Improved monitoring of inflammatory activity in patients with ulcerative colitis by combination of faecal tests for haemoglobin and calprotectin

Abstract

Faecal calprotectin (FC) tests and faecal immunological tests (FIT) for haemoglobin have been used to monitor disease activity in patients with ulcerative colitis (UC) but used alone they have some limitation concerning the predictive ability. We aimed to test if an FC test used in combination with FIT could improve the predictive ability. Consecutive out-patients with UC (n = 93) who were admitted for colonoscopy completed a single faecal sample before the start of bowel preparation. A quantitative CALPRO^® calprotectin ELISA test and a qualitative FIT (cut-off < 40 ng/mL) were analyzed. An estimated Mayo score and a score of histological inflammation was performed blinded to the result of the faecal tests. The sensitivity, specificity, negative predictive value and positive predictive value for endoscopic inflammation (Mayo score > 1) was for FIT 85%, 83%, 96%, 57% and for FC > 186 µg/g 73%, 87%, 87%, 54%. Corresponding results for FIT*FC > 186 µg/g (at least one test positive) were 92%, 69%, 97%, 43%. For detecting moderate/severe histological inflammation the results were for FIT 69%, 79%, 92%, 43%, for FC > 75 µg/g 95%, 62%, 98%, 41%, and for FIT*FC > 75 µg/g 100%, 60%, 100%, 36%. None of the markers alone or in combination were useful to predict deep remission (Mayo score = 0 and no histological inflammation). We conclude that using the combination of an FC test and FIT shows minor improvement in predictive ability for inflammatory activity and remission in patients with UC.

Keywords:

• Faecal calprotectin
• faecal immunochemical haemoglobin tests
• inflammatory bowel disease
• ulcerative colitis

Introduction

In the last decade, faecal markers to monitor inflammatory bowel disease (IBD) in clinical practice have reduced the need for endoscopy monitoring of disease activity [1]. The most common used marker is calprotectin. Calprotectin is a Ca- and Zn-binding protein, which is located in the neutrophil leucocytes. It is protease resistant, stable in the GI-tract [2] and faecal calprotectin (FC) is widely used in diagnosis and evaluation of both ulcerative colitis (UC) and Crohn’s disease [1].

There are several FC tests on the market using different methods (ELISA, fluorescence enzyme immunoassay, immunochromatography), antibodies (monoclonal or polyclonal), standardization and decision limits, which makes its hazardous to translate test results between different methods [3]. For example, the Bühlman method shows higher values of calprotectin than the other FC methods [4]. There are also other factors that influence the outcome of a faecal test such as accidental dilution of water/urine [3] and the incubation time in colon [5]. There is still uncertainty of the most accurate cut-off value for FC and how to interpret intermediate values [3].

Quantitative and qualitative immunological faecal haemoglobin tests (FIT) are used to screen for colorectal cancer but have also been tested for IBD [6]. FIT has the advantage of being easier to perform and at a lower cost compared to an FC test. Still there is lack of standardization, and different cut-offs.

However, both the FC tests and FIT used alone have some limitations concerning predictive ability [7,8]. In general, the FC tests have shown good sensitivity and FIT good specificity [8,9] in predicting inflammatory activity in patients with IBD. Therefore, we hypothesized that using a combination of tests might improve accuracy in predicting inflammatory activity in patients with UC.

In this study, we aimed to test the predictive ability of the combination of an FC test (CALPRO) and a qualitative FIT to predict endoscopic and histological inflammatory activity in patients with an established UC disease.

Materials and methods

Patients

This study is based on the Faecal and Endoscopic Colorectal study Umeå (FECSU) cohort, collected during 2007–2013 at the University Hospital Umeå, Sweden [10]. Out-patients scheduled for colonoscopy were invited to participate in the study. The patients were after informed consent asked to collect a sample for an FC test and FIT before the start of bowel preparation. Patients with dementia, and low performance status were excluded in the FECSU. Overall 1997 patients were invited to participate in the study and 1263 patients accepted and were included in the study. Of the subjects who participated in the study 104 patients had an established UC and had their colonoscopy performed due to UC. Three patients were excluded due to reported use of nonsteroidal anti-inflammatory drugs (NSAIDs) in the last week and eight patients had no faecal test performed leaving 93 patients with UC included in the present study. Sixty-nine patients with UC had performed FIT, 88 patients a FC test and 64 patients had both tests performed form the same stool.

Faecal markers

Samples for FC were taken 2 days prior to the colonoscopy and before the cleansing procedure started. The 2-day-old samples were then immediately transferred to the Laboratory Medicine, clinical chemistry, Umeå University Hospital and frozen. Before extraction and testing the samples were thawed and equilibrated to room temperature. The CALPRO^® Calprotectin ELISA Test (ALP) was used (Calpro AS, Norway). According to the measuring range, values lower than 20 µg/g faeces were set as <20 µg/g and values above 10,000 µg/g were set as >10,000 µg/g. Coefficients of variation were 4.0% and 16.6% at levels 372 and 563 µg/g, respectively. The laboratory is enrolled in the external quality assessment programme for FC, EQUALIS AB, Uppsala, Sweden.

The samples for FIT were analyzed using immunological Analyz F.O.B Test (ANL products AB, Sweden), according to the manufacturer. A positive test indicates that the sample contains >40 ng/mL of human haemoglobin. Due to the study design, only one single sample per patient was analyzed.

Endoscopy

The endoscopy was performed by scheduled endoscopists in clinical routine, the colonic mucosa was biopsied according to cause for colonoscopy (activity or surveillance). The outcome of the endoscopy was routinely noted in the patient medical record. All clinical findings were retrospectively verified by studying the patient medical records, including the pathologist reports. Mayo score was used to estimate ‘endoscopic’ inflammatory activity [11]. Inflammatory activity on colon biopsies was defined as inactive (no granulocytic reaction was found either in epithelial or stromal compartments), as mild (existence of pericryptitis and eventually few granulocytes in crypt, and/or surface epithelium) and as moderate/severe (signs of significant active inflammation, i.e. crypt abscesses, abundant of inflammatory cells and mucin depletion) [12]. For Mayo score and inflammatory activity on colon biopsies the area with most inflammation features was valued. A patient in deep remission was defined as having a Mayo score of 0 and no inflammation on colon biopsies (inactive). All endoscopists and pathologists were blinded by the outcome of the faecal samples.

Statistical analysis

Descriptive statistics is used for baseline characteristics. A receiver operating characteristic (ROC) curve was used to define the diagnostic test ability (area under the curve (AUC)). The most favourable cut-off FC value for each outcome was the maximum sum for sensitivity and specificity in the ROC analysis. The most favourable cut-off values for FC was used to calculate sensitivity, specificity, negative predictive value (NPV) and positive predictive value (PPV) for FC and for the combination of FC and FIT according to defined outcome of inflammatory activity. All statistics was calculated using SPSS version 23.0 (IBM, Armonk, NY, USA).

Ethics

The study protocol was approved by the Research Ethics Committee of Umeå University, Umeå, Sweden (Dnr 08-184 M + 07-045 M) and the data handling procedures were approved by the National Computer Data Inspection Board. The patients gave informed consent prior to the study

Results

Patient characteristics

Table 1 shows the characteristics of the patients with UC who participated in the study. A majority of the patients had an extensive disease with a long duration of the disease (median 13 years, 25th–75th percentile 4–25 years) and was admitted for colonoscopy due to surveillance for dysplasia.

Table 1. Basal characteristics of the study participants with ulcerative colitis (n = 93).

Variable	Result	n
Age, years	50 (35–62)	93
Women	44%	41
The cause of colonoscopy
Disease activity/disease extension	31%	29
Surveillance	69%	64
Montreal classification
A1 (onset <17 years)	17%	16
A2 (onset 17–40 years)	53%	49
A3 (onset >40 years)	30%	28
E1 (proctitis)	3%	3
E2 (left-sided colitis)	34%	32
E3 (extensive colitis)	62%	58
Median faecal calprotectin (FC), µg/g (25th and 75th percentile)	75 (23–425)	88
Proportion of patients with an FC
>25 µg/g	81%	71
>50 µg/g	59%	52
>100 µg/g	48%	42
>200 µg/g	39%	34
Proportion of patients with a positive faecal haemoglobin immunological test (n = 69)	30%	21
Endoscopic activity (Mayo score)
Normal (Mayo 0)	36%	34
Mild (Mayo 1)	35%	32
Moderate/severe (Mayo > 1)	29%	27
Inflammation on colonic biopsies (histological activity score)
Normal (0)	30%	28
Mild (1)	48%	45
Moderate/severe (>1)	22%	20

Median (25th–75th percentile) for continuous variables and proportions (%) for categorical variables.

The FC test versus FIT in predicting inflammatory activity

Figure 1 shows the ROC analysis, the AUC values with 95% confidence intervals and the most favourable FC cut-off values for different outcomes of inflammatory activity in IBD. Forty-nine percent of the patients with FC > 50 µg/g also showed a positive FIT (and 60% of the patients with FC > 100 µg/g). Overall FC with the CALPRO method had higher sensitivity than FIT for both endoscopic and histological inflammatory scores except for moderate/severe endoscopic activity (Mayo score > 1) where FIT performed better (Tables 2–4). Contrary, the FIT showed slight better specificity for inflammatory activity than the FC test. However, in general only minor differences in NPV and PPV was shown between the FC test and FIT.

Figure 1. Receiver operating characteristic (ROC) curve for faecal calprotectin (FC) and for different outcomes of inflammatory activity. The most favourable FC cut-off (the maximum sum for sensitivity and specificity in the ROC analysis) and the area under the curve (AUC) with 95% confidence interval is shown for each outcome.

Table 2. The proportion of patients with ulcerative colitis classified into different Mayo scores and histological inflammation grades for faecal calprotectin (FC) alone and in combination with a positive faecal immunological faecal haemoglobin test (FIT).

	FC > 50 µg/g	FC > 100 µg/g	FC > 200 µg/g	Positive FIT	FC > 50 µg/g and/or positive FIT	FC > 100 µg/g and/or positive FIT	FC > 200 µg/g and/or positive FIT
Mayo score
0 (n = 34)	42%	24%	18%	7%	48%	30%	22%
1 (n = 32)	55%	48%	34%	26%	58%	50%	42%
>1 (n = 27)	85%	77%	69%	86%	92%	92%	92%
Histological activity on colonic biopsies
None (n = 28)	26%	7%	4%	4%	36%	14%	9%
Mild (n = 45)	64%	55%	48%	33%	63%	57%	50%
Moderate/severe (n = 20)	95%	90%	68%	69%	100%	100%	92%

Table 3. The sensitivity, specificity, negative predictive value (NPV) and positive predictive value (PPV) to detect any inflammation (‘non-deep remission’).

	Sensitivity	Specificity	NPV	PPV
Endoscopic inflammation
(Mayo score > 0)
FC > 63 µg/g (n = 88)	68%	67%	56%	77%
Positive FIT (n = 69)	46%	93%	54%	90%
Positive FIT and/or FC > 63 µg/g (n = 64)	70%	63%	61%	72%
Histological inflammation (>0)
FC > 65 µg/g (n = 88)	70%	89%	57%	93%
Positive FIT (n = 69)	56%	96%	52%	96%
Positive FIT and/or FC > 65 µg/g  (n = 64)	71%	82%	60%	88%
Endoscopic (Mayo score > 0) and/or histological inflammation (>0)
FC > 65 µg/g (n = 88)	65%	90%	43%	96%
Positive FIT (n = 69)	38%	94%	31%	95%
Positive FIT and/or FC > 65 µg/g  (n = 64)	65%	81%	43%	91%

FIT: immunological faecal haemoglobin test; FC: faecal calprotectin.

Table 4. The sensitivity, specificity, negative predictive value (NPV) and positive predictive value (PPV) to detect moderate/severe inflammation (‘non-remission’).

	Sensitivity	Specificity	NPV	PPV
Endoscopic inflammation (Mayo score > 1)
FC > 186 µg/g (n = 88)	73%	87%	87%	54%
Positive FIT (n = 69)	85%	83%	96%	57%
Positive FIT and/or FC > 186 µg/g  (n = 64)	92%	69%	97%	43%
Histological inflammation (>1)
FC > 75 µg/g (n = 88)	95%	62%	98%	41%
Positive FIT (n = 69)	69%	79%	92%	43%
Positive FIT and/or FC > 75 µg/g  (n = 64)	100%	60%	100%	36%
Endoscopic (Mayo score > 1) and/or histological inflammation (>1)
FC > 75 µg/g (n = 88)	84%	70%	89%	61%
Positive FIT (n = 69)	74%	86%	90%	67%
Positive FIT and/or FC > 75 µg/g  (n = 64)	94%	65%	97%	52%

FIT: immunological faecal haemoglobin test; FC: faecal calprotectin.

The use of the combination of FIT and FC test in predicting inflammatory activity

When using the combination of FIT and an FC test a minor improvement in sensitivity and NPV for endoscopic inflammation (Mayo > 1) and for moderate histological inflammation was seen compared when using each marker alone (Table 4). For example, 12 out of 13 patients with Mayo score > 1 and all patients with a moderate/severe inflammatory activity had a positive FIT and/or an increased FC test.

Patients in ‘deep remission’

Twenty-one patients (23%) were in ‘deep remission’ (no inflammation on endoscopy and on biopsies). The median FC value for the patients in deep remission was 34 µg/g (25th–75th percentile 20–52 µg/g). The proportion of patients in deep remission with a positive FIT was 6% (one out of 16 subjects). Twenty patients in deep remission had FC test performed. The proportion of patients in deep remission with an FC > 25 µg/g, FC > 50 µg/g, FC > 100 µg/g and FC > 200 µg/g was 65%, 30%, 10% and 5% respectively. The only subject in deep remission with an FC > 200 µg/g was a 71-year-old male with a history of constipation and who also was on proton pump inhibitor therapy. The chance for a faecal marker test to predict deep remission (‘NPV for any inflammatory activity’) was poor (Table 3). Also, a strict cut-off for FC (<25 µg/g) was insufficient to predict deep remission (41%).

Patients with Mayo score < 2 and no/mild histological inflammation

When using a more moderate criteria for remission (Mayo score ≤ 1 and no/mild histological inflammation) 62% (n = 63) of the subjects with IBD in the study were in remission. The median FC value for these patients was 41 µg/g (29–165). The chance for a faecal marker test to predict remission using these criteria was well both for FIT and FC (cut-off 75 µg/g) used alone or in combination (Tables 2 and 4). Sixteen patients out of 17 (94%) with an FC < 25 µg/g was in remission using these criteria.

Discussion

In the last decade, faecal haemoglobin tests have been evaluated both for discriminating IBD from non-organic gastrointestinal disease [13–19] and for monitoring inflammatory activity in patients with IBD [6,20–25]. A recent meta-analysis [8] of six studies with a total of 625 patients with UC found a pooled sensitivity for FIT of 77% and a pooled specificity of 81%. In general, faecal haemoglobin tests perform well to ‘rule-in’ inflammatory activity (PPV) but are more moderate to ‘rule-out’ (NPV). On the other hand, FC tests show a higher degree of sensitivity but tend to lack specificity and there is still uncertainty regarding what FC cut-off levels to apply for different grades of disease activity [9]. Therefore, we aimed in this study to test if the combination of FIT and an FC CALPRO test could improve the predictive ability when monitoring disease activity in patients with UC.

In our study, using a combination of FIT and FC CALPRO test a minor improvement in prediction of both endoscopic and histological colonic inflammation was seen in comparison to when using one of the tests alone. However, there is uncertainty if this minor gain in using the combination of FIT and FC compared to using each test alone has clinical significance when applied in daily clinical practice.

Mooiweer et al. performed a study on patients with IBD who performed surveillance colonoscopy [20]. They also found that a combination of FIT and an FC test (both positive test) performed better than a single marker test in predicting inflammatory activity in patients with UC. Also, in studies with the aim to discriminate IBD from non-organic disease the use of a combination of FIT and FC tests improve accuracy [15,17,19].

The accuracy of a test to predict inflammation depends on the method used, the cut-off used and how endoscopic/histological remission is defined. There are several FC tests on the market and the decision levels differ significantly [3]. In addition, there is an uncertainty what definition of remission of UC to recommend for clinical practice [26]. Up to 40% of patients with UC have despite clinical and endoscopic remission an inflammation histological activity [27] and in the present study only 20% of the patients fulfilled the criteria of ‘deep remission’ despite that the majority of patients had their colonoscopies performed as surveillance for colorectal dysplasia. Although, histological remission in patients with UC is associated with fewer clinical relapses [28], the cost-benefit of aiming of ‘deep remission’ still needs to be evaluated [29–31]. When setting treatment goals in clinical practice the costs of biological and newer treatment for UC, and the costs of colonoscopic examinations must also be considered before introducing more strict criteria for remission.

In the present study when applying a very strict definition of remission (‘deep remission’) with no endoscopic (Mayo score = 0) and no histological inflammation, unfortunately, none of the tests (including the combination of FIT and FC test) performed well enough to predict deep remission.

Therefore, the use of faecal markers to evaluate disease activity may only be applicable for the goal of remission using a more pragmatic moderate definition for endoscopic (Mayo score ≤ 1) and histological remission (no moderate/no severe inflammation). However, a recently presented Danish study using a combined endoscopic and histological score (Mayo and Geboes score) to define deep remission showed good performance using the FC CALPRO method for this purpose [32].

There are limitations in the present study. First, the patients in our study do not mirror the total UC population because the majority of patients in our study were patients with a long duration of the disease and that performed the colonoscopy in surveillance for dysplasia. In addition, the size of the cohort is limited. The results cannot automatically be transferred to centres using different methods, as there is no strict standardization for either faecal test used, and cut-offs, decision limits and methodological differences exist

The strengths of our study are that the subjects performed the stool sample only days before the endoscopic examination and that the endoscopists and pathologists were blinded by the outcome of the faecal tests.

To conclude, a combination of FIT and FC CALPRO shows minor improvement over using each test alone in predicting endoscopic and histological inflammatory activity. FC tests and FIT perform poor to predict deep remission, defined as a normal Mayo score (0) and no histological inflammation.

Acknowledgments

We would like to extend our thanks to all the staff on the Endoscopic unit at Umeå University hospital, Umeå, Sweden.

Disclosure statement

The authors declare that there are no conflicts of interest regarding the publication of this article.

Data availability statement

On special interest, data will be available from the corresponding author.

Additional information

Funding

The study was funded by the Västerbotten County Council, Umeå, Sweden.