the International Society for Magnetic Resonance in Medicine. Berkeley: Calif: International Society for Magnetic Resonance in Medicine; 2004. p. 2064.