nature publishing group

ORIGINAL CONTRIBUTIONS

941

see related editorial on page x

The Diagnostic Accuracy of Fecal Calprotectin During the Investigation of Suspected Pediatric Inammatory Bowel Disease
Paul Henderson, MBChB, MRPCH1,2, Aoife Casey, MBChB2, Sally J. Lawrence, MBChB, MRCPH2, Nicholas A. Kennedy, MBBS3, Kathleen Kingstone, BSc, MSc, MPhil4, Pam Rogers, RGN, RSCN2, Peter M. Gillett, MBChB, FRCP, FRCPH2 and David C. Wilson, MD, FRCP, FRCPCH1,2 OBJECTIVES:

Fecal calprotectin (FC) is elevated in patients with inammatory bowel disease (IBD). Studies evaluating FC during the initial investigation of children with suspected IBD have been limited, especially with regard to their small patient groups. We aimed to evaluate the diagnostic accuracy of FC in a large regional cohort of children undergoing full upper and lower endoscopy for suspected IBD, comparing FC with six common blood parameters. Using a retrospective casecontrol design all FC measurements carried out between 2005 and 2010 in children < 18 years old were obtained. All IBD and non-IBD patients who had a FC measurement available before full endoscopic evaluation for suspected bowel inammation were examined. FC was measured using the PhiCal Test. Multivariate analyzes and receiver operating characteristic curve generation were used to derive signicance. A total of 190 patients (91 IBD and 99 non-IBD controls) met the inclusion criteria. Median FC at diagnosis for the IBD group was 1,265 g/g (interquartile range (IQR) 7342,024 g/g), compared with 65 g/g (IQR 20235 g/g) in controls (P < 0.001). FC levels did not vary signicantly between patients with Crohns disease, ulcerative colitis, and IBD unclassied and were not inuenced by age or disease location. FC was found to be far superior to commonly utilized blood parameters such as C-reactive protein and white cell count (both P < 0.01), with an area under the curve of 0.93 (95% condence interval 0.890.97). endoscopy for suspected IBD, and elevated values should prompt further investigation.

METHODS:

RESULTS:

CONCLUSIONS: This study demonstrates that FC is an invaluable tool in determining those children who may require

SUPPLEMENTARY MATERIAL is linked to the online version of the paper at http://www.nature.com/ajg

Am J Gastroenterol 2012; 107:941949; doi:10.1038/ajg.2012.33; published online 28 February 2012

INTRODUCTION
The chronic inflammatory bowel diseases (IBD) encompass the major types Crohns disease (CD) and ulcerative colitis (UC) in addition to IBD unclassified (IBD-U). Up to 25% of the IBD patients present in childhood or adolescence, with epidemiological and natural history studies demonstrating that the incidence of pediatric-onset IBD (PIBD) is rising, especially with regard to early-onset CD (1,2). The initial investigation of children with suspected bowel inflammation includes both serum and fecal biomarkers to identify

those patients warranting endoscopic evaluation (3). One marker, fecal calprotectin (FC), is a calcium-binding protein found in neutrophilic granulocytes. FC has previously been shown to be markedly raised in children and adults with IBD (4,5), with its stability allowing the convenient collection of this non-invasive marker in both inpatient and outpatient settings (6). A recent meta-analysis evaluating the role of FC during the initial investigation of suspected IBD concluded that FC was a useful screening tool to identify patients requiring endoscopic assessment (7), though the discriminative power to safely exclude IBD

1 Child Life and Health, University of Edinburgh, Edinburgh, UK; 2Department of Pediatric Gastroenterology and Nutrition, Royal Hospital for Sick Children, Edinburgh, UK; 3Gastrointestinal Unit, Institute of Genetics and Molecular Medicine, University of Edinburgh, Edinburgh, UK; 4Department of Biochemistry, Western General Hospital, Edinburgh, UK. Correspondence: Paul Henderson, MBChB, MRPCH, Child Life and Health, University of Edinburgh, 20 Sylvan Place, Edinburgh EH9 1UW, UK. E-mail: paul.henderson2@nhs.net Received 4 November 2011; accepted 17 January 2012

2012 by the American College of Gastroenterology

The American Journal of GASTROENTEROLOGY

PEDIATRICS

942

Henderson et al.

was significantly better in adults than in children. However, the included pediatric studies presented FC levels in only 226 IBD patients; the median number of PIBD patients within the individual studies was only 31 (range, 1360), and not all were new diagnoses. Therefore, larger studies including only new patient referrals are still required to fully assess the utility of this biomarker during the initial assessment of suspected PIBD. Furthermore, the comparative value of FC measurement with commonly used blood parameters in pediatric patients has not yet been fully determined. We therefore hypothesized that the diagnostic accuracy of FC in suspected PIBD would be equivalent to endoscopy and superior to six commonly used blood parameters. We also aimed to describe the differences in FC levels between IBD types (CD, UC, and IBD-U) and non-IBD disease categories.

IBD patients

All incident cases of PIBD diagnosed by standard clinical, histological, and radiological findings (8,9) since August 1997 have been collected prospectively and recorded on a departmental database using Microsoft Access 2003 (Microsoft, Redmond, WA). From this database we identified patients who had a FC measured as part of their initial diagnostic work-up during the 6-year period of FC data from 2005 to 2010 (hereafter referred to as the IBD group). Detailed phenotypic characteristics (10) for these patients were also available through our Scotland-wide Medical Research Council funded Paediatric Inflammatory Bowel Disease Cohort and Treatment Study database.
Non-IBD (control) patients

PEDIATRICS

METHODS
Setting

The pediatric gastroenterology department based at the Royal Hospital for Sick Children in Edinburgh provides a regional service for a population of ~274, 000 children aged < 18 years in SouthEast Scotland. This tertiary center functions as a referral center for the three district general (community) hospitals with pediatric services, in addition to two adult academic gastroenterology services in Edinburgh and all adult gastroenterology departments within district general hospitals. Children presenting or referred with suspected bowel inflammation currently have their primary investigations and initial follow-up carried out at this center by experienced pediatric gastroenterologists. The biochemistry department based at the Western General Hospital in Edinburgh has routinely processed all FC samples in this region since October 2004, with access to all tests carried out in primary care and all hospital-based services.
Fecal calprotectin measurements

Through the departmental records containing the details of over 5,600 children referred to pediatric gastroenterology services since 2001, a computerized search using the remaining FC sample list determined those patients who had previously had contact with the service. These hospital records and departmental endoscopy lists were used to identify all patients undergoing both upper and lower endoscopy for the clinical suspicion of bowel inflammation, but where PIBD was subsequently excluded (hereafter referred to as the control group).
Exclusion criteria

Exclusion criteria for both the IBD and control groups were: (1) insufficient stool sample provided; (2) aged < 1 year or >18 years of age on the endoscopy date; (3) greater than a 6-month delay between the FC sample and the endoscopy date; (4) FC sample taken after endoscopy; (5) any previously known, hospital diagnosed, GI disease; and (6) previous upper or lower GI endoscopy.
Blood parameters

All FC measurements carried out between January 1 2005 and December 31 2010 in patients born after January 1 1987 (to ensure the inclusion of all patients potentially undergoing endoscopy before 18 years of age) in SouthEast Scotland were obtained retrospectively from the biochemistry department laboratory records. These data included patient demographics, the unique patient identifier and unique specimen number, the location code that specifies the sample origin (e.g., general practice, outpatient department), and the sample date and FC concentration in g/g of stool for all patients. The data also specified if samples were taken but were insufficient for processing. FC was measured by the PhiCal Test (Calpro AS, Lysaker, Norway) according to the manufacturers instructions; the local assay analytical range is 202,500 g/g. A normal FC value was taken to be < 50 g/g stool with the biochemistry laboratory routinely reporting FC results as possible gastrointestinal (GI) inflammation if between 51 and 100 g/g, GI inflammation if between 101 and 200 g/g and active GI inflammation if >200 g/g. In order to aid analysis of FC levels, samples reported as < 20 and >2,500 g/g were converted to 20 and 2,500 g/g, respectively.
The American Journal of GASTROENTEROLOGY

Blood results (taken within 6 months of endoscopy and closest to the date of the FC sample) were also obtained to compare the diagnostic utility of FC with commonly used blood parameters, namely: hemoglobin, platelet count, total white cell count, erythrocyte sedimentation rate (ESR), serum albumin, and C-reactive protein (CRP). Our regional pediatric biochemistryhematology laboratory normal values by age and sex range are shown in Supplementary Table 1. Within one referral center, plasma viscosity is routinely used as an alternative to ESR, therefore in six patients within the control group plasma viscosity results were used as a proxy for ESR with a reference range of 1.501.72 mPa/s.
Data recording and statistics

Demographic information, details of endoscopic assessment, FC and blood results, phenotypic information, prescribed medications at the time of the FC sample, and final diagnosis were recorded electronically using Microsoft Access 2007 (Microsoft). Statistical analysis was performed using R version 2.14.1 (R Foundation for Statistical Computing, Vienna, Austria) and GraphPad Prism version 4.03 (GraphPad Software, La Jolla, CA). Pearsons 2, Kruskal-Wallis, and MannWhitney U-tests were used where appropriate; multivariate analysis was achieved using multiple
VOLUME 107 | JUNE 2012 www.amjgastro.com

Calprotectin During the Diagnosis of Pediatric IBD

943

Ethical approval

Ethical approval was sought but deemed unnecessary as this was an anonymous, observational study of patients already under the care of PIBD services and under the umbrella of the Paediatriconset IBD Scottish Audit.

RESULTS
Group characteristics

A flow diagram outlining the patient selection process is shown in Figure 1. In total, 91 IBD patients and 99 non-IBD controls met the inclusion criteria; the baseline characteristics of both groups are outlined in Table 1. On univariate analysis the IBD group demonstrated an older age at endoscopy, had a shorter time between their FC sample and endoscopy and a higher terminal ileum (TI) intubation rate. However, only age at endoscopy and time between FC sample and endoscopy remained significant on multivariate analysis, suggesting that differences in the TI intubation rate between the groups was a result of a higher age at endoscopy in the IBD group (with endoscopy being less technically demanding

Children with IBD have a signicantly elevated FC at diagnosis compared with controls undergoing endoscopy

The median FC at diagnosis for the IBD group was 1,265 g/g (IQR 7342,024 g/g, range 262,500 g/g), which was higher
4,155 FC results identified

257 Insufficient samples 1,117 Repeat results

Excluded 2,781 Individual patients 845 Seen by tertiary pediatric GI services 333 Underwent endoscopy
Aged <1 or >18 years >6 Month delay between FC sample and endoscopy FC sample taken after endoscopy Previous GI diagnosis Previous endoscopy

Prospective departmental IBD database 113 IBD patients 0 2 12 0 8 91

220 non-IBD controls 0 42 18 12

Complete upper and lower endoscopy

14 134

Exclusions

99

Figure 1. Flow diagram showing the retrospective selection of study participants. FC, fecal calprotectin; GI, gastrointestinal; IBD, inammatory bowel disease.

2012 by the American College of Gastroenterology

The American Journal of GASTROENTEROLOGY

PEDIATRICS

logistic regression. The R packages epiR (11) and pROC (12) were used for further analysis. Hemoglobin and total white cell counts were standardized to the reference range used for 1218 year olds for the purpose of receiver operating characteristic (ROC) curve generation. Youden index was calculated as sensitivity + (specificity-1) (13). Statistical testing between receiver operating characteristic curves was performed using the DeLong (14) and bootstrap (15) methods. Statistical significance was taken to be a two-tailed P value of < 0.05.

in older children). In addition, the TI intubation rate was not significantly different between those children with and without a FC result available at endoscopy (P = 0.437). The IBD group consisted of 62 CD, 21 UC, and 8 IBD-U patients. In IBD patients without a TI biopsy obtained 30/34 (88%) had small bowel imaging carried out within a median time of 46 days (interquartile range (IQR) 2271 days) from endoscopy. All but one of the IBD group (99%) are currently recruited to our Paediatric Inflammatory Bowel Disease Cohort and Treatment Study cohort; the remaining patient died of a non-IBD-related illness before approach for possible consent. The diagnostic categories of the control group are shown in Table 2 and reflect their definitive diagnosis after follow-up of at least 12 months from the date of endoscopy (median follow-up 41 months (IQR 2453 months)). All children presented with one or more symptoms suggestive of bowel inflammation (Table 3); 37% had two recorded symptoms, and 23% had three or more. All controls were followed-up to at least definitive diagnosis or to discharge from pediatric services. Of those with no pathology identified (n = 11), all have been discharged from further follow-up on no medications. To our knowledge none of the control group have subsequently been diagnosed with PIBD by the end of December 2011. Unless they are now in adult services or have moved away from SouthEast Scotland, all potential PIBD diagnoses will have been re-referred to our service.

944

Henderson et al.

Table 1. Characteristics of the inammatory bowel disease and control groups

Table 3. Symptoms and signs in control patients suggestive of bowel inammation/inammatory bowel disease
Symptom/sign Altered bowel habit Abdominal pain Percentage 76 55 48 10 10 3 2

PEDIATRICS

Characteristic Number of cases Male sex (n (%)) Age at endoscopy (years (IQR)) Terminal ileal biopsy obtained (n (%)) Median time from FC result to endoscopy (days (IQR)) Median time from FC result to blood result (days (IQR)) Median time from blood result to endoscopy (days (IQR))

IBD group 91 56 (62) 12.6 (9.514.0) 57 (63) 18 (744) 3 (09) 13 (237)

Control group 99 55 (56) 9.3 (5.212.7) 46 (46) 48 (2987) 9 (330) 56 (2993)

Difference between groups (P value)*

0.403 < 0.001 0.037 < 0.001 NA NA

Rectal bleeding Growth distortion Vomiting Recurrent mouth ulcers Iron decient anemia

FC, fecal calprotectin; IBD, inammatory bowel disease; IQR, interquartile range; NA, not applicable. *P values as determined with 2 or MannWhitney U-tests.

Table 2. Non-IBD (control) diagnostic categories and median fecal calprotectin values
Diagnostic category Irritable bowel syndrome Non-specic colitis No pathology identied Post-infectious enteropathy Cows milk/wheat intolerance Pinworms Allergic enteropathy Celiac disease Miscellaneous Total Number 32 12 11 11 8 7 5 3 10 99 Median FC (IQR) 48 (21123) 68 (25258) 20 (20275) 45 (20440) 67 (41115) 110 (29275) 80 (501508) 350 (NA) 177 (61292)

FC, fecal calprotectin; IBD, inammatory bowel disease; IQR, interquartile range; NA, not applicable.

(P < 0.001) than the control group with a median FC of 65 g/g (IQR 20235 g/g, range 202,500 g/g). Within the IBD group only two patients had a FC < 50 g/g recorded at initial presentation, an 11-year-old girl with colonic CD (Montreal L2 at diagnosis) who has required immunomodulators and adalimumab, and a 4-year-old boy with pancolitic IBD-U who is currently stable on maintenance mesalamine. Only one child in the control group had a FC >2,500 g/g, a 2-year-old girl with allergic enteropathy who settled on dietary restrictions and who continues to have no evidence of IBD during ongoing review. The diagnostic accuracy for FC at different cutoff levels is shown in Table 4. Using the manufacturers normal cutoff of >50 g/g
The American Journal of GASTROENTEROLOGY

gives an excellent sensitivity (0.98), but a poor specificity (0.44) for IBD. This specificity increases steadily with increasing FC levels until the pay-off between sensitivity and specificity (calculated by the Youden index) plateaus at around 200300 g/g. As discussed above it can be seen that the IBD group had a higher rate of TI biopsy, therefore to ensure that this was not confounding the results regarding diagnostic accuracy, only those children with an available TI biopsy (n = 103) were analyzed separately. This demonstrated that using a cutoff of >50 g/g gave almost identical results as when the entire cohort was evaluated (i.e., sensitivity 0.98 (95% confidence interval (CI) 0.910.99); specificity 0.46 (95% CI 0.310.61); negative predictive value 0.95 (95% CI 0.770.99); positive predictive value 0.69 (95% CI 0.580.79)). It is important also to note that 34% (n = 63) of FC results returned after the endoscopic assessment was performed, and that 14% (n = 27) of FC results were not known in the preceding 2 weeks before endoscopic assessment. Given our minimum 2 week delay to elective GI endoscopy over the full 6-year period, these 48% (n = 90) of FC results could not have influenced our decision making with regard to performance of endoscopy, nor the occurrence of TI intubation in 34% of patients. Furthermore, even for the 52% of patients where FC results were known before 2 weeks before the endoscopic assessment, none were cancelled after confirmation of procedure based on FC result, nor had procedures expedited based on FC resultthis occurred only for rapid clinical deterioration. Additional analysis within our complete cohort (n = 190) showed that the pre-test probability of IBD was 0.48 (i.e., 91/190), and that utilizing a positive result of >200 g/g provided a post-test probability of 0.77 (111 patients had a FC >200 g/g with 85 having IBD), an increase of 60%.
FC levels in children with IBD are not inuenced by sex, age, IBD type, or disease location

Including all 91 children with IBD and categorizing the entire IBD group by sex showed no difference between median FC levels in males (1,265 g/g (IQR 6581,864 g/g)) and females (1,250 g/g (IQR 8902,070 g/g)) (P = 0.695). There were also no differences between the sexes when the CD (P = 0.508), UC (P = 0.859), and IBD-U (P = 0.999) groups were analyzed separately. Although comparing age with FC level for the entire IBD group demonstrated a significant correlation (P = 0.037, Spearmans rho = 0.245),
VOLUME 107 | JUNE 2012 www.amjgastro.com

Calprotectin During the Diagnosis of Pediatric IBD

945

Table 4. Measures of diagnostic accuracy for increasing levels of fecal calprotectin and commonly measured blood parameters in children with suspected inammatory bowel disease
Sens (95% CI) Fecal calprotectin Cutoff (g/g) >50 >100 >200 >300 >800 0.98 (0.921.00) 0.97 (0.910.99) 0.93 (0.860.98) 0.89 (0.810.95) 0.73 (0.620.81) 0.44 (0.340.55) 0.59 (0.480.68) 0.74 (0.640.82) 0.83 (0.740.90) 0.95 (0.890.98) 0.96 (0.850.99) 0.95 (0.860.99) 0.92 (0.840.97) 0.89 (0.810.95) 0.79 (0.710.86) 0.62 (0.530.70) 0.68 (0.590.76) 0.77 (0.670.84) 0.83 (0.740.90) 0.93 (0.840.98) 0.42 (0.270.55) 0.55 (0.380.68) 0.67 (0.500.80) 0.72 (0.540.84) 0.68 (0.520.80) 1.8 (1.52.1) 2.3 (1.83.0) 3.6 (2.55.0) 5.2 (3.38.0) 14.5 (6.134.4) Spec (95% CI) NPV (95% CI) PPV (95% CI) Youden Indexa (95% CI) LR + ve (95% CI)

Blood parameters (using normal pediatric values outlined in Supplementary Table 1) Parameter Hemoglobin Total WCC Platelets ESR Albumin CRP 0.64 (0.530.73) 0.09 (0.040.17) 0.43 (0.320.54) 0.67 (0.570.77) 0.22 (0.140.33) 0.55 (0.440.66) 0.79 (0.630.87) 0.99 (0.940.99) 0.92 (0.850.97) 0.89 (0.810.95) 0.99 (0.940.99) 0.91 (0.830.96) 0.69 (0.590.78) 0.53 (0.450.60) 0.62 (0.530.70) 0.74 (0.640.82) 0.56 (0.490.64) 0.67 (0.580.76) 0.75 (0.640.84) 0.89 (0.521.00) 0.84 (0.710.94) 0.86 (0.750.93) 0.95 (0.761.00) 0.86 (0.740.94) 0.43 (0.230.61) 0.08 ( 0.02 to 0.18) 0.35 (0.170.50) 0.56 (0.370.72) 0.21 (0.080.33) 0.46 (0.280.62) 3.1 (2.04.7) 8.2 (1.06.4) 5.6 (2.611.8) 6.1 (3.411.2) 20.4 (2.8149.1) 6.3 (3.112.5)

CI, condence interval; CRP, C-reactive protein; ESR, erythrocyte sedimentation rate; LR + ve, positive likelihood ratio; NPV, negative predictive value; PPV, positive predictive value; sens, sensitivity; spec, specicity; WCC, white cell count. a Youden Index, sensitivity + (specicity-1).

this was lost when analyzed in a multivariate model that included ESR, CRP, and albumin (P = 0.375), likely reflecting the fact that disease severity was a confounder. To determine whether disease type or location influenced FC levels, detailed phenotypic information was collected from our Paediatric Inflammatory Bowel Disease Cohort and Treatment Study database; their Montreal (16) classification for location at diagnosis is shown in Supplementary Table 2. There was no difference (P = 0.710) between CD, UC, and IBD-U patients, with these three types having median FC levels of 1,258 g/g (IQR 710 1,671 g/g), 1250 g/g (IQR 9252,200 g/g), and 1,463 g/g (IQR 8982,125 g/g), respectively. There was neither any difference in the median FC levels between those with upper intestinal CD location (1,440 g/g (IQR 8852,034 g/g)) and those without (1,220 g/g (IQR 400 1,340 g/g)) as defined as the presence or absence of Montreal L4 disease (P = 0.077), nor any difference observed when comparing the median FC levels in CD patients with any ileal (L1L4 or L3L4) disease (1,266 g/g (IQR 8751,757 g/g)) and those without (1,295 g/g (IQR 4941,832 g/g)) (P = 0.694). To allow meaningful numeric comparison of UC and IBD-U disease location both groups were combined and each patient re-classified according to the newly described Paris classification of PIBD (17). This demonstrated no difference (P = 0.536) between those with extensive pancolonic disease (Paris E4, disease proximal
2012 by the American College of Gastroenterology

to the hepatic flexure) and those with more limited disease (Paris E1E3, all with disease of varying extent from the rectum yet distal to the hepatic flexure) who had median FC levels of 1,480 g/g (IQR 9782,135 g/g) and 963 g/g (IQR 6912,135 g/g), respectively. Similarly, to evaluate all colonic IBD, CD patients with isolated colonic or ileo-colonic disease (L2 or L3 only) had a similar median FC level of 1,230 g/g (IQR 4081,421 g/g) compared with those with UC and IBD-U combined (1,300 g/g (IQR 925 2,200 g/g)) (P = 0.324).
FC does not vary signicantly between the diagnostic categories within the control group

The median (IQR) for each of the control group diagnostic categories is shown in Table 2. The miscellaneous group comprised children with functional abdominal pain (n = 2), colonic polyps (n = 2), gastritis (n = 2), Meckels diverticulum (n = 1), pancreatic insufficiency (n = 1), and perianal abscess (n = 1) and an as yet undiagnosed growth restriction syndrome (n = 1). There was no difference in the median FC values between the diagnostic categories (P = 0.575) with all median values < 200 g/g (except for the celiac disease group that only had three members). The high third quartile within the allergic enteropathy group was as a result of the high FC level obtained in the 2-year-old girl mentioned above. As bleeding per rectum often leads to a differential diagnosis of PIBD we assessed the median FC in controls presenting with (n = 48)
The American Journal of GASTROENTEROLOGY

PEDIATRICS

946

Henderson et al.

PEDIATRICS

and without (n = 51) a history of per rectum blood, revealing no difference in median FC values of 60 g/g (IQR 20218 g/g) and 65 g/g (IQR 25275), respectively (P = 0.512). In addition, no difference was demonstrated when correlating age with FC levels across all control group diagnostic categories (P = 0.051, Spearmans rho = 0.197).
Sensitivity

1.0

0.8

Medications, including PPIs, do not seem to inuence FC levels

0.6

Within the IBD group 24% of PIBD cases and 33% of controls were on any oral medication at the time of FC sampling (P = 0.219); however, no children were currently receiving medication directly related to their suspected bowel inflammation. Combining all patients, there was no difference (P = 0.519) in median FC between those currently prescribed (275 g/g (IQR 401,265)) or not prescribed (385 g/g (IQR 601,275 g/g)) any medication. Within the control group the median FC level of patients prescribed PPIs (n = 10) was 108 g/g (IQR 20240 g/g) that was similar to those not on PPIs (n = 89) (60 g/g (IQR 21240 g/g)) (P = 0.906).
FC performs better than commonly used blood parameters as a diagnostic biomarker during the evaluation of children with suspected IBD

0.4

0.2

0.0 1.0 0.8 0.6 0.4 0.2 0.0

Specificity
Variable (AUC (95% CI)) Calprotectin/Albumin combined (0.96 (0.930.99)) Fecal calprotectin (0.93 (0.890.97)) Albumin (0.91 (0.860.95)) ESR (0.84 (0.770.90)) CRP (0.83 (0.770.89)) Platelets (0.79 (0.730.86)) Standardized Hb (0.78 (0.710.85)) Standardized WCC (0.70 (0.630.78))

To compare the performance of FC with six commonly used blood parameters, blood results taken at a similar time as the FC measurement were analyzed (median time difference 6 days (IQR 128 days); see Table 1). The availability of each blood parameter for each group is shown in Supplementary Table 3. The diagnostic accuracy of each blood parameter in comparison to FC is outlined in Table 4. Figure 2 demonstrates that the area under the curve for FC was greater than all six blood parameters at 0.93 (95% CI 0.890.97), and significantly higher than ESR (P = 0.011), CRP (P = 0.006), total white cell count (P < 0.001), hemoglobin (P < 0.001), and platelet count (P < 0.001), but was not significantly greater than albumin (P = 0.374). Further analysis of albumin as a predictor for IBD revealed that the optimum threshold was in fact 41g/l (within our normal pediatric reference range of 3350 g/l), with the relevant diagnostic specificity using our normal lower limit of normal being far inferior (Table 4). However, by combining FC and serum albumin (using an optimized formula of (60 + FC/100 g/g) (serum albumin in g/L)) it can be seen that the area under the curve is improved at 0.96 (95% CI 0.930.99); however, this was not significantly different to FC alone (P = 0.227). Using the above formula and utilizing a cutoff of 20 produced the following diagnostic accuracy indicators: sensitivity 0.97 (95% CI 0.900.99), specificity 0.81 (95% CI 0.710.89), negative predictive value 0.96 (95% CI 0.880.99), and positive predictive value 0.84 (95% CI 0.760.91).

Figure 2. Receiver operating characteristic (ROC) curves and corresponding area under the curve (AUC) for fecal calprotectin and commonly measured blood parameters in pediatric patients with suspected IBD. CI, condence interval; CRP, C-reactive protein; ESR, erythrocyte sedimentation rate; Hb, hemoglobin; WCC, white cell count.

DISCUSSION
Our results demonstrate that FC is a highly useful biomarker during the initial investigation of suspected PIBD. Using only children presenting with suspected bowel inflammation, without any known GI diagnosis, and subsequently undergoing their first full
The American Journal of GASTROENTEROLOGY

endoscopic investigation, we clearly show that FC is markedly raised in those with IBD, with no influence of IBD type or location. We have also provided evidence that FC performs better than all commonly used blood parameters with an area under the curve of 0.93 (95% CI 0.890.97). Although some would argue that FC is not needed in a classical presentation of CD or UC, our data show the utility of FC across the whole clinical spectrum of all types of IBD (e.g., those without luminal CD or with mainly extraintestinal symptoms) and, even more importantly, show the potential utility of FC in deciding which children and teenagers presenting with possible gut inflammation do not need endoscopic assessment. Calprotectin is a 24-kDa heterodimer composed of two calcium-binding proteins belonging to the S100 group of proteins (S100A8 and S100A9) (18), constituting 60% of the cytosolic protein in human neutrophils (6). FC has both bacteriostatic and fungistatic properties (19,20), mainly mediated by zinc chelation via histidine-rich regions of the calprotectin molecule (21). In PIBD FC has been shown to correlate with endoscopic severity at colonoscopy (22), disease activity in UC (23), and has also demonstrated usefulness in predicting disease relapse (24). There are currently only a limited number of studies using FC during the
VOLUME 107 | JUNE 2012 www.amjgastro.com

Calprotectin During the Diagnosis of Pediatric IBD

947

initial investigation of pediatric bowel inflammation (i.e., before endoscopic confirmation of IBD) (4,2529). These studies have often included small numbers of IBD patients. Although some have reported similar findings to our study with regard to median FC levels at diagnosis (28), they often failed to provide a detailed analysis of sub-phenotypic characteristics in a truly representative group of potential IBD patients (30). Our extensive clinical use of FC as a biomarker in GI-related disease for >7 years leads to this study having several strengths in relation to study design and analyzes performed. First, our regional IBD cohort is representative of our larger Scottish nationwide cohort with regard to demographic composition (1) and disease location at diagnosis (10). Previous studies often excluded younger children (23,30), those with a high suspicion of IBD (26) and those on medications (28), potentially skewing subsequent analyzes. This is reflected in our modest pre-test probability of 0.48, which is lower than reported in two recently published relevant papers on FC usage (28,29). In addition, previous groups often used a control group of patients with no known GI symptoms or signs to generate sensitivity and specificity (23,27), which is certainly not applicable in a real world clinical setting of a pediatric GI service. In fact within our study a selection bias does occur from the entire cohort assessed using FC levels, but this bias works against the test as only sicker children with increased FC levels are likely to be referred to specialist GI services and undergo endoscopy. Second, only including those undergoing full endoscopy, and our reporting of small bowel investigation within the IBD group, has ensured robust phenotyping of all patients; many studies include only patients undergoing colonoscopy (4) or provide no details of endoscopic investigation (27). Third, our large study group (n = 190) has allowed further meaningful examination of particular sub-groups. Previous studies have frequently combined those with previously confirmed IBD (23) (therefore presenting a heterogeneous group with both de novo disease and established IBD), or presented small numbers of each IBD type (28). The retrospective design of our study does, however, produce potential limitations. First, as the clinical utility of FC was not evaluated during the initial decision to perform endoscopic assessment, we were unable to determine whether or not the FC level contributed to the gastroenterologists choice to perform endoscopy. Similarly, the influence of the various blood parameters on the ultimate decision to perform endoscopy could not be elucidated. This is an important factor when assessing the usefulness of any biomarker in this context as prior knowledge of the FC result may have led to the avoidance of unnecessary endoscopic procedures, or conversely the delay in IBD diagnosis (which may have occurred in two of our patients if FC level had been used in isolation at diagnosis). Second, during the study we collected data within a time period of 6 months before endoscopy with other prospective studies standardizing the sampling time (e.g., 1 week before endoscopy) to potentially eliminate a confounding effect of variable disease activity on their results (22). Although useful during the initial phases of determining the diagnostic accuracy of a certain biomarker, our approach is more comparable with
2012 by the American College of Gastroenterology

current clinical practice, with children often attending general practice or a general pediatrician with non-specific GI symptoms before assessment by a pediatric gastroenterologist. Delay to endoscopic assessment of suspected pediatric IBD has been a major issue in many countries including the United Kingdom, although changes in the National Health Service service design in Scotland has markedly reduced this waiting time over the period from 2005 to 2011. Third, another potential difference from previous studies is the use of a relatively low upper limit of FC assay. For example Perminow et al. (29) were able to demonstrate a significantly higher FC level in those with CD vs. UC (1,181 g/g, range 116,123 g/g; 1,250 g/g, range 138,625 g/g, respectively) due in all likelihood to their (undisclosed) higher reference range. Finally, two potential confounding factors in this study were the differing rates of TI biopsy in the IBD and control groups and the effect of oral medications. With regard to these aspects our analysis has shown that TI intubation rates were likely a result of the higher age of the IBD group, that removal of those without a TI biopsy did not change our main results of diagnostic accuracy and that TI intubation rates did not differ between those with and without a FC level available at endoscopy; we, however, acknowledge that this study was not powered to look at the effects of drugs on FC levels. Acknowledging the relative weakness of retrospective study design, we are confident that (i) our robust choice of inclusions (i.e., only children presenting with suspected bowel inflammation, without known GI diagnosis, and subsequently undergoing their first ever endoscopic investigation); (ii) our evaluation of all of the FC levels performed in our region over the 6-year-study period; (iii) our ability to review and follow-up all suspected pediatric cases of gut inflammation in a defined geographical region (one child was initially placed in the no pathology identified control group but was very recently diagnosed as non-specific colitis and was re-assigned as such); and (iv) our knowledge of all new regional cases of PIBD (within pediatric services in primary, secondary, and tertiary care) to the end of December 2011, when taken together, add considerable confidence to the generalizability of these findings to all PIBD services worldwide. On the basis of our clinical experience, laboratory guidance and relevant literature (23), we have routinely used a FC cutoff of 200 g/g as our threshold for the suspicion of a new IBD diagnosis, accepting the need for full clinical history, examination, and other blood tests. This has been validated by our results with the Youden index, suggesting a FC level between 200 and 300 g/g providing an optimum sensitivity/specificity. Although several newly described serum markers have been identified as possible markers of IBD, these are often present at higher levels in certain sub-phenotypes (31) or are only available as research tools (32). Commonly measured blood parameters remain the investigations of choice, with ESR (33) and CRP (34) currently providing the best indication of possible IBD, but have a relatively poor diagnostic accuracy, especially specificity. Previous work by our group demonstrated that the use of common blood tests could be enhanced significantly by the inclusion of FC in a panel of inflammatory
The American Journal of GASTROENTEROLOGY

PEDIATRICS

948

Henderson et al.

markers during the investigation of suspected IBD (35). This has been further enhanced with the development of a combined FC and serum albumin score described above, which provided a better (although not statistically significantly) area under the curve than FC alone. GI endoscopic assessment is difficult for children, with the need for hospital or day-case admission. For the children this involves fasting, bowel preparation (if undergoing lower GI evaluation), and anesthesia/sedation (admission to a day-case unit and total intravenous anesthesia usage is our current design for elective procedures). For their parents, anxiety and time away from employment or dependent younger children is also a potential issue. It is also expensive, with a wide variation between different countries. By comparison FC can be obtained by providing the family with instructions, a sample pot, a prefilled laboratory form and suitable packaging for postage to the appropriate IBD unit. It is relatively cheap in the United Kingdom; currently, our National Health Service laboratory pays US$785 per kit (PhiCal Test) equating to US $10.90 per sample. The National Health Service requests are charged at US $40 per test as this currently used test remains relatively labor intensive (K Kingstone, personal communication). In conclusion, we have shown that FC is significantly raised in children with IBD compared with non-IBD, scoped controls, and have also demonstrated that FC provides greater diagnostic accuracy than other commonly used blood parameters. We feel that the characteristics of both groups and the timing of their investigations represent a true reflection of the investigative procedures carried out in children with suspected bowel inflammation and that FC should now be used routinely during the initial assessment of these children. Further studies are now required to fully determine the effect of FC measurement on endoscopy rates, with the potential to reduce the number of children undergoing endoscopic assessment for suspected IBD, therefore reducing costs, streamlining pediatric GI endoscopy services, and reducing both child and family distress and inconvenience.
ACKNOWLEDGMENTS

Study Highlights
WHAT IS CURRENT KNOWLEDGE

PEDIATRICS

3Pediatric inammatory bowel disease (PIBD) represents a phenotypically distinct subset of disease. 3A recent meta-analysis concluded that the discriminative 3However, the median number of PIBD patients within
power of fecal calprotectin (FC) to safely exclude IBD was signicantly better in adults than in children. previous studies evaluating FC was only 31 and not all represented new diagnoses. accuracy of FC with other commonly measured blood parameters has not yet been performed. WHAT IS NEW HERE

3A large, comprehensive study comparing the diagnostic 3The median fecal calprotectin (FC) value during the 3FC at diagnosis is not inuenced by age, sex, PIBD type, or disease location. 3FC provides the clinician with a signicantly greater degree 3The routine use of FC in the pediatric setting should
signicantly enhance our ability to more accurately screen children for IBD. of diagnostic accuracy than other commonly measured blood parameters. diagnosis of pediatric inammatory bowel disease (PIBD) is signicantly higher than non-IBD controls undergoing upper and lower endoscopy.

REFERENCES
1. Henderson P, Hansen R, Cameron FL et al. Rising incidence of pediatric inflammatory bowel disease in Scotland. Inflamm Bowel Dis 2011. Published online first:17 June 2011. doi:10.1002/ibd.21797. 2. Benchimol EI, Fortinsky KJ, Gozdyra P et al. Epidemiology of pediatric inflammatory bowel disease: a systematic review of international trends. Inflamm Bowel Dis 2011;17:42339. 3. Beniwal P, Harrell L. The status of diagnostic markers for inflammatory bowel disease. Curr Gastroenterol Rep 2010;12:47984. 4. Diamanti A, Panetta F, Basso MS et al. Diagnostic work-up of inflammatory bowel disease in children: the role of calprotectin assay. Inflamm Bowel Dis 2010;16:192630. 5. Schoepfer AM, Trummler M, Seeholzer P et al. Discriminating IBD from IBS: comparison of the test performance of fecal markers, blood leukocytes, CRP, and IBD antibodies. Inflamm Bowel Dis 2008;14:329. 6. Roseth AG, Fagerhol MK, Aadland E et al. Assessment of the neutrophil dominating protein calprotectin in feces. A methodologic study. Scand J Gastroenterol 1992;27:7938. 7. van Rheenen PF, Van de Vijver E, Fidler V. Faecal calprotectin for screening of patients with suspected inflammatory bowel disease: diagnostic meta-analysis. BMJ 2010;341:c3369. 8. Lennard-Jones JE. Classification of inflammatory bowel disease. Scand J Gastroenterol 1989;24 (Suppl 170): 26. 9. IBD working group of the European Society for Paediatric Gastroenterology Hepatology and Nutrition. Inflammatory bowel disease in children and adolescents: recommendations for diagnosis - the Porto criteria. J Pediatr Gastroenterol Nutr 2005;41:17. 10. Van Limbergen J, Russell RK, Drummond HE et al. Definition of phenotypic characteristics of childhood-onset inflammatory bowel disease. Gastroenterology 2008;135:111422. 11. Stevenson M. epiR: Functions for analysing epidemiological data. (R package version 0.9-33). Available from: http://cran.r-project.org/web/packages/ epiR/index.html (accessed: 2 November 2011). 12. Robin X, Turck N, Hainard A et al. pROC: an open-source package for R and S+ to analyze and compare ROC curves. BMC Bioinformatics 2011;12:77. 13. Youden WJ. Index for rating diagnostic tests. Cancer 1950;3:325.

We thank Dr John Morrice, Mrs Hazel Drummond, and Dr Carol Dryden for their help with data collection.
CONFLICT OF INTEREST

Guarantor of the article: David C. Wilson, MD, FRCP, FRCPCH. Specific author contributions: P.H. and D.C.W. prepared the manuscript with additions, comments, and corrections by all the authors. P.H., S.J.L., A.C., K.K., and P.R. collected the data. P.H. and N.A.K. analyzed the complete data set. P.M.G. and D.C.W. are the IBD clinical leads within the study center. Financial support: P.H. is funded by a Medical Research Council project grant for Paediatric Inflammatory Bowel Disease Cohort and Treatment Study (no. G0800675). N.A.K. is funded by grants from the Chief Scientist Office in Scotland (no. ETM/75) and Cure Crohns Colitis. Potential competing interests: None.
The American Journal of GASTROENTEROLOGY

VOLUME 107 | JUNE 2012 www.amjgastro.com

Calprotectin During the Diagnosis of Pediatric IBD

949

14. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 1988;44:83745. 15. Hanley JA, McNeil BJ. A method of comparing the areas under receiver operating characteristic curves derived from the same cases. Radiology 1983;148:83943. 16. Silverberg MS, Satsangi J, Ahmad T et al. Toward an integrated clinical, molecular and serological classification of inflammatory bowel disease: report of a working party of the 2005 Montreal World Congress of Gastroenterology. Can J Gastroenterol 2005;19 (Suppl A): 536. 17. Levine A, Griffiths A, Markowitz J et al. Pediatric modification of the Montreal classification for inflammatory bowel disease: the Paris classification. Inflamm Bowel Dis 2011;17:131421. 18. Fagerhol MK, Dale I, Anderson T. Release and quantitation of a leucocyte derived protein (L1). Scand J Haematol 1980;24:3938. 19. Lusitani D, Malawista SE, Montgomery RR. Calprotectin, an abundant cytosolic protein from human polymorphonuclear leukocytes, inhibits the growth of Borrelia burgdorferi. Infect Immun 2003;71:47116. 20. Murthy AR, Lehrer RI, Harwig SS et al. In vitro candidastatic properties of the human neutrophil calprotectin complex. J Immunol 1993;151:6291301. 21. Loomans HJ, Hahn BL, Li QQ et al. Histidine-based zinc-binding sequences and the antimicrobial activity of calprotectin. J Infect Dis 1998;177:8124. 22. Bunn SK, Bisset WM, Main MJ et al. Fecal calprotectin: validation as a noninvasive measure of bowel inflammation in childhood inflammatory bowel disease. J Pediatr Gastroenterol Nutr 2001;33:1422. 23. Bremner A, Roked S, Robinson R et al. Faecal calprotectin in children with chronic gastrointestinal symptoms. Acta Paediatr 2005;94:18558. 24. Walkiewicz D, Werlin SL, Fish D et al. Fecal calprotectin is useful in predicting disease relapse in pediatric inflammatory bowel disease. Inflamm Bowel Dis 2008;14:66973. 25. Fagerberg UL, Loof L, Myrdal U et al. Colorectal inflammation is well predicted by fecal calprotectin in children with gastrointestinal symptoms. J Pediatr Gastroenterol Nutr 2005;40:4505.

26. Canani RB, de Horatio LT, Terrin G et al. Combined use of noninvasive tests is useful in the initial diagnostic approach to a child with suspected inflammatory bowel disease. J Pediatr Gastroenterol Nutr 2006;42:915. 27. Bonnn Toms A, Vila Vidal M, Rosell Camps A. Fecal calprotectin as a biomarker to distinguish between organic and functional gastrointestinal disease. Rev Esp Enferm Dig 2007;99:68993. 28. Sidler MA, Leach ST, Day AS. Fecal S100A12 and fecal calprotectin as noninvasive markers for inflammatory bowel disease in children. Inflamm Bowel Dis 2008;14:35966. 29. Perminow G, Brackmann S, Lyckander LG et al. A characterization in childhood inflammatory bowel disease, a new population-based inception cohort from South-Eastern Norway, 200507, showing increased incidence in Crohns disease. Scand J Gastroenterol 2009;44:44656. 30. Fagerberg UL, Loof L, Lindholm J et al. Fecal calprotectin: a quantitative marker of colonic inflammation in children with inflammatory bowel disease. J Pediatr Gastroenterol Nutr 2007;45:41420. 31. Russell RK, Ip B, Aldhous MC et al. Anti-Saccharomyces cerevisiae antibodies status is associated with oral involvement and disease severity in Crohn disease. J Pediatr Gastroenterol Nutr 2009;48:1617. 32. Lehrke M, Konrad A, Schachinger V et al. CXCL16 is a surrogate marker of inflammatory bowel disease. Scand J Gastroenterol 2008;43: 2838. 33. Mack DR, Langton C, Markowitz J et al. Laboratory values for children with newly diagnosed inflammatory bowel disease. Pediatrics 2007;119:11139. 34. Tsampalieros A, Griffiths AM, Barrowman N et al. Use of C-reactive protein in children with newly diagnosed inflammatory bowel disease. J Pediatr 2011;159:3402. 35. Quail MA, Russell RK, Van Limbergen JE et al. Fecal calprotectin complements routine laboratory investigations in diagnosing childhood inflammatory bowel disease. Inflamm Bowel Dis 2009;15: 7569.

2012 by the American College of Gastroenterology

The American Journal of GASTROENTEROLOGY

PEDIATRICS

Copyright of American Journal of Gastroenterology is the property of Nature Publishing Group and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.