Antibodies against integrin αvβ6 have high diagnostic accuracy for ulcerative colitis

Abstract

Background: Anti-integrin αvβ6 IgG autoantibodies showed good sensitivity and optimal specificity in ulcerative colitis (UC) compared to controls. We aim at confirming the diagnostic accuracy of anti-integrin αvβ6 autoantibodies in an Italian multicentric cohort.

Methods: This observational multicentric study included adult and pediatric patients with inflammatory bowel disease and controls. Data on demographics, disease extension, partial Mayo score, fecal calprotectin, endoscopic Mayo score, and the time to the composite outcome including hospitalization or colectomy were collected. A new commercial ELISA kit was used to measure anti-integrin αvβ6 in the serum of the enrolled patients. Receiver operating curve (ROC) was used to identify the optimal cutoff to discriminate UC cases from other patients. Kaplan-Meier curves and log-rank test were used to analyze the composite outcome hospitalization and need of colectomy.

Results: A total of 228 patients were enrolled, including 36 controls (13 healthy donors and 24 diseased controls), 34 irritable bowel syndrome (IBS) patients, 50 Crohn's disease (CD) patients, and 107 UC patients. The UC patients presented higher values of anti-integrin αvβ6 IgG compared to CD, IBS, and controls (Kruskal-Wallis test and post-hoc Holm's correction: p < 0.001). The ROC of anti-integrin αvβ6 IgG performed optimally with an area under the curve of 0.93. The optimal cutoff to distinguish UC from controls was 1.68 U/mL, with a sensitivity of 87.9% and a specificity of 86.8% compared to non-UC patients with a specificity of 94.4% to non-IBD and 76% to CD, with very similar values to a recent multicentric study. A higher threshold up to 13 U/mL may be useful to make a differential diagnosis between UC and CD with a specificity of 90%. Anti-integrin αvβ6 did not correlate with clinical disease activity but weakly with fecal calprotectin (R = 0.28, p = 0.36) and moderately with endoscopic disease activity reported at the last colonoscopy (R = 0.60, p = 0.03). Despite the low number of events, the log-rank test showed the potential predictive performance of high levels of anti-integrin αvβ6 IgG (i.e., >17 U/mL) for the composite outcome (p = 0.02).

Conclusions: This study validates a new anti-integrin αvβ6 ELISA kit and confirms its high diagnostic accuracy in UC also in a European population, with particular utility in the differential diagnosis of specific forms of IBD.

Keywords: Crohn’s disease; diagnosis; inflammatory bowel disease; integrin αvβ6; ulcerative colitis.

Figure 1

Levels of anti-integrin αvβ6 IgG antibodies in the different groups of patients. The levels of anti-integrin αvβ6 IgG antibodies are displayed in base 2 logarithmic scale (log2). The p-values were computed with post-hoc Dunn’s test. The dashed line represents the optimal cutoff to distinguish ulcerative colitis cases from other patients. IBS, irritable bowel disease; CD, Crohn’s disease; ns, not significant; ***p < 0.001; ****p < 0.0001.

Figure 2

Receiver operating curve of anti-integrin αvβ6 IgG antibodies and diagnosis of ulcerative colitis considering other conditions (A) and only Crohn’s disease (B). As stated in the “Methods” section, we found the optimal cutoff that maximized sensitivity and specificity according to Youden’s point and the top-left point method. In (A), Youden’s point and the top-left point correspond to the same value (red dot, corresponding to 1.68 U/mL). In (B), the red dot and the orange dot represent the closest top-left point and Youden’s point, respectively. The sienna dot represents a threshold that maximized the specificity (see the main text). AUC, area under the curve (and 95% confidence interval).

Figure 3

Levels of anti-αvβ6 IgG antibodies based on disease location according to Montreal classification (A) in Crohn’s disease (CD) and in ulcerative colitis (UC). The levels of anti-integrin αvβ6 IgG antibodies are displayed in base 2 logarithmic scale (log2). The p-values were computed with post-hoc Dunn’s test. L1, ileal involvement; L2, colonic involvement; L3, ileo-colonic involvement; E1, proctitis; E2, left side colitis; E3, pancolitis; ns, not significant.

Figure 4

Spearman’s correlations among continuous variables analyzed in ulcerative colitis patients. PMS, partial Mayo score, EMS endoscopic Mayo score.

Figure 5

Survival curve of the composite outcome including hospitalization and colectomy. (A) represents the whole cohort. In (B), the cohort is split in two groups with the threshold of 17 U/mL. The two groups were compared with log-rank test.