The Promise and Pitfalls of Serologic Testing in Inflammatory Bowel Disease

Serologies have well-established roles in the diagnosis of a variety of immunologically driven diseases; rheumatologists rely on an alphabet soup of antibodies, hepatologists depend on serologic tests to diagnose primary biliary cirrhosis (PBC) and autoimmune hepatitis, and celiac experts have used a battery of tests for diagnosis, adherence, and a better understanding of pathophysiology. In these specialties, serologies provide a robust diagnostic accuracy either by themselves or with a minimum of other studies to establish a diagnosis or guide therapeutic decision making.

Serologic testing is a natural consideration in another immunologically driven disease, inflammatory bowel disease (IBD). If a relatively simple blood draw could provide an equally accurate but less expensive, less laborious, and less invasive test than the present gold standard of endoscopy, it would be a major advance in our approach to diagnosis and treatment of IBD. There is a 5-decade history of antibodies in ulcerative colitis; the first report of antibodies in human ulcerative colitis date from 1959. Over the last 2 decades, there has been an intensification of the interest in IBD-related antibodies.

In this review, we examine 4 areas in which IBD serologies might be important: (1) to provide insight into the etiopathogenesis of IBD, (2) to establish the diagnosis of IBD, (3) to differentiate Crohn disease (CD) from ulcerative colitis (UC), and (4) to stratify risk of developing a complicated course that might dictate earlier more aggressive treatment.

There are considerable data to suggest that IBD serologies may provide important information in all of these areas. However, it is not clear that serologic testing can adequately replace endoscopy or other simpler clinical indicators or surrogate markers of inflammation. If they can serve only a limited, adjunctive role in diagnosis and management of IBD, then one would need to carefully evaluate the specific circumstances and added value of additional testing.

Understandably, most studies have examined IBD serologies in well-defined IBD patient populations. Few studies attempt to characterize prospectively the utility of these serologies in patients who have yet to have IBD diagnosed or who are early in the course of IBD.

In 2011, there is limited evidence to recommend routine use of the current panels of serologic markers for diagnosing, stratifying, or monitoring IBD in clinical practice. However, this is a rapidly evolving field, and, perhaps, new findings may establish a role for IBD serologies in the future. As always, more studies are needed.

Pathogenesis

What is the connection between IBD antibodies and the pathophysiology of the disease? In some diseases, there is a strong link between an identified antibody and the underlying immunopathogenesis. For example, in celiac disease, antibodies directed against tissue transglutaminase provide insight into the molecular pathogenesis of the disease. Similarly, anti mitochondrial antibodies target a specific pathway important in the development of PBC.

The identification of an increasing number of diverse antimicrobial antibodies in IBD suggests that none of these are likely to be integral to the pathogenesis but are surrogate markers for a change in the interaction of the gut epithelium and the gut microbiome. Two possible changes that have been implicated in IBD are (1) a “leaky gut” with subsequent development of antibodies secondary to increased permeability, or (2) a generalized loss of tolerance to commensal organisms in the intestinal lumen in Crohn that may, for unknown reasons, lead to a more targeted microbial and auto antigen response.

If IBD-associated antibodies are a reflection of increased permeability, then one might expect an increased incidence of antibodies against dietary antigens. This does not appear to be the case. The plethora of antibodies associated with CD suggests a more generalized immune response. However, there is no indication of what is specific about these antigens that would lead to IBD. A higher rate of multiple antibody production may indicate a more global loss of tolerance to intestinal flora. Indeed, Adams and colleagues suggested that immunoglobulin (IgG) antibodies against common bacteria are more diagnostic for CD than IgG against mannan (anti-Saccharomyces cerevisiae antibody [ASCA]) or flagellin (CBir1).

Although there has been a greater focus on antimicrobial antibodies associated with IBD, there are some less-characterized (auto) antibodies directed against components of the intestinal epithelium, including goblet cells and microfilament proteins. These might conceivably be more directed to the etiology of disease. Further studies are needed to clarify a possible connection.

The observations that there can be “preclinical seropositivity” in individuals in whom IBD develops years later and that there is a significant familial incidence of positive IBD-related antibodies in clinically unaffected, first-degree relatives of IBD patients raises questions about whether genetic factors may also play a role. To date there does not seem to be an IBD gene locus that predicts seropositivity. This is an important area for future investigation.

There is an interesting inverse relationship between IBD serologies and seroprevalence of Helicobacter pylori infection. Although one might speculate whether H pylori may have a protective effect on IBD, it is more probable that both antibody responses may be linked to environmental effects during childhood that have been synthesized as the hygiene hypothesis.

Characteristics of IBD Antibodies

Identification, description, and commercialization of 2 antibodies, antineutrophilic cytoplasmic antibody (ANCA) and ASCA invigorated research in IBD serologies. Over the last several years, there has been a steady increase in the number of IBD antibodies identified. Eventually, this range of antibodies should provide a better understanding of the etiopathogenesis of IBD and lead to more judicious clinical management of patients with ulcerative colitis and Crohn. However, at this point, the array of antibodies may leave the clinician more confused than enlightened.

We will classify these antibodies into 3 somewhat arbitrary groups: (1) what may be considered the first-generation of IBD-related antibodies that are widely available now through Prometheus Labs, (2) a group of antiglycan antibodies, and (3) a group of orphan autoantibodies with diverse targets in the gastrointestinal mucosa ( Table 1 ).

Table 1

IBD-associated antibodies, presumed targets, and select population prevalences

Antibody	Target	Prevalence (%)
Antibody	Target	Ulcerative Colitis	Crohn Disease	Healthy Controls
pANCA	Unidentified protein of the nuclear envelope of neutrophils	41–73	6–38	0–8
ASCAgASCA	Carbohydrate epitopes present in the phophopeptidomannan of the cell wall of Saccharomyces cerevisiae Covalently bound mannin	0–29	29–69	0–16
OmpC	Omp-C transport protein of E coli	2–24	24–55	5–20
I2	Pseudomonas-associated sequence I2	42	38–60	15
Cbir1	Bacterial flagellin CBir1	6	50–56	8
ALCA	Laminaribioside (Glc[b1,3]Glc[b])	3–8	19–27	2
ACCA	Chitobioside (GlcNAc[b1,4]GlcNAc[b])	5–7	8–25	0.5–12
AMCA	Mannobioside (Man[a1,3]Mana)	7	12–28	9
Anti-L	Laminarin (Glc[b 1,3])3n(Glc[ b 1,6])n	3–7	18–26	1–10
Anti-C	Chitin (GlcNAc[ b 1,4])n	2–11	10–25	2–12
Antigoblet cell	Goblet cells of the intestines
Tropomyosin	Microfilament protein tropomyosin	79	12
Pancreatic autoantibody	GP2 a major zymogen granule membrane glycoprotein		30

Data from Prideaux L, De Crux P, Ng SC, et al. Serological antibodies in inflammatory bowel disease: a systematic review. Inflamm Bowel Dis 2011. DOI:10.1002/ibd.21903 . [Epub ahead of print.]

First-Generation Antibodies

ANCA

ANCAs were first associated with IBD, specifically UC, in the 1980s. ANCAs describe a class of autoantibodies, typically of the IgG subclass, directed against antigens in the cytoplasm of neutrophils and monocytes. B cells, in the intestinal mucosa, produce p-ANCA presumably against a self epitope or resident pathogen.

In 1983, Nielson and colleagues described the presence of ANCA in the sera of patients with UC. They used an indirect immunofluorescence technique to identify a perinuclear pattern to the ANCA in UC. Subsequently, numerous groups went on to further characterize p-ANCA in IBD, with groups noting a certain percentage of CD patient’s would express ANCAs.

Vasiliauskas and coworkers in 1996, performed a study comparing ANCA-positive CD patients with negative controls to further characterize clinical subtypes. They found that p-ANCA–positive CD patients more frequently have left-sided colitis and symptoms of left-sided inflammation than patients negative for p-ANCA.

Sandborn and colleagues showed in a study comparing the tests properties of p-ANCA and ASCA across several laboratories that the sensitivity of testing for p-ANCA varies between 0 and 63%. They attributed this variation to the tests potentially targeting different unidentified antigens. Vidrich and coworkers showed that the use of DNase with UC sera disrupts its reactivity to p-ANCA, potentially improving the sensitivity of detecting UC-specific p-ANCA.

Reese and coworkers examined the diagnostic ability of p-ANCA and ASCA to identify UC and CD, respectively. They performed a meta-analysis of 60 studies, which included 3,841 UC patients and 4,019 CD patients. They found that p-ANCA positivity translated to a sensitivity and specificity of 55.3% and 88.5%, respectively, for the diagnosis of UC.

Taylor and colleagues in 2001 described the relationship between p-ANCA positivity and response to anti-tumor necrosis factor (TNF) therapy in a cohort of CD patients. They examined the response of 59 patients to anti-TNF therapy based on serologic and genetic factors. They found that patients with p-ANCA had a similar response to anti-TNF therapy as they did to placebo, with ASCA patients responding significantly better. Ferrante and coworkers found similar results when they examined the response of 100 UC patients to infusion with infliximab. The subgroup with pANCA+ and ASCA– serologies had a statistically lower response rate (55% vs 76%, P = .049).

The relationship between surgical interventions for UC and p-ANCA was also examined. Aitola and colleagues examined the sera of 15 UC patients before surgery and followed up with them for a median duration of 24 months postoperatively to study p-ANCA titers. They found that preoperatively, 13 of 15 patients had positive results for p-ANCA, and postoperatively titers decreased in 10 patients and became negative in 2.

Mow and coworkers examined the prognostic value of p-ANCA in a cohort of 303 CD patients, correlating phenotypes to serologic abnormalities. They found that p-ANCA–positive patients had a statistically significantly decreased risk of small bowel disease, fibrostenosing disease, and small bowel surgery. They did, however, have a higher risk of exhibiting UC-like features.

ASCA

ASCAs are produced in response to yeast cell wall components, particularly mannan. It is commonly known as baker’s or brewer’s yeast and is common in diets. Main and colleagues in 1988 described the presence of this antibody in a small cohort of IBD patients. They found that the sera of CD patients expressed a larger proportion of IgG and IgA antibodies directed against Saccharomyces cerevisiae when compared with UC and normal controls.

Giaffer and colleagues further characterized the antibody responses with particular strains of Saccharomyces cerevisiae yeast. They tested for IgG and IgA antibody levels in 49 CD patients, 43 UC patients and 21 controls through ELISA. Both IgG and IgA levels were significantly increased in the CD cohort, versus both the UC cohort and the healthy controls.

Vasiliauskas and colleagues studied the ability of ASCA to define clinical phenotypes and natural history in CD. They studied the sera of 156 consecutive established CD patients for ASCA and then correlated sera to their clinical phenotype in a blinded fashion. They found that higher ASCA titers correlated with earlier age of onset, fibrostenosing, and internal penetrating disease.

In the previously described study by Reese and coworkers, ASCA was also examined for its sensitivity and specificity in diagnosing CD. They found positive ASCA combined with negative p-ANCA yielded a sensitivity and specificity of 54.6% and 92.8%, respectively. Mow and colleagues also showed the presence of ASCA can be used to describe phenotype and possibly assist in prognosis, because these patients have a higher risk for small bowel surgery.

Recent literature has described gASCAs, which are antibodies against covalently immobilized mannan. Although there may be some subtle differences between the 2 assays, they seem to perform basically the same in identification of patients with CD.

Outer membrane porin C

Outer membrane porin C (Omp-C) is a protein belonging to Escherichia coli that is used to regulate metabolite and toxin transport. It was first associated with IBD in 2000 through bacterial library screening, which identified E coli . Further studies showed the Omp-C protein was immunogenic, with a proportion of UC patients generating antibodies to Omp-C. In CD, Omp-C has been correlated with increased risk of fibrostenosing disease, internal penetrating disease, and need for small bowel surgery.

In a pilot study performed by Mow and coworkers, they examined the possible role the presence of Omp-C may have in identifying CD patients with better chances of clinical response to antibiotic therapy. They compared patients taking budesonide with patients taking budesonide plus metronidazole and ciprofloxacin. They found that in patients expressing higher levels of Omp-C, patients responded better, defined as a decrease in the Crohn Disease Activity Index score of 150 or greater, with the addition of antibiotics. These results, however, did not reach statistical significance.

I2

I2 was first discovered by cloning fragments of bacterial DNA recovered from lamina propria mononuclear cells present in active CD tissue but not uninvolved tissue or UC. Further genetic analysis showed this bacterial sequence belongs to Pseudomonas fluorescens . I2 correlates, in CD patients, to small bowel disease, fibrostenotic disease, and small bowel surgery. I2 was also shown, in the previously mentioned pilot study by Mow and colleagues, to select CD patients, who might benefit from treatment with antibiotics.

Spivak and coworkers found that I2 was a predictor of response to fecal diversion in CD. Seventeen of 27 (63%) patients achieved clinical response from fecal diversion; however, 15 of 16 patients who were I2 positive (94%) responded.

CBir1

CBir1, a flagellin, is a microbial antigen that presumably induces colitis through an adaptive immune response. Targan and coworkers evaluated serum responses to CBir1 in CD patients, compare this with other known IBD-related serologies and CD phenotypes. They tested sera from 484 CD patients and found CBir1 present in all previously described antibody subgroups. The expression of CBir1 increased with combinations of other IBD-related serologies. CBir1 was associated with small bowel disease, fibrostenotic disease, and internal penetrating disease. Tables 1 and 2 summarize the previously described antibodies in relation to their prevalences and clinical characteristics.

Table 2

IBD-associated antibodies, predominant disease location in Crohn disease patients, and disease behavior

Antibody	Predominant Disease Location	Disease Behavior/Significance
pANCA	Colon	Distal colitis (inflammatory phenotype)
ASCA	Small bowel	FS, IP
Omp-C	Small bowel (+/–) In combination with other Abs	FS, IP
I2	Small bowel	FS
CBir1	Small bowel	FS, IP DDx pANCA+ CD vs UC
ALCA	Small bowel in combination with other Abs	FS
ACCA	?Small bowel in combination with other antibodies	FS, IP
AMCA	?Small bowel in combination with other antibodies	FS
Anti-L	Small bowel	FS, IP
Anti-C	?Small bowel in combination with other Abs	FS, IP

Data on localization and disease phenotype are complicated by the fact that multiple antibody positivity is more common than isolated single antibodies. Additionally, not all antibodies are measured in all studies.

Abbreviations : FS, fibrostenosing; IP, internal penetration.

Data from Refs.

Antiglycan Antibodies

Glycan is a generic term for all molecules bearing glycosidic bonds, which are major constituents of cell surface components of, among others, immune cells and micro-organisms. Glycans may modulate the immune system through stimulation of cell proliferation, phagocytosis, and cytokine secretion.

ASCA was the first antiglycan antibody identified. However, several novel antibodies have subsequently been demonstrated. Dotan and colleagues first reported the association of these antibodies with CD.

The antibody targets are laminarbioside (ALCA), chitobioside (ACCA) and mannobioside (AMCA). Seroreactivity against oligosaccharides is common in patients with CD: ALCA 37.5% and ACCA 36%. Less than 10% of patients with UC or miscellaneous inflammatory conditions demonstrate seropositivity. However, another intestinal inflammatory process, celiac disease, has a high rate of ACCA positivity.

Recently, 2 additional antiglycan antibodies have been described: anti-laminarin IgA (Glc[β1,3])3n(Glc[β1,6])n carbohydrate antibody (Anti-L) and anti-chitin GlcNAc(β1,4)n IgA carbohydrate antibody (Anti-C).

In general, the incidence of these antiglycan markers is a bit lower than the first-generation antibodies. However, much like the first-generation antibodies, when multiple antiglycan antibodies are positive, there is improved sensitivity and specificity for diagnosis of CD. Increasing titers of the antibodies appear to be associated with more complicated disease and earlier surgery, similar to the first-generation antibodies.

Despite a lower rate of seropositivity in Crohn, they may have clinical relevance in that ALCA or ACCA may be positive in almost half of the ASCA-negative patients.

The first-generation antibodies are generally stable over time. However, there may be an effect of increasing antiglycan antibody positivity with disease duration. This will be an important question to define carefully if they are to be used for disease stratification. Although not as commonly used in the United States, they are available commercially ( http://www.ibdx.net/index.html ).

Orphan Antibodies

Antibodies directed toward some specific intestinal epithelial antigens have been described. They have not been as widely tested as the more commercialized antibodies described above; therefore, their clinical role has yet to be defined. Antigoblet cell antibodies may occur in up to 40% of patients with IBD (both UC and CD) and in a significant number of first-degree relatives. Although healthy controls rarely exhibit antigoblet cell antibodies, inflammatory controls, such as celiac disease, may do so frequently. Given the obvious change in goblet cells and mucin in IBD, the existence of antigoblet cell antibodies suggests a potential pathogenic mechanism. Clinical applicability of anti goblet cell antibodies remains to be determined.

Das and colleagues isolated a UC-associated antibody and eventually identified its target, the microfilament protein, tropomsyosin. The major colonic tropomyosin isoform (hTm5) induces both humoral and T cell responses in UC patients. A higher percentage of patients with UC (79%) had an enhanced antibody response compared with 12% of Crohn. This tropomyosin isoform is expressed on colonic epithelium but not small bowel. If this autoantibody indeed does have a causative role in UC, this may explain why disease activity is restricted to the colon.

Pancreatic autoantibodies (targeted to zymogen granule membrane glycoprotein) may occur in about 30% of patients with CD. They are not clearly correlated with clinical features but perhaps with early-onset disease.

Sophisticated technologies such as protein microarrays may lead to the identification of novel autoantigens in IBD. As the number of autoantibodies and antibodies directed against microbes continue to increase, it will be necessary to sort through which ones are clinically relevant and pathophysiologically important. Given the large gaps in our knowledge in IBD, this area of research may hold considerable promise.

IBD Serologies in Diagnosing IBD and Distinguishing UC From CD: Statistical Considerations

The diagnostic power of serologic testing in many diseases is well established. However, the promise of IBD serologies for screening in a general population, for distinguishing IBD from other gastrointestinal diseases, and for separating UC from CD has not been fully realized.

Sorting through the statistics on IBD serologies is akin to wandering through a labyrinthine maze. For the statistically challenged, some insights and guidance are provided by Austin and colleagues in an informative evaluation of IBD serologic testing. Although sensitivity and specificity of an individual test are relatively fixed, the relevant clinical parameters are positive and negative predictive values (PPV, NPV, respectively). The PPV and NPV depend on the disease prevalence in the test population. If the disease prevalence is low, the positive predictive value falls.

An enabling calculation helps illustrate the dilemma of screening for a disease with low prevalence. If we assume that there are 1 million patients with CD in the United States and half are Omp-C positive, then there are perhaps 500,000 Omp-C–positive IBD patients. If the prevalence in a healthy population (“false-positive” rate) is only 1% and the US population is now 300,000,000, this would suggest there are 3 million Omp-C–positive people in the general population, a 6:1 ratio of non-IBD to IBD cases.

However, few studies start with a relatively undefined population to evaluate the role of serologies in individuals without an established diagnosis. Instead, they compare patients with known IBD and different control populations, already diagnosed. This may include healthy controls or individuals with a variety of gastrointestinal diseases; there may be changes depending on the comparator.

IBD prevalence in most of these studies may range from 40% to 60%. In this range, the positive predictive value is high (>90%), but the negative predictive value is limited and less impressive, about 50%. The frequently quoted sensitivity (>80%) and positive predictive value (>90%) for the serologic markers for CD depend on a high prevalence of the disease in the tested population; this may be as high as 38%. Austin and colleagues point out that as more tests are added to a panel, the sensitivity of the overall test will increase as long as only 1 test needs to be positive. However, there will be a concomitant decrease in specificity. Thus, as we see panels of more antibodies added to serologic testing, it is important to keep in mind the impact on diagnostic accuracy.

In distinguishing between UC and CD, several studies have consistently shown that the pANCA+/ASCA– profile for UC and the pANCA-/ASCA+ pattern provide accurate discrimination between the two.

This can serve to confirm the clinical diagnosis. Multiple studies have linked a specific antibody positivity to disease location or phenotype (see Table 2 ). Broad patterns hold true. For example, ASCA positivity is associated with small bowel location and fistulizing disease. However, untangling associations when multiple antibodies are positive has led to conflicting observations about whether 1 antibody is associated with fistulizing or penetrating disease. In general, both the antiglycan and antimicrobial antibodies are associated with complicated small bowel disease and surgery (see Prognosis and Stratification). There have been no antibody associations with perianal disease. In the more clinically challenging matter of indeterminate colitis, the diagnostic capability is less helpful (or predictive; see below).

The interpretation of the Prometheus panel of IBD serologies has become somewhat opaque with the development of pattern recognition technology that is trained to recognize subtle patterns and relationships among serum biomarkers. This may increase the sensitivity of the assay but sometimes yields a diagnosis of IBD when all the individual biomarkers are normal.

Serologic testing is neither necessary nor sufficient for the primary diagnosis of IBD. The serologic markers cannot replace a conventional clinical evaluation. However, after a thorough clinical workup in cases of indeterminate differentiation of CD versus UC, the use of this antibody panel may be a potential method of discrimination.

Characteristics of IBD Antibodies

Identification, description, and commercialization of 2 antibodies, antineutrophilic cytoplasmic antibody (ANCA) and ASCA invigorated research in IBD serologies. Over the last several years, there has been a steady increase in the number of IBD antibodies identified. Eventually, this range of antibodies should provide a better understanding of the etiopathogenesis of IBD and lead to more judicious clinical management of patients with ulcerative colitis and Crohn. However, at this point, the array of antibodies may leave the clinician more confused than enlightened.

We will classify these antibodies into 3 somewhat arbitrary groups: (1) what may be considered the first-generation of IBD-related antibodies that are widely available now through Prometheus Labs, (2) a group of antiglycan antibodies, and (3) a group of orphan autoantibodies with diverse targets in the gastrointestinal mucosa ( Table 1 ).

Table 1

IBD-associated antibodies, presumed targets, and select population prevalences

Antibody	Target	Prevalence (%)
Antibody	Target	Ulcerative Colitis	Crohn Disease	Healthy Controls
pANCA	Unidentified protein of the nuclear envelope of neutrophils	41–73	6–38	0–8
ASCAgASCA	Carbohydrate epitopes present in the phophopeptidomannan of the cell wall of Saccharomyces cerevisiae Covalently bound mannin	0–29	29–69	0–16
OmpC	Omp-C transport protein of E coli	2–24	24–55	5–20
I2	Pseudomonas-associated sequence I2	42	38–60	15
Cbir1	Bacterial flagellin CBir1	6	50–56	8
ALCA	Laminaribioside (Glc[b1,3]Glc[b])	3–8	19–27	2
ACCA	Chitobioside (GlcNAc[b1,4]GlcNAc[b])	5–7	8–25	0.5–12
AMCA	Mannobioside (Man[a1,3]Mana)	7	12–28	9
Anti-L	Laminarin (Glc[b 1,3])3n(Glc[ b 1,6])n	3–7	18–26	1–10
Anti-C	Chitin (GlcNAc[ b 1,4])n	2–11	10–25	2–12
Antigoblet cell	Goblet cells of the intestines
Tropomyosin	Microfilament protein tropomyosin	79	12
Pancreatic autoantibody	GP2 a major zymogen granule membrane glycoprotein		30