Twenty-five years ago, the discovery of antibodies to bakers’ yeast (Saccharomyces cerevisiae; ASCA) in the serum of patients with Crohn’s disease (CD) sparked great interest in the concept that this and other circulating antibodies, otherwise known as serologies, reflected the underlying pathogenesis of inflammatory bowel disease (IBD).1 Indeed, an increase in intestinal permeability and loss of tolerance leading to a humoral response to translocating microbes are hallmarks of these conditions, and over subsequent years numerous other antibodies have been found to be associated with IBD. Interestingly, the earliest antibodies described, ASCA along with pANCA,2 have proved to be of greatest value, in that their presence as well as their titer patterns in combination may serve as markers of IBD, with ASCA positivity being associated with CD and pANCA positivity being associated with UC.3
Another potential use of serologies has been in discriminating between the 2 IBD phenotypes, ulcerative colitis (UC) and CD, in cases of diagnostic uncertainty, which still remains a challenge in clinical practice. Typically, antibody testing is performed in a patient who clearly has a chronic colitis, consistent with IBD, but whose features do not allow for separation into either UC or Crohn’s colitis. Such patients have been variably described as having IBD unclassified, undetermined, or indeterminate colitis (IC). Despite the common practice of using serologies for this purpose, very little research has, in fact, been published demonstrating the ability of serologies to accurately distinguish UC versus CD starting with a cohort of IC patients. Instead, prior serology research has focused on diagnostics and phenotypic differentiation by comparing a well-defined IBD cohort with control populations of healthy adults, as well with those with diarrhea or irritable bowel syndrome.4
With this in mind, in this issue of JCG, Sura et al5 have reported on the utility of serologic testing when applied specifically to an IC cohort. This was a real-world, single-center, retrospective review of 117 patients with IC who had at least 1 year of follow-up and had undergone ASCA, pANCA, and anti-OmpC serology testing at inception. The purpose of the follow-up was to allow time for the true diagnosis of either UC or CD to reveal itself on clinical grounds. The authors then looked back at the initial serology testing to determine how well these results predicted the ultimate diagnosis. The data were derived beginning with patients who presented in 2001; thus, more contemporary serologies such as anti-CBir1 did not factor significantly into the analysis. Furthermore, genetic testing (eg, NOD2 status) and the diagnostic algorithm offered by Prometheus Laboratories Inc. were also not incorporated into this study. The authors first calculated the sensitivity and specificity of each of the 3 serology tests for diagnosing UC or CD as determined after 1 year of follow-up. Then, for each individual serology, they calculated likelihood ratios (LR) expressing the probability that a serology test could identify UC (or CD) when positive or rule out UC (or CD) when negative.
After 1 year, half (58) of the patients had been reclassified from IC to UC, 42% (49) were reclassified as CD, whereas the remaining 9% (10) did not develop any distinguishing characteristics and therefore retained their original IC diagnosis. How did serologic testing fare in predicting these subsequent diagnoses? A positive pANCA test had a sensitivity of 78% and a specificity of 44% for a diagnosis of UC. ASCA and anti-OmpC revealed low sensitivity (∼20%) but better specificity (∼80%) for CD. With regard to LRs, the results showed that only pANCA had any diagnostic value. The LR of a positive pANCA test for a diagnosis of UC was a modest 1.4, which corresponds to an increased probability of disease of about 10%. The LR for a negative pANCA test excluding CD was 2.0 (or an approximately 15% decrease in the probability of disease). Neither ASCA nor anti-OmpC demonstrated a significant likelihood ratio (positive or negative) for either diagnosis.
What do these results mean for the busy gastroenterologist and how can he/she use them in a practical sense? First, the results reinforce that the sensitivity and specificity of serologies in distinguishing between UC and CD leave much to be desired and must be used with caution, as explained in greater detail below. Second, on the basis of this study, the likelihood ratio of 1.4 for a positive pANCA test adds ∼10% confidence to the pretest clinical suspicion for UC (if you think that a patient has a 50% chance of having UC, a positive pANCA increases that chance to perhaps 60%). One can argue that better results could be achieved with the addition of other serologies and genetic profile data—especially NOD2 status—but, as the authors point out, there is little evidence to support this view because of the lack of published studies in which serologies were obtained exclusively from a cohort of IC patients. In 2002, the IBD group in Leuven, Belgium, conducted a study similar to the current work and found that ASCA/pANCA analysis was predictive of CD and UC in 80% and 63% of IC patients, respectively.6 Unfortunately, these numbers are not adequate for serologies to be recommended routinely to assist in categorizing this indeterminate group. In a follow-up study by the same investigators, adding anti-OmpC and I-2 serology testing to their original IC cohort did not improve the predictive accuracy.7 Other studies have similarly shown that a diagnostic algorithm incorporating multiple serologies, as offered commercially, may not be advantageous beyond the individual results provided by ASCA and pANCA testing alone.8
The investigation by Sura et al5 should also serve as a reminder of some of the broader issues in IBD that limit the value of serology testing in general. As with any tool at the physician’s disposal, there exists the potential for misuse and/or misinterpretation. Most practicing gastroenterologists can attest to numerous instances in which antibody testing confuses rather than clarifies a patient’s diagnosis. Most typical is the case where serologies are ordered—often by a nongastroenterologist—for a patient with vague intestinal complaints. When the serologic pattern is reported as being consistent with IBD, a long and expensive search inevitably ensues, as caregivers attempt to prove the suspicion raised by the test. Another scenario is the postsurgical patient who develops complications such as an enterocutaneous fistula. In due course, serology testing once again suggests IBD, prompting an unavoidable gastrointestinal consult and costly workup. In both of these examples, antibody testing is probably not justified. In the first case, the prevalence of IBD in this patient population is too low to make serology testing worthwhile. One should always bear in mind that the positive predictive value of a serology test drops precipitously as the prevalence of IBD in the population being studied declines. In the second case, there are a host of clinical conditions in which serologies are detected but should be considered falsely positive for IBD.9 Any clinical event, such as a postsurgical complication, in which bacterial translocation would be expected to occur, can lead to circulating antibodies. Serologies, in short, cannot distinguish among multiple causes of increased intestinal permeability.
Another point germane to this topic is that, just as a case of IC is provisional, even a formal diagnosis of UC must always be considered provisional. This is because in every case of what appears to be a clear-cut UC there is a small chance that eventually CD will be revealed as the true diagnosis, sometimes years after colectomy. One might designate this as the uncertainty principle of IBD. The gastroenterologist with awareness of this principle will find it difficult to avoid a sense of unease when sending a UC patient for a colectomy.
Finally, it is important to remember that sensitivity and specificity data for serology testing were derived from a cohort of IBD patients who were diagnosed in the traditional manner, using clinical criteria of history, physical examination, imaging, endoscopy, and biopsy, all of which in totality should lead the practitioner to the correct diagnosis in the vast majority of cases. This will remain the gold standard for the foreseeable future. As shown in the study by Sura et al,5>90% of patients originally diagnosed with IC were reclassified as CD or UC on clinical grounds, obviating the need for serology testing in the first place. In the remaining small percentage of patients who cannot be reclassified after 1 year, we are currently relegated to a feeling that the experienced clinician lives with frequently: uncertainty.
1. Main J, McKenzie H, Yeaman GR.Antibody to Saccharomyces cerevisiae
(bakers’ yeast) in Crohn’s disease.Brit Med J.1988;297:1105–1106.
2. Saxon A, Shanahan F, Landers C.A distinct subset of antineutrophil cytoplasmic antibodies is associated with inflammatory bowel disease.J Allergy Clin Immunol.1990;86:202–210.
3. van Schaik FD, Oldenburg B, Hart AR, et al..Serological markers predict inflammatory bowel disease years before the diagnosis.Gut.2013;62:683–688.
4. Reese GE, Constantinides VA, Simillis C, et al..Diagnostic precision of anti-Saccharomyces cerevisiae antibodies and perinuclear antineutrophil cytoplasmic antibodies in inflammatory bowel disease.Am J Gastroenterol.2006;101:2410–2422.
5. Sura SP, Ahmed A, Cheifetz AS, et al..Characteristics of inflammatory bowel disease serology in patients with indeterminate colitis.J Clin Gastroenterol.2014;48:351–355.
6. Joossens S, Reinisch W, Vermeire S, et al..The value of serologic markers in indeterminate colitis: a prospective follow-up study.Gastroenterol.2002;122:1242–1247.
7. Joossens S, Colombel J-F, Landers C, et al..Anti-outer membrane of porin C and anti-I2 antibodies in indeterminate colitis.Gut.2006;55:1667–1669.
8. Shirts B, von Roon AC, Tebo AE.The entire predictive value of the prometheus IBD sgi diagnostic product may be due to the three least expensive and most available components.Am J Gastroenterol.2012;107:1760–1761.
9. Sellin JH, Shah RR.The promise and pitfalls of serologic testing in inflammatory bowel disease.Gastroenterol Clin N Am.2012;41:463–482.