Are anti-neutrophil cytoplasmic antibodies (ANCA) clinically useful in inflammatory bowel disease (IBD)?

Abstract

Since the first detection of ANCA in IBD, numerous studies have dealt with their prevalence, antigenic specificities, clinical significance, pathophysiological role, and their induction. This review summarizes the information obtained from those studies and shows that ANCA are not directly useful as diagnostic and prognostic factors in IBD. ANCA were detected in 50–85% of patients with ulcerative colitis (UC) and 10–20% of patients with Crohn's disease (CD). Multiple target antigens are recognized by these autoantibodies, including both cytoplasmic and nuclear proteins. A pathophysiological role for ANCA in IBD is far from clear. On the one hand, it is suggested that ANCA are genetic markers of susceptibility for IBD, and on the other hand, the induction of ANCA in those diseases may just be an epiphenomenon of chronic inflammation. We discuss recent evidence that ANCA may be induced by a break-through of tolerance towards bacterial antigens.

INTRODUCTION

ANCA are autoantibodies directed against constituents of neutrophil granulocytes. ANCA were originally detected in serum from patients with Wegener's granulomatosis (WG) [1], a disease characterized by necrotizing granulomatous inflammation of the upper and lower airways in conjunction with systemic vasculitis and necrotizing crescentic glomerulonephritis. Samples from patients with WG reveal a typical cytoplasmic staining pattern when tested by indirect immunofluorescence on ethanol-fixed neutrophils (cytoplasmic or c-ANCA). The antigen recognized by c-ANCA proved to be proteinase 3 (PR3), a constituent of the azurophilic granules of the neutrophil [2].

A second type of ANCA has been detected in several other idiopathic forms of systemic vasculitis and glomerulonephritis [3]. This type of ANCA is characterized by a perinuclear staining pattern on ethanol-fixed neutrophils (perinuclear or p-ANCA). The first antigen recognized by these p-ANCA was identified as myeloperoxidase (MPO), another constituent of the azurophilic granules [4]. The perinuclear staining proved an artefact caused by ethanol fixation. MPO, a highly cationic protein, apparently moves and attaches to the negatively charged nuclear membrane during the fixation procedure [5].

Following the detection of ANCA in the systemic vasculitides, it became clear that ANCA also occur in other idiopathic inflammatory disorders [6]. ANCA have been detected in the inflammatory bowel diseases (ulcerative colitis (UC) and Crohn's disease (CD)) [7,8], in autoimmune-mediated liver diseases [9–11], in rheumatoid arthritis (RA) [12,13], and in systemic lupus eythematosus (SLE) [14,15]. Usually, the p-ANCA type is found, but the antigen is not MPO [7]. A third type of ANCA, producing a diffuse cytoplasmic staining on ethanol-fixed neutrophils (atypical ANCA or a-ANCA), has also been described in these diseases [16].

ANCA IN IBD

IBD (UC and CD) are chronic, relapsing and tissue-destructive idiopathic inflammatory conditions limited to the large bowel (UC) or occurring anywhere along the alimentary tract (CD). The aetiology of these diseases has not yet been fully elucidated [17]. Autoimmune processes may play a role in their pathogenesis, since several types of autoantibodies have been found in these diseases, such as antibodies to goblet cells [18], antibodies to the cytoskeletal protein tropomyosin [19], and antibodies to endothelial cell antigens [20].

In 1990, two different groups reported the presence of ANCA in serum from 84% and 59% of patients with UC, respectively [7,8]. Since then, numerous reports have confirmed this finding. ANCA were also detected in CD, although the prevalence (10–20%) in this disease is much lower than in UC (50–90%) [7,8], and in primary sclerosing cholangitis (50–85%), a chronic cholestatic liver disease that is strongly associated with IBD [9,11,21].

The standard method for detection of ANCA in the systemic vasculitides, i.e. indirect immunofluorescence on ethanol-fixed neutrophils, has also been used widely in IBD. Other methods, however, have been developed as well, such as a fixed neutrophil ELISA [7,22], in which 85% of patients with UC and 20–28% of patients with CD were reported to be positive for ANCA, and an immunoalkaline phosphatase staining method [23,24], in which 33–42% of patients with UC and 0–5% of patients with CD were found positive for ANCA. Thus, the different methods used for ANCA detection in IBD apparently vary in detecting ANCA. Bansi et al. [25] reported that the immunoalkaline phosphatase method was more sensitive than the indirect immunofluorescence method for the detection of ANCA in patients with primary sclerosing cholangitis. However, prevalences of ANCA in IBD as detected with the immunoalkaline phosphatase method were lower than the prevalences in IBD reported in most studies using the indirect immunofluorescence method [25]. Large comparative studies on the three techniques using samples from different laboratories have not been performed.

ANTIGENIC SPECIFICITIES OF ANCA IN IBD

Identification of the antigens recognized by ANCA in IBD showed less homogeneous results than in the vasculitides, where antibodies to PR3 are highly specific for WG and antibodies to MPO are associated with microscopic polyangiitis, the Churg–Strauss syndrome, or idiopathic necrotizing crescentic glomerulonephritis [3]. Antibodies to PR3 and MPO are virtually absent in IBD. Since PR3 and MPO are constituents of the neutrophil granules, several other neutrophil constituents were tested for their reaction with ANCA from patients with IBD by ELISA. Lactoferrin, bactericidal/permeability-increasing protein, cathepsin G, elastase, lysozyme, and β-glucuronidase have all been reported as target antigens for ANCA in IBD and primary sclerosing cholangitis (Tables 1 and 2) [26–35]. Lactoferrin is a constituent of the specific granules of the neutrophil, whereas the other antigens are constituents of the azurophilic granules, as are PR3 and MPO (Table 1). Besides ELISA, immunoblotting with extracts of neutrophils as antigen source has been used to detect antibodies to granule-derived antigens [21,27]. Using this approach, we recently were able to identify the cytosolic enzymes catalase and α-enolase as antigens for ANCA in both UC and CD (Tables 1 and 2) [36]. Another group detected antibodies to these cytosolic antigens in primary sclerosing cholangitis [37].

Table 1.

Cellular localization of ANCA antigens

Table 2.

Prevalences of ANCA directed against specific antigens in inflammatory bowel disease

Although the term ‘ANCA’ implies that these antibodies are directed against cytoplasmic antigens, several nuclear antigens have been shown to contribute to p-ANCA staining of neutrophils. Conventional indirect immunofluorescence on ethanol-fixed neutrophils does not allow to distinguish perinuclear staining from nuclear staining, especially not when antibody titres are high. By using confocal laser scanning microscopy and immunoelectron microscopy, Billing et al. [38] showed that, at least in some UC sera, antibodies were present that localized to the inner side of the nuclear membrane. These antibodies, however, were not directed against double-stranded DNA. Vidrich et al. [39], from the same group, showed that DNase treatment of methanol-fixed neutrophils before performing indirect immunofluorescence resulted in loss of staining by p-ANCA⁺ sera from UC patients, but not by p-ANCA⁺ sera from patients with hepatobiliary disorders. The loss of p-ANCA antigenic recognition suggested either that the antigen recognized is a protein–DNA complex or that the presence of intact DNA is necessary for maintaining the integrity of the antigen. Since then, antibodies to several nuclear antigens have been identified in sera from patients with IBD. Eggena et al. produced two p-ANCA MoAbs from a patient with UC [40] and subsequently showed that the antigen recognized by these antibodies was histone H1 (Table 1) [41,42]. However, studies on the frequency of antibodies to histone H1 in UC are lacking thus far (Table 2). Interestingly, ANCA directed against histone H1 have been detected in T cell receptor-alpha (TCR-α)-deficient mice with spontaneous enterocolitis [43]. Using immunoblotting techniques, a Japanese group [44] characterized antibodies to the nuclear non-histone chromosomal proteins high mobility group (HMG)1 and HMG2 in UC (Tables 1 and 2). HMG1 and HMG2 are distributed in the nuclei, but also in the cytoplasm, of eukaryotic cells and act as transcription factors [44]. Terjung et al. [45] showed by confocal laser scanning and immunoelectron microscopy that nuclear lamina proteins (lamins A, C, and B1, and lamin B receptor), but not histones H1–H4, colocalized with the antigen(s) recognized by p-ANCA from patients with UC and hepatobiliary disorders (Tables 1 and 2).

However, despite the identification of many antigens for ANCA in IBD, a substantial number of patients are positive for ANCA as detected by indirect immunofluorescence, whilst no specific antigen has been detected until now [36]. Remarkably, the opposite is also frequently true. We [11,36] and others [28,32,46–48] have shown that ANCA detection by indirect immunofluorescence may fail to detect antibodies to some of these antigens. The reason for this phenomenon is unknown, but may be related to instability of the antigens in ethanol-fixed neutrophils [47,49] and in the relatively low intracellular concentration of cytosolic antigens [36]. Thus, indirect immunofluorescence, though the standard initial screening technique for ANCA detection in the systemic vasculitides, may not be the method of choice to detect ANCA in IBD and hepatobiliary disorders. The development of antigen-specific assays is necessary to bypass this problem. However, the still increasing number of target antigens for ANCA makes it hardly possible to develop a complete panel of antigen-specific assays for detection of ANCA in IBD.

CLINICAL SIGNIFICANCE OF ANCA IN IBD

In contrast to the systemic vasculitides, in which presence and titres of ANCA are related to disease activity and can be used as diagnostic and prognostic tools (reviewed in [50]), conflicting results have been reported on the relation between ANCA and clinical expression of IBD.

The higher prevalence of ANCA in UC in comparison with CD initially suggested that ANCA might be useful as a diagnostic marker to differentiate between UC and CD [8]. However, later studies showed that ANCA were also present in serum from patients with CD, although their prevalence (10–40%) was much lower than in UC (reviewed in [51]). Recent reports showed that combined testing for ANCA and for antibodies to the yeast Saccharomyces cerevisiae, which are relatively specific for CD, may be a sensitive and specific way to distinguish UC from CD [52].

Although ANCA are not very useful for distinction between UC and CD, they can be used as a diagnostic marker to distinguish IBD from other colitides and diarrhoeal illnesses [22]. However, their occurrence in several hepatobiliary [11,53] and rheumatic [54,55] disorders makes a careful diagnosis necessary.

Many studies have tried to find relations between ANCA and clinical features of IBD. In UC, ANCA⁺ patients were shown to have a more refractory type of disease than ANCA⁻ patients [56,57]. Several studies showed that the presence of the diseased organ is not necessary for the presence of ANCA in the serum from patients with UC. ANCA were detected after colectomy [58–60], although in some studies titres declined [61]. An association was reported between ANCA and the development of pouchitis in patients with UC who underwent surgery [60,62,63], but more recent studies failed to confirm this association [64,65]. Some UC patients with an ileal pouch reservoir develop chronic episodes of pouchitis, unresponsive to treatment. It has been hypothesized that this refractory pouchitis is an indication for an unrecognized CD in these patients [66]. Aisenberg et al. [67] utilized ANCA to test if refractory pouchitis represents underlying CD. However, ANCA were not less frequently present in patients with refractory pouchitis than in UC patients without pouchitis or with successfully treated pouchitis, suggesting that these patients were correctly diagnosed as having UC.

In a Californian study of CD, ANCA were suggested as markers of involvement of the colon in the inflammatory process. ANCA were especially present in CD patients with ‘ulcerative colitis-like features’, characterized as left-sided colonic inflammation, rectal bleeding, mucus discharge, urgency, and treatment with topical agents [68]. A French study, however, was not able to confirm these results in their study population [69].

In contrast to the associations between ANCA and refractory UC, pouchitis, and left-sided CD, as found in the aforementioned studies, most other studies failed to detect clinical differences between ANCA⁺ and ANCA⁻ patients in both UC and CD [7,23,32,70,71].

In the ANCA-associated vasculitides, several studies have shown a relation between disease activity and ANCA titre (reviewed in [50]). Rises in ANCA titre may occur during clinical relapse [72], but may also precede a clinical relapse [73]. Cohen Tervaert et al. [74] showed that treatment with immunosuppressive medication based on rises in ANCA titre prevented the occurrence of relapses in patients with WG. Thus, serial quantification of ANCA levels may be useful for follow up and treatment of patients with ANCA-associated vasculitides. In IBD, the relation between disease activity and ANCA titre is less clear. In UC, several studies reported that high titres of ANCA were especially found in patients with active disease [8,27,75], whereas other studies failed to detect a relation between disease activity and ANCA titre [7,24]. However, these were all cross-sectional studies. In a recent longitudinal study [76], we demonstrated that in patients with UC or CD ANCA titres are not related to disease activity. Thus, in contrast to the ANCA-associated vasculitides, serial quantification of ANCA levels in patients with IBD does not contribute to the monitoring of the disease process.

The disappointing results with respect to the clinical value of ANCA in IBD may be related to the heterogeneity of the antibodies in these diseases. They are directed against a wide variety of antigens. Few studies only have analysed clinical associations of the various antigenic specificities in IBD [26,29,32,56,77]. Sobajima et al. [56] reported an association between antibodies to cathepsin G and HMG1 and a refractory type of UC. Subsequently, they reported that titres of HMG1/2 antibodies were related to UC disease activity [48]. We [77] and others [26] showed that antibodies to lactoferrin are related to CD limited to the colon. In addition, Walmsley et al. [29] described antibodies to bactericidal/permeability-increasing protein as markers for colonic involvement in CD. In UC, antibodies to this antigen were related to a lower serum albumin concentration and a higher erythrocyte sedimentation rate, suggesting a severe course of the disease. However, other associations with clinical features of UC or CD were not found in any of these studies, suggesting that analysis of the antigenic specificities of ANCA does not substantially contribute to the delineation of particular subsets within the spectrum of IBD.

ANCA AS GENETIC MARKERS IN IBD

ANCA may be markers of genetic susceptibility to UC or primary sclerosing cholangitis. This is suggested by the finding of an increased prevalence of ANCA in unaffected first-degree relatives of patients with these diseases. A North-American report showed that ANCA were present in 15% of relatives of UC patients [78]. This was confirmed by a German study, in which even 30% of relatives of UC patients were found to have ANCA, as were 25% of relatives from patients with primary sclerosing cholangitis [79]. However, other studies on European populations failed to detect ANCA in unaffected relatives [80–83].

Susceptibility to IBD may be influenced by genes within the HLA complex, which are involved in the regulation of the immune response. Distinct associations of HLA class I and II antigens with both UC and CD have been reported [17]. Interestingly, a relationship between ANCA status and HLA alleles was demonstrated in a mixed Jewish/non-Jewish population from Los Angeles [84]. ANCA positivity in UC was associated with an increased frequency of the HLA-DR2 allele, whereas ANCA⁻ UC was associated with the HLA-DR4 allele. However, in Pennsylvanian and in European patients with UC, these associations were not confirmed [85–87]. Other associations between ANCA positivity and carriage of certain cytokine and adhesion molecule alleles have been published as well [88,89].

The associations between ANCA and genetic markers are thus not very consistent and seem to be highly dependent on the ethnic population that is studied.

PATHOPHYSIOLOGICAL SIGNIFICANCE OF ANCA IN IBD

ANCA with specificity for PR3 and MPO are likely to play a role in the pathophysiology of the vasculitic disorders associated with these autoantibodies. In vitro, ANCA are able to bind their target antigens on the surface of cytokine-primed neutrophils and thus stimulate neutrophil degranulation and oxygen radical production [90,91]. In addition, it has been shown that antibodies against PR3 and MPO are able to interfere with physiological functions of their target antigens [92,93]. Finally, animal models have also suggested a role for these antibodies in vivo (reviewed in [94]).

However, the pathophysiological significance of ANCA in IBD is less clear. Gionchetti et al. [95] were not able to induce oxygen radical production by primed neutrophils using ANCA of undefined specificity from patients with UC. Although rabbit polyclonal antibodies to lactoferrin were able to induce oxygen radical production of neutrophils [96], only some patients with ANCA directed against lactoferrin had antibodies that were able to induce oxygen radical production [91,96]. We were able to induce high levels of oxygen radical production by neutrophils using rabbit polyclonal antibodies to lactoferrin, bactericidal/permeability-increasing protein, and catalase, but not with antibodies to these antigens derived from patients with IBD, autoimmune liver disease, or RA (unpublished results). The difference in results between antibodies to PR3 and MPO on the one hand and antibodies to the aforementioned antigens on the other hand may be due to the relatively low titres of ANCA in patients with IBD compared with those in the systemic vasculitides.

Interference of ANCA from patients with IBD with the function of their target antigens, as has been shown for ANCA directed against PR3 and MPO in systemic vasculitis, has not been studied in detail. The presence of ANCA has been analysed in two animal models of IBD: ANCA were absent in a spontaneous colitis model in cotton-top tamarins [97], but were present in 70% of TCR-α-deficient mice [43].

Thus, a clear pathophysiological role for ANCA in IBD has not been established so far.

DEVELOPMENT OF ANCA: BREAKTHROUGH OF TOLERANCE?

It is not clear whether p-ANCA initially develop after exposure to neutrophil antigens or whether their development is due to cross-reactivity with environmental antigens that show homology to human neutrophil antigens, e.g. food antigens or bacterial antigens.

Human lactoferrin has high sequence homology (68%) with bovine lactoferrin [98], which is present in bovine milk and may thus, under certain conditions, be presented to the human immune system. Peen et al. [26] showed that antibodies against bovine lactoferrin do not cross-react with antibodies against human lactoferrin. We were also not able to find a relation between the presence of anti-bovine lactoferrin antibodies and anti-human lactoferrin antibodies in sera from patients with IBD (unpublished results).

The hypothesis of antigenic mimicry has been further investigated for autoantibodies directed against lactoferrin. Lactoferrin has high sequence homology with the mycobacterial heat shock protein (hsp)65, and rabbit polyclonal antibodies directed against lactoferrin react with mycobacterial antigens [99]. Antibodies to hsp65 have been detected in patients with IBD [100], although levels were not higher than in healthy controls [101]. Peen et al. [102] have tried to raise antibodies against human lactoferrin by immunizing rats with mycobacterial hsp65. Although the immunization resulted in high titres of antibodies against hsp65, no detectable antibodies to human lactoferrin were found, suggesting that antibodies to hsp65 do no cross-react with lactoferrin.

Cross-reactivity between ANCA and bacterial antigens has also been observed for autoantibodies against histone H1. A human MoAb derived from a UC patient [40] did react with histone H1 but also with bacterial antigens [103].

A pathogenic role for bacteria in IBD has been clearly suggested by experiments in several animal models of colitis. Mice that were deficient in IL-10 developed spontaneous colitis when maintained under conventional conditions [104]. However, when these animals were born and kept under specific pathogen-free conditions, colitis did not develop [104]. Seibold et al. [105] have shown that 67% of IL-10-deficient mice that develop spontaneous colitis produce ANCA. Sera from ANCA⁺ mice reacted with caecal bacterial antigens, whereas sera from ANCA⁻ mice did not react with these antigens, suggesting that ANCA in those sera were responsible for reaction with the bacterial antigens. Both in mouse and human p-ANCA⁺ sera, absorption with murine enteric bacterial antigens greatly reduced or abolished the p-ANCA staining [105]. The bacteria responsible for these effects have not been characterized. A recent study by Yang et al. [106] showed abolishment of p-ANCA staining of sera from UC patients by live Escherichia coli and Proteus mirabilis. However, their experiments suggested that this was not due to classic antigen–antibody interations, but rather to decomposition of the antigenic substrate of the neutrophils by enzymatic factors present in the supernatants of the live bacteria. Formalin-fixed or heat-killed bacteria did not influence p-ANCA staining.

Cross-reactivity between bacterial antigens and ANCA thus may be involved in the development of these autoantibodies. Newly developed animal models, such as the IL-10-deficient mouse model [104], will provide valuable tools to study the role of specific bacteria in the pathogenesis of IBD. These models will be useful as well for the comprehension of autoimmune processes in these disorders.

CONCLUSIONS

The diagnostic and prognostic role of ANCA in the systemic vasculitides has been well established. Since the detection of ANCA in IBD, many studies have dealt with the clinical value of ANCA in these disorders. However, it has to be concluded from these studies that the significance of ANCA in IBD is limited.

In contrast to the restricted number of antigens recognized by ANCA in the vasculitides, ANCA in UC and CD are directed against a variety of antigens. Some of these antigens are neutrophil granule proteins, but other antigens are cytosolic or nuclear proteins, that are usually not specific for neutrophil granulocytes but are also present in other cell types (Table 1). These antigens are not exclusively recognized by ANCA from patients with IBD but also by ANCA from patients with autoimmune liver diseases and rheumatic diseases, which limits their diagnostic specificity for UC and CD. Although ANCA are less prevalent in CD than in UC, the antigens recognized are identical and the distinction between the two disorders cannot be made based on the presence or absence of ANCA.

Several facts argue against a significant role for ANCA as prognostic markers in IBD. Although some studies described ANCA as markers for separate disease entities, most studies were not able to detect any clinical differences between ANCA⁺ and ANCA⁻ patients with IBD. Furthermore, in ANCA⁺ patients, titres of ANCA are not related to disease activity and thus cannot be used as a prognostic factor. Since ANCA are absent in a substantial proportion of patients with IBD and since they are also present in unaffected relatives of these patients, a pathogenic role is unlikely. Removal of the diseased organ does not lead to the disappearance of ANCA. Analysis of the antigenic specificities of ANCA in IBD does not substantially contribute to further insight into the delineation of particular subsets within this spectrum of diseases. ANCA may just be an epiphenomenon of chronic inflammation, and their development may be triggered by dysregulation of the immune system together with accidental cross-reactivity with environmental antigens.

In conclusion, the results of several years of research on ANCA in IBD have not shown promising evidence that ANCA are clinically useful. At this moment, testing for ANCA in daily clinical practice will not help the physician to obtain a definite diagnosis of ulcerative colitis or Crohn's disease, nor to predict the prognosis of inflammatory bowel disease in individual patients.