Fecal calprotectin level in microscopic colitis: a systematic review and meta-analysis

Abstract

Background

There are no specific biomarkers for microscopic colitis (MC), and the diagnosis depends on histopathological tissue obtained during colonoscopy. Studies on the role of fecal calprotectin (FCP) in MC are limited. A literature search of PubMed, Embase, and Scopus was conducted from each database’s inception through September 2022.

Methods

A DerSimonian-Liard random-effects meta-analysis was performed to examine the standardized mean difference (SMD) in FCP levels between patients with MC and control patients with chronic diarrhea.

Results

Six studies with 96 active MC patients and 200 controls were included in the meta-analysis. Random effects meta-analysis revealed that FCP was significantly and moderately elevated in patients with MC compared to the control group (SMD = 0.6 [95% CI 0.3, 1.0], P = 0.001). Imputing one effect size confirmed the observation that FCP is significantly higher in patients with MC than control subjects with chronic diarrhea (SMD = 0.5 [95% CI 0.2, 0.9], P = 0.004). Study heterogeneity between the SMDs was not significant (I² = 41%, P = 0.1).

Conclusion

These results indicate that FCP levels in patients with chronic diarrhea can help identify patients with MC. More studies with serial measurements of FCP would provide a better understanding of its utility in MC.

Microscopic colitis (MC) is an inflammatory bowel disease (IBD) that is either increasing in incidence or more frequently recognized.¹ There is no specific biomarker for MC, and the diagnosis depends on histopathological tissue obtained during colonoscopy, which is an invasive and costly procedure.² MC is classified as either collagenous colitis (CC) or lymphocytic colitis (LC).

Calprotectin is a protein derived from neutrophils. Fecal calprotectin (FCP) is often measured in patients with suspected IBD and is used for diagnosis and disease monitoring. It can be used to screen for inflammatory vs noninflammatory gastrointestinal conditions, such as irritable bowel syndrome (IBS). It is particularly useful in patients with IBD.³ The utility of FCP in diagnosing or monitoring MC has not been established, and studies on the role of FCP in MC are limited. In this meta-analysis, we analyzed the association of FCP levels in patients with MC compared to control subjects with chronic diarrhea.

Method

We conducted this meta-analysis according to Cochrane’s manual of diagnostic test accuracy, and the manuscript was prepared according to the preferred reporting items for systematic reviews and meta-analysis of diagnostic test accuracy (PRISMA-DTA) guidelines.^4,5 A literature search of PubMed, Embase, and Scopus was conducted from database inception through September 2022. Search terms included (a) microscopic colitis, collagenous colitis, and lymphocytic colitis, (b) fecal calprotectin. Only studies evaluating an adult population were included. Case series and case reports were excluded from this study. Non-English publications were also excluded if data could not be extracted. The titles and abstracts were reviewed by two independent authors (B.S. and C.K.). Discrepancies were resolved through discussion between them and the senior author (K.N.). Two independent authors compiled data from each study including study characteristics, study population characteristics, and study results. Study characteristics included author, year of publication, start and end dates for data collection, country, and type of study design. Quality assessment was performed using QUADAS-2 by two independent authors (B.S. and C.K.).⁶ Any discrepancies were resolved through discussion between them with the senior author (K.N.). Interobserver agreement was evaluated by Cohen’s kappa coefficient.

A DerSimonian-Liard random effects meta-analysis was performed to examine the standardized mean difference (SMD) of FCP between patients with MC and control participants using the ‘meta’ package (version 5.0-1) in R (version 4.1.2).⁷ When means and standard deviations were not reported, but medians and ranges or interquartile ranges (IQR) were reported, the means and standard deviations were imputed.⁸ The consistency of the findings of the meta-analysis was confirmed by leave-one-out sensitivity analyses. The likelihood of publication bias was explored using funnel plots, and the effect sizes of missing (i.e., unpublished/unreported) studies were imputed via the trim-and-fill method.^9,10 Heterogeneity of effect sizes was quantified by calculating the Higgins I² statistic.^4,5 The magnitude of the effect sizes was interpreted based on Cohen.¹¹ A SMD of >0.5 was considered clinically significant. A receiver operating characteristic curve was not calculated due to inadequate data for two-by-two tables.

Results

The PRISMA flow chart indicating the study selection process is shown in Figure 1. Six studies collectively recruiting 96 active MC patients and 200 controls were included in the meta-analysis (Table 1).^12–17 The mean age of the participants included in the study was 57.7 ± 2.1 years, and the proportion of females to males was 1 to 3.4. The mean weighted FCP level in patients with MC was 214.3 ± 176.7 μg/g vs 51.0 ± 14.0 μg/g in control subjects (P < 0.01).

Figure 1.

PRISMA for systematic reviews included searches of databases and registers.

Table 1.

Characteristics of studies evaluating utility of fecal calprotectin for detection of microscopic colitis

Author	Year	Country	Publication type	Study type	MC (n)	CC: LC (n)	Control (n)	Start date	End date	Mean age (years)	Female (n)	Mean FCP in MC (μg/g)	Mean FCP in control (μg/g)	Reference range of FCP (μg/g)
Batista	2015	Spain	Full paper	Retrospective	30	13:12*	64	Jan 2019	Dec 2016	56.4	73	169.7	51.7	0–50
Larsson	2014	Sweden	Full paper	Prospective	15	10:5	43	Jan 2007	Jun 2009	56	60	67.8	58.6	0–50
Lyutakov	2021	Bulgaria	Full paper	Prospective	14	12:2	14	Oct 2017	Apr 2019	60	11	221	43	NR
von Arnim	2016	Germany	Full paper	Prospective	23	19:4	20	May 2010	Oct 2010	63.8	19	56.3	29.2	NR
Wagner	2016	Sweden	Full paper	Prospective	13	9:4	46	2006	2009	57	50	74 (CC), 42.7 (LC)	61	0–50
Wildt	2007	Denmark	Full paper	Prospective	21	21:0	13	NR	NR	60**	17	516.3	29.4	0–50

*5 patients were diagnosed with incomplete colitis.

**Median.

CC indicates collagenous colitis; FCP, fecal calprotectin; LC, lymphocytic colitis; MC, microscopic colitis; NR, not reported.

Study characteristics

Tables 1 and 2 demonstrate the characteristics of each study. All studies except that of Wildt et al evaluated FCP in both CC and LC patients.¹⁷ Five of the six studies included in the meta-analysis prospectively included patients who presented with chronic watery diarrhea.^13–17 Two of the eight papers were reviewed but excluded from the analysis.¹⁸ The study by Abukhiran et al was excluded because it did not report standard deviations or any measures of dispersion.¹⁸ The study by Davie et al was retrospective and reported the value of FCP in CC and LC, but did not have a control group and was excluded from the analysis.¹⁹ All studies had the same standard reference test in diagnosing MC, which was colonoscopy with biopsy and histology.

Table 2.

Abstracted data from each study on the utility of FCP in MC

Author	Patients (n)	FCP measured in MC (n)	FCP measured in control (n)	FCP technique	Reference range FCP (μg/g)	Prevalence of MC (CC, LC, iMC) (%)	Mean FCP in MC (μg/g)	Mean FCP in control (μg/g)	SMD	95% CI
Batista	94	30	64	ELISA	0–50	46.2 (20, 18.5, 7.7)	169.7	51.7	0.78	0.3, 1.2
Larsson	78	12	43	NA	0–50	19.2 (12.8, 6.4, 0)	67.8	58.6	0.14	−0.5, 0.8
Lyutakov	100	14	14	Rapid ELISA test (Quantum blue)	NA	14 (12, 2, 0)	221	43	0.87	0.1, 1.7
von Arnim	43	6	20 (IBS)	ELISA and rapid test (Quantum blue)	NA	NA	56.3	29.2	0.87	−0.1, 1.8
Wagner	63	13	46	ELISA	0–50	20.6 (14.3, 6.3, 0)	74 (CC), 42.7 (LC)	61	0.09	−0.5, 0.7
Wildt	34	21	13	ELISA	0–50	NA	516.3	29.4	1.20	0.4, 2.0

CC indicates collagenous colitis; CI, confidence interval; ELISA, enzyme-linked immunosorbent assay; FCP, fecal calprotectin; LC, lymphocytic colitis; MC, microscopic colitis; SMD, standardized mean difference.

Performance of FCP in detecting MC

Random effects meta-analysis revealed that FCP was significantly and moderately elevated in patients with MC as compared to controls with chronic diarrhea and IBS-D (SMD = 0.6 [95% confidence interval (CI) 0.3, 1.0], P = 0.001; Figure 2). The observed SMD also exceeded the predetermined threshold for minimal clinically important difference. Leave-one-out sensitivity analyses did not significantly change the pooled estimate obtained from the random effects meta-analysis (Supplementary Table 1). Imputing one effect size using the trim-and-fill method and reanalyzing the data using the imputed effect size also confirmed our observation that FCP is significantly higher among patients with MC vs controls with chronic diarrhea (SMD = 0.5 [95% CI 0.2, 0.9], P = 0.004). Between-study heterogeneity between the SMDs pooled in the random effects meta-analysis was not significant (τ² = 0.083; I² = 41%, P = 0.13). Therefore, meta-regression analyses were not attempted.

Figure 2.

Forest plot depicting pooled standardized mean differences of patients with active microscopic colitis vs control participants with chronic diarrhea. MC indicates microscopic colitis; SD, standard deviation; SMD, standardized mean difference.

Quality assessment

During the article screening, there was high agreement between the two authors. A QUADAS-2 is shown in Supplementary Table 2. Overall, the concerns regarding patient selection bias, index test, reference standard, and flow and timing were low. Publication bias assessed by funnel plot of the effect sizes indicated minimal publication bias, as shown in Supplementary Figure S1.

Discussion

This meta-analysis demonstrates that FCP levels were significantly increased in patients with MC compared to control subjects; the SMD was 0.6 (95% CI 0.3, 1.0; P = 0.001). Study heterogeneity was moderate (I² = 41%). The total number of patients in this meta-analysis was 412, including 99 patients with MC, 200 control patients, and 113 patients with other diagnoses not included in the analysis. The total number of FCP levels measured in the MC group was 96, because the paper of Larsson et al included 15 patients with MC but only 12 had FCP measured.¹³ The overall mean prevalence of MC, CC, and LC in the studies included in this meta-analysis was 25%, 14.8%, and 8.3%, respectively. This percentage is consistent with the reported prevalence in the study of Olesen et al, which reported that MC was found in 10% to 20% of patients with chronic watery diarrhea who underwent colonoscopy.²⁰ Five out of six studies were prospective studies and compared MC with other diseases that can present with chronic watery diarrhea.^13–17

MC can be difficult to differentiate from IBD and IBS-D in patients who initially present with chronic diarrhea. Colonoscopy with histologic tissue remains the gold standard and is used to distinguish CC from LC.²¹ However, colonoscopy has several possible disadvantages, such as bleeding, infection, adverse reactions from sedation, and perforation.²² In MC patients, the absence of neutrophils in colonic mucosa but the presence of neutrophils only in the lamina propria can affect the FCP levels.²³ There are several benefits of FCP compared with colonoscopy. For example, it is a convenient tool that can be used at home or in outpatient clinics, is noninvasive, has no adverse side effects, and is inexpensive.²⁴ Currently, the 2020 European guidelines on MC state that FCP is not useful to exclude or monitor MC due to low predictive value,²⁵ and these guidelines note that FCP has been used to differentiate IBS from IBD since 2013.²⁶

In this meta-analysis, three studies also recruited MC patients in clinical remission.^12,15,17 Batista et al demonstrated that the FCP level is significantly decreased in patients in clinical remission compared to patients with active disease, with a median of 436 μg/g (IQR 189–1300) vs 56 mcg/g (IQR 30–167) (P = 0.004).¹² Similar findings were found in the study of Wildt et al, which only focused on CC patients and demonstrated a median FCP in active CC of 80 μg/g compared with 26 μg/g in CC in remission (P = 0.025).¹⁷ In the von Arnim et al study, FCP levels were compared in active MC vs MC in remission and showed a significant elevation of FCP in active MC vs MC in remission and IBS (P < 0.0001); mean FCP levels in active, remission, and IBS were 47.9, 20.9, and 1.9 μg/g, respectively.¹⁵ The use of the FCP rapid test could not differentiate active MC and MC in remission based on the study of von Arnim et al.¹⁵ However, in our study, when FCP was used for detection of MC in the setting of chronic watery diarrhea, the level was significantly higher in patients with MC compared to control patients with chronic diarrhea (SMD = 0.5 [95% CI 0.2, 0.9], P = 0.004).

This meta-analysis suggests that the measurement of FCP is potentially useful in the diagnosis of MC. However, studies trying to establish this conclusion clearly will depend on study design and the colonic disorders of the patients used in the comparison group(s). The sensitivity and specificity of FCP in diagnosing MC vary significantly among studies, with, for example, reported sensitivities ranging from 60% to 79%.^12,18 These studies often use different cutoff values for FCP (50 to 150 mcg/g) due to the lack of established criteria for the diagnosis of MC.^12,14,15,27 Recent European guidelines on MC published in 2021 state that FCP is not useful to either exclude or monitor patients with MC.²⁵ Other studies have focused on the use of other fecal biomarkers, such as fecal neutrophil (fecal lactoferrin) and fecal eosinophil (fecal eosinophilic protein x, fecal cation protein), and reported similar findings with low sensitivity and specificity in patients with MC.²⁸ The explanation for these various study results likely involves the fact that these fecal neutrophil markers can increase in several disorders, including colorectal neoplasia, infectious diarrhea, adverse effects of medications (nonsteroidal anti-inflammatory drugs, proton pump inhibitors), and other inflammatory disorders (diverticulitis, MC, celiac disease).³ Therefore, clinicians should not rely on these biomarkers without additional diagnostic testing, such as colonoscopy, due to the low predictive value of FCP in diagnosing MC.

Multiple studies have demonstrated that the measurement of FCP is a useful marker in patients with gastrointestinal inflammation. FCP can be used to screen for inflammatory vs noninflammatory gastrointestinal disorders, such as IBS. It is particularly useful in evaluating and monitoring patients with IBD.³ It has a high negative predictive value, which helps rule out disease, and a high sensitivity to guide the need for an endoscopy.²⁹ Therefore, this test is a useful tool in differentiating IBD from IBS and avoiding unnecessary and expensive procedures, such as endoscopy, in patients with IBS.

Several limitations in this meta-analysis should be noted. First, the total number of patients in the meta-analysis is relatively small. Second, the studies had different reference FCP values, different definitions, and different population characteristics of the control group; control subjects were defined as patients who presented with chronic diarrhea (functional diarrhea) or IBS and had a normal colonoscopy.¹⁵ Third, the studies used different definitions of chronic watery diarrhea, including duration and number of bowel movements per day, in defining MC.

Conclusion

This study demonstrates that patients with MC have increased levels of calprotectin in their stool. This laboratory test provides a simple method to differentiate patients with IBD from patients with noninflammatory disease. It can identify patients with chronic inflammation, including IBD and MC. However, given FCP’s nonspecific nature, clinicians should not rely on FCP alone in diagnosing MC, and the diagnosis should be based on colonoscopy with histological tissue. Patients with low normal levels may not require extensive evaluation at their initial presentation.

Disclosure statement

No funding or potential conflict of interest was reported by the authors.