Inflammatory bowel disease (IBD) is a chronic disorder of the gut and includes two main entities: ulcerative colitis (UC) and Crohn’s disease (CD). Their pathogenesis results from the combination of genetic, environmental, and immune factors. UC is the most frequent IBD, and an autoimmune component in the pathogenesis has been postulated (1). A continuous, diffuse, and circumferential pattern of inflammation in colonic mucosa and submucosa represents the typical endoscopic presentation, usually affecting the rectum with a variable proximal extension up to the ileocecal valve; the involvement of the last tract of ileum (“backwash ileitis”) is also possible (2). CD may involve any section of the gastrointestinal tract, with prevalent location to the terminal ileum, and presents a typically segmental mucosa involvement. The ECCO-ESGAR guidelines indicate that ileo-colonoscopy with biopsies is the preferred method to confirm the diagnosis of IBD, differentiating between UC and CD, and to assess disease activity (3). In a minority of cases with exclusive colonic involvement, the diagnosis of a specific type of IBD may still remain undefined or reclassified over time (4). Thus, the discovery of sensible, reliable, and economic serological markers that support the diagnosis of UC over CD or vice versa is warranted to aid the differential diagnosis in this particular setting. The best serological markers currently available in the clinical practice for the differential diagnosis of IBD are ASCA (anti-Saccharomyces antibodies) for CD and the atypical p-ANCA for UC (5). However, the ECCO-ESGAR guidelines do not recommend them for routine use considering their low accuracy (3). In 2021, Kuwada et al. described the presence of anti-integrin αvβ6 IgG autoantibodies in the sera of Japanese patients with UC (6). Shortly after, the prevalence of anti-integrin αvβ6 IgG autoantibodies was described in several cohorts around the world (7–10). These autoantibodies are mainly of the IgG1 class and inhibit integrin-αvβ6 fibronectin binding. This integrin is expressed in the colonic mucosa but not in the small intestine (6). Most of the previous literature on anti-integrin αvβ6 IgG autoantibodies was based on homemade ELISA kit, which is more time-consuming and not suitable for routine diagnostic. Here, we aim at confirming the diagnostic accuracy of anti-integrin αvβ6 IgG autoantibodies in an Italian multicentric cohort of IBD patients, investigating also their potential associations with UC clinical activity and extension and validating a new commercial ELISA kit for absolute quantification of anti-integrin αvβ6 IgG autoantibodies also in the European population. Our results confirm the optimal performance of the test in discriminating UC from controls. For the first time, a higher threshold to increase the specificity of UC over CD is proposed to increase diagnostic performance in case the endoscopy and histology do not allow the discrimination of the two entities. In the future, integration of anti-integrin αvβ6 IgG with endoscopic features may allow the reclassification of most IBD-U cases.

This is an Italian multicentric study including patients referred to the University Hospital Ca’ Foncello of Treviso (TV) and the Hospital Molinette of Torino (TO), two tertiary referral centers for the treatment of IBD. The study was approved by the Ethics Committee of Treviso (CET ANV 2024-56) and University of Torino (0056924) and complies with the Helsinki Declaration of 1964.

In our multicentric observational study involving two Italian centers, patients with IBD and various controls groups were enrolled, including healthy donors and patients with autoimmune diseases, celiac disease, and IBS. In our study, we could validate the optimal performance of anti-integrin αvβ6 IgG in identifying UC patients with an AUC of 0.93 and set the threshold to a cutoff of 1.68 U/mL with very good sensitivity (87.9%) and good specificity to non-UC patients (86.8%). The sensitivity for UC is slightly lower compared to the original description of the anti-integrin αvβ6 IgG (92%) (6), but it is higher than in Swedish and Italian cohorts (76% and 52%, respectively) (7, 9). Intriguingly, a recent multicentric Japanese study including 1,241 UC, 796 CD, and 206 controls with other gastrointestinal conditions found a very similar threshold of positivity for anti-integrin αvβ6 IgG (1.64 U/mL) and with strikingly similar results of sensitivity for UC at 87% and specificity at 82% for CD and 88% for other gastrointestinal disorders (14). We could confirm the optimal specificity of anti-integrin αvβ6 IgG toward healthy donors and non-IBD conditions (94.4%), including different rheumatological disorders, celiac disease, and IBS (6, 7, 15). A recent work confirmed that anti-integrin αvβ6 IgG performed better than calprotectin and C-reactive protein (CRP) in distinguishing UC from diseased controls (including IBS, celiac disease, and other conditions). In the same publication, the diagnostic performance of anti-integrin αvβ6 IgG could be further increased by combining anti-integrin αvβ6 IgG and calprotectin but not CRP (16). Considering the low number of controls with measured calprotectin at the time of sampling, we did not perform this analysis in our cohort.

Anti-integrin αvβ6 IgG presented a lower specificity for UC versus CD. In our cohort, 11 adult CD patients presented positive antibodies. These patients presented more frequently with colonic involvement (L2-3). False positive results in CD have been reported in 17%–21% (6, 8, 13) and up to 32% pediatric CD patients (15). Colonic involvement has been described as the only independent factor associated with false positive results in CD (14). Detectable anti-integrin αvβ6 IgG in CD has been associated with more frequent endoscopic and pathologic UC-like colitis (15). Intriguingly, a subgroup (31%) of patients with immune checkpoint inhibitor-induced colitis (ICI-colitis) presented anti-integrin αvβ6 IgG. This was associated with endoscopic UC-like characteristics, grade 3 colitis, and steroid resistance (17). All things considered, these observations make us speculate that anti-integrin αvβ6 IgG antibodies are associated with a specific pattern of colonic mucosa inflammation resembling UC rather than be unique for UC. However, anti-integrin αvβ6 IgG antibodies have also been shown not only in patients with primary sclerosing cholangitis (PSC) with IBD (92%) but also without IBD (74%) (18). The rationale of their serological presence in the case of liver disease may be found in the selective expression of integrin αvβ6 in the disease-associated epithelia of both bile duct and colon (18, 19). Such data were not confirmed in all reports: more recent publications indicate a lower positivity of anti-integrin αvβ6 autoantibodies in patients with PSC without IBD (39%) (19) or even comparable levels to controls (20). In our cohort, unfortunately there were no patients with PSC, so we cannot significantly contribute to this debate.

In clinical practice, the utility of anti-integrin αvβ6 IgG antibodies in differentiating UC vs. CD is of arguable importance considering that endoscopy (and histology) is generally sufficient to separate UC from CD. We think that the most important role of anti-integrin αvβ6 IgG antibodies would be in helping classify the patients with IBD-U, where endoscopy and histology have not definitively discriminated the two entities (4). Therefore, every attempt to classify IBD-U should be prompted, considering the implications on medical treatments and surgical options—for instance, colectomy is considered a definitive treatment option for UC, whereas sparing surgery is indicated for CD. A false positive may lead to colectomy for the wrong patient with irremediable consequences, so we think that, in this setting, the risk of the wrong classification weights more for false positive than false negative. Therefore, maximizing specificity (i.e., reducing the false positive) is more important than sensitivity. In this regard, we weighted our analysis to find a higher threshold of positivity to distinguish UC from CD. Here we propose an optimal threshold of 13 U/mL to confirm a UC diagnosis versus CD with specificity of 90% and sensitivity of 70%. This test performance is better than atypical p-ANCA or ASCA in discriminating the two main forms of IBD (5) despite the fact that we could not directly compare the performance of these two markers in our cohort. Thus, we propose a high cut-off to improve the ability to distinguish UC from CD, especially in patients with IBD-U. Prospective studies on IBD-U patients are needed to confirm if this strategy may be applied in this context and determine if the disease course differs based on the presence of detectable antibodies. This work may pave the way for a prospective study aiming at finding a clinical score integrating endoscopic, histological, and serological data to reduce the cases of IBD-U. A recent publication showed indeed that most IBD-U patients do have detectable antibodies (16). We also analyzed the possible correlation of anti-integrin αvβ6 levels with UC characteristics. We observed that anti-integrin αvβ6 IgG antibodies presented similar levels when grouping the patients according to disease extension. However, very high levels of anti-integrin αvβ6 IgG antibodies were seen only in patients with colonic involvement (E2–E3) rather than E1 (disease limited to the rectum). The association between the presence of antibodies and the extent of colonic involvement in UC is still debated. While some authors have described this association (6, 10), others (9), including our group, have not confirmed it. Similarly, we found no correlation with partial Mayo score or active disease state (PMS ≥2) similar to the findings of Marafini et al. (9) and opposite to that of Rydell et al. (7) We found a moderate correlation with endoscopic activity and a weak correlation with FCP, albeit data on endoscopic activity were available only in a small group of patients (19). These differences may be related to different recruitment criteria among studies. Our cohort is heterogeneous, and the patients were receiving different treatments at the time of sampling, possibly affecting the analysis. However, we did not observe a statistical difference in patients receiving immunosuppressive or advanced therapies, probably because of the relatively small sample size. In addition, the sampling was rarely performed at the time of diagnosis. Recently, Pertsnidou et al. showed a possible prognostic role of anti-integrin αvβ6 IgG in recently diagnosed UC. The UC patients with positive anti-integrin αvβ6 IgG presented with a more aggressive disease course (i.e., greater disease extent and endoscopic activity). The anti-integrin αvβ6 IgG levels remained stable at 3 months in patients with aggressive disease and decreased significantly in patients with an indolent course of disease (16).

With the limitation of the relatively low number of subjects and events, we explored if anti-integrin αvβ6 IgG could predict the composite outcome of hospitalization and/or colectomy in patients with at least 1 year of follow-up. Indeed all patients reaching the outcome presented levels of anti-integrin αvβ6 above 17 U/mL. The survival curve showed a significant difference between the group of patients with higher (≥17 U/mL) and lower antibodies (<17 U/mL). Livanos et al. showed that higher levels of anti-integrin αvβ6 IgG were associated with a composite outcome that included also a need for biological drugs or switch (10). In our cohort, the patients enrolled in Torino were all candidate to start an advanced therapy at the time of sampling. Therefore, we could not include the need for biological drug in the composite outcome for statistical bias. Thus, larger prospective studies are needed to confirm this observation.

Integrin αvβ6 is restricted to the epithelia and is upregulated during epithelial repairs and tumorigenesis (21). Therefore, it has been suggested that the formation of anti-integrin αvβ6 IgG is secondary to the epithelial colonic damage (6, 17). These antibodies have been demonstrated to inhibit the RGD binding motif, thereby blocking the interaction with several ligands, including fibronectin, in a dose-dependent manner (6, 17). A similar interaction is required for the liberation of transforming growth factor β1 (TGF-β1) from the latent associated peptide (LAP) to exert its function (22). In mouse models, the absence of RGD motif in TGF-β LAP replicated the features of knockout mice for integrin β-6 (itgb6 ^-/-) or TGF-β1 (tgfb1 ^-/-) (22). Consequently, it can be hypothesized that anti-integrin αvβ6 IgG may diminish TGF-β activation, reducing epithelial–mesenchymal transformation and tissue repair. Nevertheless, the data regarding TGF-β in IBD are somewhat contradictory—for instance, Sedda et al. reported that restoring TGF-β signaling through SMAD-7 inhibition suppresses inflammation in CD patients (23), whereas Ghorbaninejad et al. reported that inhibition of TGF-β signaling suppresses epithelial–mesenchymal transformation and preserves tight junction integrity (24). However, we have to consider that the majority of evidence of the role of TGF-β were collected in CD rather than UC. Furthermore, very-early-onset colitis and brain abnormalities have been described in three families harboring an integrin alpha-v variant (ITGAV) (25) and in a patient with homozygous integrin β-6 polymorphism (ITGB6) (26). The abnormalities resemble the phenotype of TGF signal pathologies (25). Future studies should be conducted to ascertain whether anti-integrin αvβ6 IgG is associated with diminished intestinal TGF-β signaling and to investigate the implication of this association of inflammation and fibrosis.

Uzzan et al. showed a skewed IgG response with a marked increase of IgG-producing short-lived plasma cells in the gut mucosa and peripheral blood of UC patients (8). This, in addition to the evidence of lymphoplasmacytic infiltrate, anti-microbial antibodies, and the protective role of the single nucleotide polymorphism in FCGR2A (rs1801274), which encodes an activating IgG Fcg receptor (FcgR) expressed by myeloid cells, further supports the possible involvement of B cells in UC pathogenesis (27). However, rituximab, an anti-CD20 monoclonal antibody that depletes B cells, was ineffective in treating UC in a small placebo-controlled trial involving patients with refractory disease (28). There is even a study reporting that rituximab increases the risk of IBD (29). Therefore, further studies are needed to explore the role of plasma cells in IBD pathogenesis and the possibility of targeting B cells with other therapies in UC.

Our study presents several limitations. First, this is an observational study, and data on colonoscopy and fecal calprotectin were limited to a subgroup of patients because they were not routinely measured on the same day of sampling. The sampling was performed rarely at the time of diagnosis, and therefore some seronegative UC may have been sampled after years of remission. On the other hand, this allows us to generalize our results to all time points of the patients’ history, also considering that the two centers included IBD patients who were both naïve and experienced with advanced therapies. Finally, other forms of acute colitis (such as infectious or diverticular disease) were not included.

In conclusion, this study validates the “anti-integrin αvβ6 ELISA kit” and confirms the diagnostic role of anti-integrin αvβ6 IgG in UC with an optimal cutoff of 1.68 U/mL and with good sensitivity and optimal specificity compared to the controls. The presence of antibodies supports the diagnosis of UC over CD. However, a group of CD patients with colonic involvement may present positive antibodies. Therefore, we propose a higher threshold of 13 U/mL to maximize the specificity for UC over CD. This threshold may be helpful to reduce unclassified colitis (IBD-U). Further studies are needed to build composite scores integrating endoscopic and serological data to obtain the best results. The availability of a commercial ELISA kit will allow the inclusion of anti-integrin αvβ6 IgG assessment in the routine diagnostic algorithm for IBD in Europe. Further studies are needed to evaluate longitudinally the levels of anti-integrin αvβ6 IgG and explore if these antibodies correlate with clinical and/or endoscopic activity of disease and the possible influence of specific treatments on the presence of antibodies. If the pathogenic role of anti-integrin αvβ6 IgG in blocking mucosal healing in UC patients was confirmed, potential therapies targeting TGF-β and/or mucosal IgG-producing plasma cells could be explored in the future.