Consistency of Trans-Abdominal and Water-Immersion Ultrasound Images of Diseased Intestinal Segments in Crohn’s Disease

The aim of this study is to clarify whether trans-abdominal ultrasound (TAUS) can reflect actual intestinal conditions in Crohn’s disease (CD) as effectively as water-immersion ultrasound (WIUS) does. This retrospective study enrolled 29 CD patients with 113 intestinal lesions. Five ultrasound (US) parameters (distinct presence/indistinct presence/disappearance of wall stratification in the submucosal and mucosal layers; thickened submucosal layer; irregular mucosal surface; increased fat wrapping around the bowel wall; and fistula signs) that may indicate different states in CD were determined by TAUS and WIUS for the same lesion. Using WIUS as a reference standard, the sensitivity, specificity, and accuracy of TAUS were calculated. The degree of agreement between TAUS and WIUS was evaluated by the kappa coefficient. All US parameters of TAUS had an accuracy >70% (72.6–92.7%). The highest efficacy of TAUS was obtained for fistula signs (sensitivity, specificity, and accuracy values were 63.6%, 96.0%, and 92.7%, respectively). All US parameters between TAUS and WIUS had a definitive (p ≤ 0.001) and moderate-to-substantial consistency (kappa value = 0.446–0.615). The images of TAUS showed substantial similarity to those of WIUS, suggesting that TAUS may function as a substitute to evaluate the actual intestinal conditions of CD.

Taken together, using WIUS as a reference standard, we determined whether TAUS is equivalent to WIUS for diagnosis. This analysis may help assess whether images obtained by TAUS correspond to histological alterations in CD.

Patients
Overall, 113 consecutive intestinal surgery specimens from 29 patients were included in this study. This retrospective study was approved by the Institutional Review Committee and complied with the principles of the Helsinki Declaration. Informed consent was obtained from each participating patient (patient consent statement no. B190200027).
The main inclusion criteria for the present study were as follows: A reliable or definite diagnosis of CD could be obtained, and the diseased intestinal segment was resected in surgery. According to the criteria of the Japanese Research Committee on Inflammatory Bowel Disease [15], a diagnosis of CD was made by gastroenterologists on the basis of history, physical exam, laboratory/radiological studies, and gastrointestinal histology. From December 2006 to December 2017, enterectomy was performed in 396 patients with a definite diagnosis of CD in our hospital.
TAUS was performed within 2 weeks before enterectomy. Of the 396 patients, 123 underwent TAUS within the 2 weeks prior to their operation.
Most patients did not undergo TAUS within two weeks. One reason was that the diagnosis obtained from TAUS two weeks previously already showed the necessity for a resection operation (such as serious stenosis, refractory, and fistula); therefore, there was no need for a re-examination by TAUS before surgery. Another reason was that emergency intestinal resection was performed in some patients because of unexpected intestinal perforation or massive accidental bleeding. A routine preoperative TAUS examination would cause a delay at that critical moment.
WIUS was performed within 30 min after enterectomy. Twenty-nine patients met these conditions (Table 1). Table 1. Baseline characteristics of enrolled patients.

Preoperative TAUS Examinations
The patient was examined in the supine position to help relax the abdominal muscles and reduce the antero-posterior diameter of the abdomen. The scan started with a longitudinal view of the entire intestine, providing a broad overview to assess the extent of the affected bowel. The terminal ileum was clearly identified in all patients by confirming its connection with the cecum, and the ileum was identified as bowel loops in the middle-right lower quadrant. When any abnormality was found in the longitudinal section, the examiners used a cross-section scan to make a specific observation. Aplio XG, XV and Aplio 500 US systems (Canon Medical Systems, Otawara, Japan) equipped with native tissue harmonic grayscale imaging were used. Generally, a convex probe with a frequency of 3.5-6.0 MHz was used for TAUS examination. A more localized and detailed scan of the affected area was performed using a 7.5-12.0 MHz linear probe. To avoid operator discrepancy, TAUS was conducted and lesions were determined by two experienced sonographers (NS and AH) who had more than 5 years of experience in bowel US. For surgery convenience and accuracy, the examiners recorded the location of lesions in the TAUS examination report. TAUS images and videos were stored under the patient's ID record. The examiners were not blinded to the clinical parameters of the patient.

Resection of Intestinal Segments
Patients who underwent intestinal surgery fasted 1 day before resection. At laparotomy, with reference to the results of preoperative endoscopy and imaging examinations, surgeons sequentially palpated all segments of the patient's intestine to identify serious stiffness and stricture segments that should be removed. The excision was performed by a highly experienced surgeon who had been working for 20 years in the IBD field (HK). The resected specimens were incised longitudinally along the antimesenteric border, and the observed macroscopic findings were described in the surgical report. To better determine the anatomical location of the target lesion, the doctor performing WIUS in the next step was present during surgery and photographed the appearance of the intestinal segment before and after resection under the condition of laparotomy.

WIUS Examination
To prevent denaturation of the resected tissue, WIUS examinations were performed on diseased bowel segments within 30 min after surgical resection. Resected segments were pinned in a specimen container filled with cold physiological saline solution before being examined by WIUS.
To ensure that the lesions observed in vivo were the same as those that were surgically resected, we took the following measures in the WIUS examination. For lesions in and around the ileocecum, we measured the distance between the target lesion and the ileocecal valve. Lesions located in the colon, which is far from the ileocecum, were identified according to a comparison of preoperative TAUS findings and surgical photos. If it was not certain whether the surgically resected segment was in the same location as that targeted by US, the sample was excluded from WIUS examination and subsequent analysis. The segments analyzed by WIUS in this study were not necessarily the primary lesions that necessitated surgery. For example, even if surgery was performed for an internal fistula, we included the macroscopically intact areas and/or longitudinal ulcer scars in the resected intestinal specimens.
US scans of each segment were performed using an Aplio XG and Aplio 500 US system with a 5.0-11.0 MHz high-frequency linear probe. The US parameters were set as follows: gain: 85-95; depth: 4-6 cm; dynamic range: 65 dB; and focus position: around the region of interest. Differential tissue harmonic imaging and spatial compound images were used to obtain sufficient quality for evaluation. To avoid operator dependence, WIUS examinations were performed in cooperation with a sonographer with 5 years of experience in IBD ultrasound examination (TS) and a gastroenterologist (RK) who specialized in IBD. The intestinal tract, which was incised and opened on the anti-mesenteric border, was carefully scanned in open view (observed from the mucosal surface) and closed view, Diagnostics 2020, 10, 267 5 of 16 from longitudinal and transverse views of the bowel section, respectively. WIUS images and videos were stored under the patient's ID record.

Baseline Data Processing and Ultrasonic Feature Extraction
Before evaluating the usefulness of TAUS, the two doctors who examined WIUS in the previous step (TS, RK) matched images of the same lesion by carefully reviewing the images and the video of TAUS and WIUS and photos of the appearance of the intestinal segment before and after laparotomy. Then, they identified and recorded the US changes from the images and videos of TAUS and WIUS according to previously published US parameters [14,[16][17][18] and our clinical experience in CD diagnosis, including the following: (1) Distinct presence ( Figure 1, Figure 2d,e on TAUS)/indistinct presence ( Figure 3, Figure 2b,c on WIUS)/disappearance (Figures 4 and 5) of wall stratification in the submucosal and mucosal layer. As we show in Figure 6, the US image allows for the visualization of the five layers of the bowel: the border between the lumen and mucous layer (hyperechogenic), the mucous layer (hypoechogenic), the submucosa (hyperechogenic), the muscular layer (or the muscle membrane proper) (hypoechogenic), and the serosa layer (hyperechogenic) [19]. We used stratification-related US parameters to observe the border line between the mucosal and submucosal layers because discriminating between these two layers is meaningful for the evaluation of CD. We did not judge the border line between the submucosal and muscular layers.
(2) Thickened submucosal layer (Figures 1 and 3). Thickened submucosal layer was defined as submucosal thickness accounting for more than 40% of the total intestinal wall. This US parameter was evaluated using lesions in the "presence of wall stratification in the submucosal and mucosal layers" in the first parameter, but not lesions with a "disappearance of wall stratification in the submucosal and mucosal layers".
(4) Increase in fat wrapping (Figures 2-4). Bright white hyperechoic mesenteric fat thickened and extended to the surface of the contiguous intestine, enclosing the intestinal tube. In our study, "increased" was determined as "over two-thirds of the circumference of the intestine" in transverse view.
(5) Presence of fistula signs ( Figure 5). Fistula signs were defined as a hypoechoic, duct-like, peri-intestinal lesion with a diameter smaller than 2 cm, sometimes with internal echoic spots related to the presence of air, debris, or intestinal material [20].
Any discrepancies were resolved, and a final definite diagnosis was made by consultation with a senior sonographer (HY) with 10 years of experience in IBD ultrasound examination.
According to a review of our hospital's electronic medical record system, a gastroenterologist (KY), who was blinded to the previous US examinations, created an anonymous file for each patient. The gastroenterologist collected baseline clinical information (gender, age when diagnosed as CD, duration of illness) and surgical details (indication and methods of operation). After lesion information was obtained from doctors who matched the images between TAUS and WIUS, the location of the lesions and the macroscopic changes described in the surgical report were recorded (Table 1). Then, a statistical analysis of baseline data and US features between TAUS and WIUS was performed. Distinct wall stratification as a five-layer structure in an ileal segment. The resected ileum was incised and opened on the anti-mesenteric border, scanned from the mucosal surface (open view) in transverse view. The right panel shows the image obtained from WIUS, and the left panel is a sketch of the stratification of the bowel wall. The first-to-fifth layers represent the border between (a) the lumen and mucosal layer (hyperechogenic); (b) the mucosal layer (hypoechogenic); (c) the submucosa (hyperechogenic); (d) the muscular layer (or the muscle membrane proper) (hypoechogenic); (e) and the serosa layer (hyperechogenic). No mucosal defects such as ulcers and erosion are seen. "Intact" indicates the macroscopically normal part of the surgical specimen between stenosis or a fistula, which was the reason for surgery. (b) Longitudinal view of TAUS. The wall stratification is clearly demonstrable in the submucosal and mucosal layers. The mucosal and submucosal layers show slight homogeneous thickening (＊1). The mucous surface is regular in appearance. There is slight hypertrophy of adipose tissues around the intestinal wall, but the extent is limited (no more than half of the circumference of the intestine) and has not yet reached the degree of "fat wrapping". Fistulae are not detected. Transverse (c) and longitudinal (d) views of the intestinal section by WIUS (closed view). The wall stratification of submucosal and mucosal layers is clearly visible. The mucosal and submucosal layers show mild homogeneous thickening (＊2). The mucous surface is regular. Mild hypertrophy of the adipose tissue of the mesentery is present in the transverse view (c). The longitudinal view (d) shows that the width of the intestine is uniform, and there is liquid echogenicity in the intestinal cavity. No fistula signs are The wall stratification is clearly demonstrable in the submucosal and mucosal layers. The mucosal and submucosal layers show slight homogeneous thickening (*1). The mucous surface is regular in appearance. There is slight hypertrophy of adipose tissues around the intestinal wall, but the extent is limited (no more than half of the circumference of the intestine) and has not yet reached the degree of "fat wrapping". intestinal cavity did not produce clear images. The continuity of the intestinal wall is good, and there is no interruption in the serous layer or the formation of hypoechoic fistulae. For this lesion, TAUS and WIUS were consistent in determining some US parameters: (1) disappearance of wall stratification in the submucosa and mucosal layers; (2) a change in thickness of the submucosal layer could not be judged; (3) TAUS could not distinguish the mucous surface, whereas WIUS showed the mucous surface was irregular; (4) TAUS and WIUS showed agreement in detecting increased fat wrapping around the bowel wall; and (5) absence of fistula signs.   . The images show that the mucosal layer is blurred while the submucosal layer is heterogeneous by echogenicity with an uneven thickness (＊2). The serous and muscular layers are clearly displayed with normal thicknesses. The mucous surface is irregular in appearance. There is hyperechoic adipose tissue around the intestine wall with mild hypertrophy. The continuity of the intestinal wall is good, and there is no interruption

Statistical Analysis
Taking WIUS as a standard, the sensitivity, specificity, and accuracy of TAUS were calculated according to different US parameters. The accuracy was defined as the ability to differentiate cases correctly according to the reference or standard; this was the proportion of true positives and true negatives in all evaluated cases [21]. The sensitivity, specificity, and accuracy of TAUS were calculated according to a previously published formula [21]. Descriptive statistics of baseline data included the mean and standard deviation. A two-tailed Fisher's exact test and chi-square test were used to compare the frequencies of all categorical variables for differences between WIUS and TAUS, when appropriate. The level of significance was set at p < 0.05.
The agreement of image parameters between TAUS and WIUS was calculated using kappa statistics, which measure agreement independent of that obtained by chance. Kappa ranges from −1 (complete disagreement) to +1 (perfect agreement). A kappa value < 0 indicates no agreement, 0-0.20 indicates slight agreement, 0.21-0.40 indicates fair agreement, 0.41-0.60 indicates moderate agreement, 0.61-0.80 indicates substantial agreement, and 0.81-1.00 indicates almost perfect

Statistical Analysis
Taking WIUS as a standard, the sensitivity, specificity, and accuracy of TAUS were calculated according to different US parameters. The accuracy was defined as the ability to differentiate cases correctly according to the reference or standard; this was the proportion of true positives and true negatives in all evaluated cases [21]. The sensitivity, specificity, and accuracy of TAUS were calculated according to a previously published formula [21]. Descriptive statistics of baseline data included the mean and standard deviation. A two-tailed Fisher's exact test and chi-square test were used to compare the frequencies of all categorical variables for differences between WIUS and TAUS, when appropriate. The level of significance was set at p < 0.05.
The agreement of image parameters between TAUS and WIUS was calculated using kappa statistics, which measure agreement independent of that obtained by chance. Kappa ranges from −1 (complete disagreement) to +1 (perfect agreement). A kappa value < 0 indicates no agreement, 0-0.20 indicates slight agreement, 0.21-0.40 indicates fair agreement, 0.41-0.60 indicates moderate agreement, 0.61-0.80 indicates substantial agreement, and 0.81-1.00 indicates almost perfect agreement [22]. Because it is rarely plausible that agreement might be lower than that expected by chance, we estimated the strength of agreement by constructing a confidence interval for kappa.

Baseline Information
A total of 29 patients were enrolled in this study, of whom 22 were male and seven were female. The mean age at resection was 32.3 ± 10.39 (26.0-39.0) years. The median (interquartile range) of illness duration was 6.0 (2.4-12.5) years. The reason for intestinal surgery was stenosis (48.3%), refractory disease (3.4%), stenosis and fistula (27.6%), stenosis and refractory disease (17.2%), and stenosis and abscess (3.4%). The median (interquartile range) of surgery resection length was 23 (5-73) mm. Overall, 113 lesions were compared between WIUS and TAUS. The most commonly affected site was the non-terminal ileum (49.5%). The three most frequent changes in the gross morphology of specimens were fibrous stenosis (42.5%), openly longitudinal ulcerations (33.6%), and openly irregular ulcerations (20.4%). The basic characteristics of the subjects and samples enrolled in the study are shown in Table 1.

Evaluation of the Diagnostic Efficiency of TAUS
The consistency of images obtained from preoperative TAUS and postoperative WIUS is summarized in Table 2. Only 30 lesions were evaluated as the "existence of wall stratification" by TAUS and WIUS; therefore, only these 30 lesions were examined for the presence of submucosal thickening. The US parameter of increased fat wrapping yielded the highest sensitivity of 89.4%. The highest values of specificity (96.0%) and accuracy (92.7%) were related to the presence of fistula signs. The kappa coefficient for all parameters (fat wrapping, existence or disappearance of the intestinal wall, irregularity of the mucosal surface, thickening of the submucosa, and fistula signs) ranged between 0.446 and 0.615, which indicated that images of preoperative TAUS and postoperative WIUS had moderate-to-good consistency. Table 3 shows the consistency of the detailed parameters of TAUS and WIUS for the submucosal and mucosal parietal layers, with a kappa coefficient of 0.454.

Discussion
High sensitivity and high penetration power are the key advantages of WIUS [23]. WIUS has successfully been used to describe the US features of gastric carcinoma [24] and the intestine [14,15]. WIUS can directly scan tissues shortly after gastrointestinal tract resection, providing a clear image that avoids the possibility of errors caused by inexperienced examiners. Nevertheless, before an intestinal WIUS scan, feces and blood in the intestinal cavity should be removed, which is a difficult and time-consuming task. TAUS is commonly used for the clinical diagnosis of IBD. As doctors who work in the Inflammatory Bowel Disease Center and who have years of experience in intestinal WIUS, we wondered whether TAUS, an easier and more commonly used method than WIUS, could achieve a diagnostic efficiency similar to that of WIUS. More importantly, we investigated whether TAUS could be used to accurately examine the state of CD via complex histopathological changes. Therefore, using WIUS as a standard, we evaluated the agreement between TAUS and WIUS for several commonly used US parameters in CD.
US grayscale imaging allows for the visualization of wall stratification as a five-layer structure ( Figure 6) in the normal intestine [19]. Changes in wall stratification in the submucosal and mucosal layers are not specific for the diagnosis of CD but are commonly used for CD evaluation. The earliest change caused by CD occurs in the mucosa and submucosa and consists of hyperemia and edema [25]. On the basis of many valuable positive results in the literature, the EFSUMB (European Federation of Societies for Ultrasound in Medicine and Biology) Recommendations and Clinical Guidelines for Intestinal Ultrasound in Inflammatory Bowel Diseases concluded that the loss of stratification was associated with many aspects of the disease, such as CD activity, histological inflammation, and increased surgical risk [26]. Of note, a previous study on stenosis segments of CD indicated that a loss of stratification at the segment of stenosis predicted inflammation with a low degree of fibrosis, while the presence of a stratified echogenicity pattern suggested a higher degree of fibrosis [27]. In terms of the multiple significance of intestinal wall stratification, TAUS may be an alternative tool to assist in the diagnosis and evaluation of the disease state of CD.
US has shown that, during acute inflammation, the submucosal layer is thickened by fat deposition, edema, or hemorrhage [28,29]. According to our experience and the literature [30], CD begins as inflammation in the intestinal submucosa and then crosses the intestinal wall to involve the mucosa and serosa. Even if the mucosa, as shown by endoscopy, is completely intact, potential inflammatory changes may occur in the submucosa. Therefore, thickening of the submucosal layer is sensitive for predicting potential early or subclinical lesions. A previous study reported that submucosal thickening was the initialization of chronic fibrosis. With continuous thickening of the submucosa, which might be caused by collagen deposition, the accumulation of collagen gradually leads to the loss of compliance between the submucosa and the muscularis propria [31]. Eventually, all wall layers are thickened, contributing to chronic fibrosis of the entire intestinal wall [31]. The ECCO (the European Crohn's and Colitis Organization) Guideline/Consensus Paper in 2018 reported that bowel wall thickening, as measured by US one year after surgery, was a strong independent risk factor for early postoperative recurrence [32]. Regarding bowel wall thickening, our study yielded a high detection rate by TAUS (20/30, 66.7%) and WIUS (21/30, 70.0%). Furthermore, it had the highest kappa value (0.615) among all parameters, indicating that TAUS has a good ability to detect underlying acute inflammation and early fibrosis in CD.
Unlike the US parameters of intestinal wall stratification and submucosal thickening, few US-related studies have reported in detail the irregular mucosal surface pattern in CD. This sign has been generally obtained by the colonoscopic observation of a patient's intestines in vivo [33] or by the microscopic observation of intestinal specimens in vitro [34]. The interruption or unevenness of mucosal surface continuity is thought to be mainly caused by ulcers on the mucosal surface. The British Society of Gastroenterology reporting guidelines state that an irregularity of the mucosal surface is a typical feature used to distinguish IBD from non-IBD/infective colitis [34]. However, it is not specific for a diagnosis of CD. Encouragingly, in our study, this US pattern was successfully detected by TAUS and WIUS, both with a high incidence (TAUS: 72/110, 65.5%; WIUS: 79/110, 71.8%). A moderate concordance between TAUS and WIUS (kappa: 0.557) and a good accuracy of TAUS (80.9%) were also obtained. Therefore, we concluded that TAUS reflects the presence of inflammation by observing irregular mucosal surfaces in the bowel wall.
In addition to detecting changes in the intestinal wall, TAUS also easily detected abnormalities beyond the intestinal wall. When CD occurs, the interface of the intestinal serosa, where the mesentery and intestinal areas merge, is blurred. Mesenteric fat is inflamed and extends beyond its normal anatomical distribution, covering the surface of the adjacent intestine. This phenomenon, termed "fat wrapping" or "creeping fat", is a specific sign of CD [35]. The degree of fat wrapping correlates with the severity of intestinal inflammation in the contiguous intestine [35]. Chronic inflammatory stimulation produces fibrofatty proliferation of the mesenteric fat [36]. Li et al. demonstrated that when quantified using computed tomography images, mesenteric fat correlated significantly with CD disease activity, which was determined by the CD activity index and C-reactive protein [37]. Because our study shows acceptable consistency between TAUS and WIUS (0.528), the detection of "fat wrapping" by TAUS might reflect chronic inflammation of the intestinal tract.
Fistulae, which are common but important extra-intestinal complications, form as an abnormal tunnel through the intestine and into the surrounding tissue. Because of its high resolution and flexibility, endoscopic US rather than TAUS is widely used for the study of CD perianal fistulae [38]. However, CD-associated fistulae may occur in any segment of the affected bowel and exist as internal fistulae (enterocutaneous, enteroenteric, enterovaginal, and enterovesical fistulae) [39] with an incidence of 5-10% in CD patients. Endoscopic US may not be able to enter through the stenosis or obstructive segment or display the internal fistula in deep and distant intestinal segments. In our study, there were 11 cases of fistula, of which eight were in the non-terminal ileum. TAUS successfully detected most fistulae in this study (63.6%, 7/11), including those in the non-terminal ileum segment (75%, 6/8), where endoscopic US may have difficulty in scanning. Encouragingly, TAUS achieved its highest accuracy (92.7%) and moderate accordance (kappa = 0.596) for fistula detection compared with other US parameters. These results suggest that TAUS is useful for detecting fistulae, especially internal fistulae in CD.
Although TAUS and WIUS yielded acceptable-to-good agreement, WIUS detected more anomalies than TAUS for almost all US parameters in our study, except increased fat wrapping (Table 2). It was reported that TAUS has difficulty in detecting retroperitoneal lesions, lesions behind air-distended intestinal loops, and intestinal loops located deep in tissues [36] because the visibility is limited by sound absorption, attenuation, and artifacts in tissues such as muscle and fat. In this study, the US parameter with the worst consistency between TAUS and WIUS was the disappearance of intestinal wall stratification (kappa = 0.446, Table 2). Similarly, TAUS and WIUS had a relatively low consistency (kappa = 0.454, Table 3) for "distinct presence/indistinct presence/disappearance" of wall stratification than other US parameters, suggesting that examination of the appearance of wall stratification might easily be misdiagnosed by TAUS. A potential reason for this is that linear stratification by US is subtle and needs to be carefully identified, whereas other US parameters (submucosal thickening, irregularity of mucosal surfaces, increased fat wrapping, and fistulae) can easily be detected and determined. Therefore, we speculated that the location of lesions and the fineness of US parameters may be related to the small number of misdiagnosed cases by TAUS.
There were several limitations in this study. First, all cases enrolled were in the complex situation of refractory disease or stenosis or had fistulae, which suggest a deteriorating stage of CD. Conservatively treated cases with mild and early-stage CD might have different US appearances from those in the current study. Second, vascularity and stiffness of the intestinal wall were not evaluated in our study. It was reported that the emerging technologies of TAUS-based contrast-enhanced US and elastography US can improve the diagnostic efficiency of US by detecting the vascularity and stiffness of the intestinal wall, which helps to predict intestinal inflammatory activity and fibrosis progress [40,41]. However, because they cannot be evaluated by in vitro WIUS, it is not clear whether TAUS can accurately measure these two US parameters.

Conclusions
The results of our study demonstrate that when properly performed, the results obtained from imaging with TAUS closely approximate those with WIUS. Moreover, TAUS can be used to evaluate the relationship between US and complex intestinal conditions (acute/chronic inflammation, mild/serious fibrosis, extra-intestinal complications) in CD. Nevertheless, because it is still imperfect and has a small discrepancy compared with the actual disease characteristics obtained by WIUS, the application of TAUS should be improved further.