Faecal Biomarkers in Inflammatory Bowel Diseases: Calprotectin Versus Lipocalin-2—a Comparative Study

Abstract

Background and Aims

Faecal biomarkers, particularly calprotectin [FCAL], have become important diagnostic and monitoring tools in inflammatory bowel diseases [IBD]. As FCAL is mainly produced by neutrophils, we hypothesised that faecal lipocalin-2 [FLCN2], also expressed by intestinal epithelial cells [IEC], could be beneficial in specific clinical situations.

Methods

We compared clinical and endoscopic activity-related correlations between FCAL and FLCN2, assayed from the same sample, in a cohort of 132 patients (72 Crohn’s disease [CD]) and 40 controls. A detailed analysis of cellular origins was done by confocal microscopy and flow cytometry. To evaluate the potential to detect low-grade inflammation, we studied faecal and tissue concentrations in a cohort with clinical, endoscopic, and histological remission.

Results

There was an excellent correlation between FCAL and FLCN2 [r_S = 0.87, p <0.001] and comparable sensitivity and specificity to predict clinical and endoscopic disease activity, with optimal thresholds for endoscopic activity of 73.4 and 1.98 µg/g in ulcerative colitis [UC] and 78.4 and 0.56 µg/g in Crohn’s disease for FCAL and FLCN2, respectively. Strong co-expression of both proteins was observed in granulocytes and macrophages. IECs expressed LCN2 but not CAL. In our IBD cohort in deep remission neither FCAL nor FLCN2 was different from controls; yet mucosal LCN2 but not CAL expressions remained elevated in the rectum of UC and the ileum of CD patients.

Conclusions

This study corroborates the diagnostic equivalence of FLCN2 and FCAL in IBD. In remission, persistent mucosal overexpression renders LCN2 an attractive candidate for molecular inflammation warranting further investigation.

Inflammatory bowel disease, IBD, faecal calprotectin, lipocalin 2, faecal biomarker

1. Introduction

Inflammatory bowel diseases [IBD] involve the two major clinically defined entities Crohn’s disease [CD] and ulcerative colitis [UC].^1,2 Epidemiological data clearly indicate a sustained increase in their incidence and prevalence worldwide,³ particularly in urban areas, highlighting the importance of environmental risk factors as a driving force.^3,4 Poor disease control in IBD patients results in unfavourable quality of life and drives the risk of intestinal and extraintestinal manifestations and complications.⁵ The past years have seen the approval of several new treatment options in IBD.⁶ Nevertheless, there is as yet no consensus on how to implement these options into specific treatment algorithms. A treat-to-target approach that combines early access to efficacious therapies in combination with a tight follow-up of patients, combining clinical, endoscopic, and biochemical treatment targets as well as a structured algorithm of how to escalate or deescalate therapy if treatment targets are or are not reached, appears prudent.⁷ As such, faecal biomarkers will gain further importance.

With a sensitivity of 89.5% and a specificity of 89.5%,⁸ faecal calprotectin [FCAL] has evolved as the most widely used marker to distinguish between inflammatory and functional gastrointestinal disease, and its value in the management and treatment guidance in IBD is undisputed.^8,9 Calprotectin [CAL] is produced mainly by neutrophils, making up to 60% of their cytosolic protein content.^10,11 The biological role of CAL is related to its structure, with S100A8 and S100A9 forming heterooligomers capable of binding zinc.¹² FCAL indicates neutrophilic influx from the mucosal lining into the gut lumen during acute intestinal inflammation.¹³ Despite many advantages of FCAL, such as its inherent stability, there are also some limitations. Day-to-day and even daytime variations of FCAL concentrations have been described^14,15 as well as a degree of age-variability.¹⁶ Furthermore, FCAL represents neutrophilic infiltration which is not specific for IBD. Particularly in patients with low-grade inflammation, there is considerable overlap between organic and functional gastrointestinal disease; furthermore, FCAL may underestimate disease activity in isolated small bowel disease.¹⁷ Particularly in remittent IBD, cell types other than neutrophils may be responsible for maintaining a pro-inflammatory tone that drives disease flares when trying to pause therapy.¹⁸ This underscores the need for additional markers of molecular inflammation.

Lipocalin-2 [LCN2], also known as 24p3 or neutrophil gelatinase-associated lipocalin [NGAL], is a secreted glycoprotein and a member of the lipocalin superfamily.¹⁹ Despite its name and initial description as a neutrophil protein,²⁰ LCN2 is produced by various cell types including myeloid and intestinal epithelial cells, which seems particularly important in the context of IBD. LCN2 is strongly induced in response to a wide variety of pro-inflammatory stimuli, like IL-1β, IL-22, or Toll-like receptor [TLR] activation,^21,22 and is secreted into the gut lumen in high concentrations.²³ Here this 25 kD protein acts as an antimicrobial peptide by quenching iron-loaded bacterial siderophores, and we and others have demonstrated that luminal LCN2 has significant effects on control of both gut microbial composition and host inflammation.^24,25Outside the gut, LCN2 turned out a useful biomarker for the early detection of acute and chronic renal failure.²⁶ De Bruyn et al. demonstrated that serum LCN2, which occurs in complex with matrix metalloproteinase-9, correlates with endoscopic activity in both CD and UC.^27,28

Despite increasing evidence that LCN2 may be useful as a biomarker in IBD,^23,29,30 in the present study we sought to examine commonalities and differences, potential advantages, and disadvantages, of faecal LCN2 [FLCN2] over FCAL in human IBD.

2. Methods

2.1. Study subjects

This study comprised three subprojects, each of them with separate patient cohorts. First, for a head-to-head comparison of FLCN2 and FCAL we prospectively collected stool samples from 132 patients with known IBD [60 UC and 72 CD] as well as from 40 healthy controls. The Harvey-Bradshaw index [HBI] and the partial Mayo score [pMayo] were assessed in CD and UC patients, respectively. In a subgroup of these patients who underwent ileocolonoscopy, the Simple Endoscopic Score for Crohn’s Disease [SES-CD] and Mayo Endoscopic Score [eMayo] were assessed.

In a second cohort, an in-depth analysis of cell lineage-specific distribution of CAL and LCN2 was performed in patients with active CD and UC, using real-time quantitative polymerase chain reaction [RT qPCR], immunohistochemistry [IHC], immunofluorescence double stainings, and flow cytometry in intestinal biopsies.

To assess the ability of LCN2 and CAL to predict histological inflammation, we performed an endoscopy study in 23 IBD patients in clinical remission [14 UC and 9 CD] and 9 healthy controls. Sequential biopsies from the ileum to the rectum, as well as stool samples, were collected. After assessing histological disease activity, LCN2 was systemically compared with CAL in the faeces and in intestinal biopsies [on mRNA levels using RT-qPCR and on protein levels using IHC].

This study was performed with approval from the ethics committee of the Medical University Innsbruck [EK-No. 1005/2019] and informed consent was obtained from all study subjects.

2.2. Blood sampling and clinical scoring

A routine blood test including C-reactive protein [CRP] and blood count was performed in all individuals and disease activity scores [pMayo for UC and HBI for CD³¹] were calculated. Clinical remission was defined as a pMayo of ≤1 in UC patients and an HBI of ≤4 in CD.

2.3. Endoscopy

For bowel cleansing patients received Pleinvue. Endoscopy was performed by experienced IBD endoscopists [ARM, RK] who also assessed endoscopic activity according to the simple endoscopic score for Crohn’s disease [SES-CD] and the Mayo Endoscopic Score [eMayo]. Biopsies were taken in each segment, namely [i] ileum, [ii] ascending, [iii] transverse, [iv] descending, [v] sigmoid colon, and [vi] the rectum, and transferred into formalin 10% and RNAlater [Qiagen, Hilden, Germany] [Figure 3B in section 3.3].

Histological disease activity was evaluated in a blinded fashion by an experienced gastrointestinal [GI] pathologist [GO] according to adapted variations of the Colonic and Ileal Global Histologic Disease Activity Score [CGHAS] for CD,³² and the Geboes Score [OGS] for UC.³³

2.4. Processing of faecal samples and stability tests

Faecal samples were collected into Sarstedt stool containers and weighed aliquots were immediately stored at -80°C. For the stability experiment, initial processing was performed within 1 h of sample collection by splitting the sample into nine aliquots; further processing was initiated either immediately or after storage at room temperature or at 4°C for 12, 24, 48, and 96 h. Supernatants were stored at -80°C until measurement.

2.5. Enzyme-linked immunosorbent assay

FCAL and FLCN2 were measured in the same faecal samples with commercially available enzyme-linked immunosorbent assay [ELISA] kits according to the manufacturer’s instructions. Briefly, 100 mg of faeces were diluted in 5 ml of the provided extraction buffer, vortexed for 30 s and homogenised for 30 min using an Intelli-Mixer RM-2 vortex mixer. The suspension was centrifuged for 20 min at 10 000 × g. The standard dilutions for FCAL and FLCN2 were 1:50 and 1:30, respectively, and high concentration samples were further diluted as required. Calprest ELISA [REF 9031, Eurospital, Trieste, Italy] was used for FCAL measurement. FLCN2 was assayed using the Lipocalin-2/NGAL Duoset ELISA [RD-DY1757, RandD Systems, USA]. Results were calculated from standard curves.

2.6. Immunofluorescence

Tissue sections were deparaffinised in xylene and rehydrated through graded ethanol. Antigens were unmasked using a pressure cooker and a citrate-based unmasking solution [VectorLabs, CA, USA]. Endogenous peroxidase was blocked using Dako peroxidase block [Agilent, CA, USA]. Slides were stained using specific anti-LCN2 [#HPA002695, Atlas Antibodies, Sweden] and anti-CAL antibodies [#LS-B9953, LSBio, WA, USA]. Targets were visualised with secondary fluorophore conjugated antibodies. Nuclei were counterstained using ProLong Gold antifade reagent with DAPI [Invitrogen, CA, USA]; 200 μm × 200-μm images were acquired on a Zeiss Axioobserver Z1 in combination with a LSM700 confocal laser scanning system containing four lasers with 405, 488, 555, and 654 nanometer wavelengths.

2.7. Preparation of intestinal single cells and flow cytometry

Biopsies were collected from macroscopically inflamed mucosa of patients with IBD and uninflamed mucosa from healthy controls, in RPMI-buffer on ice. After a quick and gentle spin- down, biopsies were transferred into a new tube containing Ca/Mg-free HBSS, 0.5% BSA, 2mM EDTA, and 50 μM DTT, and were shaken at room temterature [RT] for 20 min. After vortexing, the supernatant was poured through a 100-μm strainer into a fresh 50-ml falcon. After several washing steps, cells representing the intraepithelial lymphocyte and intestinal epithelial cell fractions were spun down and resuspended in complete medium. The remaining fragments were picked off the strainer with a pipette, washed, and then digested at 37°C on a shaking platform in RPMI/0.5% BSA containing 128 U/ml collagenase IV [Sigma C1889] and 10 U/ml DNAse II [type V Sigma D8764; both Sigma-Aldrich, Missouri, USA]. After straining and washing, this lamina propria lymphocyte fraction was spun down and resuspended in complete medium. Cell numbers were quantified by a LUNA automated cell counter [Logos Biosystems] before blockage of non-specific binding sites with FC block for 10 min at 4°C [TruStain FcX].

Cells and OneComp eBeads [single-colour compensation controls, eBiosciences] were stained with respective antibodies [see Supplementary Table 1, available as Supplementary data at ECCO-JCC online] and isotype controls according to standard procedures for combined intracellular/surface stainings. Typically, 60 000-150 000 cells were analysed using a FACSVerse [Becton Dickinson, NJ, USA] and FlowJo Software version 10.1.r3 [FlowJo LLC, Ashland, OR, USA]. Antibodies and gating strategy are outlined in Supplementary Table 1] and [Supplementary Figure 2, available as Supplementary data at ECCO-JCC online].

2.8. Gene expression analysis

RNAlater [Qiagen, Hilden, Germany] preserved biopsy specimens were homogenised in lysing buffer using a Precellys® 24 Homogenisator and a Precellys Tissue RNA kit [both Bertin, Montigny-le-Bretonneux, France]. Lysates were applied on RNAeasy columns [Qiagen, Hilden, Germany] and RNA extraction was carried out according to the manufacturer’s instructions. Total RNA concentration was quantified at 260 nm right after isolation, using a nanodrop 1000 [Peqlab, Erlangen, Germany]. The RNA isolation was done in the same week as the biopsies were collected, and RNA was stored at -20°C. RNA was converted to cDNA using an M-MLV reverse transcriptase in combination with hexamer primers [Thermo Fisher Scientific, Waltham, MA]. cDNA sequences were amplified by polymerase chain reaction using gene-specific primers in combination with SYBR-green chemistry. The primers for LCN2 [5’ CTA CGG GAG AAC CAA GGA GC 3‘; 5’ CAC TGG TCG ATT GGG ACA GG 3‘] and the two subunits of CAL S100A8 [5’ TTC TGT TTT TCA GGT GGG GCA 3‘; 5’ ACG TCT GCA CCC TTT TTC CTG 3‘] and S100A9 [5’ TCC TCG GCT TTG AGA GAG TG 3‘; 5’ TGC CCC AGC TTC ACA GAG TA 3‘] were produced by Microsynth [Balgach, Switzerland] and real-time PCR was performed on a Mx3000P cycler [Stratagene, California, USA] instrument.

For denaturation, primer annealing and elongation following PCR conditions were chosen: 95°C for 2 min followed by 40 cycles of 95°C for 30 s, 60°C for 30 s, and 72°C for 30 s. Data were analysed on Stratagene’s MxPro software [Stratagene, California, USA]. Relative expressions were calculated using the delta CT method, using glucuronidase beta as a housekeeping gene.

2.9. Immunohistochemistry

Tissue processing was identical to the procedure described under immunofluorescence. As primary antibodies, anti-LCN2 [#HPA002695, Atlas Antibodies, Sweden] and anti-CAL [#LS-B9953, LSBio, WA, USA] were incubated at 4°C overnight. Targets were visualised using an ImmPRESS HRP-conjugated anti-rabbit, anti-rat, or anti-goat polymeric system [Vector Laboratories, Burlingame, CA] together with 3-amino-9-ethylcarbazole [AEC] as a chromogen. Slides were counterstained with haematoxylin QS [Vector Laboratories, Burlingame, CA] and mounted aqueously. Slides were scanned on an IntelliSite Ultra Fast Scanner [Philips digital pathology, Amsterdam, The Netherlands], exported, and quantified with an ObjectiveView software module [Objective View Pathology, ON, Canada]. An IHC score was assigned to each slide according to the following criteria: 3+, intense diffuse reaction; 2+, moderate diffuse reaction; 1+ intense focal or diffuse weak reaction; 0, no reaction or focal weak reaction. LCN2 was scored in all cells as well as in IEC only. CAL, not present in epithelial cells, was scored in all cells.

2.10. Statistical analysis

Statistical analysis was performed using SPSS 24 [IBM], GraphPad Prism 8.1 software package, Excel 2016 [Microsoft, Washington, USA] and FlowJo software version 10.1.r3 [FlowJo LLC, Ashland, OR, USA]. As our data were not normally distributed, comparison between groups was performed by Mann-Whitney and Kruskal—Wallis tests with Bonferroni post hoc tests. Differences between multiple groups were analysed using one-way analysis of variance [ANOVA] with Holm-Sidak post tests for multiple comparisons of parametric data, or Kruskal-Wallis tests combined with Dunn post tests for non-parametric data. Two-way ANOVA was used for data influenced by two variables. Multiple comparison was corrected using the Sidak multiple comparison test. Descriptive results are expressed using mean and standard deviation and boxplots are used for illustration. Correlation was calculated with Spearman‘s correlation for non-parametric samples. Receiver operating characteristic [ROC] curves were used to assess the discriminatory performance of FCAL and FLCN2 expressed as area under the curve [AUC], sensitivity, and specificity. Paired ROC curves for FLCN2 and FCAL were compared using DeLong’s test.

3. Results

3.1. Comparative study of FCAL and FLCN2 in IBD and healthy controls

In a first step we compared the performance of FCAL and FLCN2, measured from the same stool sample, to distinguish between clinically active disease and remission in a cohort of 132 IBD patients and 40 healthy controls. Clinical and demographic characteristics are shown in Table 1. IBD patients were separated into UC and CD, and into active versus non-active disease based on the partial Mayo score and the Harvey-Bradshaw Index [Figure 1; pMayo ≤1 = remission; HBI ≤4 = remission]. Both FCAL and FLCN2 were higher in active UC (n = 26; FCAL: 480.3 [363.3-854.3] µg/g; FLCN2: 17.02 [6.12-44.87] µg/g) and in active CD (n = 25; FCAL: 207.6 [47.65-463.7] µg/g; FLCN2: 1.50 [0.84-11.18] µg/g) compared with UC in remission (n = 34; FCAL: 33.18 [18.17-125.5] µg/g; FLCN2: 0.49 [0.31-1.29] µg/g) and CD in remission (n = 47; FCAL: 91.90 [25.3-173.5] µg/g; FLCN2: 0.94 [0.48-1.79]µg/g) [Figure 1A and B]. Low concentrations of both faecal markers were measured in healthy individuals (n = 40; FCAL: 16.00 [16.00-17.13] µg/g; FLCN2: 0.36 [0.31-0.47] µg/g), lower than in active IBD but also significantly lower than in IBD in clinical remission [Figure 1A and B].

Figure 1.

Comparison of faecal calprotectin [FCAL] and faecal lipocalin-2 [FLCN2] in patients with inflammatory bowel disease [IBD] [n = 132] and healthy controls [n = 40]. [A] FCAL and [B] FLCN2 in patients with ulcerative colitis [UC] and Crohn’s disease [CD] in remission (partial Mayo score [pMayo] ≤1; Harvey-Bradshaw index [HBI] ≤4) or with active disease compared with healthy controls [Ctrl]. [C, D] FCAL and FLCN2 resolved by endoscopic disease activity using [C] the endoscopic Mayo [eMayo] score for UC and [C] the simple endoscopic score for CD [SES-CD]. [E] Correlation of FCAL and FLCN2 in all IBD patients; blue triangles represent individuals with UC and yellow squares with CD. Correlation coefficients were calculated by Spearman rank method. [F] Stability assay for FCAL and FLCN2 in n=13 IBD stool samples kept at room temperature [RT; squares] or at 4°C [circles] for the indicated time points. Data are expressed as percentage change from baseline. Box plots represent values as median [bold horizontal line], interquartile range [box], and minimum/maximum values [whiskers]. Differences were analysed by Kruskal-Wallis test followed by Dunn’s multiple comparisons test. *p <0.05; ***p <0.001; ****p <0.0001.

Table 1.

Clinical and demographic characteristics of IBD patients and healthy controls.

	UC	CD	Controls	p-value
Samples, n	60	72	40
Sex [female/male]	18/42	32/40	16/24
Age median [IQR]	44.0 [32.0, 53.0]	35.0 [26.0, 50.75]	30.0 [26.0, 41.0]	p = 0.001
BMI median [IQR]	23.44 [20.95, 27.06]	22.29 [20.16, 24.12]	22.49 [20.32, 23.80]	p = 0.28
Disease activitya
Remission	34	47
Mild	9	10		.
Moderate	10	11
Severe	7	4
Disease extentb
Proctitis [E1]	0
Left-sided-colitis [E2]	16
Pancolitis [E3]	44
Disease locationb
Ileum [L1]		13
Colon [L2]		12
Ileum + colon [L3]		46
Upper GI [L4]		1
Behaviourb
Non-constricting, non-penetrating [B1]		24
Stricturing [B2]		16
Penetrating [B3]		15
Stricturing and penetrating [B2 + 3]		16
Medication
5-ASA	36	9
Azathioprine	17	30
Anti-TNFα	13	31
Anti-a4b7	1	0
Steroid	10	7
Antibiotics	4	3
Others	13	9

IBD, inflammatory bowel disease; UC, ulcerative colitis; CD, Crohn’s disease; IQR, interquartile range; GI, gastrointestinal; 5-ASA, 5-aminosalicylate.

^aAssessment of clinical disease activity by using partial Mayo score [remission: 0–2 points, mild disease: 3–4 points, moderate disease: 5–6 points, severe disease: 7–9 points] for UC and Harvey-Bradshaw Index [HBI] [remission: 0–4 points, mild disease: 5–8 points, moderate disease: 9–14 points, severe disease >=15 points] for CD.

^bAssessment of disease extent, disease location, and disease behaviour by using the Montreal classification.^34,35

As FCAL is a useful surrogate marker for endoscopic disease activity,³⁶ we next assessed the extent to which FCAL or FLCN2 reflected endoscopic inflammation in a subgroup of 91 IBD patients who underwent ileocolonoscopy. Endoscopic disease activity was scored using the endoscopic [e] Mayo Score for UC and the SES-CD score for CD. Endoscopic remission was defined, as previously suggested,^37,38 as an eMayo score of 0 and 1 or an SES-CD ≤2, respectively. Both FCAL and FLCN2 concentrations were significantly lower in endoscopic remission than in patients with endoscopically active disease and increased with endoscopic severity both in UC and CD [Figure 1C D; and Supplementary Tables 2 and 3, available as Supplementary data at ECCO-JCC online].

To compare sensitivities and specificities of FCAL and FLCN2 in our cohort, we generated receiver operating characteristic [ROC] curves. When discriminating all IBD patients from healthy controls the area under the curve was 0.92 (confidence interval [CI] 0.88-0.96] for FCAL and 0.88 [CI 0.82-0.94] for FLCN2, respectively) [Supplementary Figure 1 and Supplementary Table 4, available as Supplementary data at ECCO-JCC online]. To distinguish between IBD patients and healthy controls, FCAL, at an optimal cut-off value of 22.45 µg/g, reached a sensitivity of 84.65% and a specificity of 93.75% compared with FLCN2 that demonstrated a sensitivity of 79.85% and specificity of 90.65% at an optimal cut-off value 0.45 µg/g. This difference was statistically not significant [p = 0.12; De Long’s test]. Sensitivity and specificity to discriminate between endoscopically active disease and mucosal healing in UC were 90% and 84% at a cut-off of 73.38 µg/g for FCAL, and 99% and 88% at a cut-off of 1.98 µg/g for FLCN2, respectively. This corresponded in CD to a sensitivity and specificity of 90.3% and 76.9% at an optimal cut-off value of 78.4 µg/g for FCAL, and for FLCN2 90.6% and 76.9% at an optimal cut-off of 0.56 µg/g, respectively.

To corroborate the excellent correlation between both biomarkers, we plotted FCAL against FLCN2 concentrations and found a strong correlation with an R_S of 0.87 [p <0.0001] for the entire IBD cohort [Figure 1E]. Additional correlations are outlined in Supplementary Table 5, available as Supplementary data at ECCO-JCC online. No significant influences of age, gender, body mass index [BMI], or medication on faecal biomarker levels [corrected for disease entity and activity] were detected [data not shown].

To evaluate a potential impact of biomolecule stability on the presented results, we performed a comparative stability experiment using freshly collected stool samples from IBD patients, retaining them at room temperature or at 4°C for up to 96 h and evaluating potential dynamics in FCAL and FLCN2 concentrations. As shown in Figure 1F both FCAL and FLCN2 were stable within the observation period, with a negligible effect of room temperature.

3.2. Mucosal cellular sources of CAL and LCN2 in the steady-state and during intestinal inflammation

Having demonstrated that both FCAL and FLCN2 as biomarkers show a comparable efficacy in differentiating between active and non-active IBD, regarding both clinical and endoscopic activity, as well as in healthy controls, we aimed at identifying specific cellular subsets expressing both proteins in the healthy and in the actively diseased state, in order to determine potential biological relevance.

First, we compared tissue expressions of CAL and LCN2 by real-time PCR and immunohistochemistry. CAL mRNA expression was 5.76-fold [2.42-14.01] higher in the colonic mucosa of active UC and 26.12-fold [2.98-100.6] higher in the ileal mucosa of active CD compared with steady-state expression in healthy controls. Accordingly, LCN2 mRNA expression was 18.57-fold [15.84-27.88] higher in active UC and 53.53-fold [29.39-175.2] higher in active CD [Figure 2A]. This was corroborated by immunohistochemistry, where colon and ileum sections of active UC and CD patients stained strongly for both CAL and LCN2 [Figure 2B]. Notably, immunopositivity was distributed distinctly between the two biomarkers. LCN2 showed a broader tissue distribution involving IECs.

Figure 2.

[A] Relative mRNA expression of CAL and LCN2 in biopsies of active UC [n = 7], active CD [n = 5] and healthy controls [Ctrl] [n = 9]. Data are expressed as fold-change over steady-state rectal mucosa expression. [B] Sections from active CD ileums, active UC colons, and healthy controls [Ctrl] were analysed by immunohistochemistry for CAL [top row] and LCN2 [bottom row], original magnification was 400x. [C] Representative picture captures of LCN2 [green] and CAL [purple] in a patient with CD. Nuclei are shown in blue. Double-positive cells in the merged panel appear yellow. The scale bar indicates 20 µm. Original magnification was 400x. [D] Biopsy-derived cells [lamina propria plus epithelial cell fractions] were analysed in healthy [n = 7] and inflamed rectal UC [n = 13] and inflamed ileal CD [n = 7] tissue by flow cytometry. Cells were first identified as epithelial cells or leukocytes, and the latter were then further subdivided as indicated. As depicted, numbers are given as percent of total counts and percent of total CD45+ leukocytes, respectively. [E] To identify CAL- and/or LCN2-bearing subpopulations, both targets were quantified by intracellular flow cytometry. Staining intensities were determined by mean fluorescence intensity [MFI]. UC, ulcerative colitis; CD, Crohn’s disease.

To look more closely at this observation, tissue sections of patients with active IBD were double-stained for CAL and LCN2 and analysed by confocal microscopy [Figure 2C]. LCN2 was expressed in both IEC and numerous cell types of the lamina propria, whereas CAL expression was confined to lamina propria cells. All cells positive for CAL also expressed LCN2; in contrast, LCN2 single-positive cells were also present in the lamina propria [Figure 2C].

This prompted us to characterise CAL- and LCN2-expressing cell types in biopsy-derived cells from endoscopically active UC [n = 13] and ileal CD [n = 7] and from healthy controls [n = 7] by flow cytometry. We measured intracellular CAL and LCN2 in CD324⁺ epithelial cells and various mucosal leukocytes [Figure 2D]. The gating strategy is outlined in Supplementary Figure 2, available as Supplementary data at ECCO-JCC online. Expectedly, leucocytes were more abundant in the inflamed mucosa of patients with CD and UC compared with the steady-state mucosa. This was attributable to a strong increase in granulocyte numbers both in the colonic mucosa of UC and in the ileal mucosa of CD patients [Figure 2D]. Further, we found an increase in the relative proportions of macrophages and monocytes again in active UC and CD patients [Figure 2D and Supplementary Table 6, available as Supplementary data at ECCO-JCC online]. In line with the results from confocal microscopy, CD324⁺ IECs stained positive for LCN2 but were negative for CAL [Figure 2E; and Supplementary Figure 3, available as Supplementary data at ECCO-JCC online]. Strong CAL expression was observed in CD66b⁺ granulocytes and CD14⁺/CD68⁺ macrophages, both of which also stained positive for LCN2 [Figure 2E; Supplementary Table 7, available as Supplementary data at ECCO-JCC online]. In contrast, CD14^–/CD68⁺ advanced macrophages and CD14⁺ monocytes were immune-positive for intracellular LCN2 but negative for CAL [Figure 2E]. Again, this pattern was not specific for either UC or CD but was found in both disease entities.

3.3. Evaluating the usefulness of CAL and LCN2 as markers of ‘molecular inflammation’ in IBD patients with clinical and endoscopic remission

So far, we demonstrated a comparable diagnostic value of FCAL and FLCN2 with regards to correlation with clinical and endoscopic disease activity. Experimental evidence suggests that IBD pathogenesis factors are integrated particularly by IECs, and that stressed IECs may spark mucosal inflammation.³⁹ For instance, LCN2 is strongly upregulated in epithelial scrapes of endoplasmic reticulum [ER]-stressed IEC-specific XBP-1 knockout animals.⁴⁰ Thus we hypothesised that the distinct cellular distribution of LCN2 might offer advantages, particularly in situations with low-grade mucosal inflammation where granulocytes play a subordinate role. Therefore, we conducted an additional endoscopic study on a cohort of IBD patients in clinical remission, who had a FCAL of <100 µg/g. Notably, all endoscopies were performed and scored by two experienced IBD specialists [RK and ARM]. Ultimately, 23 patients [14 UC, nine CD] with confirmed mucosal healing were enrolled along with nine healthy controls [Figure 3A]. The patients’ characteristics are demonstrated in Supplementary Table 8, available as Supplementary data at ECCO-JCC online. From all study participants, standardised biopsies [as outlined in Figure 3B] were taken from six sites, from the ileum to the rectum, for follow-up RNA extraction, histology, and immunohistochemistry. Furthermore, stool samples for the measurement of FCAL and FLCN2 were obtained.

Figure 3.

[A] Standardised biopsies were obtained from 23 IBD patients [n = 14 UC, n = 9 CD] in clinical and endoscopic remission and FCAL <100 µg/g healthy controls [Ctrl, n = 9]. Representative endoscopic picture captures of endoscopically inactive ileum and rectum of CD and UC patients and Ctrls. [B] The sampling strategy is outlined in this panel. [C] FCAL and FLCN2 were analysed by ELISA. Columns and error bars indicate mean ± SD. Statistical analyses were performed using Kruskal-Wallis testing. [D] Analyses of histological activities in UC and CD compared with Ctrls using adapted variants of the Colonic and Ileal Global Histologic Disease Activity Score [CGHAS] for CD and the Geboes Score [OGS] for UC, respectively. [E] CAL and LCN2 mRNA expressions were measured by quantitative real-time PCR of Ctrls [n = 9] and site-specifically compared in UC [n = 14] and CD [n = 9] patients. Box plots represent values as median [bold horizontal line], interquartile range [box], and minimum/maximum values [whiskers]. Statistical significance was calculated using two-way ANOVA followed by Sidak’s method for multiple comparison, n.s. = not significant *p <0.05; ***p <0.001. IBD, inflammatory bowel disease; UC, ulcerative colitis; CD, Crohn’s disease; ELISA, enzyme-linked immunosorbent assay; SD, standard deviation; PCR, polymerase chain reaction; ANOVA, analysis of variance.

As shown in Figure 3C, neither the concentrations of FCAL nor those of FLCN2 differed between our IBD patients and healthy controls, thus arguing against an advantage of FLCN2 over FCAL in detecting very low-grade mucosal inflammation.

All biopsies were assessed for histological inflammation using modified variants of the Geboes and CGHAS score, adjusted for detecting low-grade inflammation, by an experienced GI pathologist [GO]. Across all segments, histological inflammation was statistically not different between UC and CD patients and healthy controls [Figure 3D]. This prompted us to compare tissue expressions of CAL and LCN2 in mucosal biopsies by real-time PCR. In UC patients, there was a gradual and significant increase in the LCN2 expressions from the ileum to the rectum. This increase was UC-specific and not observed in healthy controls [Figure 3E]. Furthermore, such an increase was not paralleled by an increase of CAL [Figure 3E]. In CD, the biogeographical distribution of this ‘molecular inflammation’ was inverted, again with an increased tissue expression of LCN2 in the terminal ileum and unchanged concentrations in the colon compared with healthy controls. Also in CD, increased ileal LCN2 expression was not paralleled by an increase of CAL [Figure 3E]. Taken together, these data suggest that mucosal expression of LCN2 may be useful as a marker of ‘molecular inflammation’ in IBD.

To test if tissue mRNA expression would be paralleled by protein expression in immunohistochemistry, all tissue sections were stained specifically for CAL and LCN2 and were semi-quantitatively analysed for staining intensity [see Figure 4 A D]. In CD, increased mRNA expression was paralleled by a greater immunopositivity for LCN2 but not CAL in the terminal ileum [Figure 4 E G]. Notably, there was a strong LCN2 signal coming from Paneth cell granules at the bottom of the crypts of Lieberkühn, also in the steady state [Supplementary Figure 4, available as Supplementary data at ECCO-JCC online]. The elevated rectal mRNA expression for LCN2 in UC was not mirrored by protein expression on immunohistochemistry [Figure 4 E G].

Figure 4.

[A–D] Biopsies obtained from IBD patients in remission and healthy controls [Ctrl] as indicated in Figure 3 were further analysed by immunohistochemistry for lipocalin-2 [left column] and calprotectin [right column]. Representative picture captures of ileum biopsies in Ctrl [A] and CD [B] along with rectum biopsies of Ctrl [C] and UC [D] are shown. Original magnification was 200x. The small frames indicate a representative section highlighted in the right lower bottom of each picture [original magnification 400x]. [E–G] Semiquantitative analyses of LCN2 immunopositivity in epithelial cells [E] and all cells [F], as well as CAL stainings [G]. Slides were attributed arbitrary values of zero, 1, 2, or 3. Columns indicate means, n.s. = not significant. *p <0.05. IBD, inflammatory bowel disease; UC, ulcerative colitis; CD, Crohn’s disease.

4. Discussion

We present a comparative study between the well-established biomarker FCAL, which after 20 years of research has recently seen widespread adoption in monitoring recommendations,⁴¹ and a potential challenger, FLCN2.

In a first study, we compared the performance in a cohort of more than 130 IBD patients and a group of healthy controls and found an excellent and comparable ability to distinguish between active and non-active disease and between IBD patients and non-IBD controls. A particular strength of FCAL is its good correlation with the endoscopic disease activity.³⁶ Importantly, FLCN2 was equivalent to FCAL in this regard in our cohort. To be feasible as a diagnostic and research tool, a faecal biomarker needs to be reasonable stable in order to allow for variation in ‘collection-to-measurement-times’. As other members of the lipocalin superfamily,^29,42 FLCN2 was very stable, with our data suggesting comparable stability to FCAL after prolonged storage at room temperature.

To determine differences in the biological significance between the two biomarkers, we next studied cellular sources of FCAL and FLCN2 in the inflamed mucosa. Whereas FCAL is almost exclusively found in granulocytes and in macrophages, FLCN2 demonstrated a broader cellular distribution, and besides granulocytes and macrophages it was also expressed by advanced macrophages, monocytes, and particularly in IECs. To assess whether these differences in cellular expression may be an advantage particularly in clinical situations with low-grade inflammation, we compared FCAL and FLCN2 in an endoscopy study including 23 IBD patients in clinical and endoscopic remission, and studied a potential added value of LCN2 as a marker of histological or molecular inflammation.

There is ample evidence that treatment targets beyond improvement of clinical symptoms, particularly the improvement or normalisation of the mucosal appearance in endoscopy [ie mucosal healing], are associated with better long-term outcomes such as the need for surgery, hospital admissions, and days off work, both in Crohn’s disease and in ulcerative colitis.^43–45 In fact, several studies demonstrated an imperfect correlation between symptoms and mucosal inflammation, as demonstrated in an Australian cohort with more than half of patients in clinical remission exhibiting residual mucosal inflammation at endoscopy.⁴⁶ Conversely, relying solely on endoscopic readouts may also be misleading, as recently demonstrated in a phase 3 clinical trial testing etrolizumab in patients with moderate to severe UC. Here, one-third of patients with an endoscopic Mayo subscore of 1 [ie mucosal healing] still reported an elevated stool frequency, and 27% even reported ongoing rectal bleeding.⁴⁷ Thus it seems prudent to combine treatment outcomes, including symptoms, with objective measures of disease activity, particularly surrogates for mucosal healing.

Among serological biomarkers, high-sensitivity C-reactive protein [hsCRP] represents the most widely used one.⁴⁸ Noteworthy in this context, serum LCN2 in complex with matrix metalloproteinase-9 also represents an excellent surrogate marker for mucosal healing in both CD and UC.^27,28 Although elevated CRP levels have a predictive value regarding long-term outcomes such as risk for colectomy, a substantial proportion of IBD patients present with CRP levels within the normal range.⁴⁸ Notably, in our cohort we observed only a moderate correlation between circulating CRP concentrations and both FACL and FLCN2 [Supplementary Table 4]. Accordingly, faecal biomarkers may be the better choice, and FCAL represents currently the most widely used faecal biomarker in IBD which correlates well with clinical and histopathological disease activity.^13,49,50 More importantly, FCAL has been shown to be associated with mucosal healing in UC,^44,51–53 and was recently suggested as a useful monitoring marker capable of predicting relapses and guiding treatment decisions.^41,54,55

Along these lines, with an excellent biomarker such as FCAL, is there need and room for an additional faecal biomarker? Our study is confirmative of previous research published by Thorsvik et al., who also studied FLCN2 in comparison with FCAL as a potential biomarker,²⁹ showing an excellent correlation between the two markers. From a functional perspective, CAL represents a neutrophil-derived protein capable of sequestering zinc and thus inhibiting the activity of zinc-dependent enzymes such as matrix metalloproteinases.¹² Whereas the pathophysiological role of FCAL in IBD is less clear, LCN2 secreted into the gut lumen works as an antimicrobial peptide that helps to control inflammation-induced dysbiosis and colitis-driven tumourigenesis.^56,57 In this context, Stallhofer and colleagues demonstrated that homozygosity for the IL23R risk allele in CD resulted in a reduction in circulating LCN2 concentrations, raising the possibility that specific genetic variants may result in a relative LCN2 deficiency also in humans.⁵⁸ In line with Thorsvik et al.,^29,30 we highlight important spatial differences between FCAL and FLCN2, with important differences between the cellular origins of the two proteins. Using both confocal microscopy and flow cytometry, we present a detailed analysis of the cellular distribution of both CAL and LCN2. Whereas CAL is almost exclusively found in granulocytes, LCN2 is expressed by various cell types including IECs, monocytes, and advanced macrophages.

This led us to hypothesise that LCN2 as a faecal or a mucosal biomarker may be advantageous over CAL, specifically in clinical situations of low-grade inflammation. Indeed, FLCN2 was useful to detect low-grade dextran sulphate sodium [DSS]-induced intestinal inflammation as reported by Chassaing et al.⁵⁹ Therefore we examined a prospective cohort of well-characterised IBD patients in clinical and endoscopic remission with an FCAL <100 µg/g. In this ‘remission cohort’, both LCN2 and CAL as a faecal surrogate marker were not different between IBD patients and healthy controls [Figure 3C]. However, using quantitative real-time PCR, the relative expression of LCN2 in endoscopically and histologically healed biopsies of the rectum from UC and the ileum from CD patients remained significantly higher than in control patients. As this increase was not observed for CAL, LCN2 can be considered a more sensitive marker of persistent low-grade inflammation in IBD. As it has been shown that an FCAL threshold as low as 56 µg/g already predicts a risk for clinical relapse after stopping anti-tumour necrosis factor [TNF] treatment,⁵⁴ mucosal LCN2 expression could turn out a valuable predictive biomarker for such clinical decisions.

As real-time PCR may be too elaborate for clinical routine use, we aimed at reproducing this molecular information of LCN2 mRNA expression by LCN2 [and CAL] immunohistochemistry, which could be easily implemented into routine pathological workup. In spite of achieving a reasonable and reproducible staining quality, the resulting immunohistochemical data did not reflect the differences as clearly as mRNA expression. Further work is warranted to improve technical aspects of LCN2 immunohistochemistry or to apply newer techniques such as RNAscope^TM.

Taken together, we provide new and detailed evidence that in IBD FLCN2 represents a highly sensitive marker of intestinal inflammation—seemingly comparable to FCAL—with a potential advantage in detecting very low-grade intestinal inflammation, most likely due to its distinct cellular distribution. Our data suggest that the observed persistent mucosal overexpression of LCN2, despite endoscopic and histological healing, renders LCN2 an attractive candidate for molecular inflammation, which warrants further investigation.

Funding

This study was supported by the MFF Tyrol grant number 286 [to ARM]. The work was supported in part by an independent general research grant [number #55130045] to ARM from Pfizer Corporation Austria GmbH. ARM is also supported by the Christian Doppler Research Association, and we gratefully acknowledge support by the Austrian Federal Ministry of Science, Research, and Economy and the National Foundation for Research, Technology, and Development.

Conflict of Interest

ARM is receiving research support from AbbVie and Takeda under the framework of the Christian Doppler Research Society. He has received further consultation fees and/or speaker honoraria from AbbVie, Merck Sharp and Dohme, Takeda, Janssen-Cilag, Amgen, Sandoz, and Pfizer. HT has received speaker honoraria from AbbVie, Janssen-Cilag, Merck Sharp and Dohme, and Takeda. TR receives an unrestricted research grant from AbbVie, and further consultation fees and/or honoraria from Abbvie, BMS, Celgene, Ferring, Gilead, GSK, LabGenius, Janssen, Mylan, MSD, Novartis, Pfizer, Sandoz, Takeda, and UCB.

Acknowledgments

We thank all patients who consented to participate in this project, and gratefully acknowledge the support of the nursing staff of our Department of Internal Medicine 1.

Author Contributions

Concept: ARM, HT, AZ, SJR, AS. Experimental procedures: AZ, SJR, GO, AP, CW, BT. Statistical analysis: AZ, AS, SJR. Endoscopy: ARM, RK, ME, HT. Writing of original draft: ARM, AZ, SJR, AS, TR. Funding question: ARM, HT.

Author notes

These authors contributed equally (shared first authorship).