Derivation of Inflammatory Score (I-Score)
We propose a semiquantitative score on a scale 0-4,
based on the literature review,
by considering intestinal and extraintestinal findings.
To reduce variability,
we evaluate qualitative parameters on a 3-point scale (0 = absent; 2 = marked – diffuse) and minimize or standardize numeric variables.
Among intestinal findings,
mural thickening (>3 mm) has a low sensitivity (29–67%) and a good specificity (80%-89%) for detection of active CD [10; 11].
Wall enhancement has a higher sensitivity (80%-86%) and a specificity of 82% -86% for active CD [10; 12].
The layered aspect has a low sensitivity (47-60%) and a specificity of 78-89% for active inflammation in CD [11; 13].
Among extraintestinal findings,
comb sign has low sensitivity and high specificity (89-100%) for active inflammation [10; 11].
Creeping fat has a low sensitivity and a specificity of 93-96% for active disease in association with wall enhancement [10; 11; 14].
The perienteric fat stranding is considered a finding of active inflammation [3].
Discordant data are available on lymph nodes (LN) [3; 15].
The i-Score is on a 5-point scale (I0 = not involved segment; I4,
severe radiological inflammatory activity).
To detect the disease localization in a given intestinal segment (I1 vs I0),
we consider the wall thickening (>3 mm),
while the presence of inflammatory activity (I2 vs I1) is characterized by wall enhancement.
The presence of at least two wall layers is the target sign in active inflammation (I Score ≥2) [4].
The wall enhancement is absent when comparable to not involved loops and marked (grade 2) when similar to vessels.
The extraintestinal findings have the same weight in categorization of active inflammation (I2,
I3,
I4.
Comb sign is graded as 2 for increased number of enlarged vasa recta [16].
LN are pathological (grade 1) if increased in number,
diameter (short axis > 10 mm),
or both (grade 2).
If perivisceral stranding and creeping fat are diffuse,
they are graded 2.
I2 has up to 2 extraintestinal findings graded 1.
I3 has more than 2 extraintestinal findings with maximum grade of 2,
or 2 extraintestinal findings with one graded 2.
The I4 has more than two extraintestinal findings and at least two graded as 2.
The I-Score is summarized in Table 1.
Finally,
stenosis was defined by measuring the diameters of involved segments (DIS) and close non-involved segments (DNIS) as (DIS / DNIS)<0.5 [17].
Patient Selection
Inclusion criteria: patients older than 18 y.o.
with known CD who underwent to CTE between June 2017 and August 2018 and endoscopy within 2 weeks from CTE.
All patients included had also laboratory data within 2 weeks from CTE: C-reactive protein (CRP) and fecal calprotectin (FC) were recorded.
Exclusion Criteria: previous intestinal resection,
contraindication to contrast material,
lack of clinical,
laboratory or endoscopic data.
CTE Scanning Protocol
Patient preparation included the administration of polyethylene glicole 1 hour before the examination (Water solution of Macrogol per os,
Macro P,
Alfasigma,
Milan,
Italy) and a spasmolytic agent (Hyoscine Butylbromide,
20 mg iv,
Buscopan,
Boehringer Ingelheim,
Barcelona,
Spain).
CtE examinations were performed with a 64-row LightSpeed VCT (GE Healthcare,
Milwaukee,
WI) with the following parameters: 120 kVp,
modulated mA,
pitch: 0.984:1,
kernel standard,
slice thickness (ST) 2.5 mm,
slice spacing (SP) 1.25 mm.
The enterographic phase was acquired at 45-55 s after administration of contrast material (Iopamidol 370 mgI/ml,
1.5 ml/kg body weight at 2,5 – 4 ml/s; Iopamiro 370,
Bracco,
Milan,
Italy),
and completed with multiplanar reconstructions.
Image evaluation,
Reference Standard and Data Analysis
The i-Score was calculated on intestinal and extraintestinal findings (Table 1) by two radiologists in consensus (10 and 5 years of experience in gastrointestinal imaging).
Endoscopies were evaluated as per SES-CD score in each involved segment,
and active disease was considered as SES-CD ≥3.
I-Scores and stenoses were correlated with endoscopic data with ROC Curve Analysis.
I-Scores were correlated with laboratory data with non-parametric tests.
Statistical analysis was performed with MedCalc v12.5 (MedCalc Software,
Ostend,
Belgium) and significant p were set at p<0.05.