Professional Documents
Culture Documents
Background: C-reactive protein (CRP) is an acute phase reactant. Patients with pediatric inflammatory bowel disease (PIBD) differ from adult patients
with inflammatory bowel disease with regard to phenotype, inflammatory profile, and treatment response. We hypothesized that variations in CRP and
CRP genotype influence PIBD phenotype, natural history, and remission after anti-tumor necrosis factor alpha therapy.
Methods: Six single nucleotide polymorphisms tagging CRP (rs1935193, rs1130864, rs1205, rs1417938, rs11265263, and rs1800947) were genotyped
in 465 patients with PIBD (diagnosed ,17 yr). Phenotyping was serially performed until last follow-up and serum CRP levels recorded at diagnosis and
before biological therapy in a subgroup.
Results: CRP haplotype (ATGCTC) differed in those diagnosed ,10 years, with rs1205T more frequent in Crohn’s disease (CD) than ulcerative colitis
(UC) (P ¼ 0.009); the haplotype ATGCTC was less frequent in UC (P ¼ 0.002). Three single nucleotide polymorphisms (rs1205, rs1130864, and
rs1417938) showed association with elevated CRP levels at diagnosis. CRP genotype had no association with CD phenotype or natural history. CRP was
more frequently raised at diagnosis in CD than UC (63% versus 22%, P , 0.0001). Elevated CRP at diagnosis was associated with a higher risk of
progression to surgery in patients with CD (P , 0.0001) and the need for azathioprine in the overall PIBD cohort (P ¼ 0.002). There was no effect of
CRP genotype or serum CRP on the achievement of remission using anti–tumor necrosis factor alpha therapy.
Conclusions: CRP and CRP genotype differ between pediatric patients with CD and UC with a high inflammatory burden at diagnosis suggesting
a worse prognosis. Additional evaluation of CRP in inflammatory bowel disease pathogenesis and natural history is now warranted.
(Inflamm Bowel Dis 2015;21:596–605)
Key Words: inflammatory bowel disease, pediatric, C-reactive protein, polymorphism, Crohn’s disease
Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on
the journal’s Web site (www.ibdjournal.org).
Received for publication September 25, 2014; Accepted October 23, 2014.
From the *Department of Pediatric Gastroenterology and Nutrition, Royal Hospital for Sick Children, Edinburgh, United Kingdom; †Child Life and Health, University of
Edinburgh, Edinburgh, United Kingdom; ‡Gastrointestinal Unit, Institute of Genetics and Molecular Medicine, University of Edinburgh, Edinburgh, United Kingdom; §IBD
Center, Division of Pediatric Gastroenterology and Nutrition, IWK Health Center, Dalhousie University, Halifax, NS, Canada; and kDepartment of Pediatric Gastroenterology,
Royal Hospital for Sick Children, Glasgow, United Kingdom.
P. Henderson was funded by a Medical Research Council project grant for Pediatric Inflammatory Bowel Disease Cohort and Treatment Study (No. G0800675) and
a Shire Innovation Award for Specialist Registrars. D. C. Wilson, R. K. Russell and J. Satsangi are grant holders of the Medical Research Council patient research cohorts
initiative grant (G0800675) for Pediatric Inflammatory Bowel Disease Cohort and Treatment Study. N. A. Kennedy is funded by a Research Training Fellowship from the
Wellcome Trust (WT097943MA). F. L. Cameron is supported by a CICRA Research Fellowship. R. K. Russell is supported by an NHS Research Scotland Career
Fellowship Award and J. E. V. Limbergen by a NASPGHAN/CCFA Young Investigator Development Award. Additional support was provided by the GI-Nutrition
Research fund, Child Life and Health, University of Edinburgh.
The authors have no conflicts of interest to disclose.
Reprints: Paul Henderson, PhD, Child Life and Health, University of Edinburgh, 20 Sylvan Place, EH9 1UW Edinburgh, United Kingdom (e-mail: paul.
henderson2@nhs.net).
Copyright © 2015 Crohn’s & Colitis Foundation of America, Inc.
DOI 10.1097/MIB.0000000000000296
Published online 29 January 2015.
596 | www.ibdjournal.org Inflamm Bowel Dis Volume 21, Number 3, March 2015
2015
Inflamm Bowel Dis Volume 21, Number 3, March 2015 Pediatric Inflammatory Bowel Disease
kappa-light-chain-enhancer of activated B cells.5 Additionally, effective treatment also varying widely20; the reasons for these
other transcription factors at the CRP promoter site (such as person-to-person variations are not yet clear. Specifically, regard-
STAT3, cRel, Oct1, HNF-1, and HNF-3) have been shown to ing IBD, serum CRP has been shown to have some utility at
be important during induction.628 Regarding the immunological diagnosis in patients suspected of having bowel inflammation
function of CRP, the highly conserved amino acid sequence sug- and to correlate to some extent with disease activity in those with
gests a critical and nonredundant role.9 CRP has been shown to known disease.15,21 Of additional interest is the finding that adult
interact with many molecules including C1q and factor H,10 in patients with IBD with a high serum CRP level show a superior
addition to its primary ligand, phosphocholine, which is exposed response to biological therapies,21,22 in addition to a possible cor-
on damaged biological membranes and bacterial cell walls.11,12 relation of CRP levels with Crohn’s disease (CD) phenotype and
A significant role for CRP in the clearance of apoptotic cells natural history.23,24
and bacteria has also been postulated.13,14 An outline of the impor- We therefore hypothesized that variations in CRP genotype and
tant pathways involved in CRP formation, induction, and function serum CRP (particularly at diagnosis) influence pediatric IBD (PIBD)
are shown in Figure 1. phenotype (with variations between those with CD and ulcerative
In clinical practice, the predicted rise in serum CRP levels colitis [UC]), natural history, and response to biological therapies.
allows its routine use as a diagnostic biomarker,15 as an indicator
of disease activity,16 and in monitoring therapeutic efficacy.17 METHODS
However, it is commonly accepted that the absence of a rise in
serum CRP does not always indicate the lack of infectious or Subjects
inflammatory stimuli.18 Additionally, an individual patient’s Patients diagnosed with IBD before their 17th birthday
serum CRP response to these stimuli may be “inappropriately” (Montreal25/Paris26 A1) were eligible for genotyping and subse-
low despite the clinical history,19 with the CRP response to quent phenotyping. All subjects were recruited in Scotland
FIGURE 1. Outline of the pathways involved in CRP induction, formation, and function in response to infection and/or inflammation. After an infectious or
inflammatory stimulus, the production of IL-1b and IL-6 leads to CRP synthesis in the liver. A complex arrangement of transcription factors are modified at
the CRP gene promoter, with the displacement of the C/EBPz-RBP-Jk complex by the CCAAT/enhancer-binding protein beta-p50 complex the most
critical step. After the synthesis of the pentameric molecule, serum CRP binds to various ligands to alter immune modulation and apoptosis.
www.ibdjournal.org | 597
2015
Henderson et al Inflamm Bowel Dis Volume 21, Number 3, March 2015
between October 2002 and July 2009 as part of ongoing research genome-wide association study (GWAS; described below) was
and latterly through the Medical Research Council–funded Pae- also examined.34 Using the “tag SNP picker” function on the
diatric Inflammatory Bowel Disease Cohort and Treatment Study HapMap project Web site (http://hapmap.ncbi.nlm.nih.gov/),
(PICTS). Recruitment was carried out at all sites across Scotland 8 SNPs tagging 3 genes (CEBPZ [C/EBPz], HNF1A [HNF-1],
by convenience sampling of patients, usually during their initial and POU2F1 [Oct-1]) were identified. It was then ensured that
work-up or induction therapy in their respective tertiary referral these SNPs (or a suitable proxy, D0 ¼ 1 and r2 ¼ 1.0, determined
center or during outpatient reviews; there were no exclusion cri- using SNAP [http://www.broadinstitute.org/mpg/snap/index.php])35
teria. PIBD was diagnosed according to standard clinical, histo- was present on the HumanHap550 BeadChip (Illumina, San
logical, and radiological findings.27,28 Diego, CA) with full results therefore available within the GWAS
data. Brief details of all 8 SNPs are shown in Table, Supplemental
Single Nucleotide Polymorphism Selection Digital Content 2, http://links.lww.com/IBD/A700.
The CRP gene is located on chromosome 1 between posi-
tions 159,682,079 and 159,684,379 (Ensembl Release 68— DNA Extraction and Genotyping
ENSG00000132693; Build 37) flanked by 2 genes, APCS and DNA was extracted from venous whole blood from patients
DUSP23, which are 124 kb and 68 kb downstream and upstream by either a modified salting out technique36 or by using the Nucleon
of CRP, respectively. CRP tagging single nucleotide polymor- BACC2 kit (Gen-Probe Life Sciences, Manchester, United King-
phisms (SNPs) were selected using genotypic data from the Cen- dom) and resuspended in ·1 Tris-ethylenediaminetetraacetic acid
tre d’Etude du Polymorphisme Humain (Utah Residents with buffer (10 mM Tris, pH 7.5, and 1 mM ethylenediaminetetraacetic
Northern and Western European Ancestry; abbreviated CEU) acid) at a final concentration of 100 ng/mL. Primer sequences of the
study available from the HapMap project (Phase III/Release 3; 6 SNPs tagging CRP are shown in Table, Supplemental Digital
http://www.broad.mit.edu/mpg/haploview)29 and identified using Content 3, http://links.lww.com/IBD/A701; all genotyping was car-
Haploview software version 4.230 based on solid spine of linkage ried out in a single batch with random plate allocation in August
disequilibrium (r2 .0.8, haplotype frequency .5%, minor allele 2009. The mean genotyping call rate for all SNPs in all patients was
frequency .10%) (see Fig. A, Supplemental Digital Content 1, 98.6%. Additional genotyping data were also obtained from a pre-
http://links.lww.com/IBD/A699).This process identified 5 tagging vious GWAS (the major PIBD GWAS, published by Imielinski
SNPs spanning a 46 kb region: rs1935193, rs1130864, rs1205, et al34), which included a subset of the Scottish patients with PIBD.
rs1417938, and rs11265263 (see Fig. B, Supplemental Digital Con- This genotyping was performed using the Illumina Infinium II
tent 1, http://links.lww.com/IBD/A699). It is important to note that HumanHap550 BeadChip Technology (Illumina), at the Center
3 of these SNPs (rs1130864, rs1205, and rs1417938) were spe- for Applied Genomics at the Children’s Hospital of Philadelphia.
cifically chosen as tagging SNPs as they had previously been The data contained genotyping results on a representative sample of
demonstrated to influence basal CRP levels.31233 An additional 359 of the patients described above (239 CD, 91 UC, and 29
SNP previously shown to influence serum CRP levels, colonic IBD, type unclassified [IBDU]) for approximately
rs1800947,31 was also genotyped. Details of all genotyped SNPs 550,000 SNPs on the platform.
are presented in Table 1. For all patients, SNPs conformed to the
Hardy–Weinberg equilibrium. Phenotyping and Serum CRP Level
To assess the contribution of genes encoding various CRP at Diagnosis
transcription factors, SNP data from a previously performed Rigorous disease phenotyping of the PICTS cohort was
carried out as previously described.37 Briefly, phenotyping was
performed by detailed examination of case notes, computerized
records, laboratory results, and radiological imaging. As part of
TABLE 1. Details of the 6 SNPs Genotyped to Tag CRP our standard phenotyping study protocol, serum CRP levels are
Marker Chromosomal recorded for the PICTS database at the time of diagnosis
Number dbSNPa Positionb Location MAF Alleles (i.e., during the initial workup of symptoms suggestive of PIBD
and closest to the primary diagnostic endoscopic assessment).
1 rs1935193 159664090 Upstream 0.301 A:T As CRP reference ranges have differed marginally both geo-
2 rs1205 159682233 30 flanking 0.329 C:T graphically and with time throughout Scotland, in addition
3 rs1130864 159683091 30 flanking 0.299 G:A to minor variations in the biochemical assay, results are simply
4 rs1800947 159683438 Exon 2 0.063 C:G recorded as “raised” or “normal” (i.e., above normal range or
5 rs1417938 159684186 Intronic 0.298 T:A not) at diagnosis. The highest normal cutoff value in use at any
6 rs11265263 159710517 Downstream 0.072 C:A time was 10 mg/L. Last date of follow-up was defined as the last
review of clinical notes for patients still under the care of their
a
Based on NCBI Human Build 135.
b PIBD team, the date of transition/transfer to adult services or
Ensembl Release 68, July 2012.
MAF, minimum allele frequency. relocation to an area out with Scotland while in pediatric
services.
598 | www.ibdjournal.org
2015
Inflamm Bowel Dis Volume 21, Number 3, March 2015 Pediatric Inflammatory Bowel Disease
Response to Anti–TNF-a Monoclonal Antibody P values are presented unless otherwise stated. Analysis of
Physician global assessment (determining status as remis- requirement for surgery and azathioprine was carried out using
sion or mild, moderate, or severely active disease38) was used to the survival package in R. Multivariable analysis was performed
assess disease activity of patients naive to anti–TNF-a monoclonal using Cox proportional hazards and backwards stepwise regres-
antibody following their induction course, regardless of whether sion with Akaike Information Criterion as the criterion for exclud-
remission was then sustained in those who had achieved it ing variables. Variables were excluded from the final model if
(as worsening disease and other factors may contribute to loss of their P value was greater than 0.05. A 2-sided P value of 0.05
response). Remission was recorded at 10 to 12 weeks after com- was considered significant throughout.
mencement of the induction course of infliximab (3 infusions at wk
0, 2, and 6—unless infusions were discontinued for severe infusion Ethics
reaction) or adalimumab (6 fortnightly doses at wk 0–10 unless All patients and/or their parents provided full written
treatment discontinued for severe adverse effects or absolute lack consent as part of the Medical Research Council–funded PICTS
of response; UK pediatric clinical experience has been that many cohort, which was approved by local ethics committees at all
children and young people need longer than 4 weeks to judge 3 participating clinical networks: South-East Scotland (LREC/
induction response with adalimumab39). Remission was defined 2002/6/18), West of Scotland (YREC/P12/03), and North of
by physician global assessment after individual review of all case Scotland (GREC/03/0273).
notes, clinic letters, and laboratory values performed at the regional
site where anti-TNF was administered; if there was any doubt, the
case was reviewed with the clinical team members at the site.40 In RESULTS
a subcohort of these cases, the pediatric CD activity index38 or
pediatric UC activity index41 score for CD and UC/IBDU, respec- Patients
tively, were also used to complement the physician global assess- A total of 465 patients had full CRP genotyping data
ment analysis to define remission; a senior clinician with available (cohort 1); the basic demographics of this PIBD group
experience of providing the routine care of these patients (D.C. are shown in Table 2. Although convenience sampling was used
W.; lead of the Scottish Pediatric IBD Biological Registry) made for recruitment, the composition of the study cohort was com-
the final decision on whether remission was achieved based on all parable with our previously published population-based Scottish
available data and discussion to ensure that in all cases, evidence of data,43,44 with a similar proportion of IBD types (i.e., approxi-
remission (as opposed to response) was robust. Serum CRP results, mately two-thirds CD) and median age at diagnosis; there was
taken immediately before the first anti–TNF-a induction dose, were also a slight male preponderance and comparable disease loca-
also obtained. tion and behavior. In the entire cohort, 92% of patients had
phenotype data available and 75% had a recorded CRP level
Statistical Analyses at diagnosis. For the analysis of CD phenotype, a subgroup of
Haplotype and single SNP analyses were carried out using cohort 1 consisting of 232 patients with CD (74% of all patients
Haploview v4.230 (with correction for multiple testing achieved with CD), had both full phenotyping and CRP data available at
using 100,000 permutations) and additional analyzes performed diagnosis (cohort 2). A further nested cohort (a subgroup of
using SNPStats42 and R v3.1.1 (R Foundation for Statistical Com- cohort 2) consisting of 181 patients had at least 2 years of
puting, Vienna, Austria). For anti-TNF-a response, all 6 SNPs follow-up data available (median follow-up 5.2 yr) and were
were tested in a multivariate model to determine any effect of therefore used in the analyses of natural history (cohort 3). De-
CRP genotype on response, with all analyses corrected for IBD tails of cohorts 2 and 3 are shown in Table, Supplemental Digital
type (categorical) and age at diagnosis (quantitative) if appropri- Content 4, http://links.lww.com/IBD/A702. There was no differ-
ate. Chi-squared, Fisher’s exact and Mann–Whitney U tests were ence in the length of follow-up for those with or without a raised
carried out as appropriate. In the genetic analyses, corrected CRP at diagnosis (P ¼ 0.551).
TABLE 2. Characteristics of the Patients with PIBD Included in the Analysis of CRP Genotype
IBD CD UC IBDU
www.ibdjournal.org | 599
2015
Henderson et al Inflamm Bowel Dis Volume 21, Number 3, March 2015
CRP Haplotype Differs Between Pediatric genes encoding CRP transcription factors (POU2F1 [OCT1],
Patients with CD Diagnosed Less than 10 CEBPZ [C/EBPz], and HNF1A [HNF-1]) were also evaluated.
Years of Age This demonstrated that none of the 8 individual SNPs were
There were no SNPs significantly associated with either age overrepresented in either CD or UC; similarly, none of the hap-
at symptom onset or age at diagnosis when evaluated as lotype analyses demonstrated differences between CD and UC
a quantitative trait for either total IBD or any IBD type. Recent (data not shown).
consensus guidelines have highlighted that CD phenotype differs
between those diagnosed aged 0 to 9 years of age (Paris A1a) CRP Genotype Influences Serum CRP Levels
and between 10 to 16 years (Paris A1b).26 Genotypic analyses at Diagnosis
between those diagnosed in these 2 age brackets was therefore The influence of CRP genotype on the presence of an ele-
assessed revealing that the haplotype ATGCTC was associated vated CRP response at diagnosis was assessed. To achieve this,
with age at diagnosis less than 10 years in patients with CD potential confounders demonstrated above (i.e., IBD type and age
(P ¼ 0.040). at diagnosis) were used as covariates in a multivariate model.
Raised or normal serum CRP at diagnosis was used as a binary
Pediatric Patients with CD and UC Differ outcome from patients in cohort 2. In this analysis, rs1205,
Regarding Serum CRP at Diagnosis and rs1130864, and rs1417938 showed association with serum CRP
CRP Genotype elevation in a dominant model (Table 4).
Serum CRP results showed that patients with CD were
more likely to have a raised CRP at diagnosis than patients with CRP Genotype Does not Influence CD
UC (63% versus 22%, P , 0.001). Evaluating the genotypic Phenotype or Natural History
differences between patients with CD and UC only (i.e., exclud- The effect of CRP genotype on major CD phenotypes was
ing patients with IBDU), there was an increased frequency of the also evaluated regarding presenting phenotype (cohort 2) and natural
T allele of rs1205 in patients with CD (34% and 22% in CD and history (cohort 3). There were no significant differences in CRP
UC, respectively; Table 3). Haplotype analyses of these 2 IBD genotype (either individual SNPs or haplotypes) in patients with
types also demonstrated significant differences in haplotype CD with or without ileal disease (i.e., Montreal L1 or L3 versus
structure, with the strongest difference seen in the 6-marker L2), with or without upper gastrointestinal disease (i.e., 6Montreal
haplotype ATGCTC (Table 3 and see Fig., Supplemental Digital L4), or with or without pure colonic disease (i.e., Montreal L2 versus
Content 5, http://links.lww.com/IBD/A703). To further eluci- L1 6 L4 or L3 6 L4) either at diagnosis (cohort 2) or at last follow-
date the potential reasons for the differences in serum CRP at up (cohort 3). Similarly, patients with CD with stricturing or pene-
diagnosis between patients with CD and UC, tagging SNPs of 3 trating disease behavior (Montreal B2 and/or B3) did not differ with
Haplotypeb Haplotype Frequency, % Haplotype Frequency in CD, % Haplotype Frequency in UC, % Uncorrected P Corrected P
600 | www.ibdjournal.org
2015
Inflamm Bowel Dis Volume 21, Number 3, March 2015 Pediatric Inflammatory Bowel Disease
TABLE 4. Influence of SNPs on Serum CRP Levels at Diagnosis in Patients with PIBD
dbSNPa Genotype CRP ¼ Normal, % CRP ¼ Raised, % Or (95% CI) Uncorrected P
a
Based on NCBI Human Build 135.
CI, confidence interval; OR, odds ratio.
respect to CRP genotype compared with those with inflammatory In 90 patients, serum CRP levels were available in the 7
disease only (i.e., Montreal B1) either at diagnosis or last follow-up. days before the first dose of anti-TNF-a agent. There was no
There were insufficient numbers to perform comparison analyses on difference in CRP at anti-TNF induction between those who at-
patients with CD with pure colonic disease versus those with UC or tained remission and those who did not, either stratified as raised
IBDU. Similarly, need for surgical resection, exposure to azathio- ($10 mg/L) or normal (P ¼ 1.00) or when assessed as a contin-
prine, or need for biological therapy was not associated with certain uous variable (median 15 mg/L versus 14.5 mg/L, respectively,
CRP genotypes or haplotypes. P ¼ 0.334).
www.ibdjournal.org | 601
2015
Henderson et al Inflamm Bowel Dis Volume 21, Number 3, March 2015
FIGURE 2. A, Survival plot showing risk of progression to surgery in pediatric patients with CD with normal and raised serum CRP at diagnosis.
B, Survival plot showing risk of progression to azathioprine therapy in patients with PIBD with normal and raised serum CRP at diagnosis. Normal,
CRP ,10 mg/L; raised, CRP .10 mg/L.
CRP genotype/haplotype structure and serum CRP response at of IBD subgroups, in addition to the collection of laboratory
diagnosis. Finally, recent evidence has suggested that adult parameters49 and response to therapies.50 Additionally, use of
patients with IBD with a high baseline CRP level at induction correction for multiple testing ensured that significance was not
demonstrate a better primary response to biological therapies, unduly placed on individual SNPs or haplotypes.
with early normalization of CRP levels correlating with sustained Although convenience sampling was used as the recruit-
long-term response21,47; however, these results are variable.48 ment method, we are confident that although there is potential for
The strengths of this study were the large number of recruitment bias, there are several reasons why we believe this
children included and also our ability to rigorously phenotype cohort to be truly representative of the Scottish PIBD population.
a distinct population of patients with PIBD, allowing their First, we have previously published accurate Scottish incidence43
progress to be followed over time and enabling detailed analyses and regional prevalence rates,44 allowing us to determine the
602 | www.ibdjournal.org
2015
Inflamm Bowel Dis Volume 21, Number 3, March 2015 Pediatric Inflammatory Bowel Disease
www.ibdjournal.org | 603
2015
Henderson et al Inflamm Bowel Dis Volume 21, Number 3, March 2015
604 | www.ibdjournal.org
2015
Inflamm Bowel Dis Volume 21, Number 3, March 2015 Pediatric Inflammatory Bowel Disease
47. Sandborn WJ, Colombel JF, Castillo MM, et al. Elevated C-reactive pro- 52. Wensley F, Gao P, Burgess S, et al. Association between C reactive pro-
tein in anti-TNF-naive patients is associated with higher remission rates. tein and coronary heart disease: mendelian randomisation analysis based
Gut. 2011;60(suppl 3):A195. on individual participant data. BMJ. 2011;342:d548.
48. Willot S, Vermeire S, Ohresser M, et al. No association between C-reactive 53. Tilakaratne S, Lemberg DA, Leach ST, et al. C-reactive protein and disease
protein gene polymorphisms and decrease of C-reactive protein serum activity in children with Crohn’s disease. Dig Dis Sci. 2010;55:131–136.
concentration after infliximab treatment in Crohn’s disease. Pharmacogenet 54. Toedter GP, Blank M, Lang Y, et al. Relationship of C-reactive protein
Genomics. 2006;16:37–42. with clinical response after therapy with ustekinumab in Crohn’s disease.
49. Quail MA, Russell RK, Van Limbergen JE, et al. Fecal calprotectin com- Am J Gastroenterol. 2009;104:2768–2773.
plements routine laboratory investigations in diagnosing childhood 55. Kruis W, Katalinic A, Klugmann T, et al. Predictive factors for an uncom-
inflammatory bowel disease. Inflamm Bowel Dis. 2009;15:756–759. plicated long-term course of Crohn’s disease: a retrospective analysis.
50. Buchanan E, Gaunt WW, Cardigan T, et al. The use of exclusive enteral J Crohns Colitis. 2013;7:e263–70.
nutrition for induction of remission in children with Crohn’s disease dem- 56. Kiss LS, Papp M, Lovasz BD, et al. High-sensitivity C-reactive protein for
onstrates that disease phenotype does not influence clinical remission. identification of disease phenotype, active disease, and clinical relapses in
Aliment Pharmacol Ther. 2009;30:501–507. Crohn’s disease: a marker for patient classification? Inflamm Bowel Dis.
51. Danesh J, Wheeler JG, Hirschfield GM, et al. C-reactive protein and other 2012;18:1647–1654.
circulating markers of inflammation in the prediction of coronary heart 57. Markowitz J. Early inflammatory bowel disease: different treatment
disease. N Engl J Med. 2004;350:1387–1397. response to specific or all medications? Dig Dis. 2009;27:358–365.
www.ibdjournal.org | 605
2015