GASTROENTEROLOGY 1998;115:182–205
`
`SPECIAL REPORTS AND REVIEWS
`
`Inflammatory Bowel Disease: Etiology and Pathogenesis
`
`CLAUDIO FIOCCHI
`Division of Gastroenterology, University Hospitals of Cleveland, Case Western Reserve University School of Medicine, Cleveland, Ohio
`
`The field of gastroenterology offers many challenges
`
`to both the clinician and the investigator, but few
`are as complex and enigmatic as inflammatory bowel
`disease (IBD). Crohn’s disease (CD) and ulcerative colitis
`(UC) have been known for well over one half a century,
`but why affected individuals spend their lives with a
`chronic inflammatory process that relentlessly destroys
`their bowel remains a mystery. Advances, although
`unequal in different areas, have been plentiful in the last
`decade when the combined power of cellular and molecu-
`lar biology began to unveil the enigmas of IBD. No
`single agent or distinct mechanism is the sine qua non
`motive that explains all aspects of IBD, and several
`distinguishing factors are likely necessary to result in
`either CD or UC (Table 1). This review will attempt to
`discuss those that currently appear important.
`
`Conditioning IBD:
`Environment and Genes
`Environmental Factors
`Among the many puzzles of IBD pathogenesis,
`one of the less understood and most difficult to tackle is
`the role of environmental factors in the appearance and
`progression of CD and UC in this century. Environmental
`factors are probably as important as the patient’s genetic
`makeup for the risk of IBD. Potentially relevant environ-
`mental factors include prenatal events, breastfeeding,
`childhood infections, microbial agents, smoking, oral
`contraceptives, diet, hygiene, occupation, education, cli-
`mate, pollution, stress, and miscellaneous components
`such as toothpaste, appendectomy, tonsillectomy, blood
`transfusions, contact with animals, and physical activity.1
`Among these, the most established association is with
`smoking, but the reproducibility and strength of associa-
`tion with other risk factors needs further investigation.
`Smoking. The effect of smoking or, curiously, the
`opposite effect of smoking on the outcome of each form of
`IBD represents the most intriguing connection between
`environmental factors and IBD. With one exception,
`reports have shown that nonsmoking is a feature of
`patients with UC, whereas smoking is a feature of
`
`patients with CD.2–4 Smoking is an independent risk
`factor for clinical, surgical, and endoscopic recurrence in
`CD and influences disease activity after surgery. Al-
`though the components of tobacco responsible for these
`observations are uncertain, transdermal nicotine patches
`added to conventional therapy improve symptoms in
`patients with mild to moderate UC.5 A modulatory effect
`of nicotine on immune responses in vitro has been
`observed.6 However, what relationship,
`if any, exists
`between these in vitro responses and the in vivo effect on
`IBD is unclear. The inverse association in UC and CD is
`especially puzzling and may be linked to differences in
`the pathogenesis of the two disorders.
`Diet. A link between diet and IBD is logical
`because IBD affects the very site of nutrient absorption.
`Nutritional deficiencies in IBD are well documented,
`particularly that of zinc in CD with associated immuno-
`logic dysfunction.7 The effectiveness of elemental or
`special diets in reducing the symptoms or inducing
`remission of CD has been proposed but not universally
`accepted, and some studies have found that enteral
`nutrition is less effective than steroids and aminosalicy-
`lates.8,9 Moreover, the relapse rate does not appear to
`differ between nutritional and traditional approaches,10
`even when exclusion diets are implemented.11 Some data
`suggest that elemental diet may improve CD by reducing
`intestinal permeability,12 but it is not clear why nutri-
`tional therapies improve CD but not UC.
`Intestinal permeability. Increased intestinal per-
`meability may play a role in the pathogenesis of patients
`with CD. In addition to the patients themselves, in-
`creased permeability can be found in symptom-free
`first-degree relatives.13 Several studies have confirmed
`this observation, although results depend on the method-
`
`Abbreviations used in this paper: COX, cyclooxygenase; ICAM-1,
`intercellular adhesion molecule 1; IFN-␥, interferon gamma; IL,
`interleukin; PAF, platelet-activating factor; pANCA, perinuclear anti-
`neutrophil cytoplasmic antibody; PMN, polymorphonuclear leuko-
`cyte; ROM, reactive oxygen metabolite; TGF, transforming growth
`factor; TNF, tumor necrosis factor.
`r 1998 by the American Gastroenterological Association
`0016-5085/98/$3.00
`
`

`

`July 1998
`
`IBD ETIOLOGY AND PATHOGENESIS 183
`
`Table 1. Distinguishing Features of UC and CD Pathogenesis
`
`Component
`
`UC
`
`CD
`
`Environmental
`factors
`
`Beneficial effect of
`smoking
`No beneficial effect of
`diet
`Normal intestinal perme-
`ability in healthy rela-
`tives
`Largely different from
`Genetic
`CD
`associations
`Microbial agents Limited role of bacterial
`flora
`No association with M.
`paratuberculosis
`No association with
`measles virus
`Prominent antibody
`secretion
`Evidence for autoimmu-
`nity
`Strong association with
`ANCA
`Prominent neutrophil
`infiltration in the
`mucosa
`Normal/hyporeactive T
`cells
`Normal T-cell apoptosis
`(?)
`Prominent production of
`eicosanoids
`Th2-like profile
`Increased cytokine pro-
`duction limited to
`involved mucosa
`
`Humoral
`immunity
`
`Cell-mediated
`immunity
`
`Cytokines and
`mediators
`
`Detrimental effect of
`smoking
`Symptoms improved by
`selected diets
`Increased intestinal per-
`meability in healthy
`relatives
`Largely different from
`UC
`Important role of bacte-
`rial flora
`Association with M.para-
`tuberculosis
`Some association with
`measles virus
`Moderate antibody
`secretion
`Limited evidence for
`autoimmunity
`Weak association with
`ANCA
`Prominent T-cell infiltra-
`tion in the mucosa
`Hyperreactive T cells
`Resistance of T cells to
`apoptosis (?)
`
`Moderate production of
`eicosanoids
`Th1-like profile
`Increased cytokine pro-
`duction in involved and
`uninvolved mucosa
`
`direct link between an altered intestinal barrier and the
`well-documented immune aberrations of CD.
`Colonic mucus. Mucus represents a major compo-
`nent of intestinal defenses, and defects of this tough layer
`of glycoconjugates all along the gastrointestinal tract
`could predispose to IBD, UC in particular. A selective
`alteration of colonic mucin composition in involved and
`uninvolved mucosa of patients with UC but not CD was
`proposed18 but subsequently challenged by studies show-
`ing that the composition of purified mucins in UC and
`control subjects was similar.19 These discrepancies could
`be ascribed to methodological differences, but even if
`colonic mucin alterations were present, they alone would
`be insufficient to trigger inflammation because twins
`with the same defect are discordant for the occurrence of
`UC.20
`
`Familial and Genetic Factors
`Family studies. One of the most consistent obser-
`vations in studies of IBD populations is the high
`incidence of CD or UC among family members of the
`index case. Both vertical and horizontal associations
`occur, including father-son, father-daughter, mother-son,
`mother-daughter, and sibling-sibling. The frequency of
`IBD in first-degree family members may be as high as
`40%.21 Both genetic and environmental factors have been
`considered as the basis for these associations. Discriminat-
`ing between the two has been difficult because data can
`be interpreted in different ways. A large international
`cooperative study found very few statistically significant
`differences between index and control families.22 Except
`for increased incidence of eczema, no other differences
`were found between patients and controls in regard to a
`multitude of factors traditionally linked to IBD, includ-
`ing breastfeeding, childhood gastroenteritis, sugar and
`cereal consumption, and stressful events of life, among
`others. In large studies performed in Scandinavian coun-
`tries, a 10-fold increase in familial risk was interpreted to
`be strongly suggestive of a genetic cause,23 whereas
`seasonality of the cohort effect combined with the urban
`preponderance was interpreted as pointing to environmen-
`tal factors.24 A higher rate of concordance exists for
`monozygotic twins with CD or UC compared with
`dizygotic twins.25 Immunologic abnormalities such as
`higher frequency of antibodies to Escherichia coli O:14
`antigens and intestinal epithelial antigens in healthy
`first-degree relatives of patients with IBD have also been
`reported.26,27 However, these observations do not permit
`a clear distinction between genetic and environmental
`contribution. Of note, the frequency of affected first-
`degree relatives tends to be higher in CD than in UC. On
`the other hand, there is usually concordance of clinical
`
`ANCA, antineutrophil cytoplasmic antibodies.
`
`ology used, and some have failed to find altered permeabil-
`ity in relatives. Determining whether first-degree healthy
`relatives of patients with CD have increased permeability
`is of considerable pathogenic importance because it could
`represent a predisposing factor leading to overt inflamma-
`tion and clinical manifestations later in life. Of note,
`increased intestinal permeability antedating clinical mani-
`festations in CD has been reported.14 Although permeabil-
`ity is normal in the majority of relatives, those with a
`high permeability rate may represent a subgroup prone to
`developing CD.15 Indeed, when first-degree relatives are
`challenged with acetylsalicylic acid, permeability in-
`creases in some of them.16 The possible consequences of
`abnormal permeability have been shown by the fact that
`relatives have a concomitant increase in the number of
`circulating CD45RO⫹ B cells. This suggests that a leaky
`intestinal barrier intensifies antigen absorption, which, in
`turn, leads to an exaggerated immune stimulation re-
`flected at the systemic level by an excessive number of
`mature B cells.17 If correct, this finding would provide a
`
`

`

`184 CLAUDIO FIOCCHI
`
`GASTROENTEROLOGY Vol. 115, No. 1
`
`characteristics between the index case and family mem-
`bers in regard to age, site of inflammation, and type of
`clinical manifestations.28 At the very least, it is clear that
`neither form of IBD can be explained by a simple
`mendelian inheritance pattern.
`Genetic studies. In parallel with the above clini-
`cal observations, many studies investigating genetic
`factors in IBD have been performed (Table 2). An early
`investigation in a random European population found a
`significantly increased frequency of HLA-A11 and
`HLA-A7 in UC and a decreased frequency of HLA-A9 in
`CD.29 One of the most reproducible associations in a
`defined population is that with HLA-DR2 in Japanese
`subjects with UC.30 In contrast, no DR2 association is
`observed in British and American patients with IBD,
`although association with DRB1*103 and DRB1*12 is
`found in UC, including negative and positive associations
`with extent and severity of disease and extraintestinal
`manifestations.31–33 These differences show the value of
`less heterogeneous groups, such as the Japanese popula-
`tion, to minimize contrasting results among ethnically
`and genetically diverse populations in various countries.
`There is good evidence for genetic heterogeneity in IBD,
`
`Table 2. Genetic and Immunologic Associations
`Reported in IBD
`
`Association
`
`UC
`
`Decreased
`
`Increased
`
`Increaseda
`
`None, increaseda,b
`Decreaseda,b
`
`HLA-A3
`HLA-A9, HLA-B27
`HLA-A7, A11
`HLA-B12, DR1,
`DQB1*0501
`HLA-Bw52, -Bw35, DQw1,
`DPB1*0901
`DR2, DRB1*1502
`DR4, DRw6
`DRB1*01, *07, *0501,
`*1302, DRB3*0301
`DRB1*03, DQB1*0602,
`*0603
`DR3, DQ2
`
`DRB1*0103
`
`Extensive disease,
`decreased in women
`Extensive disease, extrain-
`testinal manifestations
`Chromosomes 3, 7, and 12 Linkage
`Chromosome 6
`Chromosome 16
`ANCA-positive
`IL-1 receptor antagonist
`allele 2
`TNFa2b1c2d4e1
`TNF-␣-308 allele 2
`ICAM-1, ANCA-negative
`ICAM-1, ANCA-positive
`
`No linkage
`Increased
`Extensive disease
`
`None
`Decreased
`Increased
`
`ANCA, antineutrophil cytoplasmic antibodies.
`aJapanese patients.
`bMostly Jewish patients.
`cNo linkage in Jewish patients.
`
`which is probably responsible for disparate associations
`with class II antigens, such as positive association with
`the HLA-DR2 and a negative association of DR4 and
`DRw6 in UC and a positive association with DR1 and
`DQw5 in CD.34 Studies combining genetic and immuno-
`logic markers corroborate the existence of genetic hetero-
`geneity in IBD,35 but the consistency of findings varies
`depending on the target population.32
`Evaluation of Japanese patients with CD found suscep-
`tibility and resistance
`linked to DQB1*4 and
`DQA1*0102, respectively.36 In a large cohort of French
`patients with CD, however, susceptibility was associated
`with the HLA-DRB1*01 and DRB1*07 alleles, and
`resistance was associated with the DRB1*03 allele,37
`whereas an increased risk for CD, but not UC, was
`detected in an American white population carrying the
`HLA-DRB3*0301 allele.38 In sharp contrast, no evidence
`that class II genes contribute to CD susceptibility and
`disease phenotype was reported in a British study.31
`Associations of IBD with interleukin (IL) 1 receptor
`antagonist, intercellular adhesion molecule 1 (ICAM-1),
`and tumor necrosis factor (TNF)-␣ genes have been
`described, but the strength of these associations is
`variable and generally weak.39–42 Using large aggregates
`of families with IBD, recent studies at the genomic level
`started to identify linkages to specific chromosomes. A
`French group reported that CD genetic predisposition
`resides outside of chromosome 643 and subsequently
`reported a susceptibility locus on chromosome 16,44 a
`finding confirmed by other investigators. A British group
`reported evidence for susceptibility for both CD and UC
`on chromosomes 3, 7, and 12, suggesting that these are
`distinct disorders sharing some, but not all, susceptibility
`genes.45 These observations not only provide new in-
`sights into potential IBD genes but may help discard
`older and inconsistent data obtained with flawed serologi-
`cal methods.
`A concept that further supports the relevance of genes
`to disease predisposition is that of genetic anticipation,
`defined by a progressively earlier disease onset in succes-
`sive generations.46 Multicenter screening studies using
`well-defined groups of patients and families will undoubt-
`edly provide invaluable information on the role of
`genetics in IBD. In the midst of new enthusiasm, it
`should not be forgotten that even if identical twins are
`significantly more likely to develop IBD than nonidenti-
`cal twins, this only accounts for part of the concordance.
`Furthermore, there is an irreconcilable disparity between
`the rapid increase in IBD incidence during the last few
`decades and any plausible changes in the genetic makeup
`of stable populations during the same period of time.
`Thus, nongenetic, i.e., environmental factors, are clearly
`
`CD
`
`None
`Decreased
`None
`Increasedb
`
`None
`None
`Increased
`
`Decreased
`
`Linkage
`No linkage
`Linkagec
`
`Increased
`No linkage
`
`Increased
`
`

`

`July 1998
`
`IBD ETIOLOGY AND PATHOGENESIS 185
`
`indispensable to develop IBD through a still nebulous
`interaction with multiple predisposing, but not sanction-
`ing, genes.
`
`Causing and Promoting IBD:
`Microbial Agents
`Bacteria
`An infectious etiology for IBD, with a direct
`cause-and-effect relationship between a single microorgan-
`ism and inflammation, still remains plausible. The
`relationship between microbes and defined clinical enti-
`ties is often ambiguous, but diseases of unknown etiology
`are unexpectedly proven to be infectious. The most
`striking example is peptic ulcer disease and Helicobacter
`pylori. Although traditional methods have failed to detect
`a specific pathogen associated with IBD, newer molecular
`biological guidelines are now available to establish
`microbial causation in IBD.47
`Immune reactivity to microbial antigens. A tra-
`ditional approach to identify microbial agents linked to
`IBD has been the demonstration of antibodies against
`known microorganisms. Essentially all patients with IBD
`have elevated titers against bacteria, viruses, and fungi,
`making the mere presence of serum or intestinal antibod-
`ies of limited value. Although patients with active CD
`have elevated serum antibody titers to multiple enteric
`bacterial pathogens,48 this seroreactivity likely represents
`a secondary sensitization to cross-reacting antigens or a
`nonspecific enhancement of immune reactivity.
`Cell-mediated immunity against microbes has also
`been assessed. Early studies showed that migration of
`leukocytes in vitro was inhibited by colonic homogenates
`in UC but not CD, and subsequent studies showed
`migration-inhibitory activity against enterobacterial com-
`mon antigen by peripheral blood lymphocytes of patients
`with IBD.49 These results should be interpreted in the
`context of ‘‘physiological intestinal inflammation,’’ im-
`plicit to which is tolerance to local endogenous bacteria.
`Thus, although T cells from uninvolved mucosa fail to
`proliferate in vitro on stimulation by pathogenic and
`nonpathogenic microbial antigens in CD, the heightened
`response of T cells from involved mucosa implies antigen-
`specific recognition and loss of tolerance.50
`Intestinal flora. The possibility that components
`of the normal intestinal flora could trigger, initiate, or
`somehow contribute to IBD has intrigued investigators
`for years. Differences in microbial ecology among indi-
`viduals with UC and CD and normal individuals were not
`strongly supported by early investigations, but a recent
`study found a significant decrease in the number of
`anaerobic bacteria and Lactobacillus in patients with active
`
`but not inactive IBD.51 It is possible that products of the
`commensal flora promote inflammation in the presence of
`an impaired mucosal barrier or injury to the mucosa.52
`The concept that the normal flora somehow functions as a
`modulator of
`‘‘physiological
`inflammation’’ has been
`strengthened substantially by the observations of Duch-
`mann et al.53,54 They have shown that mucosal, but not
`peripheral blood, mononuclear cells from patients with
`IBD proliferate when exposed to autologous intestinal
`bacteria. Cells from uninvolved mucosa of the same
`patients and from patients in remission failed to prolifer-
`ate to autologous IBD flora. Blood and mucosal mono-
`nuclear cells from normal individuals, however, fail to
`proliferate to autologous bacteria, but do so in response to
`sonicates of bacteria from heterologous intestine. These
`results indicate that, in health, there is tolerance to
`autologous but not allogeneic intestinal flora, and toler-
`ance is lost during inflammation.53 Evidence also exists
`that animals are tolerant to their own flora in health but
`not after colitis develops.54 Infection breaks T-cell toler-
`ance, and the hypothesis that intestinal infections abro-
`gate tolerance leading to a chronic immune response to
`self flora in susceptible individuals is appealing.
`Fecal stream and bacterial stasis. The concept
`that intestinal contents contribute to IBD pathogenesis is
`gaining momentum. Fecal stream diversion may deter-
`mine recurrence of CD.55 Patients with ileal resections
`and a diverting ileostomy excluding the neoterminal
`ileum fail to develop recurrent disease until reanastomo-
`sis is performed. Furthermore, infusion of autologous
`luminal contents in excluded normal
`ileal
`loops of
`patients with CD rapidly induces new inflammation,
`indicating that fecal component(s) may promote disease
`flare-up.56 These observations are reminiscent of pouchi-
`tis, where repopulation of the pouch by enteric flora
`seems to be essential for induction or persistence of
`inflammation.57 The common denominator of all these
`observations is a change in the quantity or quality of
`bacterial and dietary luminal content. In animals, exces-
`sive enteric bacteria lead to increased expression of cell
`adhesion molecules and granulocyte infiltration, and
`both are reduced by antibiotics or speeding fecal flow.58
`These clinical and experimental observations suggest that
`microbial
`imbalance could be a major factor in the
`development or maintenance of chronic intestinal inflam-
`mation.
`Mycobacteria. An association of CD with myco-
`bacteria has been postulated since the original description
`of terminal ileitis. This idea has been resurrected over a
`decade ago by the recovery of an unclassified acid-fast
`organism related to Mycobacterium paratuberculosis from
`CD tissues.59 Neonate goats fed with such an organism
`
`

`

`186 CLAUDIO FIOCCHI
`
`GASTROENTEROLOGY Vol. 115, No. 1
`
`showed evidence of both humoral and cellular immunity
`to it and developed granulomatous inflammation in the
`terminal ileum. This observation spurred a tremendous
`amount of investigation followed by very inconsistent
`results. Some showed that patients with CD had signifi-
`cantly increased serum antibody titers to M. paratuberculo-
`sis,60 but subsequent studies failed to find a significant
`elevation of serum immunoglobulin (Ig) A, IgG, or IgM
`to mycobacterial antigens,61 despite evidence of contact
`with other environmental mycobacteria. No evidence of
`systemic or mucosal
`cell-mediated immunity to
`M. paratuberculosis has yet been shown.
`Detecting M. paratuberculosis in involved tissues has
`been difficult. Immunohistochemistry fails to show the
`organism in tissue sections,62 and isolation of M. paratu-
`berculosis in bacterial or spheroplastic form has met with
`limited success.63 Even the sensitive polymerase chain
`reaction does not provide definitive evidence for
`M. paratuberculosis–specific DNA in CD-affected bowel.
`Up to 65% of involved tissues may contain amplification
`products compared with 4% in patients with UC and
`12% in controls,64 and they can be found in granuloma-
`tous tissue. Collectively, these studies make only a
`marginal case for an important role of M. paratuberculosis
`in CD. The similarities between CD and Johne’s disease
`in animals remain a puzzling and unresolved issue, but it
`is still plausible that M. paratuberculosis causes CD in a
`few patients.65 Inconsistent results with antituberculous
`therapy have not helped to resolve these issues.66
`Other transmissible agents. The possibility that
`a nonconventional bacteria or virus is related to CD has
`been considered after investigators induced chronic granu-
`lomatous inflammation in mice67 or cytopathic effect in
`tissue cultures with CD tissue homogenates.68 This line
`of investigation came to an end with the demonstration
`that the cytopathic effect of IBD tissue extracts was
`caused by toxic proteins.69
`
`Measles Virus and Mesenteric Vasculitis
`The complex challenges
`in defining ‘‘IBD-
`specific’’ agents are shown by studies on the measles virus.
`A vascular cause for the etiology of IBD was proposed
`more than a quarter of a century ago, but recently
`Wakefield et al. have provided more evidence for the
`involvement of the intestinal microcirculation in CD.
`Using resected specimens and an infusion technique,
`these investigators showed vascular injury, focal arteritis,
`fibrin deposition, and arterial occlusions in the inflamed
`segments.70 In follow-up studies, they suggested that the
`vascular injury was caused by a diffuse granulomatous
`vasculitis.71 Detailed examination of such intravascular
`
`granulomas showed the presence of paramyxovirus-like
`particles by transmission electron microscopy, compatible
`with the presence of the measles virus in CD tissues.72
`This suggestion was reinforced by positive hybridization
`for the measles virus N-protein genomic RNA and
`positive immunohistochemical staining for nucleocapsid
`protein in CD tissues, but only a minority of control
`intestinal tissues. Although the actual presence of measles
`virus in CD tissue has been challenged, the original group
`of investigators has provided epidemiological data indicat-
`ing that perinatal exposure to measles virus increases
`significantly the risk of CD,73 and measles vaccination
`increases the risk of developing IBD.74 Finally, they
`showed that inherited disorders of coagulation protect
`against IBD.75 Together, these observations could indi-
`cate a role for vascular abnormalities in CD pathogenesis,
`a hypothesis fitting with the increased risk of CD in
`people who smoke or use contraceptives. The notion of a
`vascular component in CD pathogenesis is credible, but
`whether it is specifically or exclusively caused by measles
`infection may be extremely difficult to prove.
`
`Mediating IBD:
`Immune and Nonimmune Systems
`Humoral Immunity
`Mucosal B cells. When investigation switched
`from systemic immune events to those occurring in
`IBD-involved intestine, comparative mapping of B-cell
`distribution showed a massive increase in the number of
`plasma cells. This increase is not uniform with IgA, IgM,
`and IgG increasing 2-, 5-, and 30-fold, respectively.76
`This pronounced humoral immune response is particu-
`larly striking for IgG, the antibody class with the highest
`pathogenic potential. The two forms of IBD can be
`distinguished based on this Ig’s subclasses: IgG1 increas-
`ing more in UC and IgG2 in CD. The reason for this
`difference is unclear, but distinct immune regulatory
`mechanisms or genetically conditioned differences may
`be responsible.77 Abnormalities of J-chain expression are
`also present, both IgA2 and IgA1 being decreased in
`IBD.78
`
`Autoantibodies and autoantigens. Much of the
`interest in studying humoral
`immunity in IBD is
`centered on whether disease-causing autoantibodies are
`produced. In a hallmark study in the late 1950s,
`Broberger and Perlmann described antibodies in the
`serum of UC patients that cross-reacted with human fetal
`colon cells in vitro.79 The anticolon antibodies were not
`cytotoxic for colon cells, but this report was the first to
`highlight the concept of autoimmunity in IBD. Patients
`with IBD possess a variety of circulating antibodies that
`
`

`

`July 1998
`
`IBD ETIOLOGY AND PATHOGENESIS 187
`
`seldom correlate with disease activity, e.g., cow’s milk
`proteins.80 Anticolon antibodies are not disease specific81
`and are not tissue specific, reacting, for instance, with
`biliary epithelium.82 Antibodies to cytoskeletal proteins
`are frequent in IBD83 but are also nonspecific, being
`found in many other disorders as well. Not all antibodies
`lack specificity. Serum antibodies to Saccharomyces cerevi-
`siae (baker’s yeast) are increased in patients with CD but
`not UC,84 and a novel antierythrocyte autoantibody
`appears to have some specificity for CD.85
`Validation for a role of autoimmunity requires the
`identification of autoantibodies that specifically recog-
`nize gut antigens and mediate tissue damage via this
`recognition. Candidate autoantigens have been isolated
`from intestinal epithelium (epithelial cell–associated
`components),86 and sensitization to them has been shown
`in patients with IBD87 and their relatives.27 Still, none of
`them has been shown to cause gut inflammation. The
`best-defined autoantigen has been described by Takaha-
`shi and Das, who characterized a colonic protein of
`40-kilodalton size exclusively recognized by IgG eluted
`from the colon affected by UC.88 Monoclonal antibodies
`against this putative autoantigen identified a shared
`epitope in human colon, skin, and biliary epithelium as
`well as the eye and joints, locations compatible with the
`extraintestinal manifestations of UC.89,90 This autoanti-
`gen appears to be the fraction 5 of the tropomyosin family
`of cytoskeletal proteins91 and is codeposited with IgG1
`and complement on UC colonocytes.92 The true relevance
`of these findings for UC pathogenesis remains to be
`established.
`Antineutrophil cytoplasmic antibodies. Antibod-
`ies against polymorphonuclear neutrophils (PMNs) or
`perinuclear
`antineutrophil
`cytoplasmic
`antibodies
`(pANCAs) have attracted the most interest.93 A series of
`reports has confirmed the high prevalence (approximately
`80%) of these antibodies in UC and an even higher
`prevalence in patients with primary sclerosing cholangi-
`tis.94 A related finding is the increased incidence of
`pANCAs in healthy relatives of patients with UC and, to
`a lesser degree, patients with CD.95 The prevalence of
`pANCAs in UC does not significantly vary in ethnically
`diverse populations,96 but it does among unaffected
`relatives in different areas of the world.97 In addition to
`being a potential marker of susceptibility and genetic
`heterogeneity in UC, pANCAs may define subsets of
`patients with IBD because recent studies suggest that
`pANCAs are more prevalent in patients with aggressive
`UC98 and in those who develop chronic pouchitis after
`ileo-pouch anastomosis.99
`A fundamental question is the true relevance of
`autoantibodies to UC in particular and IBD in general.
`
`pANCAs have yet to be proven to alter neutrophil
`function, although the recent demonstration of pANCA-
`producing B-cell clones in UC mucosa leaves the door
`open to alternate roles.100 A related question is how many
`other autoantibodies exist in IBD and what is the range of
`their specificity. Humoral autoimmune phenomena are
`common in IBD, but their relevance to disease pathogen-
`esis needs clarification, particularly because the simple
`occurrence of autoantibodies no longer indicates a patho-
`logical condition (horror autotoxicus), and it is part of
`normal immune homeostasis.
`
`Cell-Mediated Immunity
`Systemic immunity. Broberger and Perlmann
`were also the first to show the cytotoxic action of UC
`peripheral blood white blood cells against human fetal
`colon cells in vitro.101 The cells mediating cytotoxicity
`were lymphocytes. These observations showed that im-
`mune cells could recognize and destroy intestinal cells,
`implicating classical immune mechanisms in IBD patho-
`genesis. Shorter et al. confirmed the action of blood
`lymphocytes against freshly isolated colonocytes102 and
`eventually concluded that natural killer cells were respon-
`sible for this phenomenon.103 These studies raised much
`excitement but relied on fairly crude assay systems, and
`the validity of the results is questionable.
`Impaired natural killer cell activity has been described
`in IBD,104 but other reports failed to confirm this
`observation. These early studies showed the inconsistency
`that still plagues immunologic data in IBD. Using
`mitogen-induced proliferation, some investigators de-
`tected impaired responsiveness in IBD,105 a finding
`corroborated by anergic skin reactivity to dinitrochloro-
`benzene and depressed lymphocyte numbers in the
`circulation of patients with CD and UC.106 In contrast,
`similar studies found no evidence for decreased immune
`reactivity in either subjects with CD or with UC.107
`Prolonged controversy also raged in trying to determine
`whether normal or defective generation of suppressor
`function is present in IBD, with different approaches
`providing conflicting results.108,109
`Mucosal T cells. One important aspect of IBD,
`especially CD, is that the intestine can be infiltrated
`diffusely by immune cells in the absence of obvious
`morphological, clinical, and endoscopic evidence of inflam-
`mation. Aphthous ulcers are considered characteristic
`early lesions of CD, but how ‘‘early’’ such lesions really are
`in seemingly affected mucosa is debatable. Ultrastruc-
`tural alterations,110 increased numbers of CD68⫹ macro-
`phages,111 and lymphocytic accumulation can go undetec-
`ted in ‘‘unaffected’’ mucosa in patients with CD.112 Thus,
`how and when mucosal
`immunocytes go awry is a
`
`

`

`188 CLAUDIO FIOCCHI
`
`GASTROENTEROLOGY Vol. 115, No. 1
`
`fundamental but unresolved question in IBD pathophysi-
`ology. In established lesions, the presence of T-cell
`infiltrates was detected by early immunohistochemical
`studies, but in contrast to other chronic inflammatory
`diseases, the proportions of CD4⫹ and CD8⫹ T cells were
`comparable in involved and uninvolved mucosa,113 point-
`ing to quantitative rather than qualitative changes in
`IBD tissue. Although the elevated numbers of T cells in
`the mucosa affected by either form of IBD suggest a
`connection between this cell type and tissue destruction,
`direct evidence is still lacking despite experimental data
`showing that activated T cells induce mucosal damage in
`organ culture.114
`function. Techniques allowing
`Mucosal T-cell
`the isolation of immune cells directly from the mucosa
`constituted a major advance in the study of IBD.
`Although some differences were detected in the popula-
`tion of lamina propria compared with peripheral blood
`mononuclear cells, the proportions of T- and B-cell
`subsets were unexpectedly similar between IBD and
`control mucosa.115 Not surprisingly, local cells in IBD
`showed an activated phenotype. Markers of lymphocyte
`activation, including the 4F2 antigen, the T9 transferrin
`receptor (CD71), and IL-2 receptor, are increased among
`isolated intestinal B cells and T cells and the CD4⫹ and
`CD8⫹ subsets,116 as is the expression of IL-2, IL-2
`receptor ␣, and ␤ messenger RNA (mRNA).11