The advent of biologics has shifted the therapeutic goal of Crohn’s disease (CD) to mucosal healing (MH). Selecting Therapeutic Targets in Inflammatory Bowel Disease in 2015 proposed a treat-to-target strategy focusing on early intervention using an effective medication and suggested agreed-upon goals, such as clinical remission, adjunctive use of objective indicators (C-reactive protein and fecal calprotectin), and ultimately endoscopic remission.¹ In the updated Selecting Therapeutic Targets in Inflammatory Bowel Disease-II, the treat-to-target strategy became more detailed.² The former accepted cross-sectional imaging (CSI) as an alternative tool if ileocolonoscopy was inaccessible, such as for deep small bowel. In comparison, the latter recommended capsule endoscopy or balloon-assisted enteroscopy (BAE), emphasizing the evaluation of MH if it could not be reached by ileocolonoscopy. The CSI was referred to as an adjunctive tool for estimating transmural healing (TH).^2,3

CD is characterized by the transmural involvement of the intestine.⁴ There are arguments that TH should be considered a therapeutic goal, going one step further.^5-13 However, TH is not well defined due to inconsistent definitions in each study using CSI or sonography. TH was defined as without any suspicious findings of active inflammation radiologically. However, this could not be called TH in the true sense because whether MH was achieved was unknown.^{6,10-12,14,15} Furthermore, there was a limitation in that MH was assessed by ileocolnoscopy.^13,16-19 For example, in patients with small bowel CD, it is desirable to define TH when MH is verified on BAE and when transmural inflammation is improved on CSI. Therefore, the best way to find concordance or discrepancy between MH and TH is to compare BAE²⁰ and magnetic resonance enterography (MRE) findings in small bowel CD patients segment-by-segment.

MRE is one of the objective tools to measure transmural inflammation.^21-25 The magnetic resonance index of activity (MARIA) developed in 2009 is a commonly used MRE index for evaluating disease activity.^26,27 However, several limitations affect the practical application of MARIA. Relative contrast enhancement, a quantitative item of MARIA, is an involute and time-consuming task. The item of ulcer takes up more weight than other items in the MARIA formula, so inconsistency between the readers might influence the reproducibility. Reflecting this, a simplified MARIA (MARIAs) that could overcome the pitfalls of MARIA was recently constructed. MARIAs can be calculated without contrast enhancement, and the applicability of the simple formula has been proven.^28-31 However, no study has compared it to BAE in patients with small bowel CD.

Therefore, we compared simple endoscopic scores for CD (SES-CD) and MARIAs in matched segments of the ileum based on BAE and MRE in patients with small bowel CD.

1. Patients and follow-up

This was a single-center retrospective cohort study. BAE or MRE is performed every 1 to 2 years to evaluate the therapeutic response of small bowel CD patients in our institution. We enrolled 50 patients with small bowel CD who underwent BAE and MRE concurrently within 3 months from September 2020 to June 2021. The Institutional Review Board of Samsung Medical Center approved the protocol (IRB number: 2020-08-145). The requirement for obtaining informed consent from the patients was waived because only de-identified data were collected.

2. BAE and assessment of disease activity

All procedures were conducted transanally using a single-balloon enteroscope by an experienced endoscopist (S.N.H.). Deep insertion was performed to evaluate skipped lesions in the ileum. If the passage was difficult due to stenosis, the extent of stenosis was estimated using contrast dye and balloon dilatation was performed or not, considering length, combined deep ulceration, and angulation.

The ileum was divided into the terminal ileum (from ileocecal valve to 10 cm proximal), distal ileum (10–50 cm), mid-ileum (50–100 cm), and proximal ileum (proximal location of 100 cm). This study used SES-CD for scoring the endoscopic disease activity in each segment and defined it as ileal SES-CD. Ileal SES-CD was calculated for each segment (terminal ileum, distal ileum, mid-ileum, and proximal ileum) by summing the four endoscopic variables, the same as in the original SES-CD. Each variable was scored from 0 to 3 as follows: the presence and size of ulcers (none, score of 0; diameter of 0.1–0.5 cm, score of 1; 0.5–2 cm, score of 2; and >2 cm, score of 3), the extent of the ulcerated surface (none, score of 0; <10%, score of 1; 10%–30%, score of 2; and >30%, score of 3), the extent of the affected surface (none, score of 0; <50%, score of 1; 50%–75%, score of 2; and >75%, score of 3), and the presence and type of narrowing (none, score of 0; single, can be passed, score of 1; multiple, can be passed, score of 2; cannot be passed, score of 3). Two endoscopists (J.E.N. and S.N.H.) blindly re-executed the SES-CD retrospectively based on reports and endoscopic images for all enrolled segments and then made a consensus on the differences. In addition, an active score of ileal SES-CD (ileal SES-CDa) was used to represent active inflammation using only the size of the ulcer, extent of the ulcer, and ulcerated surface, except for narrowing. Ileal SES-CDa scores ranged from 0 to 9. Index scores of 5 or higher were considered to indicate active disease. Ileal SES-CD scores ranged from 0 to 12. Index scores of 7 or higher indicated severe disease (Fig. 1).

Figure 1. Endoscopic assessment of disease activity in patients with small bowel Crohn’s disease and the MARIAs formula. MARIAs, simplified magnetic resonance index of activity; ileal SES-CD, ileal simple endoscopic score for Crohn’s disease; ileal SES-CDa, active score of SES-CD.

A biopsy was performed for the most severe lesion in each segment during the procedure. If there was no mucosal lesion, a biopsy of the normal mucosa was conducted. Fluoroscopy was captured with opened forceps during the biopsy. Those captured images and endoscopically estimated locations from ileocecal valve were utilized to match the MRE area.

3. MRE and assessment of disease activity

The patients were instructed to drink one bottle of Pico Solution (170 mL; Pharmbio Inc., Seoul, Korea) and 1 L of water at around 8 PM the night before the MRE and take two bisacodyl tablets before bed. On the examination day, the patients took Colyte (1,200 mL; Taejoon Pharma Co., Seoul, Korea) as a contrast medium 1 hour before the appointment. Buscopan was injected before the examination.

Two experienced radiologists (K.D.S. and H.S.K.) matched the MRE segments with those designated by BAE. After matching, they independently measured the radiologic degree of bowel inflammation and made a consensus by mutual discussion of the cases with interobserver variations. The following parameters were collected: wall thickening >3 mm, mural edema (hyperintense signal on fat-saturated T2-weighted images), fat stranding (increased T2-weighted signal intensity in the mesenteric fat adjacent to bowel loops owing to edema/fluid in the perienteric fat), and ulcers. MARIAs was calculated as ([1×thickness >3 mm]+ [1×edema]+[1×fat stranding]+[2×ulcers]). MARIAs scores of ≥1 and ≥2 were defined as active and severe disease, respectively, based on a previous report.²⁸ Other parameters with mural hyperenhancement (G0, equivalent to the normal bowel wall; G1, increased but significantly less than the nearby vascular structures; G2, increased but somewhat less than the nearby vascular structures; G3, approaches that of nearby vascular structures), stricture (luminal narrowing with or without upstream dilatation), sacculation, and fistula/sinus tract were also measured.^32,33

4. Outcomes and data sources

The primary outcome was the correlation between ileal SES-CDa/ileal SES-CD and MARIAs based on BAE and MRE. The secondary outcome was to identify magnetic resonance parameters related to endoscopically active disease. These data and the following data were gathered through a retrospective review of medical records: age at examination, Montreal classification (age at diagnosis, location, perianal disease, and behavior), Crohn’s Disease Activity Index, C-reactive protein, concomitant treatment, and reports and images/videos of BAE and MRE.

5. Statistical analysis

Continuous variables are presented as the median and interquartile range. Categorical variables are presented as numbers and percentages (%). The Spearman rank correlation analysis was used to investigate the correlation between ileal SES-CDa/SES-CD and MARIAs. We illustrated the correlation of ileal SES-CDa/SES-CD according to MARIAs using bar plots. The predictive value and cutoff value of MARIAs for endoscopically active/severe disease were evaluated using area under the receiver operating characteristic curves, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV). Univariable and multivariable logistic regression analyses were used to explore the magnetic resonance parameters related to ileal SES-CDa scores of 5 or higher. Parameters with p-values less than 0.1 were included in the multivariable analysis. The variance inflation factors of the parameters were less than 10. Statistical significance was considered for p-values of less than 0.05. All statistical analyses were conducted using R-4.1.2.

1. Patients and segmental matching process

Fifty patients were enrolled. A total of 142 segments were matched between BAE and MRE. The baseline characteristics are shown in Table 1. The patients had a median age of 34 years on the date of the BAE and a median disease duration of 9 years. Those with the concomitant use of thiopurine and biologics accounted for 74% and 64%, respectively.

Table 1 . Baseline Characteristics of Patients with Small Bowel Crohn’s Disease.

Variable	All patients (n=50)
Male sex	36 (72)
Age at examination, yr	34 (27–42)
Age at diagnosis, yr	24 (20–28)
Disease duration, yr	9 (4–14)
Location
Ileal	22 (44)
Ileocolonic	28 (56)
Behavior
Inflammatory	15 (30)
Stricturing	18 (36)
Penetrating	17 (34)
Perianal disease	29 (58)
Crohn’s Disease Activity Index	82 (30–137)
C-reactive protein, mg/dL	0.10 (0.04–0.26)
Concomitant treatments
Mesalazine	7 (14)
Steroids	1 (2)
Thiopurine	37 (74)
Biologics	32 (64)

Data are presented as number (%) or median (interquartile range)..

Fig. 2 shows an example of matching a lesion observed on BAE to the MRE area. The segment estimated by enteroscopy and the captured fluoroscopic image with opened biopsy forceps were used to correlate with MRE.

Figure 2. An example of matching an ileal lesion between balloon-assisted enteroscopy and magnetic resonance enterography. (A) Proximal ileum, (B) mid-ileum, (C) distal ileum, and (D) terminal ileum.

2. Correlation of BAE and MRE indices

Ileal SES-CDa/SES-CD and MARIAs showed strong associations (R=0.76, p<0.001; R=0.78, p<0.001) (Fig. 3). Ileal SES-CDa ≥5 (98%) showed a higher proportion of MARIAs ≥1 than ileal SES-CDa <5 (30%). The proportion of MARIAs ≥1 was higher in the ileal SES-CD ≥7 (97%) than ileal SES-CD <7 (32%) (Fig. 4).

Figure 3. Correlation between the MARIAs and ileal SES-CDa/ileal SES-CD. MARIAs, simplified magnetic resonance index of activity; ileal SES-CD, ileal simple endoscopic score for Crohn’s disease; ileal SES-CDa, active score of ileal SES-CD.

Figure 4. Distribution of the MARIAs according to ileal SES-CDa and ileal SES-CD. MARIAs, simplified magnetic resonance index of activity; ileal SES-CD, ileal simple endoscopic score for Crohn’s disease; ileal SES-CDa, active score of ileal SES-CD.

The area under the receiver operating characteristic curve of MARIAs for ileal SES-CDa ≥5 was 0.92 (95% confidence interval [CI], 0.88 to 0.97) (Fig. 5). MARIAs index scores of ≥1 and ≥2 identified active disease (ileal SES-CDa ≥5) with a sensitivity of 98% and specificity of 70%, and a sensitivity of 93% and specificity of 75%, respectively. The cutoff value of MARIAs for detecting ileal SES-CDa ≥5 was 2.5, namely 3, due to that MARIAs has no decimal index. MARIAs scores of ≥3 showed a sensitivity of 85%, specificity of 87%, PPV of 73%, and NPV of 94% in predicting ileal SES-CDa ≥5 (Table 2). MARIAs index values of ≥1 (99%) and ≥2 (96%) consistently had a high NPV.

Figure 5. AUROC curve of the MARIAs for active disease defined as an ileal SES-CDa ≥5 and ileal SES-CD ≥7. ileal SES-CD, ileal simple endoscopic score for Crohn’s disease; ileal SES-CDa, active score of ileal SES-CD; AUROC, area under the receiver operating characteristic curve; CI, confidence interval; MARIAs, simplified magnetic resonance index of activity.

Table 2 . Performance of MARIAs for the Prediction of Active Disease Defined as Ileal SES-CDa ≥5 and Ileal SES-CD ≥7.

Endoscopic assessment	Cutoff	Sensitivity	Specificity	PPV	NPV
Ileal SES-CDa ≥5	MARIAs ≥1	40/41 (98)	71/101 (70)	40/70 (57)	71/72 (99)
	MARIAs ≥2	38/41 (93)	76/101 (75)	38/63 (60)	76/79 (96)
	MARIAs ≥3	35/41 (85)	88/101 (87)	35/48 (73)	88/94 (94)
Ileal SES-CD ≥7	MARIAs ≥1	37/38 (97)	71/104 (68)	37/70 (53)	71/72 (99)
	MARIAs ≥2	36/38 (95)	77/104 (74)	36/63 (57)	77/79 (98)
	MARIAs ≥3	33/38 (87)	89/104 (86)	33/48 (69)	89/94 (95)

Data are presented as number/number (%)..

MARIAs, simplified magnetic resonance index of activity; ileal SES-CD, ileal simple endoscopic score for Crohn’s disease; ileal SES-CDa, active score of ileal SES-CD; PPV, positive predictive value; NPV, positive predictive value..

The area under the receiver operating characteristic curve of MARIAs for ileal SES-CD ≥7 was 0.92 (95% CI, 0.87 to 0.97) (Fig. 5). MARIAs index scores of ≥1 and ≥2 detected severe disease (ileal SES-CD ≥7) with a sensitivity of 97% and specificity of 68%, and a sensitivity of 95% and specificity of 74%, respectively. The cutoff value of MARIAs for ileal SES-CD ≥7 was 2.5, the same as for ileal SES-CD ≥5. MARIAs ≥3 showed a sensitivity of 87%, specificity of 86%, PPV of 69%, and NPV of 95% in predicting ileal SES-CD ≥7 (Table 2). A high NPV of MARIAs index scores of ≥1 (99%) and ≥2 (98%) was found.

3. Magnetic resonance parameters associated with endoscopically active disease

In the multivariable analysis, segmental mural hyperenhancement (G1: odds ratio [OR], 12.89; 95% CI, 1.40 to 119.08; G2-3: OR, 22.05; 95% CI, 2.14 to 227.72), fat stranding (OR, 5.77; 95% CI, 1.52 to 21.86), and an ulcer (OR, 4.87; 95% CI, 1.46 to 16.26) were identified as independent factors for ileal SES-CDa ≥5 (Table 3).

Table 3 . MR Parameters with Independent Predictors for Ileal SES-CDa ≥5.

MR parameter	Description	Total	Ileal SES-CDa ≥5	Univariate OR (95% CI)	p-value	Multivariate OR (95% CI)	p-value
Wall thickening	No	74	2	1		1
	Yes	68	39	48.41 (10.97–213.73)	<0.001	1.09 (0.06–19.69)	0.951
Segmental mural hyperenhancement	Grade 0	68	1	1		1
	Grade 1	31	10	31.91 (3.86–264.02)	0.001	12.89 (1.40–119.08)	0.024
	Grade 2–3	43	30	154.62 (19.34–1,236.41)	<0.001	22.05 (2.14–227.72)	0.009
Mural edema	No	80	3	1		1
	Yes	62	38	40.64 (11.51–143.49)	<0.001	2.39 (0.28–20.50)	0.426
Fat stranding	Absent	115	18	1		1
	Present	27	23	30.99 (9.57–100.33)	<0.001	5.77 (1.52–21.86)	0.010
Ulcers	Absent	97	8	1		1
	Present	45	33	30.59 (11.49–81.49)	<0.001	4.87 (1.46–16.26)	0.010
Stricture	No	113	21	1		1
	Yes	29	20	9.74 (3.89–24.40)	<0.001	0.75 (0.17–3.40)	0.712
Sacculation	No	116	22	1		1
	Yes	26	19	11.60 (4.34–31.00)	<0.001	1.37 (0.31–6.00)	0.676
Fistula/sinus tract	No	139	39	1
	Yes	3	2	5.13 (0.45–58.18)	0.187

MR, magnetic resonance; ileal SES-CDa, active score of ileal simple endoscopic score for Crohn’s disease; OR, odds ratio; CI, confidence interval..

This study found that ileal SES-CDa/SES-CD based on BAE was correlated with MARIAs based on MRE in patients with small bowel CD. The sensitivity and NPV of MARIAs for endoscopically active or severe segments ranged from 85% to 99%. In comparison, the specificity and PPV ranged from 53% to 87%. These results suggest that endoscopically activity disease was correlated with MRE findings. However, transmural inflammation might remain, even in some patients with inactive disease based on BAE. In our study, the cutoff value of MARIAs for predicting ileal SES-CDa ≥5 and ileal SES-CD ≥7 was 2.5, which was rounded to 3. These are slightly higher than the known cutoff value²⁸ of 1/2 for active/severe disease.

MARIA has complex formulas and might not be calculated depending on the MRE protocol. Recently, easily applicable MARIAs was introduced²⁸ and its usefulness compared to ileocolonoscopy was verified.^29,30 However, the validation of these new indices compared to BAE in small bowel CD patients has yet to be reported. Thus, we performed this study to address this knowledge gap and demonstrate the correlation between MARIAs and ileal SES-CDa/SES-CD. Notably, depending on the high NPV of MARIAs, if there is no evidence of active inflammation based on MRE, BAE might be applied more liberally or substituted with less invasive tools, such as fecal calprotectin. However, mural hyperenhancement, an independent parameter of MRE associated with active disease in our study, is not included in MARIAs. Thus, caution is needed when applying MARIAs. Another strength of the study was the matching process. Compared to the large intestine, it is difficult to match the location exactly between BAE and MRE in a segment-by-segment manner in the small intestine. Hence, we utilized a location endoscopically estimated from ileocecal valve and captured fluoroscopic images with an indicator. Furthermore, two gastroenterologists and two radiologists (K.D.S. and H.S.K.) blindly interpreted the BAE and MRE and tuned interobserver variations.

Our findings for the correlation between MARIAs and BAE-based ileal SES-CDa/SES-CD are comparable to previous research on the correlation between MRE and BAE/capsule endoscopy in small bowel lesions. In small bowel CD patients, MRE findings suggesting active lesions (increased contrast enhancement, wall thickening, submucosal edema, deep mucosal fissure, or extra-mucosal hypervascularity) identified active lesions of the small bowel (SES-CDa ≥5) with a sensitivity of 82.4% and a specificity of 87.6%.³⁴ The sensitivity and specificity of original MARIA scores of ≥11 (severe disease) for detecting SES-CDa ≥5 were 78.3% and 98%, respectively.²³ MARIA with a cutoff value of 7.6 for predicting active disease based on video capsule endoscopy (Lewis score ≥135) showed a sensitivity of 82.9% and specificity of 58.6%.³⁵ Those relative comparisons support the usability of MARIAs for identifying endoscopically active lesions in small bowel CD patients.

Previous studies comparing MH with TH were based on ileocolonoscopy and CSI/sonography, although they included patients with small bowel involvement.^{5,9,15,16,18,36,37} Since deep small bowel lesions skipped the terminal ileum could be missed by ileocolonoscopy,^19,38 inflammation might remain in the deep small bowel even if it was judged to be MH.³⁹ This defect might cause an underestimation in MH compared to TH. Due to this inconsistency, the proportion of TH in patients with MH confirmed by ileocolonoscopy was reported to vary from 27% to 70%.^{5,9,13,16,18,36,37} Several studies reported decreased risks of hospitalization and surgery in CD patients showing radiologic response/radiologic healing.^10-14 We also found that the NPV of MARIAs was high. Although radiologic healing is defined exclusively by imaging findings without endoscopic evaluation, by relying on a high NPV of MARIAs, an endoscopically inactive disease might be suspected if radiologic healing is achieved. Further studies are needed to determine whether radiologic response/radiologic healing alone can represent MH. However, after the administration of biologics for at least 1 or 2 years, the radiologic response rate on CSI was low, at about 29% to 37%.^10,11 Radiologic healing is seen in about 3% to 29% of cases on CSI^{9,11-13,16,36,37} and 14% to 42% on sonography.^5,6,15,17,36 In addition, after a median of 13 months of biologics treatment, MH (ileal SES-CDa <5) of the small bowel was 35%, which was lower than that of the colon at 79%. That is, the actual rate of achieving radiologic healing is low and the small bowel requires more time for MH.⁴⁰ Considering this clinical progress of patients with small bowel CD, BAE is crucial for evaluating mucosal lesions in these patients.⁴¹

In our cohort, among 77 segments with ileal SES-CD <3 (MH or no definitive ulceration), 13 (17%) segments showed transmural inflammation (MARIAs ≥1). Conversely, in patients with confirmed TH (MARIAs 0), mucosal inflammation (SES-CD ≥3) was observed in eight of 72 segments (11%). Takenaka et al.²³ also observed this discrepancy (7.9% and 4.6%, respectively) when comparing BAE with MRE. Whether these discrepancies are related to long-term outcomes requires advanced studies based on BAE and CSI.

Our study had some limitations. Due to its retrospective nature, there were various intervals between BAE and MRE within 3 months. We utilized fluoroscopic images and biopsy forceps as an indicator and estimated the location based on BAE to match each segment of BAE and MRE. Still, there might have been some errors since the small bowel is not a fixed organ. The analyses were based on segments. Further studies using a large sample size and per-patient design are necessary due to our relatively small cohort size. This study also lacked long-term outcomes according to BAE and MRE findings.

In summary, we demonstrated the applicability of MARIAs compared to BAE-based ileal SES-CDa/SES-CD. MARIAs, together with BAE, is expected to help assess disease activity in small bowel CD patients.

No potential conflicts of interest relevant to this article were reported.

Study concept and design: J.E.N., S.N.H., K.D.S., H.S.K. Data acquisition, analysis, or interpretation: J.E.N., H.S.K., S.N.H., K.D.S. Drafting of the manuscript: J.E.N., H.S.K., S.N.H., K.D.S. Critical revision of the manuscript for important intellectual content: S.N.H., K.D.S., J.E.N., E.R.K., Y.H.K., D.K.C. Statistical analysis: J.E.N., S.N.H. Approval of final manuscript: all authors.