The Relationship Between the Endoscopic Healing Index, Fecal Calprotectin, and Magnetic Resonance Enterography in Crohn’s Disease

Abstract

Introduction:

The serum-based endoscopic healing index (EHI) test identifies endoscopic Crohn’s Disease (CD) activity. Data are lacking on the relationship between EHI with other endpoints. We assessed the relationship between EHI and simplified Magnetic Resonance Index of Activity (sMaRIA).

Methods:

Data were prospectively collected on patients with CD with either an EHI or fecal calprotectin (FCAL) within 90 days of magnetic resonance enterography (MRE). Diagnostic accuracy was assessed using area under the receiver operator characteristics (AUROC). Proportions with any, severe, and terminal ileum MR inflammation were compared above/below identified thresholds for both EHI and FCAL.

Results:

A total of 241 MREs paired to either EHI or FCAL from 155 patients were included. Both EHI and FCAL had similar accuracy to diagnose inflammation (AUROC: EHI:0.635–0.651, FCAL: 0.680–0.708). Optimal EHI values were 42 and 26 for inflammation on MRE and endoscopy, respectively. Patients with EHI≥42 (100% vs. 63%, p=0.002), FCAL>50μg/g (87% vs. 64%, p<0.001) and FCAL>250μg/g (90% vs. 75%, p=0.02) had higher rates of sMaRIA≥1 compared to lower values. EHI differentiated ileitis numerically more than FCAL (delta: 24–25% vs. 11–21%). Patients with FCAL≥50μg/g had higher rates of severe inflammation compared to FCAL<50μg/g (75% vs. 47%, p<0.001) whereas smaller differentiation existed for EHI threshold of 42 (63% vs. 49%, p=0.35).

Conclusion:

Both EHI and FCAL were specific in their confirmation of inflammation and disease activity on MRE in patients with CD. However, MRE detected inflammation was frequently present in the presence of low EHI and FCAL in similar proportions.

Summary:

EHI is a serum biomarker to detect endoscopic inflammation in CD. MRE measures transmural and small bowel disease activity, which are limitations of both stool studies and endoscopic evaluations. The relationship of FCAL and EHI to MRE both displayed adequate diagnostic accuracy for radiographic inflammation.

Introduction:

Crohn’s Disease (CD) is an immune-mediated, chronic inflammatory condition which can affect any part of the gastrointestinal tract. One third of patients have disease limited to the small bowel and 70% of patients have some degree of involvement of their small bowel.^1,2 Mucosal healing is a key therapeutic goal in patients with IBD rather than only resolution of clinical symptoms.³

Ileocolonoscopy is the current gold standard for objective disease assessment for both diagnosis and monitoring CD activity.⁴ Limitations of endoscopy include the limited ability to evaluate small bowel disease extent, transmural disease and the invasive nature of the procedure. The utility of radiographic imaging of the abdomen has been recognized for its ability to examine small bowel disease extent and transmural involvement. For this reason, it is routinely used in practice. Both ileocolonoscopy and biomarker surrogates are limited in assessing small bowel disease compared to magnetic resonance enterography (MRE).^5,6 The Simplified Magnetic Resonance Index of Activity (sMaRIA) is validated to detect and characterize CD luminal activity, severity, and therapeutic response with reproducibility even without gadolinium-enhanced sequences.^7,8

A multimodal approach to patients with CD is necessary taking together the patient’s clinical, endoscopic, radiological, and biomarker data. The most frequently relied upon biomarker in CD is fecal calprotectin (FCAL).⁹ However, stool-based tests often have poor patient compliance.^10–12 The endoscopic healing index (EHI) (Monitr; Prometheus Biosciences, San Diego, CA, USA) is a commercially available peripheral serum-based biomarker panel that identifies endoscopic CD activity. The EHI analyzes 13 biomarkers involved in the pathogenesis and pro-inflammatory cascade in CD; it has been validated to detect endoscopic inflammation and has performed similarly to FCAL.^6,13,14 Biomarker assessment of disease activity is inherently less invasive and time consuming than endoscopy or MRE. Due to barriers in collecting fecal specimens and patient dissatisfaction, the EHI test offers a serum-based alternative. However, there are no data on the relationship between EHI and radiographic outcomes in CD.

To address this knowledge gap, this study explored the relationship between EHI and validated radiologic disease activity assessments in CD. In addition, we analyzed the relationship between FCAL and radiologic outcomes and explored relationships between both biomarkers and endoscopic and symptom-based disease activity assessments.

Materials and Methods:

Study Population and Design:

Data were extracted on patients with CD with either an EHI or FCAL within 90 days of MRE at New York Presbyterian – Weill Cornell Medical Center. The demographic and clinical data were extracted using a prospectively maintained biobank from December 2015 to March 2021. These included prospective clinical and endoscopic scores, medication use, and laboratory data. Missing data were further collected by manual extraction of electronic medical records. Both global and segmental sMaRIA scoring were performed on MREs by a blinded expert radiologist. For a single patient, each discrete MRE was included if it was paired with either EHI or FCAL that had not been previously pair to another MRE within the same patient. Data was analyzed for all pairs (i.e., using multiple MRE/biomarker pairs) per patient and repeated using one pair per patient. Each per patient analysis was evaluated for those with the shortest interval between EHI and MRE. All therapeutic interventions were reviewed during the 90-day period and confirmed that no intervening changes were performed between biomarker and radiographic assessments. Endoscopic data were collected prospectively, and the Simple Endoscopic Score for Crohn’s Disease (SES-CD) was scored by expert IBD endoscopists.

Outcomes and Definitions

The primary objective was to assess the diagnostic accuracy of the EHI for inflammation visualized on MRE.

Any inflammation on MRE was defined as sMaRIA ≥ 1 and severe inflammation on MRE was defined as sMaRIA ≥ 2.^7,15 Radiographic ileal inflammation was defined as Terminal Ileum (TI) sMaRIA ≥ 1. The secondary objectives characterized the relationship between EHI and active endoscopic inflammation (SES-CD ≥ 5), and active clinical symptoms (Patient-Reported Outcomes 2, [CD PRO-2 ≥ 8]).^16,17,18,19 In addition, analyses were repeated for the relationship between FCAL (> 50μg/g, > 250μg/g) and all listed outcomes. Data on available variables of interest included patient characteristics (age and sex), disease characteristics (prior surgeries, disease-related complications, Montreal classification), current and prior treatments (corticosteroids, immunosuppressives, biologics), and clinical disease activity (CD PRO-2).

Magnetic Resonance Enterography

The MRE studies were performed on 1.5T (Aera, Siemens) or 3T (Skyra, Siemens) systems.

To obtain adequate bowel distension, patients were required to drink non-absorbable oral contrast media (Breeza, Beekley Media). An anti-peristaltic agent, glucagon, was also administered intravenously prior to scanning. Patients were scanned in the prone position. The routine scanning sequences include coronal single-shot fast spin echo (HASTE), axial HASTE, coronal balanced steady state free precession (TrueFISP), pre-contrast coronal fat-suppressed 3D T1-weighted gradient recalled echo (VIBE), multiple phases of dynamic contrast-enhanced coronal VIBE, axial VIBE post-contrast and axial diffusion weighted imaging.

The sMaRIA for each segment is calculated according to the formula validated by Ordás et al. [sMaRIA = (1x mural thickening >3mm) + (1x edema with presence of T2 hyperintensity) + (1x fat stranding) + (2x mucosal ulceration)]. The global sMaRIA is obtained by the summation the sMaRIA for each segment: distal ileum, ascending colon, transverse colon, descending colon, sigmoid colon, and rectum.

Statistical Analysis

We assessed the diagnostic accuracy of both EHI for any or severe inflammation identified on MRE using area under the receiver operator characteristic (AUROC) curve analyses. The optimal threshold EHI was identified using Youden’s Index. Above listed established FCAL thresholds were selected. Proportions with inflammation on MRE were compared between groups above and below the identified EHI and FCAL thresholds. Analyses were repeated between EHI and active endoscopy, fecal calprotectin, and clinical symptoms.

Data analysis included descriptive statistics for continuous variables and percentages for categorical variables. To compare outcomes between groups, either the chi-square test or the Fisher exact test was used as appropriate for the analysis. P values ≤ 0.05 were considered significant. Analysis was conducted using STATA (StataCorp. College Station, TX).

Ethics

All study parameters were approved by the institutional review board at Weill Cornell Medical College prior to study initiation. Informed consent was provided by each patient.

Results:

Patient Characteristics and Outcomes

A total of 241 MREs uniquely paired to either EHI or FCAL from 155 patients with an established diagnosis of CD were included. This included 65 unique EHI-MRE pairs from 59 patients and 176 unique FCAL-MRE pairs from 96 patients. Thirty patients had both FCAL and EHI paired to the same MRE for comparison. The mean time between EHI to MRE was 28 (± 25) days, and the mean time between FCAL and MRE was 24 (± 24) days. The mean age was 41.8 (± 15.1) years and 50.3% were female. In addition, 83.2% of patients had small bowel involvement, including L1 and L3 phenotypes, and 29.7% had penetrating disease. The mean EHI for L1–3 was 30.4, 29.1, and 30.3 respectively. The mean FCAL for L1–3 was 428.9, 437.5, and 429.4 respectively. The median number of prior biologics was 2 (IQR, 1–3, Table 1). Seventy nine percent and 57% had an EHI within 90 days of clinical or endoscopic assessments, respectively. Forty seven percent of patients had an endoscopy within 90 days of collection of the FCAL. In both groups with either EHI or FCAL, all patients had MREs available.

Table 1:

Demographic and Clinical Characteristics of Patients with Crohn’s Disease

	Total (n)	Endoscopic Healing Index [n, (%), [avg EHI]]	Fecal Calprotectin [n, (%)]
Total Subjects	155	59	96
Female	78	27 (46%) [29.6]	51 (53%)
Male	77	32 (54%) [30.3]	45 (47%)
Age average	41.85	43.60	40.70
BMI average	41.85	25.60	24.70
Montreal Classification
A1 – 16 years or younger	31	10 (17%)	21 (22%)
A2 – 16-40 years of age at diagnosis	90	29 (49%)	61 (64%)
A3 - >40 years old	34	20 (34%)	14 (15%)
L1 - Terminal Ileum	54	20 (34%) [30.4]	34 (35%)
L2 - Colon	24	9 (15%) [29.1]	15 (16%)
L3 - Ileocolonic	74	29 (49%) [30.3]	45 (47%)
L4 - Upper GI	3	1 (2%) [52.0]	2 (2%)
B1 - nonstricturing, nonpenetrating	49	14 (24%)	35 (36%)
B2 - stricturing	60	21 (36%)	39 (41%)
B3 - penetrating	46	24 (41%)	22 (23%)
Perianal involvement	40	17 (29%)	23 (24%)
Social History
Smoker	51	24 (41%)	27 (28%)
Prior History
IBD related surgery history	82	34 (58%) [30.3]	48 (50%)
Ostomy history	15	5 (8%)	10 (10%)
IBD Hospitalizations, avg	1	0.9	1
Biologic History	119	48 (81%)	71 (74%)
Biologics, avg number	1	1.7	1
Currently on Corticosteroid	13	8 (14%)	15 (16%)
Currently on 5-ASA	21	8 (14%)	13 (14%)
Currently on Immunomodulator	16	14 (24%)	2 (2%)

Diagnostic Accuracy of EHI and FCAL for Active Inflammation on MRE

EHI had adequate diagnostic accuracy, with an area under the receiver operator characteristic curve (AUROC) of 0.635 (95% Confidence Interval [CI], 0.501 – 0.769) for sMaRIA ≥ 1. EHI had adequate diagnostic accuracy in the unique per patient analysis with an area under the receiver operator characteristic curve (AUROC) of 0.651 (95% CI, 0.51 – 0.79) for sMaRIA ≥ 1. The optimal threshold identified for the EHI was 42 with a sensitivity of 34% and specificity of 100%. The established threshold of FCAL ≥ 50μg/g had a 75% sensitivity and 56% specificity and FCAL ≥ 250μg/g had 37% sensitivity and 83% specificity. In the subgroup of 30 patients with both EHI and FCAL within 90 days of MRE, we found no significant difference in the proportions of active disease activity. Table 2 outlines the positive and negative predictive values for EHI and FCAL for each outcome described.

Table 2:

Relationships Between Biomarkers and Radiographic Inflammation Assessed with MR Enterography

		sMaRIA ≥ 1	TI sMaRIA ≥ 1	sMaRIA ≥ 2
EHI: 42	PPV	100%	75%	62.50%
	NPV	36.70%	49%	51%
FCAL: 50 μg/g	PPV	86.80%	70%	75%
	NPV	36.40%	51%	53%
FCAL: 250 μg/g	PPV	90%	71%	84%
	NPV	25%	40%	42%

EHI: Endoscopic Healing Index, FCAL: Fecal Calprotectin, sMaRIA: The Simplified Magnetic Resonance Index of Activity, PPV: positive predictive value, NPV: negative predictive value

Relationship between Active MR Inflammation (sMaRIA ≥ 1) and EHI and FCAL thresholds

Above an EHI of ≥ 42, a significantly higher proportion of patients had inflammation on MRE (sMaRIA ≥ 1), 100% (16/16) compared to patients with EHI < 42, 63% (31/49), p = 0.002 (delta: 37%, Figure 1).

Figure 1:

The Endoscopic Healing Index and Fecal Calprotectin Compared to Any Inflammation on Magnetic Resonance Enterography

Abbreviations: FCAL: Fecal Calprotectin, EHI: Endoscopic Healing Index, sMaRIA: The Simplified Magnetic Resonance Index of Activity

FCAL thresholds were associated with a numerically lower differentiation of active MRE inflammation as compared to EHI. For example, a 37% difference of proportions existed when using an EHI threshold of 42. For FCAL this difference ranged from 15% (FCAL 250μg/g) to 23% (FCAL 50μg/g). Specifically, higher proportions of patients had sMaRIA ≥ 1 when comparing patients with FCAL above and below 50μg/g (87% [105/121] vs. 64% [35/55], p < 0.001) and 250μg/g (90% [52/58] vs. 75% [88/118], p = 0.02). When analyzing data at a per patient level, similar conclusions were found using an EHI threshold of 42 (p = 0.002) and FCAL of 50μg/g (p = 0.006). However, FCAL of 250 μg/g did not significantly differentiate outcomes (p = 0.15).

Relationship between Ileal MR Inflammation (TI sMaRIA ≥ 1) and EHI and FCAL thresholds

EHI ≥ 42 numerically differentiated inflammation of the ileum on MRE (TI sMaRIA ≥ 1) compared to patients with EHI < 42 (75% [12/16] vs. 51% [25/49], p = 0.08, Figure 2). Similarly, when analyzing on a per patient basis, EHI ≥ 42 differentiated inflammation of the ileum on MRE compared to patients with EHI < 42 (73% (11/15) vs. 48% (21/44), p = 0.08).

Figure 2:

The Endoscopic Healing Index and Fecal Calprotectin Compared to Inflammation of the Terminal Ileum on Magnetic Resonance Enterography

Abbreviations: FCAL: Fecal Calprotectin, EHI: Endoscopic Healing Index, sMaRIA: The Simplified Magnetic Resonance Index of Activity, TI: Terminal Ileum

FCAL thresholds differentiated ileal inflammation numerically less than EHI differentiated TI MRE inflammation. For example, a 24–25% difference of proportions existed when comparing EHI above and below 42. For FCAL this difference ranged from 11% (FCAL 250) to 21% (FCAL 50). Specifically, significantly higher proportions of patients had TI sMaRIA ≥ 1 comparing patients above and below FCAL ≥ 50μg/g (all pairs: 70% [85/121] vs. 49% [27/55], p = 0.007, per patient: p = 0.02). However, proportions of patients with TI sMaRIA ≥ 1 were similar comparing patients above and below FCAL of ≥ 250μg/g (all pairs: 71% [41/58] vs. 60% [71/118], p = 0.2, per patient: p = 0.4).

Relationship between Severe MR Inflammation (sMaRIA ≥ 2) and EHI and FCAL thresholds

EHI ≥ 42 did not significantly differentiate severe inflammation on MRE (sMaRIA ≥ 2) compared to patients with EHI < 42 (63% [10/16] vs. 49% [24/49], p = 0.35, Figure 3) and similar conclusions existed for per patient analyses (67% [10/15] vs. 50% [22/44], p = 0.21).

Figure 3:

The Endoscopic Healing Index and Fecal Calprotectin Compared to Severe Inflammation on Magnetic Resonance Enterography

Abbreviations: FCAL: Fecal Calprotectin, EHI: Endoscopic Healing Index, sMaRIA: The Simplified Magnetic Resonance Index of Activity

FCAL thresholds differentiated active inflammation numerically greater than EHI differentiated MRE inflammation. For example, a 14 – 17% difference of proportions existed when comparing EHI above and below 42. For FCAL this difference ranged from 26% (FCAL 250μg/g) to 28% (FCAL 50μg/g). Specifically, higher proportions of patients had sMaRIA ≥ 2 when comparing above and below FCAL ≥ 50μg/g (all pairs: 75% [91/121] vs. 47% [26/55], p < 0.001, per patient: p = 0.002) and FCAL ≥ 250μg/g (all pairs: 84% [49/58] vs. 58% [68/118], p < 0.001, per patient: p = 0.007).

Endoscopy

A subgroup of 37 patients had an endoscopy within 90 days of EHI collection. EHI had good diagnostic accuracy with the AUC = 0.726 (95% CI, 0.524–0.928) for endoscopic disease activity. The optimal threshold for detection of endoscopic activity was EHI 26 (sensitivity 89% and specificity 58%). EHI ≥ 26 differentiated endoscopic disease activity compared to patients with lower EHI values, (40% [8/20] vs. 6% [1/17], p = 0.02).

A subgroup 45 patients had an endoscopy within 90 days of FCAL collection. FCAL had similar diagnostic accuracy with the AUC = 0.705 (95% CI, 0.515–0.895) for endoscopic disease activity. FCAL ≥ 50μg/g had a sensitivity of 67% and specificity of 45%, and FCAL ≥ 250μg/g had a sensitivity of 50% and specificity of 84%. Both FCAL ≥ 50μg/g (89% [25/28] vs. 59% [10/17], p = 0.02) and ≥ 250μg/g (100% [13/13] vs. 69% [22/32], p = 0.020) differentiated endoscopic disease activity compared to patients below these thresholds.

Clinical Scores

The proportion of patients with sMaRIA ≥1 was similar amongst patients with (68%, 17/25) and without (69%, 18/26) active clinical symptoms (p = 0.9). There is no association between patient reported active clinical symptoms (CD PRO-2 ≥ 8) and either radiographic inflammation (sMaRIA ≥ 1, p = 0.9), endoscopic disease activity (p = 0.3), inflammation measured by EHI (EHI ≥ 42, p = 0.9) or inflammation measured by FCAL (FCAL ≥ 50μg/g, p = 0.9) (Figure 4).

Figure 4:

The Relationship Between Symptom Clinical Scores and Objective Disease Activity

Abbreviations: FCAL: Fecal Calprotectin, EHI: Endoscopic Healing Index, sMaRIA: The Simplified Magnetic Resonance Index of Activity, Simplified Endoscopic Score – Crohn’s Disease (SES-CD), Crohn’s Disease – Patient Reported Outcome (CD-PRO2)

Discussion:

Stool-based testing for detection of disease activity in patients with CD is limited in detection of small bowel disease and it is underutilized due to compliance issues. The EHI was previously validated as a serum-based alternative to detect endoscopic inflammation in patients with CD of all phenotypes, including post-operative CD, in order to fill this gap.^20,21,22,23 While it is important to understand the relationship between biomarkers and endoscopy, the latter has limited access, is invasive and does not adequately measure transmural inflammation or small bowel disease extent. Radiologic measurement of intestinal inflammation addresses these issues. However, limited data exists for biomarkers such as the EHI or FCAL and radiologic disease activity assessments. To address this, this study assessed the diagnostic accuracy of both FCAL and EHI to MRE and endoscopy.

Importantly, optimal EHI and FCAL thresholds were 42 and 50μg/g, respectively. Regardless of either EHI or FCAL cutoffs utilized, a high proportion of patients below threshold also had some degree of inflammation and this was similarly high between EHI and FCAL. However, for endoscopic inflammation, those with low EHI had low false negative rates, whereas the false negative rate for FCAL persisted. This is in the setting a higher threshold for endoscopic disease activity: SES-CD of 5, which notably was the chosen threshold for clinically significant endoscopic disease activity. Both tests had similar numerical differentiation of radiographic inflammation of the terminal ileum, however, FCAL had greater differentiation for severe radiologic inflammation than the EHI. Given the small sample size of patients who had both FCAL and EHI within 90 days of the same MRE, formal comparisons were not feasible. Further studies are needed to validate comparative the diagnostic accuracies of FCAL versus EHI. Symptoms are often not representative of disease activity in CD. Consistent with previous observations, in the current study there was no correlation with active symptoms and EHI, FCAL, endoscopic activity and MRE inflammation.

There are limited data comparing FCAL and sMaRIA scoring. However, FCAL and MRE used in conjunction has adequate diagnostic accuracy for endoscopic and radiographic disease activity.^24–27 Ye et al. directly compared the accuracy of MaRIA and FCAL as surrogates for endoscopic inflammation. Both independently showed adequate diagnostic accuracy of endoscopic inflammation. However, this study also revealed that FCAL and MaRIA scores correlated poorly with one another. Alternatively, Cerillo et al. demonstrated FCAL correlates with the degree of inflammation independently, validated with histopathologic diagnosis.²⁸ The current study confirms adequate diagnostic accuracy of FCAL (both 50μg/g and 250μg/g) for active inflammation in CD on MRE, both total inflammation and that localized to the terminal ileum.²⁹ However, it should be noted that the sensitivity and specificity, 67% and 45% respectively, for FCAL in those who underwent endoscopy in this cohort is lower than previous reports in the literature, often cited as >80% sensitivity.^30,31 This difference in findings may reflect this cohort’s low sample size and the high proportion with small bowel disease. Importantly for small bowel disease, the sample size was smaller for EHI likely underpowering the analysis of EHI diagnostic accuracy of radiographic small bowel disease. It should be noted however, that EHI numerically differentiate TI MRE inflammation more often than FCAL despite only proving significance for FCAL. In addition, this study adds the largest sample size of patients with MRE disease index of activity measured with sMaRIA scoring in relationship to FCAL in CD.

EHI has been well validated for small bowel and general disease activity in CD. Our study is consistent with prior studies, in which sensitivity analyses showed disease location, phenotype, and surgical intervention does not impact the EHI.¹³ There is no prior data comparing objectively scored MRE data with sMaRIA scoring. However, evaluating radiographic inflammation and EHI, our study confirms the diagnostic accuracy at optimal threshold 42, which is similar to previous data. Together these datapoints support using biomarkers, both FCAL and EHI, and MRE to help assess disease activity.

Study limitations include different sample sizes for EHI and FCAL cohorts. Secondly, multiple paired MRE and diagnostic testing (i.e., EHI or FCAL) were used in the same patient, a minority of patients. To address this, we repeated analyses using one pair per patient and results were similarly statistically significant (EHI 42: p = 0.002 and FCAL 50μg/g: p < 0.001). Thirdly, in this study we did not assess histopathology for endoscopic healing nor longitudinal data for patient outcomes. Both must be further studied to determine the relationship with therapeutic monitoring, disease assessment, disease relapse, and radiographic and EHI disease activity assessment. Lastly, the mean time between EHI/FCAL and MRE was short (25 days) which likely did not impact results.

Strengths of the current study include the use of validated objective endoscopic and radiographic disease activity assessments matched to blinded biomarker assessments. To our knowledge, this is the first cohort to assess the relationship of the EHI to assess radiographic disease activity on MRE and the largest cohort to assess the relationship between FCAL and sMaRIA. The use of the EHI and MRE to assess small bowel disease activity creates a distinct advantage to endoscopic evaluation, as frequently the latter can be limited for small bowel disease extent and transmural inflammation. Uniquely, we have validated both the EHI and FCAL with the largest cohorts using the standardized MRE based disease index, sMaRIA scoring, to assess radiographic disease. The use of accessible biomarkers along with MRE data can help assess disease activity and aid in therapeutic management and monitoring clinically.

In summary, both EHI and FCAL were both specific to confirm inflammation on MRE with a high rate of true positive cases, however, similarly high proportions with false negative MRE results existed in both cohorts.

Key Messages:

• Ileocolonoscopy and biomarker surrogates are limited in assessing small bowel disease compared to magnetic resonance enterography.

• There is adequate diagnostic accuracy of EHI and FCAL to confirm inflammation on MRE

• EHI and MRE, in conjunction, may provide an accurate disease assessment of CD activity.