Impaired neutrophil chemotaxis in Crohn’s disease relates to reduced production of chemokines and can be augmented by granulocyte-colony stimulating factor

SUMMARY

Background

Defective neutrophil recruitment has been described as a primary pathogenic abnormality in Crohn’s disease. Cantharidin-induced blisters provide a novel investigative tool to assess cellular influx and inflammatory mediator production during acute inflammation and allows the effects of therapy on these parameters to be measured.

Aims

To determine whether reduced neutrophil tissue penetration in Crohn’s disease relates to impaired production of inflammatory mediators, and whether it can be reversed by granulocyte-colony stimulating factor (G-CSF).

Methods

Neutrophil and monocyte/macrophage populations and inflammatory mediators were measured in cantharidin blisters at 24 h. Neutrophil chemotaxis was assessed in vitro using blister fluid as the chemoattractant. The effect of s.c. G-CSF on blister phenotype was determined.

Results

Significantly fewer neutrophils migrated into blisters in Crohn’s patients. The production of neutrophil chemokines, but not other inflammatory mediators, was reduced. This significantly correlated with reduced chemotaxis in vitro. Differences were unrelated to caspase-recruitment domain 15 genotype. G-CSF significantly increased blister neutrophil concentrations in control subjects and Crohn’s patients.

Conclusions

Reduced neutrophil migration during acute inflammation in Crohn’s disease is associated with impaired production of appropriate chemoattractants. G-CSF therapy increases neutrophil tissue migration, which may partially account for its observed therapeutic effect.

INTRODUCTION

Crohn’s disease (CD) is a chronic inflammatory, granulomatous disorder primarily affecting the bowel. The cause remains unknown, although various genetic¹ and environmental² factors have been postulated. Whilst most theories focus on an excessive immune reaction as the underlying problem, these patients may in fact possess a diminished initial inflammatory response.³ Previous studies have demonstrated a reduction in the number of neutrophils migrating to the sites of skin abrasions^4-6 or intestinal biopsies⁷ in patients with CD. These leucocytes constitute the first line of defence after microbes and organic debris breach the mucosal barrier.⁸ A delay in their accumulation might lead to abnormal persistence of exogenous material within the bowel wall. Subsequent uptake and encirclement by macrophages could then produce the granulomata characteristic of this disease.⁹

The mechanism behind the failure of neutrophil accumulation was not satisfactorily explained in the original descriptions.^4-6 The neutrophils themselves behave normally during in vitro assays of chemotaxis,^{10, 11} suggesting alterations in the inflammatory environment. One proposal was the existence of an as yet undefined serum inhibitory factor, based on ex vivo experiments.¹² The effects of this potential mediator were also evident with the serum from ulcerative colitis (UC) patients (who demonstrate no impairment in neutrophil recruitment) and the potency correlated with disease activity. This suggests that it might arise as a consequence of active inflammation and not underlie the defective neutrophil recruitment reported in CD in vivo, which occurs in quiescent disease.¹⁰

An alternative explanation would be impaired production of chemotactic cytokines (chemokines) in CD patients. Release of such mediators by extravascular leucocytes and epithelial cells establishes a gradient that promotes migration of neutrophils into the tissues. Although present at high concentrations in established active CD lesions,¹³ their role in the first 24 h of acute inflammation remains largely unstudied. It was recently suggested that concentrations of interleukin (IL)-8, a potent neutrophil chemoattractant, were reduced in new inflammatory lesions in CD and that this might underlie the abnormal cell migration.⁷ This theory merits further evaluation.

Application of exogenous IL-8 has been shown to correct neutrophil numbers in Crohn’s skin abrasions.⁷ Unfortunately, this molecule is a small protein and difficult to develop as a pharmaceutical. In contrast, the immunostimulatory haematopoetic growth factors granulocyte macrophage-colony stimulating factor (GM-CSF) and granulocyte-colony stimulating factor (G-CSF) are widely used in other clinical disciplines¹⁴ and have also been trialled with some success in CD.^{15, 16} Their ability to correct the defect in neutrophil recruitment has been postulated, but not proven.

We previously developed a novel skin window technique using the topical application of cantharidin, a protein phosphatase inhibitor that causes atraumatic acantholysis and blister formation by detachment of desmosomes.¹⁷ Leucocytes (predominantly neutrophils) and inflammatory mediators (including a number of cytokines and activated complement components) rapidly accumulate within these lesions. This permits assessment of the acute inflammatory response in vivo. In this study, we examine the composition of the exudate in patients with CD and UC, and assess the effect of G-CSF administration on neutrophil tissue penetration in patients.

MATERIALS AND METHODS

Patient selection

Patients with quiescent inflammatory bowel disease and non-inflammatory controls were recruited through the out-patient clinics of University College London Hospitals and St Mark’s Hospital. Diagnoses of CD and UC were confirmed by histology and endoscopy and/or radiology, using the Lennard-Jones criteria.¹⁸ Disease activity was determined using a modified Crohn’s Disease Activity Index validated for both CD and UC ranging from 0 (no clinical activity) to 15 (severe activity),¹⁹ allowing intergroup comparisons to be made. All patients had quiescent disease (activity indices: CD = 1, IQR 1-2; UC = 0, IQR 0-1.25). For the assessment of the effect of G-CSF, in which only CD patients and healthy controls (HC) were studied, activity was determined using the Harvey-Bradshaw index²⁰ (activity index: 0.5, IQR 0-3). Age, sex, body mass index, peripheral venous blood leucocyte counts and serum C-reactive protein concentrations were similar in each group. Subjects were excluded if they showed any evidence of intercurrent infection, other inflammatory disorder, significant co-morbidity or recent surgery. The majority of patients were stable off medication, although a minority used oral steroids (3 CD, 5 UC), azathioprine (3 CD, 1 UC) or mesalazine (8 CD, 6 UC). CD patients and HC were matched for smoking history, although none of the UC patients currently smoked. Patients were genotyped for the R702W, G908R and L3020finsC polymorphisms in caspase-recruitment domain 15 (CARD15) known to predispose to CD^21-23 as previously described.⁷ Carriage of two wild type alleles is indicated as w/w, simple heterozygosity for the polymorphisms as w/m, and compound heterozygosity or homozygosity as m/m. Characteristics of individual CD patients are provided (Table 1). Written informed consent was obtained from all subjects. The local ethics committees of both hospitals approved the study.

Table 1.

Characteristics of individual Crohn’s disease (CD) patients. Disease behaviour is described according to the Vienna classification, and reflects previous or established characteristics rather than current activity. Patients who participated in both the blister and granulocyte-colony stimulating factor (G-CSF) studies did so on different occasions

Patient	Age	Sex	Disease duration (years)	Disease distribution	Disease behaviour	Caspase-recruitment domain 15 genotype	Disease activity (mCDAI)	Medications	Smoker
Blister study
1	21	M	3	Colonic	Fistulating	w/m	2	Nil	N
2	25	M	6	Colonic	Inflammatory	w/w	1	Azathioprine	N
3	25	M	10	Colonic	Inflammatory	w/w	0	Nil	Y
4	26	F	5	lleocaecal	Inflammatory	w/w	1	Nil	N
5	28	M	3	Colonic	Inflammatory	w/w	1	Mesalazine	N
6	29	F	10	lleocaecal	Inflammatory	w/m	0	Mesalazine	Y
7	31	M	8	Colonic	Fistulating	w/w	6	Mesalazine	N
8	32	F	15	lleocaecal	Stenotic	w/w	1	Nil	N
9	35	F	9	lleocaecal	Stenotic	m/m	6	Nil	Y
10	37	F	23	Ileal	Fistulating	w/m	0	Nil	Y
11	38	M	20	lleocaecal	Inflammatory	w/w	0	Mesalazine	Y
12	42	F	13	Colonic	Inflammatory	w/w	2	Prednisolone	Y
13	44	F	2	lleocaecal	Inflammatory	w/m	2	Nil	N
14	47	M	3	lleocolonic	Mixed	m/m	2	Azathioprine	Y
15	47	M	26	lleocaecal	Inflammatory	w/m	4	Azathioprine	Y
16	50	M	34	Colonic	Inflammatory	w/w	3	Mesalazine	N
17	50	M	27	lleocaecal	Fistulating	w/w	0	Prednisolone	N
18	56	F	40	lleocolonic	Inflammatory	w/w	1	Nil	Y
19	56	M	2	lleocolonic	Inflammatory	w/w	2	Mesalazine	N
20	59	M	1	Colonic	Inflammatory	w/w	1	Prednisolone	N
21	64	F	20	lleocaecal	Inflammatory	w/w	1	Mesalazine	N
22	64	M	47	lleocaecal	Inflammatory	m/m	2	Mesalazine	N
23	72	F	1	lleocaecal	Fistulating	w/w	6	Nil	N
G-CSF study
9	35	F	9	lleocaecal	Stenotic	m/m	6	Nil	Y
13	44	F	2	lleocaecal	Inflammatory	w/m	1	Nil	N
19	56	M	2	lleocolonic	Inflammatory	w/w	6	Azathioprine	N
24	38	F	17	Colonic	Inflammatory	w/w	0	Mesalazine	N
25	54	M	24	lleocaecal	Fistulating	w/w	0	Nil	N
26	30	F	3	lleocolonic	Inflammatory	w/w	0	Mesalazine	N
27	54	F	2	Colonic	Inflammatory	w/w	2	Mesalazine	N
28	51	M	33	Colonic	Inflammatory	w/w	0	Mesalazine	N

Blister formation

Cantharidin skin blisters were created in 23 patients with CD, 20 with UC and 21 control subjects. Their formation and analysis have been described previously.¹⁷ In brief, 25 μL of 0.1% cantharidin (Cantharone; Dormer Laboratories Inc., Rexdale, Canada) in acetone was applied to two 0.8 cm² diameter discs of Whatman qualitative No.1 filter paper (Whatman Ltd., Maidstone, UK) placed on the forearm. Discs were individually covered with Nescofilm (Azwell Inc., Osaka, Japan) followed by a Mefix adhesive dressing (Mölnlycke Healthcare, Sweden). After 24 h, blister fluid was collected into siliconized microfuge tubes (Novara Group Ltd., Leicestershire, UK) on ice.

Blister composition

All samples were analyzed in a blinded fashion. Blister fluid volume was determined by weight. Total cell numbers and viability were determined in duplicate by manually counting cells stained with trypan blue on a Neubauer counting slide (Hawksley, England). Blister fluid was then centrifuged (500 g, 5 min, 4 °C) and supernatants stored at −70 °C. Flow cytometry with fluorescein isothiocyanate-conjugated anti-CD16²⁴ (Becton Dickinson, San Diego, USA) and phycoerythrin-conjugated anti-CD14²⁵ (Becton Dickinson) monoclonal antibodies were used to quantify neutrophils and monocyte/macrophages, respectively. Labelled cells were identified on a FACScan flow cytometer (Becton Dickinson) and data analyzed with cellquest software (Becton Dickinson).

Blister fluid supernatants were analyzed for epithelial neutrophil-activating peptide-78 (ENA-78), growth-related oncogene-α (GRO-α), IL-1β, IL-4, IL-5, IL-8, IL-12, interferon-γ (IFN-γ), monocyte chemotactic protein-1, macrophage inflammatory protein-1α (MIP-1α) and tumour necrosis factor-α (TNF-α) (R&D Systems, Abingdon, Oxford, UK), complement factor 3a (C3a) and complement factor 5a (Pharmingen, San Diego, CA, USA), histamine (Immunotech, Marseille, France) and bradykinin (Peninsula Laboratories, Merseyside, UK), using the commercial kits according to the instructions from the manufacturer. Appropriate dilutions of blister fluid were established for quantification within the linear range of each assay.

Chemotaxis assay

Neutrophils were purified from peripheral venous blood of an independent healthy subject by dextran sedimentation and centrifugation through Lymphoprep (Nycomed, Oslo, Norway),²⁶ then resuspended in Dulbecco’s modified Eagle’s medium (Life Technologies, Rockville, MD, USA) containing 1% bovine serum albumin (Sigma Aldrich, St Louis, MO, USA). Sufficient blister fluid was available to study chemotaxis to a 10% (v/v) dilution in 14 CD, 6 UC and 9 HC. Cell migration was determined in a 96-well chemotaxis chamber (Neuroprobe, Cabin John, MD, USA) as described previously.²⁷ Migrated cells were stained with 5 μL 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide thiazolyl blue (MTT) for 3 h at 37 °C, and then absorbance at 570 nm measured on a Dynatech MR600 automatic plate reader (Dynatech, Billingshurst, UK).

G-CSF administration

In eight CD patients (two ileal, two ileocolonic and four colonic involvement) and eight HC, neutrophil numbers in new cantharidin blisters (in duplicate) were quantified by measuring the surface expression of CD16 (this remains static during the G-CSF therapy).²⁸ An s.c. dose of 5 μg/kg recombinant human G-CSF (Lenograstim; Chugai Pharma UK Ltd, London, UK) was then administered. This was repeated 24 h later, at which time a further two cantharidin blisters were induced and sampled as before. No new medication was allowed during the study.

Statistical analysis

Statistical analysis was conducted using the Intercooled Stata version 6 (Stata Corp., College Station, TX, USA). A P-value <0.05 was regarded as significant. The two-tailed Mann-Whitney test was used for single comparisons, and Kruskal-Wallis anova with Dunn post-tests for multiple comparisons as appropriate. In the G-CSF study, effects of treatment were compared within each subject group using the Wilcoxon matched pairs test. Associations between chemotaxis and inflammatory mediator concentrations were determined by Pearson’s correlation. ancova was performed to detect effects of potential prospectively recorded confounding variables.

RESULTS

Cellular infiltrate

Macroscopic appearances of the blisters, and side effects and acceptability of the technique, were similar to those previously reported in all subjects.¹⁷ Their diameter was approximately 1.5 cm and cell viability was >95% in all samples. Blister volume was similar in all groups (HC: 294 ± 42 μL; CD: 258 ± 39 μL; UC: 304 ± 41 μL; Figure 1a). In healthy subjects, the majority of cells were CD16⁺ neutrophils (63 ± 11%), followed by CD14⁺ monocyte/macrophages (14 ± 6%). In patients with CD, there was a significant reduction in total cell counts (HC: 4.55 ± 1.09 × 10⁶/mL; CD: 2.27 ± 0.56 × 10⁶/mL, P = 0.03; UC: 2.85 ± 0.85 × 10⁶/mL, p = 0.09; Figure 1b) because of a selective deficit in neutrophils (HC: 3.55 ± 1.09 × 10⁶/mL; CD: 0.95 ± 0.30 × 10⁶/mL, P = 0.01; UC: 1.95 ± 0.68 × 10⁶/mL, P = 0.12; Figure 1c). Monocyte recruitment was similar in all subjects (HC: 0.59 ± 0.18 × 10⁶/mL; CD: 0.48 ± 0.15 × 10⁶/mL; UC: 0.35 ± 0.10 × 10⁶/mL; Figure 1d). Although numbers were small, no parameter in CD was related to CARD15 genotype and responses of patients carrying two polymorphisms were indistinguishable from wild type subjects.

Figure 1.

Blister composition in healthy controls (HC), Crohn’s disease (CD) and ulcerative colitis (UC) subjects, showing the reduced neutrophil accumulation in CD blisters. (a) Blister volume. (b) Total cell concentration. (c) CD16⁺ cell concentration. (d) CD14⁺ cell concentration. Mean values and significances of differences compared with HC shown, logarithmic scales. Caspase-recruitment domain 15 (CARD15) genotypes indicated as wild type (filled circles), simple heterozygous (half circles), compound heterozygous/homozygous (open circles) or unknown (crosses).

Inflammatory mediators

Measurable blister fluid concentrations were obtained for all inflammatory mediators tested except for bradykinin, IFN-γ, IL-5 and MIP-1α, which were below the sensitivity of the assay (40, 8, 8 and 10 pg/mL, respectively). Concentrations of all other mediators were independent of CARD15 genotype.

There was no difference in the production of activated complement components, histamine or cytokines IL-1β, IL-4, IL-12 or TNF-α in any group. In contrast, the mean production of each of the three neutrophil chemokines examined (IL-8, ENA-78 and GRO-α) in CD patients was approximately 50% that was seen in controls (Table 2), although differences for individual mediators were marginally above the 5% statistical significance level. There was, however, a significant correlation between blister fluid concentrations of IL-8 (r = 0.52, P = 0.02) and GRO-α (r = 0.41, P = 0.03) and the magnitude of chemotaxis these samples induced in HC neutrophils in vitro. In this model, CD blister fluid recruited the fewest cells. Inflammatory mediator production was normal in UC except for GRO-α, which was elevated (P = 0.02).

Table 2.

Concentrations of inflammatory mediators in blister fluid, and correlations with the magnitude of chemotaxis induced in vitro. Diminished production of neutrophil chemokines in CD correlates with reduced migration

	Blister fluid concentration, ng/mL			Correlation with chemotaxis
	Healthy controls	CD	UC	r	P
Complement factor 3a	29.52 ± 2.57	27.30 ± 2.23	29.46 ± 1.72	0.25	N.S.
Complement factor 5a	39.38 ± 11.57	37.09 ± 6.00	43.66 ± 7.20	0.23	N.S.
Epithelial neutrophil activating peptide-78	17.07 ± 5.53	9.27 ± 2.04	15.47 ± 4.33	0.09	N.S.
Growth related oncogene-α	2.52 ± 1.53	0.69 ± 0.09	5.04 ± 2.02	0.41	0.03
Histamine	50.57 ± 13.3	37.3 ± 19.48	29.00 ± 6.82	0.13	N.S.
Interleukin (IL)-12	0.09 ± 0.02	0.07 ± 0.02	0.07 ± 0.01	−0.35	N.S.
IL-1β	0.17 ± 0.04	0.13 ± 0.02	0.16 ± 0.04	−0.11	N.S.
IL-5	0.48 ± 0.11	0.18 ± 0.06	0.38 ± 0.11	−0.24	N.S.
IL-8	13.71 ± 6.75	6.64 ± 2.04	12.98 ± 4.50	0.52	0.02
Monocyte chemotactic protein-1	8.77 ± 3.51	6.01 ± 3.05	11.42 ± 4.30	0.38	N.S.
Tumour necrosis factor-α	0.14 ± 0.05	0.22 ± 0.08	0.33 ± 0.14	−0.09	N.S.

Effect of G-CSF on cantharidin blister phenotype

Systemic effects of G-CSF on peripheral venous blood neutrophil counts were similar in HC and CD subjects, both showing substantial augmentation following treatment (HC: 3.4 ± 0.1 × 10⁹/L to 27.0 ± 5.3 × 10⁹/L; CD: 4.3 ± 0.6 × 10⁹/L to 28.8 ± 6.4 × 10⁹/L). The increase in peripheral venous monocyte counts was slightly lower in CD patients (HC: 0.6 ± 0.2 × 10⁹/L to 1.6 ± 0.3 × 10⁹/L; CD: 0.5 ± 0.1 × 10⁹/L to 1.2 ± 0.2 × 10⁹/L, P = 0.03). Side effects of G-CSF administration were similar in both groups, the most significant of which was bone pain. This was reported by 12 subjects, requiring the simple analgesia in four that was taken after the collection of the second set of blister fluids.

Blister volume was unchanged by the treatment (Figure 2a). Both HC and CD subjects responded to G-CSF by recruiting greater numbers of cells (Figure 2b; P = 0.02 and P = 0.02, respectively). This was due to increases largely in the neutrophil populations (Figure 2c; P = 0.01 and P = 0.02 respectively) but also marginally in the monocyte/macrophage populations in HC but not CD subjects (Figure 2d; P = 0.04). Although most CD patients administered G-CSF had quiescent disease, two had active inflammation (Harvey-Bradshaw score = 6 in each); augmentation of blister cell numbers was observed in both subsets.

Figure 2.

Effects of granulocyte-colony stimulating factor on blister fluid composition, showing the augmentation of neutrophil and monocyte recruitment. (a) Blister volume; (b) total cell concentration; (c) CD16⁺ cell concentrations; (d) CD14⁺ cell concentration. Mean values and significances of differences compared with HC shown.

Confounding factors

No confounding effect was observed with any patient characteristic within any disease group, except for GRO-α levels in UC patients that co-varied with use of oral steroids [partial r² (×100) = 7.1; P = 0.03]. Specifically, there were no other significant effects because of intercurrent use of immunosuppressive medication, or any relationship between peripheral venous blood neutrophil and monocyte counts and blister neutrophil and monocyte/macrophage counts. The reductions in blister total cell numbers and CD16⁺ cells in CD were still observed even after exclusion of patients taking medication (P = 0.05 and P = 0.02, respectively). Standard measures of disease activity were not associated with blister phenotype when analyzed by multivariate analysis. In particular, results from the five active CD patients were comparable with patients with activity scores of zero (total cells: 2.4 ± 2.8 × 10⁶/mL and 1.9 ± 0.8 × 10⁶/mL, respectively; CD16⁺: 1.1 × 10⁶/mL and 0.9 × 10⁶/mL, respectively). Moreover, mild CD activity in the G-CSF study was associated with higher blister neutrophil concentrations (r = 0.93, P = 0.003).

DISCUSSION

Cantharidin blisters have been used previously to assess tissue drug levels²⁹ and the acute inflammatory response.^{17, 30-32} We have used them to examine the accumulation of leucocytes and inflammatory mediators in newly created inflammatory lesions in subjects with inflammatory bowel disease, and to assess whether systemic G-CSF modulates the response in CD. Studying this process in the skin is appropriate in the context of inflammatory bowel disease, as neutrophils and monocyte/macrophages are non-organ specific when they constitute an early component of the innate immune response. Any cellular defect predisposing to inflammatory bowel disease is likely to be present at extra-intestinal sites, such that investigation using more accessible surrogate sites may reveal abnormalities of pathogenic relevance in the bowel. In support of this concept, it was recently shown that neutrophil migration into sites of rectal or ileal trauma in a model analogous to the skin window was also reduced in CD.⁷

In CD, fewer neutrophils were recruited into blisters by 24 h, confirming the findings of previous studies that used skin windows.^4-7 Whether this abnormality is primary or secondary is difficult to prove, although the fact that it occurred in patients with quiescent disease on no medication favours the former. Consequently, it is unlikely that reduced numbers of blister neutrophils in CD relates to their sequestration at other sites of inflammation. Mildly active disease was actually associated with increased numbers of blister cells, suggesting that the reduction in neutrophil tissue penetration would be even greater if all the patients were entirely quiescent.

In concordance with impaired neutrophil accumulation, production of secreted neutrophil chemoattractants was diminished in CD. Failure of any individual mediator to reach significance may reflect a lack of statistical power in this study, or that CD is a heterogeneous disease and recruitment of neutrophils into tissues requires a series of signals. These include a sequence of chemokines³³ in addition to changes on the vascular endothelium.³⁴ The ultimate magnitude of migration will depend on all elements in this pathway. As such, a variety of underlying lesions might lead to a similar phenotype, which this study could have been underpowered to detect in different subgroups. Alternatively, the abnormality could relate to a combination of modest defects in the production of each neutrophil chemokine. This would be consistent with the correlation between blister fluid chemokine concentrations and chemotaxis induced in vitro observed here. The relevance of our findings is further underscored by the specificity of the defect for neutrophils: blister volume, monocyte/macrophage recruitment and production of their chemoattractants, complement activation, and production of non-chemotactic cytokines were entirely normal. They also concord with previous reports of reduced IL-8 production in CD in leucocytes,^{7, 35-37} new acute inflammatory lesions,⁷ and early recurrent intestinal lesions developing 3 months after bowel resection and re-anastomosis.³⁸ The proximal cause of the abnormality remains unidentified although, whilst numbers were small, it is likely that CARD15 polymorphisms can be excluded. It is unlikely that the majority of CD patients possess abnormalities in the neutrophils themselves or their chemokine receptors, as these cells migrate normally in vitro​,¹⁰ and recruitment can be restored in vivo by addition of exogenous IL-8.⁷

Although G-CSF is normally used to increase systemic neutrophil numbers,¹⁴ little is known about its effects on their recruitment into the tissues. The only previous study to try and quantify this process used skin windows in healthy volunteers after s.c. administration of 300 μg/day G-CSF (Filgrastim, Amgen Inc., Thousand Oakes, CA, USA).³⁹ Leucocyte migration at 24 h was similar to basal levels but after 5 days was reduced by 60%, despite peripheral blood neutrophil counts >20 × 10⁹/L. Conversely, we observed a fivefold increase in neutrophil concentrations in cantharidin blisters after 48 h of G-CSF. The mechanism of this enhancement at the tissue level has not been determined, although it may simply reflect a ‘mass action’ effect secondary to increased levels of circulating granulocytes. There are three possible explanations for this disparity. Firstly, cantharidin blisters produce a greater acute inflammatory response than skin windows. Secondly, many cells accumulating after the first 48 h will have been newly synthesized, and might be immature and thus migrate poorly into inflamed tissues.³⁹ Finally, there may be enhanced migration of neutrophils induced by the physiologically relevant glycosylated form of G-CSF (Lenograstim) as used here, but not non-glycosylated G-CSF (Filgrastim). Such a difference has been observed in terms of their ability to prime the respiratory burst.⁴⁰ Despite the discrepancy with the previous study, our findings argue that glycosylated G-CSF can promote neutrophil recruitment in CD patients (although it is acknowledged that numbers studied were small). It may thus provide a rational therapeutic approach under the hypothesis of defective neutrophil recruitment, and could partially account for its observed therapeutic benefits to date.¹⁶ The same might apply to GM-CSF, which may also have additional beneficial actions on macrophages in view of the defects in their function recently reported in CD.⁷

It is plausible that CD could be caused by defects in the acute inflammatory response to bacteria and food antigens. The requirement for luminal contents to drive mucosal inflammation in CD has been elegantly demonstrated by amelioration of lesions on diverting the faecal stream and their reoccurrence on its re-introduction.^{41, 42} The concentration of bacteria is extremely high adjacent to the mucosal cells of the terminal ileum and colon, the sites of predilection for CD. In everyone, there is some degree of translocation of the luminal constituents into and across the bowel wall.⁴³ In healthy individuals, the presence of bacteria in the tissues provokes an acute inflammatory response with release of cytokines and vasoactive factors followed by neutrophil influx, phagocytosis and clearance of debris. If this process is relatively deficient, this material might instead persist, leading to a granulomatous reaction. This could then secondarily stimulate influx of chronic inflammatory cells causing further damage to the local environment and propagating a T-cell-mediated inflammation as described in active CD, occurring as a consequence and not the cause of the primary insult. This distinction is extremely important, as factors that initiate dysregulated inflammation are almost certainly distinct from those underlying its perpetuation. The clinically and pathologically very similar inflammatory bowel diseases that frequently develop in patients with rare congenital neutrophil immunodeficiencies provide an important precedent for such a mechanism.^44-49 Additionally, the recent descriptions of polymorphisms in CARD15^21-23 and deficiencies in defensin production⁵⁰ in CD patients reinforce the importance of host innate immunity.

In conclusion, we have applied the cantharidin blister technique to demonstrate major abnormalities in acute inflammation in CD. We confirmed the impairment in neutrophil recruitment and found that this was associated with reduced production of chemokines. Systemic G-CSF greatly increased the neutrophil tissue penetration, supporting its use as a potential therapy for CD. Our data indicate that the underlying molecular mechanisms are likely to be heterogeneous, consistent with the view that CD is a polygenic disease with a spectrum of clinical phenotypes. Future studies should be undertaken in patient subsets stratified according to these characteristics.