A combination of tests is used when evaluating for possible inflammatory bowel disease (IBD), and no one test can diagnose IBD with absolute certainty. 1–3 The testing starts with a detailed history, comprehensive physical examination, and a panel of laboratory studies that include nonspecific markers of inflammation. The next steps involve endoscopic examinations of the upper and lower gastrointestinal tracts, careful examination of endoscopically obtained tissue specimens, and contrast radiography of the small bowel. A diagnosis of IBD is made or excluded based on results of these tests, and, if present, the IBD is classified as Crohn's disease (CD), ulcerative colitis (UC) or, infrequently, indeterminate colitis (IC).
Various biologic markers have been studied in patients with IBD over the last several years. 4 Antineutrophil cytoplasmic antibody (ANCA) has shown promise, and studies indicate that ANCA with a perinuclear staining pattern (pANCA) is present in 60% to 80% of patients with UC versus 3% to 4% of healthy individuals. 4–6 Elevated titers of IgA and IgG antibodies directed against the phosphopeptidomannan part of the cell wall of Saccharomyces cerevisiae (baker's yeast) (anti–Saccharomyces cerevisiae antibody, ASCA) are found in 50% to 80% of CD patients and are rarely found in patients with UC (<10%) or healthy controls (<5%). 4,5 Main et al 7 initially correlated ASCA with CD in 1988.
Technologically refined serological tests for IBD are now commercially available. One such test is the IBD Diagnostic System (IDS) (Prometheus Laboratories, Inc., San Diego, CA). The IDS examined for pANCA, ASCA IgA, and ASCA IgG until recently when another component, namely αOmpC, was added to the panel. We wished to study the concordance of IDS with the clinical diagnosis established by traditional tests in children. We also examined the correlation between IDS and several patient characteristics.
- Top of page
- Patients and Methods
During the study period, 112 patients (56/112 males) had the IDS ordered with a mean age of 12.57 years (range 0.16–18.43 years); 5 patients were excluded due to incomplete evaluation. The remaining 107 patients were divided into 6 groups based on clinical diagnosis and results of the IDS test (Table 1). The results of the first IDS were used when the patients had IDS performed more than once (12 patients).
Table 1. Patients Were Divided into Five Groups Based on Clinical Diagnosis (CLD) and IBD Diagnostic System (IDS)
|Group (n)||CLD/IDS||Male % (N)||Mean Age (Years) (Range)|
|5(7)||Indeterminate colitis||28.6(2)||10.42 (1.83–15.67)|
|6(24)||Control (−/−)||50.0(12)||10.78 (0.16–15.67)|
The 27 patients in group 1 carried a clinical diagnosis of UC that was confirmed by IDS; their pANCA titer was elevated and they were negative for ASCA (IgA/IgG). The 20 patients in group 2 had a clinical diagnosis of CD and the IDS results supported this; they were negative for pANCA and their ASCA (IgA and/or IgG) was elevated. The 12 patients in group 3 were clinically felt to have UC, but their IDS test was discordant, that is, either negative for both pANCA/ASCA (9 patients) or positive for both pANCA/ASCA (3 patients, one of whom was later re-classified as CD based on pouchitis with granuloma on tissue specimens following colectomy). The clinical data in the 17 patients in group 4 were consistent with CD, but the IDS results were either negative for both pANCA/ASCA (14 patients) or pANCA+/ASCA− (three patients, all of whom were felt to have CD based on the endoscopic findings and/or the presence of granulomas on biopsies). Group 5 consisted of seven children who were clinically diagnosed with IC. The IDS was pANCA+/ASCA− in five of these children and negative for both pANCA/ASCA in the other two, who were pANCA+/ASCA− on repeat testing. Lastly, the 24 patients in group 6 served as controls. They were not thought to have IBD clinically, and this was supported by a negative IDS (pANCA-/ASCA−). Their final diagnoses included irritable bowel syndrome (n = 6), infectious illness (n = 5), nonspecific recurrent abdominal pain (n = 6), and one each with granulomatous gastritis, rectal ulcer, allergic colitis, eosinophilic ileocolitis, lactase deficiency, Helicobacter pylori infection, and hyperthyroidism. The mean age, sex distribution, and mean IBD activity scores were similar between disease groups.
In 54 of the 83 patients with IBD (Groups 1–5), a diagnosis of IBD had been established prior to the IDS being performed, including all 7 patients with IC. In the remaining 29 patients, the IDS was performed at the time of initial evaluation (ie, endoscopic studies). Of these 29 patients, 21 had a definitive clinical diagnosis of UC or CD, and their IDS was consistent with the clinical diagnosis [UC (n13; group 1) and CD (n8; group 2)]. In the remaining eight children, we established a clinical diagnosis of IBD [UC (n1) or CD (n7)], but their IDS was negative for both pANCA and ASCA.
The specificity of IDS was excellent for UC and CD, but the sensitivity was poor (Table 2). These results may be affected by the retrospective nature of the study and the potential for selection bias. The likelihood of this, however, is diminished by the fact that our results for specificity and positive and negative predictive values were similar to those provided by Prometheus Laboratories, Inc.; their sensitivity was 61.1% for UC and 67.1% for CD (personal communication). The probability of an abnormal IDS was not related to patient age, gender, absence/presence of symptoms, use of medications to treat IBD, diagnosis duration, or IBD activity score (Table 3), except that the IDS was more likely to be abnormal in a naive UC patient compared with one in whom the diagnosis had already been established (p 0.042).
Table 2. Statistical Analysis of IBD Diagnostic System for UC and CD
| ||Ulcerative Colitis||Crohn's Disease|
|+ Predictive value||90.0||90.9|
|- Predictive value||87.1||80.8|
Table 3. Selected Patient Characteristics at Time of IBD Diagnostic System Testing
| || || || || ||Diagnosis Duration in Months|| || || || |
| ||Age (Years)|| || || || ||Disease Activity ‡ (No. Patients)|
|Group (n)||≤12 (%)||>12(%)||On Medications No. (%)*||Symptomatic. No. (%)||All Patients*||Established Patients Mean Duration (Range) [n]||Inactive||Mild||Moderate||Severe|
|1(27)||10(37)||17(63)||13(48.2)||24(88.9)||20.5||39.5(4–162)[n = 14]||12||6||9||–|
|2(20)||7(35)||13(65)||12(60.0)||15(75.0)||11.5||19.1(2–67)[n = 12]||6||5||8||1|
|3(12)||3(25)||9(75)||10(83.3)||10(83.3)||29.2||31.9(1–114)[n = 11]||6||3||3||–|
|4(17)||8(47)||9(53)||9(52.9)||13(76.5)||12.9||24.0(1–112)[n = 9]||6||1||8||2|
|5(7)||3(43)||4(57)||6(85.7)||6(85.7)||36.4||36.4(2–81)[n = 7]||3||3||1||–|
- Top of page
- Patients and Methods
A diagnosis of IBD may be difficult to establish, and accurate classification of the IBD subtype may be even more challenging. For example, it was the teaching that involvement of the non-colonic GI tract argues strongly against UC, but a number of patients with UC have been found to have pathology in their upper gastrointestinal tracts. 10–12 It was hoped that availability of a serological test would allow more precise identification and classification of patients with IBD.
We performed the IDS in a large cohort of children and adolescents—both with IBD and non-IBD (controls). We found the test to have a high specificity but poor sensitivity, similar to other adult and pediatric reports. 5,13–17 The sensitivity of the test is better for UC than for CD. We were not able to demonstrate any correlation between the serological markers and patient age or sex, presence or absence of symptoms, medications, diagnosis duration (except for newly diagnosed UC who were more likely to be pANCA positive compared with established UC), or IBD activity score, as in other reports. 5,18–20 Three of our 37 CD patients (8%) were pANCA+/ASCA−, which is lower than most other reports, including pediatric studies, where the incidence has been 0%–40%. 5,15,21–23 The reasons for this discrepancy are unclear, especially as the laboratory methods are similar. Furthermore, none of our three pANCA+ CD patients had a “UC-like state.” 21 This is in contrast to the results of Ruemmele et al, 5 where 10 of the 17 (59%) pANCA+ CD patients had UC–like symptoms, and Vasiliauskas et al, 21 where 38 of 69 (55%) pANCA+ CD patients had a UC-like state.
Sandborn et al 20 reported that patients with treatment-resistant left sided UC are more likely to be pANCA+ than those with treatment-responsive left-sided UC and/or UC historical controls (90% vs 62% and 60%). We were unable to reproduce these results in children; clinical features of our pANCA+ and pANCA− UC patients were similar. The discrepancy between our results and adult studies is intriguing and could be due to sample size. Other pediatric studies, however, also do not report the associations noted in adult patients. 5,15
Ten percent to 15% of children with colitis have an indeterminate form, 2,3 and 2% to 4% of patients undergoing colectomy for UC are later diagnosed with CD. 24,25 Seven of 107 patients (6.5%) in our study were thought to have IC (group 5), 5 of whom were pANCA+/ASCA− on initial IDS. Two were initially negative for both pANCA and ASCA, but became pANCA+/ASCA− on repeat testing. Our results are similar to those of Olives et al 19 who reported five patients with IC, four of whom were pANCA+. In contrast, Ruemmele et al 5 had nine patients with IC and only two were pANCA+. None of our IC patients were ASCA+. Gupta et al 26 studied 41 children with IC and 19 had histologic findings suggestive of UC. Thirteen of the 19 patients were pANCA+ and 3 were ASCA+. A recent study of adults with IC found a higher number of patients to be ASCA+ than pANCA+. 27 As such, whereas in our experience patients with IC generally test positive for pANCA, the role of ASCA and pANCA measurements in IC is yet to be settled. 1,6,28
In summary, the low sensitivity of IDS reduces its value in screening and diagnosis and precludes its ability to replace traditional studies when evaluating a patient with suspected IBD. 13–15,17 The IDS may serve as an adjunct to the clinical workup. For example, if a patient is felt to have UC and is doing poorly, it would be worthwhile to measure the serological markers to ensure that the patient does not have serological evidence of CD, which would influence further management considerations. Will the role of these serological markers be to define the clinical and genetic/immunologic heterogeneity of IBD rather than be a screening/diagnostic tool for IBD? Further studies are needed to examine these issues and assess the cost-effectiveness of using IDS in combination with other tests for screening/testing for IBD in children. As our experience with, and knowledge of, these serological markers increases, we may find them to be more useful in predicting disease patterns rather than the disease itself.