Small intestine contrast ultrasonography (SICUS) for the detection of small bowel complications in crohn's disease: A prospective comparative study versus intraoperative findings

Authors


Abstract

Background:

In Crohn's disease (CD) patients, small intestine contrast ultrasonography (SICUS) accurately assesses small bowel lesions. Its diagnostic role is not known in the assessment of intraabdominal CD complications. The aim was to assess the value of SICUS to detect intestinal complications in patients with CD.

Methods:

Forty-nine CD patients (21 female, mean age 37.7 years; range 12–78 years) underwent resective bowel surgery and were included in this study. The accuracy of SICUS to preoperatively detect number, site, and length of strictures, fistulas, and abscesses was compared with surgical and pathological findings by kappa statistics.

Results:

SICUS identified at least one stricture in 39/40 and excluded it in 9/9 (97.5% sensitivity, 100% specificity, k = 0.93); two or more strictures in 9/12 (75% sensitivity, 100% specificity, k = 0.78). The agreement by k-statistics between SICUS and surgery in identifying proximal and distal small intestine site of stricture was 1 and 0.92, respectively. The extension of strictures was 6.8 ± 5.4 cm at surgery, 6.6 ± 5.4 cm at SICUS (NS). Fistulas were correctly identified in 27/28 patients and excluded in 19/21 patients (96% sensitivity, 90.5% specificity, k = 0.88). Intraabdominal abscesses were correctly detected in 10/10 patients and excluded in 37/39 patients (100% sensitivity, 95% specificity, k = 0.89).

Conclusions:

SICUS is an accurate method for the detection of small intestinal complications in CD. Noninvasive SICUS is valuable as a primary investigative method for evaluating and planning proper treatment in patients with severe CD of the small bowel. (Inflamm Bowel Dis 2011;)

Contrast radiology, computed tomography (CT), magnetic resonance imaging (MRI), and transabdominal ultrasound (TUS) are widely and variably utilized for the diagnosis of Crohn's disease (CD) lesions of the small bowel (SB) and of strictures, fistulas, and abscesses that often complicate the lifelong course of CD. Contrast radiology has for a long time been the only diagnostic tool to detect SB stenoses but it has been shown to be inaccurate to detect fistulas and abscesses1–7 because of poor assessment of transmural and extramural extension of disease, and it is not indicated in patients with symptoms of acute obstruction. TUS proved to be valuable in detecting SB CD strictures and abscesses, whereas its sensitivity in detecting entero-enteric fistulas is still debated.8–12 Furthermore, TUS does not properly assess degree, localization, and length of strictures. MRI is valuable to detect abscesses,13, 14 but is limited to identify low-grade strictures15 and to discriminate them from muscular bowel wall contractions; finally, its accuracy in the detection of fistulas is not yet fully established.16 CT is useful for the detection of not-deep abscesses12, 17 but its diagnostic accuracy for CD strictures and fistulas is low when compared with surgical findings.12, 17, 18 Furthermore, since with the medical use of x-rays it is a significant (the second after natural terrestrial) source of radiation,19 radiological exposure should be minimized, particularly in young and child-bearing age female patients and in those who may require repetitive follow-up assessments.

In order to plan appropriate treatments for CD, accurate diagnostic tools are needed to detect and characterize these complications.20, 21 Surgery remains an important component of treatment of CD and an accurate preoperative assessment of CD lesions and associated complications is required to plan the surgical approach and intervention.22–24 Such preoperative assessment is even more important when using a laparoscopic approach that may limit the surgeons manual search for subtle lesions of the intestine.23

Four studies8, 9, 12, 18 have compared TUS and CT preoperative findings with operative findings. Two8, 9 assessed the accuracy of TUS in detecting the presence of strictures, abscesses, and fistulas, one12 comparatively assessed the diagnostic accuracy of contrast radiology, CT, and TUS in the detection of fistulas and abscesses, whereas the accuracy of CT in the diagnosis of strictures, fistulas, and abscesses was assessed retrospectively by Chiorean et al.18 However, so far no study based on any of the following single imaging techniques, i.e., contrast radiology, CT, MRI, or TUS, at the same time assessed number, site, extension of SB CD lesions, and the associated complications. In addition, no study so far has been performed on the preoperative assessment of the presence and location of mesenteric fat hypertrophy that may influence the surgical approach.25

Abdominal ultrasonography (US) performed after the ingestion of an intraluminal oral contrast (small intestine contrast US, SICUS) enables visualizing the entire SB26–29 and has been shown to be comparable to radiology in detecting site, number, and extension of SB lesions in patients with undiagnosed SB disease as well as in those with known CD.28–31 Oral contrast US, by distending the intestinal lumen, detects and characterizes SB strictures better than TUS and radiology.30, 31 So far, however, the diagnostic accuracy of SICUS is not known in the assessment of number, site, length of CD SB strictures or in the assessment and characterization of internal fistulas and of intraabdominal abscesses. Furthermore, the role of SICUS, if any, is not known in the assessment of mesenteric involvement in CD compared to surgical findings.

The primary aim of this prospective study was to assess the role of SICUS in the preoperative assessment of CD SB lesions and complications, namely: 1) the presence, site, number, and length of lesions and strictures; 2) the presence and type of internal fistulas; 3) the presence of abscesses; and 4) mesenteric involvement, having surgical and pathological findings as the reference standard.

A secondary aim was to comparatively evaluate the diagnostic accuracy of TUS and SICUS in the detection and assessment of CD complications.

PATIENTS AND METHODS

Patients

As a part of a long-term prospective follow-up study, consecutive patients with proven CD submitted to a complete diagnostic workup and undergoing surgery after US evaluation between January 2000 and October 2010 were included in the study. Medical history, including abdominal and extraabdominal complaints, associated diseases, CD behavior, smoking status at diagnosis, family history, location of CD, duration of the disease before surgery, number of surgical resections, postoperative medical treatment, and pathological studies of the resection specimen and of the section margins were assessed.

Informed consent was obtained from each subject and the study protocol was approved by the local Ethics Committee of our University Hospital (#918).

Protocol of the Study

Each patient was initially submitted to a standardized clinical interview and a physical examination performed by certified and experienced gastroenterologists. After an overnight fast, patients were consecutively submitted on the same day to TUS and SICUS by an expert intestine-dedicated sonologist (N.P.) aware of the diagnosis, history of previous bowel surgery, and current clinical symptoms but blinded to the results of the other investigations. At the end of the diagnostic investigation the operator reported on a standardized form SB abnormalities with particular reference to presence, anatomical site of intestinal wall, and lumen abnormality as well as of all CD complications. When deemed necessary, additional investigations, including upper and lower gastrointestinal (GI) endoscopy, histology, CT, or MRI were performed.

The decision to operate was made independently from the SICUS findings on the basis of clinical findings, laboratory results, and other imaging studies by a sovereign gastroenterologist and the surgeons (P.C. or C.M.).

Ultrasonography

Real-time ultrasound (US) was performed using Toshiba Tosbee (Tokyo, Japan) equipment with a 3.5 MHz convex and a 5 MHz linear array transducers. The apparatus can detect a bowel wall thickness (BWT) variation of 0.1 mm. The sonologist (N.P.) has performed more than 9000 SICUS examinations.

TUS and SICUS was performed after an overnight fast according to a previously published26 method. Briefly, after the ingestion of 250–500 mL of macrogol contrast oral solution and after the contrast was seen to flow through the terminal/neoterminal ileum into the colon, a retrograde follow-through assessment of the entire SB was performed visualizing, in a caudocranial sequence, the contrast-filled ileal and jejunal loops. Whenever required, the body position of patient was changed and abdominal compression with the ultrasound transducer was used to improve visualization of any single loop and detection of intestinal abnormalities. Wall thickness and lumen diameter were measured at several sites (proximal, middle, and distal) of the SB at the level of maximally distended, and not contracting, intestinal loops.

Contrast Solution

Before the investigation the contrast solution was freshly prepared by dissolving, in 250 mL tapwater, a granular powder containing PEG 4000 14.6 g, anhydrous sodium sulfate 1.42 g, sodium bicarbonate 0.42 g, sodium chloride 0.36 g, and potassium chloride 0.18 g (Promefarm, Milan, Italy). PEG 4000 is a nondigestible, nonabsorbable, and nonfermentable macromolecule that links water and once ingested progresses through the GI tract until rectal evacuation.26 The iso-osmolar PEG solution employed in this study was devoid of any caloric content and does not delay gastric emptying and, once in the SB, remains unchanged, filling the lumen and distending the intestinal loops.

Definitions of CD Lesions and Complications at US

US criterion for the presence of SB CD lesions was increased wall thickness (>3 mm).26, 28 The extension of the increased intestinal wall thickness (>3 mm) was reported as the average of at least three measurements. Bowel stenosis was defined as lumen diameter less than 1 cm26, 31 with or without prestenotic bowel dilatation (Fig. 1A,B). Bowel dilatation was defined as lumen diameter >2.5 cm.26, 31

Figure 1.

(A) SICUS assessment of ileal stricture at the level of terminal ileum. Calipers indicate thickness of ileal wall. Arrows indicate the extension of luminal narrowing. (B) Corresponding surgical findings of the same patient. Arrowheads indicate the extension of luminal narrowing. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Hypoechoic peri-intestinal lesions were defined fistulas when duct-like, shaped with a cross-sectional lumen diameter less than 2.0 cm (Fig. 2A,B) and abscesses when a round-like mass with a cross-sectional diameter more than 2 cm (Fig. 3A,B).8, 11 Mesenteric fat hypertrophy (MFH) was defined as homogeneous hyperechoic tissue surrounding thickened bowel32 with loss of the normal double-shaped stratification of hypoechoic and hyperechoic layers (Fig. 4A–C).

Figure 2.

(A) Longitudinal US section (3.5 MHz probe). A bowel loop is shown with a hypoechoic peri-intestinal lesion of 3.7 mm (see calipers) arising from the thickened intestinal wall (arrows) through a wrapping mesentery. (B) Corresponding surgical findings of the same patient. The forceps (see arrowheads) point fistula arising from an ileal loop through the mesentery. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Figure 3.

(A) Transversal US section (3.5 MHz probe) at the level of the lower abdominal right quadrant. A round shaped hypoechoic image (see arrows) of 4.6 cm (see calipers) is shown. (B) In the same patient intraoperative findings show the presence of an interloop abscess (see arrows) at the level of the lower abdominal right quadrant. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Figure 4.

(A) A normal SICUS appearance of mesentery at the level of jejunal loops (see arrowhead). (B,C) SICUS appearance (B) and corresponding surgical findings (C) of MFH (arrowhead), and of entero-enteric fistula (arrows). [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

At the end of the US investigation the following measurements were reported on a standardized form: 1) length of intestinal wall alterations; 2) number, site, and length of strictures; 3) presence and type of fistulas; 4) presence of abscesses; 5) presence of MFH.

Surgery and Pathology

Laparatomic surgical and pathological findings were reference standards for determining length of SB CD lesions, for detecting presence, site, number, and length of SB strictures, presence of internal fistulas, abscesses, and MFH. Surgeons with a specific experience in CD surgery were not blinded to the results of preoperative SICUS. Following laparotomy, the examination and description of SB alterations and extraluminal complications preceded any surgical maneuver. The SB was carefully examined and manually palpated by the surgeon from the ligament of Treitz to the cecum. Site, number, and extension of CD lesions were carefully considered and measured. If adhesions were present a complete lysis was made and CD complications were specifically sought. Bowel strictures were diagnosed in the presence of luminal narrowing and bowel wall thickening with or without prestenotic dilatation. Internal fistulae were defined as abnormal communications between the lumen of the gut and the mesentery and/or another hollow organ and were classified according to their anatomy into entero-enteric, entero-mesenteric, and entero-vesical. Abscesses were defined as intra-peritoneal or mesenteric collections of fluid containing necrotic debris (Fig. 3B). Abscesses were classified according to their localization into intra-peritoneal, retroperitoneal, and pelvic. Intraperitoneal abscesses were further subdivided into superficial and deep (intraloop and mesenteric). MFH or fat-wrapping was defined as thickening of fat-mesentery covering greater than 50% of the intestinal loop circumference, with loss of the bowel-mesentery angle25 and of normal transparency. Specimens of resected bowel were carefully examined by the pathologist at the end of the procedure by opening the antimesenteric site to measure the intestinal lumen diameter in millimeters and to explore each possible fistula. Lesions that had penetrated the muscolaris propria were considered fistulas.

Diagnosis and Analysis of Data

Sensitivity and specificity of TUS and SICUS in the assessment of SB CD lesions as well as of CD complications were calculated using intraoperative findings and surgical specimens as reference standards. Consistency between surgery and pathological findings and TUS and SICUS in detecting number and anatomical site(s) of CD SB lesions as well as of CD complications was evaluated using the concordance index by kappa statistics.33 The quantitative significance of k was calculated according to Landis and Koch.33 The comparative assessment of bowel lesion extension measured at surgery and TUS and SICUS was evaluated by paired Student's t-test. The correlation between the US and surgical extent of SB lesions was assessed by Pearson's correlation coefficient. The comparative assessment of the length of the strictures at SICUS and surgery was assessed by paired Student's t-test. The association between the presence of MFH and the presence of fistulas and strictures was analyzed by the chi-square test.

RESULTS

Of the 380 patients referred and followed up in our center, 49 underwent surgery of macroscopically diseased bowel. Patients underwent surgery because of the following conditions: 32 (65%) for obstructive symptoms, 10 (20.5%) for sepsis, one (2%) for suspicion of neoplasia, one (2%) for reduction of colostomy, and five (10.5%) for sepsis associated with obstructive symptoms. The main demographic and clinical characteristics of evaluated patients are reported in Table 1. The mean and the median time interval between US investigations and surgery was 59 and 24 days, respectively (range 1–180 days).

Table 1. Clinical Characteristics and Location of Crohn's Disease Lesions According to Montreal Classification
Female N (%)21 (43)
Age (yrs) mean, range37.7 (12–78)
Smoking N (%)20 (40)
Disease duration at surgery (yrs) (range)6 ± 4.6 (0–41)
Median period between SICUS and surgery (days) (range)24 (1–180)
Illness behavior N (%) 
 B16 (12)
 B224 (49)
 B319 (39)
CD intestinal location N (%) 
 L127 (55)
 L21 (2)
 L320 (40)
 L41 (2)

Surgical and Pathological Findings

At surgery eight patients had one or more jejunal and proximal ileal CD involvement, 39 one or more distal and terminal ileum/neoterminal ileum CD involvement, and one patient colic CD. Patients submitted to reduction of colostomy had adhesions. The extension of SB lesions was 31 ± 22 cm (mean ± SD). Surgical and pathological evaluations revealed at least one CD complication in 45 patients (Table 2). One or more strictures were present in 40 (82%) patients and in 14/40 patients (35%) were preceded by proximal prestenotic dilatation (Table 3). Strictures were located in the distal/terminal ileum (including the ileocecal valve or anastomosis) and in the upper SB in 33 and 7 patients, respectively (Table 3). Twenty-eight (70%) patients had one stricture, nine (22.5%) patients had two strictures, and three patients ≥3 strictures (Table 3). The extension of SB stricture was 6.8 ± 5.4 cm (mean ± SD). In all, 82% of patients had strictures either alone (n = 17) or in combination with fistulas (n = 16) and fistulas and abscesses (n = 7) (Table 2). Twenty-eight (57%) patients had fistulas. Fistulas were: 12 entero-enteric, nine entero-mesenteric, 13 entero-colic, one entero-vesical, and two entero-cutaneous (Table 3). Two patients had fistulas alone and three combined with abscesses. In four patients no SB CD intestinal complication was detected (two patients, colon CD stricture; one patient, adhesions; one patient, mucocele of appendix) (Table 2). MFH was detected in 27/49 (55%) patients and was associated with fistulas (P = 0.001) but not with strictures (P = 0.45)

Table 2. Comparative Results of Surgical and US Evaluation by k-statistics in the Assessment of CD Complications
 SurgeryTUSSICUS
TPFPFNkTPFPFNk
  • a

    Two patients CD strictures of colon; one patient mucocele of appendix; one patient with adhesion and reduction of colostomy.

  • b

    Patient with mucocele of appendix.

  • c

    Patient with abscess and fistula.

  • TP, true positive; FP, false positive; FN, false negative; k, k-statistics.

Stricture alone1713940.35316010.945
Strictures and fistula165120.29516001
Stricture, fistula and abscess76110.547001
Fistula and abscess3101200
Fistula2100200
Abscess001002bc0
No SB CD complication4a201300
Total49281290.1546210.75
Table 3. Accuracy of Ultrasound in the Detection of CD Small Bowel Intestinal Complications Compared to Surgical Findings
Surgery and PathologyTUSSICUS
Sensitivity (95% CI)Specificity (95% CI)kSensitivity (95% CI)Specificity (95% Cl)k
  1. nd, not detectable; k, k-statistics.

Strictures (n=40)80% (65–90)75% (40–93)0.4297.5% (87–99.6)100% (67.6–100)0.93
# 1 (n=28)78.5% (60–89.7)90% (70–97)0.6685.7% (68.5–94)100% (84–100)0.70
# 2 (n=9)0% (0–30)nd56% (27–81)93% (80–97.5)0.50
# ≥3 (n=3)0% (0–56)nd100% (44–100)97% (86–99.5)0.85
Upper strictures (n=7)57% (25–84)96.5% (83–99)0.42100% (65–100)100% (89–100)1
Distal/terminal strictures (n=33)87.5% (72–95)85.7% (48–97)0.3997% (84–99)100% (65–100)0.92
Presence of prestenotic dilatation (n=14)28% (12–55)88% (70–96)0.20100% (78–100)68% (48–83)0.60
Fistulas (n=28)55.5% (37–72)100% (84.5–100)0.5496% (82–99)90.5% (71–97)0.88
Entero-enteric (n=12)ndnd 100% (76–100)82% (61–93)0.77
Entero-mesenteric (n=9)ndnd 100% (67–100)100% (87–100)1
Entero-colic (n=13)ndnd 54% (29–77)100% (85–100)0.61
Abscesses (n= 10)89% (56.5–98)95% (83–98.6)0.78100% (72–100)95% (83–98.6)0.89
MFH (n=27)60% (41–76.6)90% (72–97.5)0.4996% (82–99)91% (72–97.5)0.88

TUS and SICUS Findings

A complete SICUS investigation was performed in all patients on average in 45 minutes (range 20–120 minutes) after the ingestion of on average 375 mL (range 250–500 mL) of contrast solution. TUS was performed on average in 15 minutes.

Preoperative SICUS detected one or more jejunal and proximal ileal CD involvement in nine patients, one or more distal and terminal ileum/neoterminal ileum CD involvement in 39, and in one patient colic CD. The extension of SB lesion was 30 ± 14 cm (mean ± SD). TUS detected one or more jejunal and proximal ileal CD involvement in four patients, one or more distal and terminal ileum/neoterminal ileum CD involvement in 33 patients. The extension of SB lesion was 16 ± 7 cm (mean ± SD).

Preoperative SICUS revealed at least one intraabdominal CD complication in 45 patients, TUS did so in 39. SICUS showed one or more strictures in 39 (79.6%) patients and in 14/39 (36%) were preceded by proximal dilatation. Strictures were located in the distal/terminal ileum (including the ileocecal valve or anastomosis) and in the upper SB in 32 and 7 patients, respectively. Twenty-seven (69%) patients had one stricture, eight (20.5%) patients had two strictures, and four patients ≥3 strictures. The extension of SB stricture was 6.6 ± 5.4 cm (mean ± SD). At SICUS 16 patients had strictures alone, 16 in combination with fistulas alone, and seven with fistulas and abscesses (Table 2). Four patients had fistulas alone (n = 2) or combined with abscesses (n = 2) (Table 2). Twenty-seven (55%) patients had fistulas. Fistulas were: 16 entero-enteric, nine entero-mesenteric, seven entero-colic, two entero-vesical, and two entero-cutaneous. Two patients had abscesses alone. In four patients no SB CD intestinal complication was detected (Table 2).

TUS showed one or more strictures in 35 (71%) patients and in 4/35 (11%) were preceded by proximal dilatation. Strictures were located in the distal/terminal ileum (including the ileocecal valve or anastomosis) and in the upper SB in 31 and 4 patients, respectively. Thirty-three (94%) patients had one stricture, two (6%) patients had two strictures. At TUS 22 patients had strictures alone, six in combination with fistulas alone and seven with fistulas and abscesses (Table 2).

One patient had abscesses alone and one combined with fistula. One patient had fistulas alone. In 10 patients no small intestinal CD complication was detected (Table 2). SICUS revealed MFH in 26 patients, TUS did so in 16 patients.

Comparison of US with Surgical and Pathological Findings

In the assessment of site of CD lesions the agreement between surgery and SICUS was almost perfect for the lesions located in the jejunum–proximal ileum as well as for those located in the distal/terminal ileum with a k-value of 0.93, whereas the agreement between surgery and TUS was moderate and substantial for the lesions located in the jejunum–proximal ileum and distal/terminal ileum, with k-values of 0.54 and 0.68, respectively.

There was no difference in the assessment of CD lesion length between surgery and SICUS (mean difference between paired measurements of CD lesion length between surgery and SICUS not significantly different from 0, P = 0.829 by Student's t-test; mean 0.255, 95% confidence interval [CI] for mean: −2.11; 2.625, SD 8.07, median 0.00), whereas TUS underestimated the length of CD lesions compared to surgery (mean difference between paired measurements of CD lesion length between surgery and TUS significantly different from 0, P = 0.000 by Student's t-test; mean −13, 95% CI for mean: −16.99; −9.27, SD 13, median −12) (Fig. 5A,B). In the assessment of lesion extension the correlation coefficient between surgery and TUS was 0.55 and between surgery and SICUS was 0.78.

Figure 5.

Scatterplot of difference between paired measurements of CD lesion length at SICUS (A) and at TUS (B) and surgery against their mean. (A) Overall, the average difference was 0.255 cm and the SD was 8.07 cm assuming that the probability distribution of differences follows a normal distribution, the 95% difference is expected to be between −16.35 cm and +15.85 cm (see dotted line). The mean difference was not significantly different from 0 (P = 0.829 by Student's t-test, 95% CI: −2.11; 2.625). (B) Overall, the average difference was −13.1 cm and the SD was 13.1 cm assuming that the probability distribution of differences follows a normal distribution, the 95% difference is expected to be between −39 cm and +12.9 cm (see dotted line). The mean difference was significantly different from 0 (P = 0.000 by Student's t-test, 95% CI: −16.99; −9.27).

Sensitivity and specificity of SICUS and TUS in detecting all CD complications are reported in Table 3. It was possible to identify at least one SB stricture in 39/40 patients (sensitivity 97.5%) and in 35/40 patients (sensitivity 80%) and to exclude it in 9/9 (specificity 100%) and 7/9 (specificity 75%) by SICUS and TUS, respectively. In the assessment of the presence of at least one stricture the agreement between surgery and SICUS and TUS were 0.93 (almost perfect) and 0.42 (fair), respectively (Table 3). In the assessment of the site of CD strictures the agreement between surgery and SICUS was almost perfect for the lesions located in the jejunum–proximal ileum as well as for those located in the distal/terminal ileum with k-values of 1 and 0.92, respectively. There was no difference in the assessment of CD stricture length between surgery and SICUS (mean difference in the assessment of CD stricture length between surgery and SICUS not significantly different from 0, P = 0.54 by Student's t-test; mean −0.25, 95% CI: −1.095; 0.595, SD 2.17, median 0.00) (Fig. 6). Fistulas were truly identified in 27/28 patients (sensitivity 96%) and in 15/28 (sensitivity 55.5%) patients and excluded in 19/21 patients (specificity 90.5%) and in 21/21 (specificity 100%), patients by SICUS and TUS, respectively. Intraabdominal abscesses were correctly detected in 10/10 patients (sensitivity 100%) and in 9/10 (sensitivity 89%) patients and excluded in 37/39 patients (specificity 95%) and in 37/39 (specificity 95%) patients by SICUS and TUS, respectively. Agreement by k-statistics between SICUS and surgery/pathology and TUS and surgery/pathology in detecting preoperatively the number and type of strictures, fistulas, abscesses, and MFH are reported in Tables 2 and 3.

Figure 6.

Scatterplot of difference between paired measurements of strictures length at SICUS and surgery against their mean. Overall, the average difference was −0.25 cm and the SD was 2.17 cm assuming that the probability distribution of differences follows a normal distribution, we can expect that 95% difference is expected to be between −4.6 cm and +4.1 cm (see dotted line). The mean difference was not significantly different from 0 (P = 0.54 by Student's t-test, 95% CI: −1.095; 0.595).

DISCUSSION

The accurate assessment of lesions and complications is mandatory in the diagnostic workup of patients with CD in order to plan the proper treatment. Contrast radiology for several decades, and also US, CT, and MRI subsequently, have been widely used to diagnose SB CD lesions and associated complications. Intraoperative and pathological findings at surgery represent the only reference standard to assess the diagnostic accuracy of a technique in the assessment of CD complications as well as of site and extension of CD lesions. The main result of this prospective study is that SICUS accurately detects, compared to surgical and pathological findings, site and extension of CD inflammatory lesions of the SB as well as of all associated complications. Findings of the present study confirm our previous ones showing that, in comparison with contrast radiology, SICUS is definitely superior to TUS in the detection of the site, and the measure of the extension, of noncomplicated SB CD inflammatory lesions.31 Also, MRI has been proven to detect the presence of noncomplicated SB CD inflammatory lesions, but an acceptable intestinal lumen distension is obtained with an oral load of 2000–3000 ml of solution or by enteroclysis that requires the placement of a nasojejunal tube.15 MRI fails to prolong at will the observation period and may still miss early, subtle CD intestinal lesions even in properly distended loops.34, 35 In addition, despite precautions, vomiting and inadvertent rectal evacuation may occur inside the bore of the magnet.36 SICUS requires less than 500 mL of oral macrogol solution (375 mL in the present study) because, different from MRI, the observation period of the dynamic examination enables following the filling of any single SB loop as the intraluminal contrast moves and distends the lumen.

Even if previous studies2, 37 indicate in small series of patients a high diagnostic accuracy of contrast radiology in the detection of strictures, a large study by Otterson et al22 found that in comparison to operative findings SB follow-through incorrectly predicted the number of strictures in 30% of patients. Compared to surgical findings also, the accuracy of CT enterography is low in the assessment of the number of strictures.38 However, CT enterography should be avoided in CD patients who require repetitive follow-up assessments. It has been shown, indeed, that patients with CD receive a mean of 8.1 mSv of diagnostic radiation per year of follow-up, and those with complications requiring surgery are more likely to receive a high cumulative effective dose (CED), reaching in most cases the 75 mSv threshold.39, 40 MRI is a noninvasive, radiation-sparing tool but its diagnostic accuracy in detecting SB strictures has not been assessed since no study has compared MRI findings with surgery, except for a promising report referring only to five patients.36 Despite diagnosis with TUS of CD SB strictures being the object of several reports,8, 9, 10, 30, 41–43 a single study8 used as reference standard only the surgical pathological findings, showing an excellent sensitivity (100%) and specificity (91%) in the detection of at least one stricture in patients submitted to surgery for obstructive symptoms. The present study confirms that TUS has a high sensitivity (87%) and specificity (100%) in the assessment of the presence of at least one stricture in patients with obstructive symptoms (data not shown), but in addition it shows that SICUS has a higher diagnostic accuracy than TUS to detect: 1) the presence of more than two strictures; 2) those located in the proximal tract of SB; 3) extension of strictures, independently from the presence of prestenotic dilatation and of obstructive symptoms. The presence of the prestenotic dilatation has been regarded for imaging techniques the hallmark of fibrotic strictures. On the contrary, in the present study prestenotic dilatation was present at surgery and pathology in only 35% of patients, confirming a previous report.38 SICUS appears a valuable technique to be used preoperatively in CD patients since the inaccurate detection of number, site, and length of strictures may negatively affect the operation time and even the outcome if they are unexpectedly found, or if not looked for and missed, at surgery, especially in the case of a planned laparoscopic approach. Two studies have previously assessed in comparison with surgical findings the diagnostic performances of TUS8, 12 and one12 also of contrast radiology and CT in the detection of internal fistulas. In comparison with these previous reports, the present study confirms a similar and excellent specificity and a similar, yet lower, sensitivity of TUS; but in addition it shows that SICUS has a better sensitivity than TUS in the detection and characterization of internal, entero-enteric and entero-mesenteric, fistulas. In the Maconi et al study12 the sensitivity of TUS in the detection of internal fistulas was enhanced up to 97% combining TUS with contrast radiology and CT. Similar diagnostic accuracy has been obtained in our study by the use of SICUS alone. It is likely that the presence of intraluminal anechoic contrast distending the intestinal lumen allows better visualization and characterization of fistulas. Intraabdominal abscesses are equally detected by MRI, CT, and TUS. However, the diagnostic yield is lower for small, deep, intraloop, mesenteric abscesses.12 The present study confirms the high diagnostic yield of US, SICUS being more sensitive than TUS. It is likely that by distending with contrast the intestinal lumen of the entire SB, intraloop abscesses can be better differentiated from distended loop.

MFH has been found at TUS in about 50% of CD patients and it has been correlated with clinical activity of CD, internal fistulas, and increased bowel thickness.32 So far the preoperative assessment with imaging techniques of MFH it has not been compared with surgical findings. In the present study MFH was detected in 55% of patients at surgery and it was associated with fistulas but not with strictures, confirming previous surgery findings that MFH is associated with a transmural inflammation.25 The diagnostic accuracy of SICUS for MFH is higher than TUS, with a sensitivity of 96% and specificity of 91%.

It can be argued that, TUS being performed before SICUS by the same operator, the latter would be more sensitive than the former. However, the detection at SICUS of several lesions in areas other than those with identified lesions at TUS is supportive of a real higher diagnostic yield of SICUS. In addition, the diagnostic yield of SICUS is not merely limited to detect the presence, but also enables obtaining a more accurate definition of number, site(s), and extension of such alterations. Nonetheless, it would be appropriate to compare TUS and SICUS performed by independent operators and to assess the interobserver variability of SICUS in further studies.

The high prevalence of CD intestinal complications reported in this study is not representative of the CD population at large, since the protocol was conceived to evaluate the diagnostic efficacy of US to detect CD lesions and associated complications in comparison with the objective reference standard of surgical-pathological findings and not their prevalence. The comparison between US and surgical-pathological findings has shown that, despite the high prevalence of CD complications, as expected in patients with surgical indications, TUS is not as able as SICUS to detect all CD lesions and complications. In conclusion, TUS is not sufficiently reliable to offer a complete assessment of the CD SB lesions and associated complications. The use of a luminal contrast agent increases markedly the diagnostic efficacy of US since any single loop is well delineated and dissociated from the others; in addition, the MFH, fistulas, and abscesses are made more visible through, and between, the hypoechogenic windows of the fluid-filled loops. TUS and SICUS examinations were performed during a diagnostic workup in CD patients who, either not preceded by any previous CD history or during a regular follow-up, presented the clinical indication for surgery. Except for those patients with no deferrable surgical indication, a variable time interval could elapse between US assessments and surgery due to several factors, mainly hospital waiting list or patients' personal requirements for delaying. In this study we considered those patients in whom US and other diagnostic assessment were made less than 6 months from the operation, provided no other relevant factor interfered with, or changed, the clinical course of the disease. In a few (seven) patients the TUS and SICUS assessments were performed immediately before surgery in patients with no previous history of CD. The diagnostic yield of SICUS in the detection of CD SB lesions and associated complications was not different from those with a previously known diagnosis of CD. Similar findings of diagnostic efficacy of SICUS in patients with either known or unknown CD diagnosis have been reported when comparing US with contrast radiology.31

US is an operator-dependent technique and the diagnostic accuracy of TUS improves over time. Investigation of the SB by means of US requires a relevant skill; the operator in the present study is an experienced sonographer with specific expertise in SB investigation. It is likely also that accuracy of SICUS improves as the operator accumulates experience; however, SICUS has higher diagnostic accuracy than TUS even in inexperienced operators.44

Compared to TUS, SICUS requires more time, the duration of the examination on average being 45 minutes. However, the comprehensive detailed information of the entire SB and of any single lesion and complication offered by SICUS, definitely superior to TUS, suggests using this technique in the diagnosis, follow-up, and preoperative evaluation of CD patients. Confirming our previous study,28 the small amount of oral contrast used, which did not exceed 500 mL, could be easily ingested by all patients, and did not cause discomfort or side effects in any of them.

In conclusion, SICUS is a noninvasive, radiation-free, and highly acceptable technique that can alone perform a complete assessment of CD lesions of the SB and associated complications. It thus seems appropriate to indicate SICUS as the first examination upon the suspicion of a CD diagnosis, in the follow-up of CD patients, and in planning surgical intervention.